阅读说明：本技术 面向山林防火的三维无线传感器网络数据路由融合方法 (Mountain forest fire-prevention-oriented three-dimensional wireless sensor network data route fusion method ) 是由 王皓辰 林蔚 于 2021-09-01 设计创作，主要内容包括：本发明属于三维无线传感器网络技术领域,具体涉及一种面向山林防火的三维无线传感器网络数据路由融合方法。本发明包括基于熵权综合评价与改进模拟退火的分簇算法、基于熵权综合评价与概率优选的路由算法、基于数据类型与用户需求的快速数据融合算法。本发明在一定程度上克服了二维无线传感器网络的局限性,具有一定的节能性,鲁棒性强,容错性高,能很好地满足山林防火监测的需求。(The invention belongs to the technical field of three-dimensional wireless sensor networks, and particularly relates to a mountain forest fire prevention-oriented three-dimensional wireless sensor network data routing fusion method. The method comprises a clustering algorithm based on entropy weight comprehensive evaluation and improved simulated annealing, a routing algorithm based on entropy weight comprehensive evaluation and probability optimization, and a rapid data fusion algorithm based on data types and user requirements. The invention overcomes the limitation of a two-dimensional wireless sensor network to a certain extent, has certain energy saving property, strong robustness and high fault tolerance, and can well meet the requirement of forest fire prevention monitoring.)

1. A mountain forest fire-proof-oriented three-dimensional wireless sensor network data routing fusion method is characterized by comprising the following steps:

step 1: a three-dimensional wireless sensor network facing forest fire prevention is arranged in a monitoring area;

all nodes in the three-dimensional wireless sensor network facing to mountain forest fire prevention are isomorphic, are completely same equipment and have unique numbers; the initial energy of all nodes is equal and not supplementary, and the initial energy of all nodes is enough for them to complete at least one data transmission; after all the nodes are deployed, the spatial rectangular coordinates of the nodes are known and do not change any more, and all the nodes store the numbers and the spatial coordinates of other nodes; all nodes can adjust the transmitting power, and the upper limit of the transmission distance of the nodes is d₀(ii) a The states of the nodes are divided into three types: normal operation, dormancy and death, any node is in one of the three states; a node may enter a sleep state, but not all nodes are allowed to enter the sleep state; the normal node and the dormant node can modify the state when the condition is met, and the state of the dead node is not changed; after all normal nodes transmit data, the network is regarded as finishing data transmission of one turn, and the service life of the whole network is measured by the data transmission turn; the state of the node changes only when the turn changes and is determined before starting to transmit data; the normal nodes have to transmit data in the current round, the dormant nodes do not transmit data in the current round, and the dead nodes do not transmit any data.

The monitoring area is internally provided with a movable Sink node, the Sink node can transmit instructions to all nodes, and the Sink node receives data, fuses and summarizes the data and sends the data to a base station; dividing a monitoring area into n rows and m columns, wherein n multiplied by m small blocks are provided, and at least 2 nodes exist in each small block and meet the requirement

Wherein the content of the first and second substances,are all set values;represents rounding in the x direction; TR (transmitter-receiver)₀The transmission radius of a node, the transmission radius TR of all nodes_iAre identical constants, i.e. TR_i＝TR₀；

Step 2: planning initial clustering;

step 2.1: acquiring clustering information of all nodes;

step 2.2; selecting a node which is not clustered, and judging whether a proper cluster exists or not; if no proper cluster exists, the node is clustered; otherwise, the node joins the best cluster;

with C_jDenotes the jth cluster, let C_jThe node in (1) is N_j1，N_j2，…，For node N_iIf DNC (i, j) ≦ TR is satisfied_iAnd p is_jIf < MCN, then C is determined_jIs node N_iA suitable cluster of (a); MCN is the set maximum membership of each cluster;

wherein (x)_i，y_i，z_i)、(x_jk，y_jk，z_jk) Are respectively node N_i、N_jkThe coordinates of (a);

if node N_iAll suitable clusters of (A) areFor clustersIf it is satisfied withThen it is decidedIs node N_iThe best cluster of (a);

step 2.3: judging whether the initial clustering is finished or not; if the initial clustering is not finished, returning to the step 2.2; otherwise, executing step 2.4;

the conditions for completion of the initial clustering were: all nodes are added into respective clusters, the clusters have unique numbers, the numbers of the clusters are determined according to the sequence of cluster formation, and at least 1 node is required to be arranged in any one cluster;

step 2.4: judging whether a unary cluster exists; if the unary cluster does not exist, outputting initial clustering information of all nodes; otherwise, clearing the clustering result, and returning to the step 2.1 to re-execute the initial clustering;

and step 3: selecting a cluster head;

first, C is calculated_jCenter of gravity coordinates CG of inner node_j＝(x_j，y_j，z_j) And the median CE of the residual energy of the nodes in the cluster_jAnd constructing a virtual nodeIts spatial rectangular coordinate is (x)_j，y_j，z_j) (ii) a For C_jEach of which isScreening out nodes with enough residual energy to complete a cluster head transmission task as cluster head candidates; if a cluster head candidate does not exist in a certain cluster, setting the cluster as a dormant cluster until clustering is carried out again next time; for each cluster head candidate in the cluster, 2 indexes are considered to select the most suitable cluster head:anddistance d (j) of_k，j₀)、Residual energy ofAnd CE_jIs a distance ofAndthe smaller the size, the smaller the size is considered to beThe more suitable the clustering;

wherein the center of gravity CG_jIs the center of gravity in the sense of spherical distance and limits (x)_j，y_j，z_j) In the area to be monitored,finding (x) using a modified simulated annealing algorithm_j，y_j，z_j) The exact value of (d);

and 4, step 4: planning a route;

step 4.1: acquiring information of all nodes in a cluster;

the clusters have the following three states: unplanned, planned, and dormant, with any one cluster in and only one; at least one normal node is arranged in the unplanned cluster, and the routing planning of the node is not completed; at least one normal node is arranged in the planned cluster, and the route planning of all the normal nodes is completed; the sleeping cluster has no normal nodes, namely all the nodes are sleeping nodes;

non-cluster head nodes in the cluster are divided into the following four types according to the cluster out degree of the nodes:

(1) cluster inner degree 0 nodeThe node of the class satisfies IOD (j)_u)＝0；

(2) A-type cluster interior degree 1 nodeThe node of the class satisfies IOD (j)_q) 1, the node can transmit data to the cluster head in a single-hop mode;

(3) b-type cluster interior degree 1 nodeThe node of the class satisfies IOD (j)_h) 1, the node can not transmit data to the cluster head in a single-hop mode;

