阅读说明：本技术 一种基于iaga算法的高速列车车联网拓扑优化方法 (High-speed train networking topology optimization method based on IAGA algorithm ) 是由 贺德强 孙大亮 陈彦君 陈泽前 梁晨 李先旺 于 2021-07-19 设计创作，主要内容包括：本发明公开了一种基于IAGA算法的高速列车车联网拓扑优化方法,将网络拓扑结构优化转化为网络终端设备的分配规划过程,对车载网络的各参数进行定义；在车载网络连通可靠性约束条件下初步构建列车车联网的两层拓扑结构模型；在数据通信流量的计算基础上建立以网络流量负载和通信时延为优化指标的目标函数模型根据全双工交换节点物理条件的限制确定目标函数的约束条件,使用改进的自适应遗传算法对车载终端设备的分配方式进行规划；在改进的自适应遗传算法中进行优化求解,得到优化后网络拓扑中终端设备节点的分配结果。本发明能有效减小拓扑中各交换节点子网的通信负载,从而减小了终端设备间的端到端时延,改进了高速列车车联网的实时性能。(The invention discloses a high-speed train car networking topology optimization method based on an IAGA algorithm, which comprises the steps of converting network topology optimization into a distribution planning process of network terminal equipment, and defining various parameters of a vehicle-mounted network; preliminarily constructing a two-layer topological structure model of the train-vehicle networking under the constraint condition of the communication reliability of the train-vehicle network; establishing an objective function model taking network traffic load and communication delay as optimization indexes on the basis of data communication traffic calculation, determining constraint conditions of an objective function according to the limitation of physical conditions of full-duplex switching nodes, and planning the distribution mode of vehicle-mounted terminal equipment by using an improved adaptive genetic algorithm; and carrying out optimization solution in an improved adaptive genetic algorithm to obtain an optimized distribution result of the terminal equipment nodes in the network topology. The invention can effectively reduce the communication load of each switching node subnet in the topology, thereby reducing the end-to-end time delay among terminal devices and improving the real-time performance of the high-speed train car networking.)

1. An IAGA algorithm-based high-speed train networking topology optimization method is characterized by comprising the following steps: optimizing and converting the network topology structure into a distribution planning process of network terminal equipment, and performing optimized distribution processing on the communication data of the high-speed train networking to obtain a distribution result of network topology nodes; the optimization process comprises the following steps:

step S01: optimizing and converting a network topological structure into a distribution planning process of vehicle-mounted terminal equipment, defining parameters of a vehicle-mounted network, wherein the parameter definitions comprise the definitions of communication flow, communication time delay, exchange node performance parameters, terminal nodes and link parameters, and determining a basic physical topological structure of the train car networking according to the communication reliability of the vehicle-mounted network;

step S02: preliminarily constructing a two-layer topological structure model of the train-to-train network under the condition of communication reliability constraint of a vehicle-mounted network, calculating communication flow of process data and message data communicated between nodes of vehicle-mounted terminal equipment in the two-layer topological structure model, and constructing a communication weight matrix and an adjacent mapping matrix of the nodes of the terminal equipment according to the calculated communication flow value;

step S03: establishing an objective function model taking network traffic load and communication delay as optimization indexes on the basis of data communication traffic calculation, and establishing a function taking the minimum communication traffic load, the minimum load balance and the minimum transmission delay among network terminal devices in all vehicle layer subnets in the internet of vehicles network topology structure as optimization objectives according to a traffic weight matrix and an adjacent mapping matrix;

step S04: establishing an evaluation function model of multi-objective optimization, and determining constraint conditions of a target function according to the limitation of physical conditions of full-duplex switching nodes;

step S05: planning the distribution mode of the vehicle-mounted terminal equipment by using the improved adaptive genetic algorithm for the objective function model optimized in the step S03, thereby determining a coding scheme, a fitness function, adjusting a genetic operator and designing the whole algorithm flow;

step S06: and (4) carrying out optimization solution in an improved adaptive genetic algorithm by using the traffic of the high-speed train car networking obtained in the step (S02) to obtain the distribution result of the terminal equipment nodes in the optimized network topology.