(4) other nodesThe node of the class satisfies IOD (j)_k)≥2；

Step 4.2: judging whether the cluster head can complete the transmission task; if the transmission task cannot be completed, enabling the full cluster nodes to sleep; otherwise, executing step 4.3;

step 4.3: let Counter be 0;

step 4.4: judging whether the Counter is equal to the number of the inspected routing schemes; if the Counter is equal to the number of the inspected routing schemes, determining the index weight through an entropy weight comprehensive evaluation method, and selecting and executing the optimal routing scheme; otherwise, executing step 4.5;

considering RNs independent of each other_jA route planning which is set asThe following four indicators of these route plans are considered to select the best routing scheme: full cluster residual energyVariance of residual energy of full cluster nodesFull cluster data trafficAnd number of sleeping nodesWherein the total cluster residual energyFull cluster data trafficVariance of residual energy of full cluster nodes as benefit indexNumber of sleeping nodesIs a cost-type index;

step 4.5: awakening nodes in the cluster;

step 4.6: judging whether all nodes in the cluster complete routing planning or not; if all nodes in the cluster finish route planning, storing the route scheme, and returning to the step 4.4 after the Counter is enabled to increase by 1; otherwise, executing step 4.7;

step 4.7: judging whether a node with the degree of 0 exists in the cluster; if the cluster has a degree 0 node, enabling the degree 0 node to be dormant; otherwise, executing step 4.8;

step 4.8: judging whether a B-type degree 1 node exists in the cluster; if the cluster has the node with the type B degree 1, the routing is planned for the node with the type B degree 1 at random, and the step 4.7 is returned; otherwise, executing step 4.9;

step 4.9: recording the serial numbers of the nodes with the type A degree 1 in the cluster;

step 4.10: judging whether other nodes exist in the cluster; if other nodes exist in the cluster, selecting one other node as a sending node sender and planning a route for the other node, and returning to the step 4.8; otherwise, executing step 4.11;

after ensuring that the node with the degree 0 and the node with the type B degree 1 are not in the cluster and the node with the type A degree 1 is removed, the rest nodes are other nodes and are set as the other nodesAnd considering the following three indexes of other nodes to select one node as a sending node sender and plan a route for the node: distance d (j) from cluster head_i，j_CH) Residual energy ofDegree of inter-cluster run out Wherein the distance d (j) to the cluster head_i，j_CH) Is a benefit type index, the residual energyAnd intra cluster out degree IOD (j)_i) Is a cost-type indicator; other nodes with far distance from cluster head, less residual energy and small cluster internal emergenceThe point should plan the route preferentially;

inspecting the three index data values of all other nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation method Selecting a node to plan a route according to the probability, which comprises the following steps:is provided withIs scored asThen selectThe probability of planning a route is

Other nodes to be planned are selectedThen, a receiving node receiver is selected for the receiving node receiver); the other nodes are nodes with an in-cluster out-degree of at least 2, and thusAt least 2 receivable nodes; is provided withThe receivable nodes areOne receiving node is selected considering the following four criteria of a receivable node: anddistance d (j) of_i，j_q) Residual energy ofDistance d (j) from cluster head_q，j_CH) Amount of data to be transmitted Andthe nodes with short distance, more residual energy, short distance with cluster head and less data amount to be transmitted should be preferentially taken asA receiving node of (1); examining the above four index data values of the receivable nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation method Also selected according to probabilityThe receiving node of (2) is specifically as follows:is provided withIs scored asThen selectAs the probability of the receiving node is

To this end obtainThe routing planning of (2);

step 4.11: planning routes for all nodes with type A degree 1 in the cluster;

step 4.12: planning a route for the cluster head, storing a route scheme, and returning to the step 4.4 after the Counter is increased by 1;

and 5: transmitting and fusing;

is provided withThe collected smoke concentration data, temperature data, humidity data and wind speed data are sequentiallyThe following fusion algorithm is used: let a, B ∈ (0,1), A, B ∈ R⁺To and from

If and only if one of the following two conditions is met, the cluster head transmits alarm information and dangerous node numbers to the Sink node:

(1) in thatIn the presence of at least k_jA (without setting them asSatisfy the requirement of

(2) In thatIn the presence of at least u_jA (without setting them asSatisfy the requirement of

Wherein a, B, A and B are artificially set according to actual conditions;

when the two conditions are not met, the cluster head transmits to the Sink nodeTo provide the information in the mountain forest most favorable to the fire to happen;

wherein the content of the first and second substances,the upper quartile of (d);is composed ofThe upper quartile of (d);is composed ofThe lower quartile of (d);the upper quartile of (d);

step 6: judging whether clustering needs to be carried out again; if the clustering is not needed, returning to the step 4; otherwise, executing step 7;

and 7: judging whether the three-dimensional wireless sensor network facing to mountain forest fire prevention is alive or not; if the three-dimensional wireless sensor network facing to mountain forest fire prevention survives, planning non-initial clustering, and returning to the step 3; otherwise, the operation is terminated;

the method for planning the non-initial clustering specifically comprises the following steps:

step 7.1: acquiring clustering information of all nodes;

let N₁，N₂，…，N_mIs all non-dead nodes in the network, N_i∈{N₁，N₂，…，N_m}; in N₁，N₂，…，N_mIf there are t nodes N₁，N₂，…，N_tSatisfy the requirement ofk is not equal to i and d (i, k) is less than or equal to TR_iIf it is true, then N_iThe global out-degree of (d) is t, which is denoted as eod (i) ═ t; if EOD (i) ═ 0, N_iIs a global out 0 node;

in N₁，N₂，…，N_mIf there are q nodes N₁，N₂，…，N_qSatisfy the requirement ofl ≠ i and d (i,l)≤TR_lif it is true, then N_iThe global degree of (a) is q, and is recorded as EED (i) ═ q; if EED (i) is 0, N_iIs a global in-degree 0 node;

step 7.2: judging whether nodes which are not planned to be clustered exist or not; if the nodes which are not planned into clusters do not exist, the clustering information and the priority cluster numbers of all the nodes are output; otherwise, executing step 7.3;

step 7.3: examining the global out-degree and the global in-degree of all unplanned clustered nodes; judging whether a node with the global out degree of 0 exists or not; if no node with the global out-degree of 0 exists, executing an initial clustering algorithm on all unplanned clustering nodes, and returning to the step 7.2; otherwise, executing step 7.4;

step 7.4: judging whether a node with the global in-degree of 0 exists in the nodes with the global out-degree of 0; if the node with the global in-degree of 0 exists, setting the node with the global out-degree and the node with the global in-degree of 0 as a dead node, and returning to the step 7.2; otherwise, executing step 7.5;

step 7.5: searching cluster head winners and cluster head failure winners in the nodes with the global out-degree of 0 and the global in-degree of not 0, setting the cluster head failure winners as dead nodes, executing a priority cluster algorithm, and returning to the step 7.2;

for the node N with the global out degree of 0 and the global in degree of not 0_i，E_iIs N_iIf E is satisfied_i≥E(DD，d_Sink) Then N is_iIs the winner of cluster head, otherwise N_iSelecting the cluster head; the DD is the default data transmission quantity of the nodes, namely the bit number of data collected by each node;

where v is the number of bits of the transmission data; s is the transmission distance;E_elec＝50nJ/bit；ε_fs＝10pJ/(bit·m²)，ε_mp＝0.013pJ/(bit·m⁴)；

the priority cluster algorithm is executed according to the principle of firstly selecting cluster heads and then clustering; the cluster head of the priority cluster is fixed, namely the cluster head winner of the priority cluster is established; the priority clusters can absorb the nodes with the global out-degree not being 0 to enter the clusters as much as possible, and after all the priority clusters are planned to be clustered, if other nodes with the global out-degree not being 0 exist at the moment, initial clustering is carried out on the nodes.

2. The mountain forest fire-prevention-oriented three-dimensional wireless sensor network data routing fusion method according to claim 1, characterized in that: step 3 is to search (x) by using an improved simulated annealing algorithm_j，y_j，z_j) The method for the accurate value of (2) is specifically as follows:

step 3.1: initializing and setting an initial temperature, a termination temperature, the maximum iteration times of an outer loop and the maximum iteration times of an inner loop; initializing to generate an initial solution as a current solution of a first iteration;

step 3.2: adding disturbance to the current solution to obtain a new solution;

step 3.3: judging whether to receive a new solution as a current solution according to a Metropolis criterion; if the new solution is received as the current solution, directly executing the step 3.5; otherwise, abandoning the new solution, increasing the iteration number of the inner loop by 1, and executing the step 3.4;

the Metropolis criterion is as follows: assuming that the current solution is (x, y, z) and the new solution is (x ', y ', z '), the probability of accepting the new solution as the current solution is

Wherein Δ G ═ G (x ', y ', z ') -G (x, y, z); k ∈ (0,1 ];

step 3.4: judging whether the maximum iteration number of the inner loop is reached; if the maximum iteration times of the internal loop is not reached, returning to the step 3.2; otherwise, executing step 3.5;

step 3.5: resetting the iteration times of the inner loop to 0, increasing the iteration times of the outer loop by 1, and cooling;

step 3.6: judging whether the maximum iteration times of the external loop is reached or whether the current temperature is less than the termination temperature; if the maximum iteration times of the outer loop is reached or the current temperature is lower than the termination temperature, outputting the current solution as a global optimal solution; otherwise, return to step 3.4.

3. The mountain forest fire-prevention-oriented three-dimensional wireless sensor network data routing fusion method according to claim 1 or 2, characterized in that: the comprehensive evaluation method of entropy weight described in step 4.4 and step 4.10 specifically comprises the following steps:

step (1): carrying out standardization processing on the data;

let x be a benefit index, and index values of m items to be evaluated be x₁，x₂，…，x_mAnd is further provided with m_x＝min{x₁，x₂，…，x_m}，M_x＝max{x₁，x₂，…，x_mInstruction of

Wherein, w_xIs the weight of the index, x is obtained₁′，x₂′，…，x_mIs simply x₁，x₂，…，x_mNormalized data;

let y be a cost index and index values of m items to be evaluated as y₁，y₂，…，y_mAnd is further provided with m_y＝min{y₁，y₂，…，y_m}，M_y＝max{y₁，y₂，…，y_mInstruction of

Wherein, w_yIs the weight of the index, and the obtained y₁′，y₂′，…，y_mIs simply y₁，y₂，…，y_mNormalized data;

step (2): form the original data matrix a ═ a_ij)_m×n：

Wherein, a_ijThe index value of the ith item under the jth index;

and (3): calculating the proportion of the index value of the ith item under the jth index;

and (4): calculating the entropy value of the j index;

wherein k is 1/ln m;

and (5): calculating the entropy weight of the j index;

Technical Field

The invention belongs to the technical field of three-dimensional wireless sensor networks, and particularly relates to a mountain forest fire prevention-oriented three-dimensional wireless sensor network data routing fusion method.

Background

The wireless sensor network has the characteristics of large scale, wireless, self-organization, multi-hop and no infrastructure support, the sensor nodes have low cost, small volume and isomorphism, most of the nodes are fixed in position and randomly distributed in a monitoring area, and the network system is required to work for as long as possible.

Based on the above features and limitations of the wireless sensor network, the routing algorithm becomes the key point of the wireless sensor network research. The routing algorithm, also known as a routing algorithm, is an algorithm for finding an optimal transmission path for a sensor node, and aims to improve the processing capacity of a network and reduce energy consumption.

Clustering algorithms (also called hierarchical routing algorithms) are a hot spot in the research of routing algorithms. In the clustering algorithm, nodes are grouped and clustered into clusters, each cluster generally selects one node as a leader (cluster head), cluster member nodes only communicate with the respective cluster heads, and the cluster heads receive data of the cluster member nodes and then send the data of the whole cluster to an observer or a sink node.

Data fusion refers to a process of processing multiple data or information to obtain data more effective and more in line with user requirements. Data fusion can improve the robustness of the system, help users make decisions and reduce the scale of data, thereby saving energy.

At present, most algorithms are designed based on a two-dimensional wireless sensor network, and have certain limitation when being applied to fireproof monitoring of mountains and forests with relatively steep terrain. Therefore, it is an urgent need to solve the problem of providing a layout method of a three-dimensional wireless sensor network and a related route fusion method.

Disclosure of Invention

The invention aims to overcome the limitation of a two-dimensional wireless sensor network and provides a mountain forest fire-proof three-dimensional wireless sensor network data routing fusion method.

The purpose of the invention is realized by the following technical scheme: the method comprises the following steps:

step 1: a three-dimensional wireless sensor network facing forest fire prevention is arranged in a monitoring area;

all nodes in the three-dimensional wireless sensor network facing to mountain forest fire prevention are isomorphic, are completely same equipment and have unique numbers; the initial energy of all nodes is equal and not supplementary, and the initial energy of all nodes is enough for them to complete at least one data transmission; after all the nodes are deployed, the spatial rectangular coordinates of the nodes are known and do not change any more, and all the nodes store the numbers and the spatial coordinates of other nodes; all nodes can adjust the transmitting power and the transmission distance of the nodesThe upper limit is d₀(ii) a The states of the nodes are divided into three types: normal operation, dormancy and death, any node is in one of the three states; a node may enter a sleep state, but not all nodes are allowed to enter the sleep state; the normal node and the dormant node can modify the state when the condition is met, and the state of the dead node is not changed; after all normal nodes transmit data, the network is regarded as finishing data transmission of one turn, and the service life of the whole network is measured by the data transmission turn; the state of the node changes only when the turn changes and is determined before starting to transmit data; the normal nodes have to transmit data in the current round, the dormant nodes do not transmit data in the current round, and the dead nodes do not transmit any data.

The monitoring area is internally provided with a movable Sink node, the Sink node can transmit instructions to all nodes, and the Sink node receives data, fuses and summarizes the data and sends the data to a base station; dividing a monitoring area into n rows and m columns, wherein n multiplied by m small blocks are provided, and at least 2 nodes exist in each small block and meet the requirement

Wherein the content of the first and second substances,are all set values;represents rounding in the x direction; TR (transmitter-receiver)₀The transmission radius of a node, the transmission radius TR of all nodes_iAre identical constants, i.e. TR_i＝TR₀；

Step 2: planning initial clustering;

step 2.1: acquiring clustering information of all nodes;

step 2.2; selecting a node which is not clustered, and judging whether a proper cluster exists or not; if no proper cluster exists, the node is clustered; otherwise, the node joins the best cluster;

with C_jDenotes the jth cluster, let C_jThe node in (A) isFor node N_iIf DNC (i, j) ≦ TR is satisfied_iAnd p is_jIf < MCN, then C is determined_jIs node N_iA suitable cluster of (a); MCN is the set maximum membership of each cluster;

wherein (x)_i,y_i,z_i)、(x_jk,y_jk,z_jk) Are respectively node N_i、N_jkThe coordinates of (a);

if node N_iAll suitable clusters of (A) areFor clustersIf it is satisfied withThen it is decidedIs node N_iThe best cluster of (a);

step 2.3: judging whether the initial clustering is finished or not; if the initial clustering is not finished, returning to the step 2.2; otherwise, executing step 2.4;

the conditions for completion of the initial clustering were: all nodes are added into respective clusters, the clusters have unique numbers, the numbers of the clusters are determined according to the sequence of cluster formation, and at least 1 node is required to be arranged in any one cluster;

step 2.4: judging whether a unary cluster exists; if the unary cluster does not exist, outputting initial clustering information of all nodes; otherwise, clearing the clustering result, and returning to the step 2.1 to re-execute the initial clustering;

and step 3: selecting a cluster head;

first, C is calculated_jCenter of gravity coordinates CG of inner node_j＝(x_j,y_j,z_j) And the median CE of the residual energy of the nodes in the cluster_jAnd constructing a virtual nodeIts spatial rectangular coordinate is (x)_j,y_j,z_j) (ii) a For C_jEach of which isScreening out nodes with enough residual energy to complete a cluster head transmission task as cluster head candidates; if a cluster head candidate does not exist in a certain cluster, setting the cluster as a dormant cluster until clustering is carried out again next time; for each cluster head candidate in the cluster, 2 indexes are considered to select the most suitable cluster head:anddistance d (j) of_k,j₀)、Residual energy ofAnd CE_jIs a distance ofd(j_k,j₀) Andthe smaller the size, the smaller the size is considered to beThe more suitable the clustering;

wherein the center of gravity CG_jIs the center of gravity in the sense of spherical distance and limits (x)_j,y_j,z_j) In the area to be monitored,finding (x) using a modified simulated annealing algorithm_j,y_j,z_j) The exact value of (d);

and 4, step 4: planning a route;

step 4.1: acquiring information of all nodes in a cluster;

the clusters have the following three states: unplanned, planned, and dormant, with any one cluster in and only one; at least one normal node is arranged in the unplanned cluster, and the routing planning of the node is not completed; at least one normal node is arranged in the planned cluster, and the route planning of all the normal nodes is completed; the sleeping cluster has no normal nodes, namely all the nodes are sleeping nodes;

non-cluster head nodes in the cluster are divided into the following four types according to the cluster out degree of the nodes:

(1) cluster inner degree 0 nodeThe node of the class satisfies IOD (j)_u)＝0；

(2) A-type cluster interior degree 1 nodeThe node of the class satisfies IOD (j)_q) 1, the node can transmit data to the cluster head in a single-hop mode;

(3) b-type cluster interior degree 1 nodeThe node of the class satisfies IOD (j)_h) 1, the node can not transmit data to the cluster head in a single-hop mode;

(4) other nodesThe node of the class satisfies IOD (j)_k)≥2；

Step 4.2: judging whether the cluster head can complete the transmission task; if the transmission task cannot be completed, enabling the full cluster nodes to sleep; otherwise, executing step 4.3;

step 4.3: let Counter be 0;

step 4.4: judging whether the Counter is equal to the number of the inspected routing schemes; if the Counter is equal to the number of the inspected routing schemes, determining the index weight through an entropy weight comprehensive evaluation method, and selecting and executing the optimal routing scheme; otherwise, executing step 4.5;

considering RNs independent of each other_jA route planning which is set asThe following four indicators of these route plans are considered to select the best routing scheme: full cluster residual energyVariance of residual energy of full cluster nodesFull cluster data trafficAnd number of sleeping nodesWherein the total cluster residual energyFull cluster data trafficVariance of residual energy of full cluster nodes as benefit indexNumber of sleeping nodesIs a cost-type index;

step 4.5: awakening nodes in the cluster;

step 4.6: judging whether all nodes in the cluster complete routing planning or not; if all nodes in the cluster finish route planning, storing the route scheme, and returning to the step 4.4 after the Counter is enabled to increase by 1; otherwise, executing step 4.7;

step 4.7: judging whether a node with the degree of 0 exists in the cluster; if the cluster has a degree 0 node, enabling the degree 0 node to be dormant; otherwise, executing step 4.8;

step 4.8: judging whether a B-type degree 1 node exists in the cluster; if the cluster has the node with the type B degree 1, the routing is planned for the node with the type B degree 1 at random, and the step 4.7 is returned; otherwise, executing step 4.9;

step 4.9: recording the serial numbers of the nodes with the type A degree 1 in the cluster;

step 4.10: judging whether other nodes exist in the cluster; if other nodes exist in the cluster, selecting one other node as a sending node sender and planning a route for the other node, and returning to the step 4.8; otherwise, executing step 4.11;

after ensuring that the node with the degree 0 and the node with the type B degree 1 are not in the cluster and the node with the type A degree 1 is removed, the rest nodes are other nodes and are set as the other nodesAnd considering the following three indexes of other nodes to select one node as a sending node sender and plan a route for the node: distance d (j) from cluster head_i,j_CH) Residual energy ofDegree of inter-cluster run out Wherein the distance d (j) to the cluster head_i,j_CH) Is a benefit type index, the residual energyAnd intra cluster out degree IOD (j)_i) Is a cost-type indicator; other nodes which are far away from the cluster head, have less residual energy and small in-cluster outturn degree should preferentially plan the route;

inspecting the three index data values of all other nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation method Selecting a node to plan a route according to the probability, which comprises the following steps:is provided withIs scored asThen selectThe probability of planning a route is

Other nodes to be planned are selectedThen, a receiving node receiver is selected for the receiving node receiver); the other nodes are nodes with an in-cluster out-degree of at least 2, and thusAt least 2 receivable nodes; is provided withThe receivable nodes areOne receiving node is selected considering the following four criteria of a receivable node: anddistance d (j) of_i,j_q) Residual energy ofDistance d (j) from cluster head_q,j_CH) Amount of data to be transmittedAndthe nodes with short distance, more residual energy, short distance with cluster head and less data amount to be transmitted should be preferentially taken asA receiving node of (1); examining the above four index data values of the receivable nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation methodAlso selected according to probabilityThe receiving node of (2) is specifically as follows:is provided withIs scored asThen selectAs the probability of the receiving node is

To this end obtainThe routing planning of (2);

step 4.11: planning routes for all nodes with type A degree 1 in the cluster;

step 4.12: planning a route for the cluster head, storing a route scheme, and returning to the step 4.4 after the Counter is increased by 1;

and 5: transmitting and fusing;

is provided withThe collected smoke concentration data, temperature data, humidity data and wind speed data are sequentiallyThe following fusion algorithm is used: let a, B ∈ (0,1), A, B ∈ R⁺To and from

If and only if one of the following two conditions is met, the cluster head transmits alarm information and dangerous node numbers to the Sink node:

(1) in thatIn the presence of at least k_jA (without setting them as) Satisfy the following requirements

(2) In thatIn the presence of at least u_jA (without setting them as) Satisfy the following requirements

Wherein a, B, A and B are artificially set according to actual conditions;

when the two conditions are not met, the cluster head transmits to the Sink nodeTo provide the information in the mountain forest most favorable to the fire to happen;

wherein the content of the first and second substances,is composed ofThe upper quartile of (d);is composed ofThe upper quartile of (d);is composed ofThe lower quartile of (d);is composed ofThe upper quartile of (d);

step 6: judging whether clustering needs to be carried out again; if the clustering is not needed, returning to the step 4; otherwise, executing step 7;

and 7: judging whether the three-dimensional wireless sensor network facing to mountain forest fire prevention is alive or not; if the three-dimensional wireless sensor network facing to mountain forest fire prevention survives, planning non-initial clustering, and returning to the step 3; otherwise, the operation is terminated;

the method for planning the non-initial clustering specifically comprises the following steps:

step 7.1: acquiring clustering information of all nodes;

let N₁,N₂,…,N_mIs all non-dead nodes in the network, N_i∈{N₁,N₂,…,N_m}; in N₁,N₂,…,N_mIf there are t nodes N₁,N₂,…,N_tSatisfy the requirement ofAnd d (i, k) is less than or equal to TR_iIf it is true, then N_iThe global out-degree of (d) is t, which is denoted as eod (i) ═ t; if EOD (i) ═ 0, N_iIs a global out 0 node;

in N₁,N₂,…,N_mIf there are q nodes N₁,N₂,…,N_qSatisfy the requirement ofAnd d (i, l) is less than or equal to TR_lIf it is true, then N_iThe global degree of (a) is q, and is recorded as EED (i) ═ q; if EED (i) is 0, N_iIs a global in-degree 0 node;

step 7.2: judging whether nodes which are not planned to be clustered exist or not; if the nodes which are not planned into clusters do not exist, the clustering information and the priority cluster numbers of all the nodes are output; otherwise, executing step 7.3;