2. The IAGA-algorithm-based high-speed train networking topology optimization method according to claim 1, characterized in that: in the parameter definition of step S01, the parameter definition of the communication traffic includes data length, data period and data priority weight, the definition of the switching node performance parameter includes the number of switching nodes, the number of switching node ports and port transmission rate, and the definition of the end node and link parameter includes the number of end nodes, the number of links and link communication bandwidth in the car networking network topology.

3. The IAGA-algorithm-based high-speed train networking topology optimization method according to claim 1, characterized in that: for step S02, two types of real-time data, i.e., process data and message data, are fully considered, the communication traffic of the two types of data communicated by two vehicle-mounted terminal devices in the car networking topology structure is calculated, and a corresponding traffic weight matrix and a communication traffic q are constructed according to the calculated traffic flow value_ijThe calculation of (a) satisfies:

where i and j represent two terminal devices communicating in the networkNode, beta_ijFlow mean, τ, representing periodic real-time process data in a network_ijRefers to communication traffic of aperiodic message data in the network,andweights respectively representing priorities of process data and message data can construct a corresponding traffic weight matrix A according to the calculated traffic flow value, and the traffic weight matrix A meets the requirements;

in which the matrix element q_ijThe directional communication traffic weight value representing data between a source device i and a target device j, i, j is {1,2, …, m }, m is the total number of nodes of communication devices in the network, the size of a matrix element is defined as the ratio of the actual traffic of two device nodes to the minimum traffic of all communication devices in the network, and if the traffic of round-trip communication between the two device nodes is equal and no communication exists between the device nodes, a diagonal element in the matrix is 0.

4. The IAGA-algorithm-based high-speed train networking topology optimization method according to claim 3, characterized in that: after constructing the corresponding traffic weight matrix A, constructing the corresponding adjacency mapping matrix, wherein the construction process comprises the following steps: first, element X of the adjacency matrix is defined according to the relationship between the device node and the sub-network of switching nodes_ikRespectively as follows:

in the formula, i represents a device node for communication, k represents a subnet number of a switching node at a vehicle layer, k is {1,2,3, …, N }, and N represents the total number of switching nodes at the vehicle layer, and the formula indicates that when a terminal device node i is in the switching node subnet k, a corresponding matrix element is 1, and otherwise, the corresponding matrix element is 0;

secondly, constructing a corresponding equipment node adjacency mapping matrix B:

the conditional formula used to determine communication between device nodes of two different subnets can be expressed as:

H_ij＝X_ik(1-X_jk)

in the formula H_ijIndicating the communication status of a pair of device nodes in different switch node subnets, when the source device i and the target device j are in the same switch node subnet, H_ij0, when the source device i and the target device j are not in the same subnet, H_ij＝1。

5. The IAGA-algorithm-based high-speed train networking topology optimization method is characterized in that constraint conditions of a multi-objective optimization objective function are determined according to the limitation of physical conditions of full-duplex switching nodes, and the multi-objective optimization objective function is established by a communication traffic weight matrix and an adjacent mapping matrix;

step 501: the first objective is to minimize the communication load among all the vehicle layer switching node subnets in the network, and the objective function is expressed as:

in the formula, a device node i, j of communication is {1,2, …, M }, where M is a total number of device nodes participating in communication in the network; k is a subnet number of the switching nodes in the vehicle layer, k is {1,2,3, …, N }, and N represents the total number of the switching nodes in the vehicle layer;

step 502: the second objective is to minimize the difference in load traffic transmitted between the switching node subnets, whose objective function is expressed as:

wherein k represents the subnet serial number of the switching node on the vehicle layer, and the total communication load flow ω (k) of any switching node subnet k is defined as:

in the formula, the device node i, j of communication is {1,2, …, M }, M is the number of communication device nodes in the topology network, and n is_kThe number of the communication equipment nodes in the switching node subnet k is represented, and the expression represents the sum of the communication flow of all the equipment nodes in a certain switching node subnet and all the communication nodes except the subnet in the network;

step 503: a third objective is to minimize the data transmission delay in the network, whose objective function is expressed as:

in the formula, D_ijThe number of switching nodes through which data streams communicated by the devices i and j pass is M, the number of total nodes of the communication terminal device is M, and D is the total number of switching nodes in the network topology;

step 504: for objective function f₁、f₂、f₃And planning the distribution mode of the vehicle-mounted terminal equipment by using an improved adaptive genetic algorithm.