step 7.3: examining the global out-degree and the global in-degree of all unplanned clustered nodes; judging whether a node with the global out degree of 0 exists or not; if no node with the global out-degree of 0 exists, executing an initial clustering algorithm on all unplanned clustering nodes, and returning to the step 7.2; otherwise, executing step 7.4;

step 7.4: judging whether a node with the global in-degree of 0 exists in the nodes with the global out-degree of 0; if the node with the global in-degree of 0 exists, setting the node with the global out-degree and the node with the global in-degree of 0 as a dead node, and returning to the step 7.2; otherwise, executing step 7.5;

step 7.5: searching cluster head winners and cluster head failure winners in the nodes with the global out-degree of 0 and the global in-degree of not 0, setting the cluster head failure winners as dead nodes, executing a priority cluster algorithm, and returning to the step 7.2;

for the node N with the global out degree of 0 and the global in degree of not 0_i，E_iIs N_iIf E is satisfied_i≥E(DD,d_Sink) Then N is_iIs the winner of cluster head, otherwise N_iSelecting the cluster head; the DD is the default data transmission quantity of the nodes, namely the bit number of data collected by each node;

where v is the number of bits of the transmission data; s is the transmission distance;E_elec＝50nJ/bit；ε_fs＝10pJ/(bit·m²)，ε_mp＝0.013pJ/(bit·m⁴)；

the priority cluster algorithm is executed according to the principle of firstly selecting cluster heads and then clustering; the cluster head of the priority cluster is fixed, namely the cluster head winner of the priority cluster is established; the priority clusters can absorb the nodes with the global out-degree not being 0 to enter the clusters as much as possible, and after all the priority clusters are planned to be clustered, if other nodes with the global out-degree not being 0 exist at the moment, initial clustering is carried out on the nodes.

The present invention may further comprise:

step 3 is to search (x) by using an improved simulated annealing algorithm_j,y_j,z_j) The method for the accurate value of (2) is specifically as follows:

step 3.1: initializing and setting an initial temperature, a termination temperature, the maximum iteration times of an outer loop and the maximum iteration times of an inner loop; initializing to generate an initial solution as a current solution of a first iteration;

step 3.2: adding disturbance to the current solution to obtain a new solution;

step 3.3: judging whether to receive a new solution as a current solution according to a Metropolis criterion; if the new solution is received as the current solution, directly executing the step 3.5; otherwise, abandoning the new solution, increasing the iteration number of the inner loop by 1, and executing the step 3.4;

the Metropolis criterion is as follows: assuming that the current solution is (x, y, z) and the new solution is (x ', y ', z '), the probability of accepting the new solution as the current solution is

Wherein Δ G ═ G (x ', y ', z ') -G (x, y, z); k ∈ (0,1 ];

step 3.4: judging whether the maximum iteration number of the inner loop is reached; if the maximum iteration times of the internal loop is not reached, returning to the step 3.2; otherwise, executing step 3.5;

step 3.5: resetting the iteration times of the inner loop to 0, increasing the iteration times of the outer loop by 1, and cooling;

step 3.6: judging whether the maximum iteration times of the external loop is reached or whether the current temperature is less than the termination temperature; if the maximum iteration times of the outer loop is reached or the current temperature is lower than the termination temperature, outputting the current solution as a global optimal solution; otherwise, return to step 3.4.

The comprehensive evaluation method of entropy weight described in step 4.4 and step 4.10 specifically comprises the following steps:

step (1): carrying out standardization processing on the data;

let x be a benefit index, and index values of m items to be evaluated be x₁,x₂,…,x_mAnd is further provided with m_x＝min{x₁,x₂,…,x_m},M_x＝max{x₁,x₂,…,x_mInstruction of

Wherein, w_xIs the weight of the index, x is obtained₁′,x₂′,…,x_mIs simply x₁,x₂,…,x_mNormalized data;

let y be a cost index and index values of m items to be evaluated as y₁,y₂,…,y_mAnd is further provided with m_y＝min{y₁,y₂,…,y_m},M_y＝max{y₁,y₂,…,y_mInstruction of

Wherein, w_yIs the weight of the index, and the obtained y₁′,y₂′,…,y_mIs simply y₁,y₂,…,y_mNormalized data;

step (2): form the original data matrix a ═ a_ij)_m×n:

Wherein, a_ijThe index value of the ith item under the jth index;

and (3): calculating the proportion of the index value of the ith item under the jth index;

and (4): calculating the entropy value of the j index;

wherein k is 1/lnm;

and (5): calculating the entropy weight of the j index;

the invention has the beneficial effects that:

the invention provides a mountain forest fire-prevention-oriented three-dimensional wireless sensor network data routing fusion method which comprises a clustering algorithm based on entropy weight comprehensive evaluation and improved simulated annealing, a routing algorithm based on entropy weight comprehensive evaluation and probability optimization, and a rapid data fusion algorithm based on data types and user requirements. The invention overcomes the limitation of a two-dimensional wireless sensor network to a certain extent, has certain energy saving property, strong robustness and high fault tolerance, and can well meet the requirement of forest fire prevention monitoring.

Drawings

Fig. 1 is a schematic view of a monitoring area space of a three-dimensional wireless sensor network facing mountain forest fire prevention according to the present invention.

Fig. 2 is a general flow diagram of the present invention.

FIG. 3 is a flow chart of the initial clustering in the present invention.

FIG. 4 is a flow chart of a simulated annealing algorithm in accordance with the present invention.

FIG. 5 is a flow chart of non-initial clustering in the present invention.

FIG. 6 is a flow chart of a priority clustering algorithm in the present invention.

Fig. 7 is a flow chart of a routing algorithm in the present invention.

Detailed Description

The invention is further described below with reference to the accompanying drawings.

The invention aims to overcome the limitation of a two-dimensional wireless sensor network and provides a set of algorithm based on a three-dimensional wireless sensor network, which comprises a clustering algorithm based on entropy weight comprehensive evaluation and improved simulated annealing, a routing algorithm based on entropy weight comprehensive evaluation and probability optimization, and a rapid data fusion algorithm based on data types and user requirements.

Firstly, the following requirements are put forward for the wireless sensor network and the operation thereof:

(1) all nodes are homogeneous, they are identical devices, have unique numbers, and do not change.

(2) The initial energy of all nodes is equal and not supplementary. The initial energy of all nodes is sufficient for them to complete at least one data transmission.

(3) After all the nodes are deployed, the spatial rectangular coordinates of the nodes are known and do not change any more, and all the nodes store the numbers and the spatial coordinates of other nodes.

(4) All nodes may adjust transmit power to transmit data to the receiving node with as little energy as possible. The transmission power of the node has an upper limit, so that when the energy is sufficiently large, the upper limit d of the transmission distance of the node can be derived from the upper limit of the transmission power₀。

(5) The states of the nodes are divided into three types: normal operation, dormancy and death. Any one node is in, and can only be in, one of these three states. Nodes may enter a sleep state but not all nodes are allowed to enter the sleep state. The normal node and the dormant node may modify the state when the condition is satisfied. The status of the dead node is no longer changed.

(6) After all normal nodes transmit data, the network is considered to complete one round of data transmission, and the data transmission time is considered to be extremely short. The lifetime of the entire network is measured in turns of transmitted data.

(7) The state of the node changes only when the turn changes and is determined before starting to transmit data. The normal nodes have to transmit data in the current round, the dormant nodes do not transmit data in the current round, and the dead nodes do not transmit any data.

(8) A plurality of movable Sink nodes (Sink nodes) exist in a monitoring area, the Sink nodes have extremely high energy and a strong communication function, and can transmit instructions to all the nodes. And the Sink node receives the data, fuses and summarizes the data and sends the data to the base station.

(9) The energy consumption of the node for transmitting data is calculated according to the following formula (energy consumption formula).