6. The IAGA-algorithm-based high-speed train networking topology optimization method according to claim 5, characterized in that: for the objective function f₁、f₂、f₃Allocation of vehicle terminals using improved adaptive genetic algorithmsThe method is to plan by converting a multi-objective problem into a single objective planning problem by using a linear weighting method, before determining a total objective function model, consistency processing is firstly carried out on dimensions of the three objective functions, then the influence of difference in relative importance degree and magnitude of each objective function is comprehensively considered, and a weight coefficient is determined, wherein the objective functions are restricted by the physical conditions of full-duplex switching nodes, and the total objective evaluation function and the restriction conditions are as follows:

wherein, in the objective function F, F_t ^maxAnd f_t ^minRespectively representing the t-th function f_tT ═ 1,2,3}, η₁、η₂、η₃Representing weight coefficients corresponding to the three objective functions, wherein M is the total number of equipment nodes participating in communication in the network topology in constraint condition 1, and N is the total number of exchange nodes in the vehicle layer; constraint 2 indicates that each device node in the network topology is assigned to a sub-network of switching nodes; constraint 3S_kRepresenting the port number of the vehicle layer switching node k, wherein the condition restricts that the number of the node devices connected in the switching node subnet cannot exceed the port number of the switching node; in constraint 4 and constraint 5, q_cThe two conditions restrict that the transmission rate of the uplink and downlink traffic of any switching node subnet cannot exceed the maximum transmission rate of the switching node port in the subnet.

7. The IAGA algorithm-based high-speed train networking topology optimization method according to claim 5 or 6, characterized in that: planning the distribution mode of the vehicle-mounted terminal equipment by using an improved adaptive genetic algorithm, and optimizing the distribution mode of the terminal equipment in the network topology by referring to an objective function, wherein the method specifically comprises the following substeps:

s0501: an integer coding mode is adopted in a genetic algorithm, and a device allocation mode in a switching node subnet is described by a chromosome in the genetic algorithm;

s0502: the optimization target of the objective function model minimizes the transmission flow load and transmission delay in the network, namely, the equipment distribution mode with the best real-time performance is found, and the objective function is directly used as the fitness function of the algorithm; finally, the weight relation of each index is measured, and the weight coefficient eta is preliminarily determined₁、η₂、η₃；

S0503: calibrating and adjusting the fitness of each individual again, and determining a calibration formula of the individual fitness value according to the maximum fitness value and the minimum fitness value of the individuals in the current population:

wherein F' is the adjusted fitness value, F is the original fitness value, F_maxIs the maximum fitness value in the current population, F_minIs the minimum fitness value in the current population, and phi is an adjusting factor;

s0504: selecting an individual with a large fitness value before next operation and directly copying the individual into the next generation by adopting a roulette mode;

s0505: in the cross operation, the cross operator adopts a double-point cross method, the value of the self-adaptive cross probability is segmented, and the adjusting formula is as follows:

in the formula, z₁、z₂Is the adjustment coefficient of the adaptive cross probability, F_maxIs the maximum fitness value in the current population, F_avgThe average fitness value of all individuals in the current population is obtained;

s0506: the mutation method used in the mutation operation is bit-flipping mutation, the rationality of the mutated individuals is checked, if the constraint conditions cannot be met, the individuals need to be mutated again, and the adaptive mutation probability adjustment meets the following requirements:

in the formula, v₁、v₂Is the adjusting coefficient of the self-adaptive variation probability, and F represents the adaptability value of the variation individual;

s0507: and setting the number of the initialized population, the generation ditch and the iteration number, judging whether the iteration termination condition is met, and returning to the substep S0503 if the iteration termination condition is not met.

8. The IAGA-algorithm-based high-speed train networking topology optimization method according to claim 1, characterized in that: and carrying out optimization solution in a designed adaptive genetic algorithm by utilizing transmission data in the existing high-speed train car networking control system to obtain a final optimization result, wherein the optimization result comprises the position distribution condition of the optimized terminal equipment nodes in the switching node sub-network, the total communication traffic among the switching node sub-networks, the maximum difference of the communication loads among the switching node sub-networks and the end-to-end communication delay of data transmission.

Technical Field

The invention belongs to the technical field of train communication networks, and particularly relates to a high-speed train networking topology optimization method based on an IAGA algorithm.

Background

In recent years, the track traffic operation mileage of China is continuously increased, the control and monitoring technology of high-speed trains is improved, and the passenger service requirement information is improved, so that the types and the number of network terminal devices in a train network control system are continuously increased, and the high-speed train networking needs to bear more information exchange, so that the requirement of higher data transmission rate is met. The transmission rate of the bus type network on the high-speed train cannot meet the requirement of real-time transmission of a large amount of data, and the industrial Ethernet is a main solution for development of the high-speed train car networking in the future due to higher transmission rate and good compatibility. The real-time performance is one of the most remarkable characteristics of the high-speed train car networking, and the upper limit of transmission delay must be strictly controlled when information transmission such as car control, braking, fault diagnosis and monitoring is involved, so that the driving safety can be ensured. From the analysis of the communication transmission delay of the industrial ethernet, the topological structure optimization is an important way to ensure the real-time performance of the network, so that the method has important significance for the research on the real-time performance of the industrial ethernet applied to the high-speed train.

In the research of the high-speed train networking topology optimization method based on the industrial Ethernet, some researches start from the optimization of a physical topology structure of a network under the economic constraint condition, most researches only consider a single network flow index or a communication delay index, and do not fully consider the relationship of a plurality of optimization targets; the network topology optimization effects obtained by research under different optimization objectives are also different. When the optimization model of the high-speed train car networking is solved, various heuristic search algorithms can be adopted, the convergence rate and the solving precision of each search algorithm are different, and the genetic algorithm can be well used for solving the problem as a random and uncertain search algorithm. However, the traditional genetic algorithm is often slow in convergence speed and premature in engineering application, and research on the improved genetic algorithm with better performance also becomes a research focus of network topology optimization, so that the proposed high-speed train-vehicle networking topology optimization method based on the IAGA algorithm has important practical significance for more efficiently improving the real-time performance of the network.

Disclosure of Invention

The invention aims to provide a high-speed train car networking topology optimization method based on an IAGA algorithm, which aims at the problem of poor real-time performance of high-speed train car networking based on industrial Ethernet and reduces the communication load of each exchange node subnet in the topology by optimizing the distribution mode of vehicle-mounted terminal equipment of a physical topological structure of the high-speed train car networking, thereby reducing the end-to-end time delay among the vehicle-mounted terminal equipment in the high-speed train car networking and achieving the aim of improving the real-time performance of the high-speed train car networking. In order to achieve the purpose, the technical scheme adopted by the invention is as follows:

according to one aspect of the invention, the invention provides a high-speed train networking topology optimization method based on an IAGA algorithm, which is characterized in that: optimizing and converting the network topology structure into a distribution planning process of network terminal equipment, and performing optimized distribution processing on the communication data of the high-speed train networking to obtain a distribution result of network topology nodes; the optimization process comprises the following steps:

step S01: optimizing and converting a network topological structure into a distribution planning process of vehicle-mounted terminal equipment, defining parameters of a vehicle-mounted network, wherein the parameter definitions comprise the definitions of communication flow, communication time delay, exchange node performance parameters, terminal nodes and link parameters, and determining a basic physical topological structure of the train car networking according to the communication reliability of the vehicle-mounted network;

step S02: preliminarily constructing a two-layer topological structure model of the train-to-train network under the condition of communication reliability constraint of a vehicle-mounted network, calculating communication flow of process data and message data communicated between nodes of vehicle-mounted terminal equipment in the two-layer topological structure model, and constructing a communication weight matrix and an adjacent mapping matrix of the nodes of the terminal equipment according to the calculated communication flow value;