Where v is the number of bits of the transmission data, s is the transmission distance,E_elec＝50nJ/bit，ε_fs＝10pJ/(bit·m²)，ε_mp＝0.013pJ/(bit·m⁴)。

(10) by using the coordinate system in the following formula,

the following formula is an equivalent coordinate transformation formula thereof,

the following formula represents the monitoring region, whereinIs set manually.

If getThe aerial image of the monitored area is obtained as shown in fig. 1, which is a portion of a sphere. The advantages of using such a monitoring area are: the monitored area in one or more of the graphs may be used to approximate the actual contour of the mountain forest.

(11) With N_iRepresenting the ith node. The distance between the nodes is calculated by adopting the spherical distance, namely N is set_iHas the rectangular coordinate of (x)_i,y_i,z_i)，N_jHas the rectangular coordinate of (x)_j,y_j,z_j) Then the distance between them is

When the area of the monitoring area is larger, the spherical distance is adopted, and the reality is better. When the area of the monitoring region is small, the spherical distance may be replaced with the euclidean distance. If not stated, a spherical distance is adopted as the distance between the nodes.

(12) The maximum distance maxs which can be transmitted by the node can be calculated by an energy consumption formula and by utilizing the residual energy and the transmission data volume of the node. Note N_iHas a transmission radius of

TR_i＝min{maxs,d₀}.

Because the nodes are isomorphic, the transmission radius of all nodes is the same constant, denoted as TR, before the network is operated₀。

(13) The monitoring area is divided into n rows and m columns for a total of n x m patches, see fig. 1. At least 2 nodes exist in each small block and satisfy

This ensures that before operationAll nodes may establish communication in some manner.Indicating that x is rounded up.

The above-mentioned 13 requirements include a layout method of a three-dimensional wireless sensor network.

The operation flow of the whole network is shown in fig. 2. Wherein, the re-clustering condition is set as follows: the number of normal nodes in the current round is not more than 90% of the number of normal nodes when clustering is completed. The network is considered dead when all nodes in the network die or the remaining energy of none of the nodes is sufficient to complete the cluster head transmission task.

The clustering algorithm (hereinafter referred to as clustering algorithm) based on entropy weight comprehensive evaluation and improved simulated annealing comprises the following 3 algorithms: an initial clustering algorithm, a cluster head selection algorithm and a non-initial clustering algorithm.

The initial clustering algorithm flow is shown in fig. 3. The cluster member nodes transmit data to the cluster heads in a single-hop or multi-hop (self-adaptive) mode, the cluster head nodes perform data fusion after receiving the data, and the fused data are transmitted to the movable Sink nodes. The Sink node can move to the position near each cluster head to receive data, and the distance between the Sink node and the cluster head when the Sink node receives the data is not more than d_Sink。d_SinkThe value of (A) can be set manually according to the requirement.

Clustering is defined as all nodes joining their respective clusters. Optionally, a node that is not clustered refers to: randomly selecting one node from all nodes which are not clustered, wherein the probability of selecting any node which is not clustered is equal to the probability of selecting any node which is not clustered.

The clusters also have unique numbers, the cluster numbers are dependent on the order in which the clusters are formed, and there must be at least 1 node in any one cluster. With C_jDenotes the jth cluster, let C_jThe node in (A) isN is defined as follows_iAnd C_jDistance DNC (i, j):

if it is satisfied with

DNC(i,j)≤TR_i,

And is

p_j＜MCN,

Then C is_jIs N_iA suitable cluster of (a). Where MCN is the maximum number of members per cluster and can be set artificially as desired. Let N_iAll suitable clusters of (A) areIf it is notThenIs N_iThe best cluster of (a).

Sometimes, a unary cluster (a cluster with only one node) occurs, and the occurrence of the unary cluster is not beneficial to saving energy, so the occurrence of the unary cluster is to be avoided as much as possible. After clustering is finished and before a cluster head is selected, a link for checking whether a monadic cluster appears is set, and if the monadic cluster appears, a clustering scheme of all nodes needs to be emptied, and clustering is planned again. After the initial clustering is completed, cluster heads are selected for each cluster according to the following cluster head selection algorithm.

The cluster head selection algorithm is an algorithm based on comprehensive evaluation and has the right to set some clusters to sleep until the next time to be clustered again. The algorithm first calculates C_jCenter of gravity coordinates CG of inner node_j＝(x_j,y_j,z_j) And the median CE of the residual energy of the nodes in the cluster_jAnd constructing a virtual nodeIts spatial rectangular coordinate is (x)_j,y_j,z_j). For C_jEach of which isThe algorithm screens out nodes (cluster head candidates) with enough residual energy to complete the cluster head transmission task. If there is no cluster head candidate in a cluster, the cluster is set as a dormant cluster until the next re-clustering. For each cluster head candidate in the cluster, 2 indexes are considered to select the most suitable cluster head:anddistance d (j) of_k,j₀) Andresidual energy ofAnd CE_jIs a distance ofCenter of gravity CG herein_jIs the center of gravity in the sense of spherical distance and limits (x)_j,y_j,z_j) In the monitored area, i.e.Finding (x) using a modified simulated annealing algorithm_j,y_j,z_j) The exact value of (c). To save energy consumption and balance energy consumption, d (j)_k,j₀) Andthe smaller the size, the smaller the size is considered to beThe more suitable it is to become a cluster head.

The flow of the simulated annealing algorithm is shown in fig. 4. In the simulated annealing algorithm, the initialization step completes the setting of a plurality of parameters including initial temperature, termination temperature, maximum iteration number of outer loop, length of Markov chain (maximum iteration number of inner loop) and the like, and generates an initial solution as the current solution of the first iteration. In the inner loop, the algorithm applies a disturbance to the current solution to obtain a new solution, substitutes the new solution into the objective function to judge whether the new solution is superior to the current solution, and judges whether to accept the new solution as the current solution according to the Metropolis criterion. Taking the minimum value of the objective function G (x, y, z) as an example, the current solution is set to (x, y, z), and the new solution is set to (x ', y ', z '). The Metropolis criteria are: the probability of accepting the new solution as the current solution is

Where Δ G ═ G (x ', y ', z ') -G (x, y, z), k ∈ (0,1] can be set on demand, if a new solution is accepted, the inner loop is skipped, otherwise the inner loop iteration number is incremented by 1.

In the outer loop, after the algorithm jumps out of the inner loop, the current temperature is cooled once, and the adopted cooling method is to directly multiply the current temperature by an artificially set cooling coefficient to obtain a new temperature as the current temperature.

The improved simulated annealing algorithm limits the current solution and disturbance according to the boundary of the closed rectangle with the minimum area in the coordinate system of the cluster, namely the closed rectangle with the minimum area containing all nodes in the cluster in the coordinate system can be found

If the new solution is not within D, it is discarded. Limiting the perturbation vector toSatisfy the requirement of

In addition, the simulated annealing algorithm is independently executed for u times (u ≧ 2), and one with the optimal objective function among the u approximately optimal solutions is found as the approximately optimal solution of the improved simulated annealing algorithm. Therefore, the algorithm operation efficiency can be greatly improved, and the efficiency and the precision of the calculation method can be better balanced.

The entropy weight comprehensive evaluation method comprises the following steps:

(1) form the original data matrix a ═ a_ij)_m×n:

Wherein a is_ijThe index value of the ith item under the jth index is obtained.

(2) Calculating the specific gravity of the index value of the ith item under the jth index

(3) Calculating entropy of jth index

Wherein k is 1/lnm.

(4) Calculating the entropy weight of the jth index

In the process of solving the optimal problem by using a comprehensive evaluation method, data needs to be standardized, and the specific method is as follows:

(1) let x be a benefit index, and index values of m items to be evaluated be x₁,x₂,…,x_mAnd is further provided with m_x＝min{x₁,x₂,…,x_m},M_x＝max{x₁,x₂,…,x_mInstruction of

Wherein w_xIs the weight of the index. Thus obtained x₁′,x₂′,…,x_mIs simply x₁,x₂,…,x_mNormalized data.

(2) Let y be a cost index and index values of m items to be evaluated as y₁,y₂,…,y_mAnd is further provided with m_y＝min{y₁,y₂,…,y_m},M_y＝max{y₁,y₂,…,y_mInstruction of

Wherein w_yIs the weight of the index. Obtained y₁′,y₂′,…,y_mIs simply y₁,y₂,…,y_mNormalized data.

By utilizing an entropy weight comprehensive evaluation method, the nodes which are most suitable to be synthesized into cluster heads in the clusters can be found. The comprehensive evaluation method of the entropy weight has strong objectivity.

When the non-initial clustering algorithm is executed, all nodes have completed at least one round of data transmission, and various extreme situations may occur. The non-initial clustering algorithm is designed for various extreme conditions, and robustness and energy conservation of the algorithm can be greatly improved. If no extreme condition occurs, the non-initial clustering algorithm is very similar to the initial clustering algorithm. The flow of the non-initial clustering algorithm is shown in fig. 5. The non-initial clustering algorithm is the only algorithm that can set a node as a dead node.

Let N₁,N₂,…,N_mIs all non-dead nodes in the network, N_i∈{N₁,N₂,…,N_m}. In N₁,N₂,…,N_mIf there are t (do not assume that they are N)₁,N₂,…,N_t) Satisfies the following conditions:k≠i、d(i,k)≤TR_iif it is true, then N_iThe global out-degree of (d) is t, denoted eod (i) ═ t. If EOD (i) ═ 0, N_iIs a global out 0 node. In N₁,N₂,…,N_mIf there are q (provided that they are N)₁,N₂,…,N_q) Satisfy the requirement ofl≠i，d(i,l)≤TR_lIf it is true, then N_iIs denoted as EED (i) ═ q. If EED (i) is 0, N_iIs a global in 0 node. If N is present_iIs both a global out 0 node and a global in 0 node, then N_iAre two degree 0 nodes. Let N_iIs a global out 0 node but not a global in 0 node, E_iIs N_iThe remaining energy of (c). If E is satisfied_i≥E(DD,d_Sink) Then N is_iIs the winner of cluster head, otherwise N_iFor the cluster head failure selector, DD is the default data transmission amount of the node, i.e. the number of bits of data collected by each node. The global out-degree 0 node can not transmit data to any node, the utilization value of the node is only the cluster head, when the global in-degree is 0 or the cluster head is selected, the node does not have any utilization value any more, and the node is set to be a dead node.

Let N_iIs the winner of the cluster head, with N_iEstablishing a cluster C as a cluster head_j(priority cluster). Is provided with C_jFor a priority cluster, hold C_jThe cluster head winner is N_j1，N_iIs a node with global out-degree not 0, if | C is satisfied_j|＜MCN、d(j₁,i)≤TR_iThen C is_jIs N_iOne feasible priority cluster. The priority cluster algorithm is executed according to the principle of firstly selecting cluster heads and then clustering. The cluster head of the priority cluster is fixed, i.e. the cluster head winner who established the priority cluster. The flow of the priority clustering algorithm is shown in fig. 6. The priority cluster can absorb the nodes with the global out-degree not being 0 as much as possibleAnd (4) clustering. After all the priority clusters are planned into clusters, if other nodes with the global out-degree not being 0 exist at the moment, performing initial clustering on the nodes. The initial clustering algorithm performed here will not consider any one of the priority clusters as a suitable cluster. And after the non-initial clustering algorithm is executed, executing the same cluster head selection algorithm on the non-priority cluster, and obtaining the clustering scheme of all the non-dead nodes.

A routing algorithm (hereinafter referred to as a routing algorithm) based on entropy weight comprehensive evaluation and probability optimization plans routes for all nodes of all clusters. The main idea of the algorithm is as follows: and classifying the nodes in the cluster, and adopting different planning strategies for the nodes in different classes. And independently executing routing planning for the same cluster for a plurality of times, and selecting one scheme with the best effect as a final transmission scheme. This may effectively increase the probability of finding a better routing scheme. Before the algorithm is executed, the nodes in the network have completed clustering, and a cluster head is elected. The algorithm operates according to fig. 7. The algorithm has the following principle settings:

(1) all non-dead nodes must be routed in the present algorithm.

(2) During the running process of the algorithm, the state of the node is not changed into death. If the state of a certain node is adjusted to be dormant in a certain routing scheme, all transmitting nodes of the node become dormant nodes in the routing scheme.

(3) The algorithm plans routes for all the nodes of the clusters one turn in sequence, namely, any node of any one cluster is selected as a receiving node or becomes a dormant node in the next turn, and the problem is solved after the algorithm is executed. Shown in fig. 7 is a flow chart for planning routes for all nodes of any one cluster.

(4) The node will not transmit data to nodes of other clusters.

(5) After the receiving node of any one node is established, no change is made in the routing scheme, and the node becomes a planned node, which is not used by other unplanned nodes in the cluster.

(6) The clusters have the following three states: unplanned, planned, and dormant. Any one cluster is in and only one of them. At least one normal node is arranged in the unplanned cluster, and the routing planning of the node is not completed. And at least one normal node is arranged in the planned cluster, and the route planning of all the normal nodes is completed. The sleeping cluster has no normal nodes, namely all the nodes are sleeping nodes. The dormant cluster occurs in an extreme case, which is caused by that the energy of the cluster head is insufficient to transmit the fused data of the cluster to the Sink node.

If C is present_jThere are t nodes (including cluster head, do not set them as) Satisfy the requirement of Is a normal node,Is established, thenThe cluster out-degree of (c) is t, the number of receivable nodes is t, and is marked as IOD (j)_k)＝t。

Non-cluster head nodes in the cluster are divided into the following four types according to the cluster out degree of the nodes:

(1) cluster inner degree 0 nodeThe node of the class satisfies IOD (j)_u)＝0.

(2) A-type cluster interior degree 1 nodeThe node of the class satisfies IOD (j)_q) 1, and the node can transmit data to the cluster head in a single-hop mode.

(3) B type cluster interior 1 sectionDotThe node of the class satisfies IOD (j)_h) 1, and the node cannot transmit data to the cluster head in a single-hop manner.

(4) Other nodesThe node of the class satisfies IOD (j)_k)≥2.

Since the intra-cluster degree of a node is not considered here, the four types of nodes are sequentially called a degree 0 node, an a-type degree 1 node, a B-type degree 1 node, and other nodes without causing ambiguity.

When the cluster head can complete the transmission task, the method adopts a manual setting mode to set C_jTo investigate the number of routing schemes RN_jThe algorithm will be C_jIndependently planning a routing RN_jAnd then, selecting the optimal routing scheme from the routing schemes. Each time is C_jWhen planning the route, the router will firstly wake up C_jNode in (1), i.e. setting C_jAll nodes in (1) are normal nodes.

C_jIn the possible existence degree 0 nodeNeither data is sent nor received in this round, settingIs a sleeping node. After any node plans a route, the cluster out-degree of other unplanned nodes may be reduced, so that the degree 0 nodes in the cluster need to be continuously searched and set as dormant nodes until no degree 0 node exists in the cluster.

C_jThere may be node of B type degree 1They tend to be very far from the cluster head compared to non-type B degree 1 nodes.Is unique and the receivable node is not a cluster head.The following two cases are only possible for route planning: or eitherTransmitting data to its only receivable node, orBecoming a dormant node in the current turn. The sleeping node is not beneficial to data transmission, so that all the node B type degree 1 nodes can work normally as far as possible. After the degree 0 node is set as the dormant node, the algorithm plans the route for the type B degree 1 node preferentially. Similarly, since the cluster out-degree of other nodes may be reduced after any one node plans the route, there may be a case where a new degree 0 node appears after a route is planned for one B-type degree 1 node, and it is necessary to detect the degree 0 nodes again and set them as sleeping nodes after a route is planned for any one B-type degree 1 node until no B-type degree 1 node exists in the cluster.

C_jThere may be a type degree 1 nodeThey tend to be very close to the cluster head compared to non-type a degree 1 nodes.Is unique and the receivable node is the cluster head.The following two cases are only possible for route planning: or eitherTransmitting data to the cluster head, orBecoming a dormant node in the current turn. On one hand, the dormant node is not beneficial to data transmission; on the other hand, becauseOften very close to the cluster head, it should take on more transmission tasks. Therefore, the algorithm plans routes for type a degree 1 nodes only when only type a degree 1 nodes in the cluster are unplanned.

After ensuring that the node with the degree 0 and the node with the type B degree 1 are not in the cluster and the node with the type A degree 1 is removed, the rest nodes are other nodes and are not set as the nodesAnd considering the following three indexes of other nodes to select one node as a sending node (sender) and plan a route for the node: distance d (j) from cluster head_i,j_CH) Residual energy ofDegree of inter-cluster run out Other nodes which are far away from the cluster head, have less residual energy and small cluster outrun degree should preferentially plan the route, namely d (j)_i,j_CH) Is an index of the benefit type,is a cost-type indicator, IOD (j)_i) Is a cost-type indicator. Inspecting the three index data values of all other nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation methodIn order to enhance the flexibility of the algorithm, a node is selected according to the probability to plan the route, which specifically comprises the following steps:is provided withIs scored asThen selectThe probability of planning a route is

Other nodes to be planned are selectedThereafter, a receiving node (receiver) is also selected for it. The other nodes are nodes with an in-cluster out-degree of at least 2, and thusThere are at least 2 receivable nodes. Without being provided withThe receivable nodes areOne receiving node is selected considering the following four criteria of a receivable node: anddistance d (j) of_i,j_q) Residual energy ofDistance d (j) from cluster head_q,j_CH) Amount of data to be transmittedAndthe nodes with short distance, more residual energy, short distance with cluster head and less data amount to be transmitted should be preferentially taken asI.e. d (j)_i,j_q) Is a cost-type index,Is a benefit type index, d (j)_q,j_CH) Is a cost-type index,Is a cost-type indicator. Examining the above four index data values of the receivable nodes, and obtaining the score of each node by using an entropy weight comprehensive evaluation methodAlso selected according to probabilityThe receiving node of (2) is specifically as follows:is provided withIs scored asThen selectAs the probability of the receiving node is

To this end obtainRouting planning of (2). After a route is planned for any node, the cluster out-degree of other unplanned nodes may be reduced, that is, the cluster out-degree is reducedAfter routing is planned, a new node of type B degree 1 may appear, and therefore after routing is planned for any other node, the node of type B degree 1 needs to be detected again and planned. In addition, the type a degree 1 node may also become a degree 0 node or a type B degree 1 node, because the type a degree 1 node is planned last, and before that, there is often a non-type a degree 1 node to transmit data to them, which will cause their transmission radius to decrease, and thus there is a possibility that data cannot be transmitted to the cluster head.

Considering RNs independent of each other_jA route planning which is set asThe following four indicators of these route plans are considered to select the best routing scheme: full cluster residual energy(benefit type index), variance of residual energy of full cluster nodes(cost index), full cluster data transmission volume(benefit type indicator) and number of sleeping nodes(cost-type index),the four indexes of all route plans are examinedAnd according to the value, selecting the routing plan with the highest score as the optimal routing scheme by using an entropy weight comprehensive evaluation method, and transmitting data according to the optimal routing scheme in the turn.

And fusing the collected data by using a rapid data fusion algorithm (hereinafter referred to as a data fusion algorithm) based on the data type and the user requirements, wherein data level fusion is mainly used. The data fusion algorithm can effectively reduce energy consumption and help a user to make a decision.

The early stage of fire is mainly characterized by generating invisible smoke and temperature change, and all nodes are provided with the following four sensors: smoke sensor, temperature sensor, humidity transducer, wind speed sensor. Wherein the smoke sensor monitors the shielding degree or the dimming rate, and the unit is obs/m; the temperature sensor monitors the temperature, and the unit is; the humidity sensor monitors the relative humidity and is dimensionless; the wind speed sensor monitors wind speed, and the unit is m/s. In practical applications, the user is concerned about the occurrence of a fire. The earlier the fire is discovered, the more beneficial the people to put out the fire in time. The higher the smoke concentration, the higher the temperature, the lower the humidity and the higher the wind speed, the more beneficial to the fire. At C_jAmong all data collected by the cluster head, the data with high smoke concentration, high temperature, low humidity and high wind speed are the data which are most needed by the user, and the other data which are not beneficial to the occurrence of fire are not concerned by the user.

Is provided withThe collected smoke concentration data, temperature data, humidity data and wind speed data are sequentiallyThe following fusion algorithm is used: let a, B ∈ (0,1), A, B ∈ R⁺To and from

The cluster head transmits alarm information and dangerous node numbers to the Sink node, and if and only if one of the following two conditions is met:

(1) in thatIn the presence of at least k_jA (without setting them as) Satisfy the following requirements

(2) In thatIn the presence of at least u_jA (without setting them as) Satisfy the following requirements

Here, a, B can be set artificially according to actual conditions. When the two conditions are not satisfied, recordingIs composed ofUpper quartile of (2), not provided with

Memo

Note the bookIs composed ofUpper quartile of (2), not provided with

Memo

Note the bookIs composed ofLower quartile of (2), not provided with

Memo

Note the bookIs composed ofUpper quartile of (2), not provided with

Memo

Transmitting cluster head to Sink nodeTo provide the information in the mountain forest that is most favorable for the occurrence of fire.

Compared with the prior art, the invention has the beneficial effects that: the method overcomes the limitation of a two-dimensional wireless sensor network to a certain extent, has certain energy conservation, strong robustness and high fault tolerance, and can well meet the requirement of forest fire prevention monitoring.

Example 1:

the parameters are set as follows:the maximum cluster member number MCN is 30, the initial energy DE of the node is 1J, the default transmission data volume DD of the node is 64bit, and the quantity RN of the routing schemes is considered_j3; the parameters of the simulated annealing algorithm are set as follows: initial temperature of 10⁶The temperature reduction coefficient is 0.98, and the termination temperature is 10⁵The maximum number of iterations of the inner loop is 10⁵The maximum iteration number of the outer loop is 5000, k in Metropolis criterion is 1, and the execution number of the simulated annealing algorithm isWherein p is_jIs C_jThe number of nodes in (1). In the example, 4 nodes are uniformly distributed in the small block with the smallest area, and 483 nodes are uniformly distributed in other small blocks according to the area ratio. When the node residual energy is sufficient, the average energy consumed by transmitting data by one node in a round is 4.518 multiplied by 10^-5J, if data is transmitted 24 rounds per day, it can be used for at least 2 years. Node density is increased, the execution times u of the simulated annealing algorithm are increased, and the quantity RN of routing schemes is inspected_jEnergy savings may be further improved, but some algorithm operating efficiency may be sacrificed accordingly.

The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.