step S03: establishing an objective function model taking network traffic load and communication delay as optimization indexes on the basis of data communication traffic calculation, and establishing a function taking the minimum communication traffic load, the minimum load balance and the minimum transmission delay among network terminal devices in all vehicle layer subnets in the internet of vehicles network topology structure as optimization objectives according to a traffic weight matrix and an adjacent mapping matrix;

step S04: establishing an evaluation function model of multi-objective optimization, and determining constraint conditions of a target function according to the limitation of physical conditions of full-duplex switching nodes;

step S05: planning the distribution mode of the vehicle-mounted terminal equipment by using the improved adaptive genetic algorithm for the objective function model optimized in the step S03, thereby determining a coding scheme, a fitness function, adjusting a genetic operator and designing the whole algorithm flow;

step S06: and (4) carrying out optimization solution in an improved adaptive genetic algorithm by using the traffic of the high-speed train car networking obtained in the step (S02) to obtain the distribution result of the terminal equipment nodes in the optimized network topology.

Further preferably, in the parameter definition of step S01, the parameter definition of the communication traffic includes data length, data period and data priority weight, the definition of the switching node performance parameter includes the number of switching nodes, the number of switching node ports and port transmission rate, and the definition of the end node and link parameter includes the number of end nodes, the number of links and link communication bandwidth in the car networking network topology.

Preferably, for step S02, two real-time data, namely process data and message data, are fully considered, and two vehicle-mounted terminal devices in the car networking topology are communicatedCalculating the communication flow of the two types of data, constructing a corresponding traffic weight matrix according to the calculated traffic flow value, and obtaining the communication flow q_ijThe calculation of (a) satisfies:

where i and j represent two terminal equipment nodes communicating in the network, beta_ijFlow mean, τ, representing periodic real-time process data in a network_ijRefers to communication traffic of aperiodic message data in the network,andweights respectively representing priorities of process data and message data can construct a corresponding traffic weight matrix A according to the calculated traffic flow value, and the traffic weight matrix A meets the requirements;

in which the matrix element q_ijThe directional communication traffic weight value representing data between a source device i and a target device j, i, j is {1,2, …, m }, m is the total number of nodes of communication devices in the network, the size of a matrix element is defined as the ratio of the actual traffic of two device nodes to the minimum traffic of all communication devices in the network, and if the traffic of round-trip communication between the two device nodes is equal and no communication exists between the device nodes, a diagonal element in the matrix is 0.

Preferably, the method further includes constructing a corresponding adjacency mapping matrix after constructing the corresponding traffic weight matrix a, and the constructing process includes: first, element X of the adjacency matrix is defined according to the relationship between the device node and the sub-network of switching nodes_ikRespectively as follows:

in the formula, i represents a device node for communication, k represents a subnet number of a switching node at a vehicle layer, k is {1,2,3, …, N }, and N represents the total number of switching nodes at the vehicle layer, and the formula indicates that when a terminal device node i is in the switching node subnet k, a corresponding matrix element is 1, and otherwise, the corresponding matrix element is 0;

secondly, constructing a corresponding equipment node adjacency mapping matrix B:

the conditional formula used to determine communication between device nodes of two different subnets can be expressed as:

H_ij＝X_ik(1-X_jk)

in the formula H_ijIndicating the communication status of a pair of device nodes in different switch node subnets, when the source device i and the target device j are in the same switch node subnet, H_ij0, when the source device i and the target device j are not in the same subnet, H_ij＝1。

The scheme is further preferable, the constraint condition of the objective function of the multi-objective optimization is determined according to the limitation of the physical condition of the full-duplex switching node, and the objective function of the multi-objective optimization is established by using the communication traffic weight matrix and the adjacent mapping matrix;

step 501: the first objective is to minimize the communication load among all the vehicle layer switching node subnets in the network, and the objective function is expressed as:

in the formula, a device node i, j of communication is {1,2, …, M }, where M is a total number of device nodes participating in communication in the network; k is a subnet number of the switching nodes in the vehicle layer, k is {1,2,3, …, N }, and N represents the total number of the switching nodes in the vehicle layer; (ii) a Step 502: the second objective is to minimize the difference in load traffic transmitted between the switching node subnets, whose objective function is expressed as:

wherein k represents the subnet serial number of the switching node on the vehicle layer, and the total communication load flow ω (k) of any switching node subnet k is defined as:

in the formula, the device node i, j of communication is {1,2, …, M }, M is the number of communication device nodes in the topology network, and n is_kThe number of the communication equipment nodes in the switching node subnet k is represented, and the expression represents the sum of the communication flow of all the equipment nodes in a certain switching node subnet and all the communication nodes except the subnet in the network;

step 503: a third objective is to minimize the data transmission delay in the network, whose objective function is expressed as:

in the formula, D_ijThe number of switching nodes through which data streams communicated by the devices i and j pass is M, the number of total nodes of the communication terminal device is M, and D is the total number of switching nodes in the network topology;

step 504: for objective function f₁、f₂、f₃And planning the distribution mode of the vehicle-mounted terminal equipment by using an improved adaptive genetic algorithm.

The above solution is further preferred, for the objective function f₁、f₂、f₃The distribution mode of the vehicle-mounted terminal equipment is planned by using an improved self-adaptive genetic algorithm, and a multi-target problem is converted into a single target plan by using a linear weighting methodBefore determining a total objective function model, carrying out consistency processing on dimensions of the three objective functions, then comprehensively considering the influence of differences in relative importance degree and magnitude of each objective function, and determining a weight coefficient, wherein the objective functions are limited by the constraint of physical conditions of a full-duplex switching node, and the total objective evaluation function and the constraint condition are as follows:

wherein, in the objective function F,andrespectively representing the t-th function f_tT ═ 1,2,3}, η₁、η₂、η₃Representing weight coefficients corresponding to the three objective functions, wherein M is the total number of equipment nodes participating in communication in the network topology in constraint condition 1, and N is the total number of exchange nodes in the vehicle layer; constraint 2 indicates that each device node in the network topology is assigned to a sub-network of switching nodes; constraint 3S_kRepresenting the port number of the vehicle layer switching node k, wherein the condition restricts that the number of the node devices connected in the switching node subnet cannot exceed the port number of the switching node; in constraint 4 and constraint 5, q_cThe two conditions restrict that the transmission rate of the uplink and downlink traffic of any switching node subnet cannot exceed the maximum transmission rate of the switching node port in the subnet.

Preferably, the above-mentioned scheme further plans the allocation mode of the vehicle-mounted terminal device by using an improved adaptive genetic algorithm, and optimizes the allocation mode of the terminal device in the network topology by referring to an objective function, and specifically includes the following substeps:

s0501: an integer coding mode is adopted in a genetic algorithm, and a device allocation mode in a switching node subnet is described by a chromosome in the genetic algorithm;

s0502: the optimization target of the objective function model minimizes the transmission flow load and transmission delay in the network, namely, the equipment distribution mode with the best real-time performance is found, and the objective function is directly used as the fitness function of the algorithm; finally, the weight relation of each index is measured, and the weight coefficient eta is preliminarily determined₁、η₂、η₃；

S0503: calibrating and adjusting the fitness of each individual again, and determining a calibration formula of the individual fitness value according to the maximum fitness value and the minimum fitness value of the individuals in the current population:

wherein F' is the adjusted fitness value, F is the original fitness value, F_maxIs the maximum fitness value in the current population, F_minIs the minimum fitness value in the current population, and phi is an adjusting factor;

s0504: selecting an individual with a large fitness value before next operation and directly copying the individual into the next generation by adopting a roulette mode;

s0505: in the cross operation, the cross operator adopts a double-point cross method, the value of the self-adaptive cross probability is segmented, and the adjusting formula is as follows:

in the formula, z₁、z₂Is the adjustment coefficient of the adaptive cross probability, F_maxIs the maximum fitness value in the current population, F_avgThe average fitness value of all individuals in the current population is obtained;

s0506: the mutation method used in the mutation operation is bit-flipping mutation, the rationality of the mutated individuals is checked, if the constraint conditions cannot be met, the individuals need to be mutated again, and the adaptive mutation probability adjustment meets the following requirements:

in the formula, v₁、v₂Is the adjusting coefficient of the self-adaptive variation probability, and F represents the adaptability value of the variation individual;

s0507: and setting the number of the initialized population, the generation ditch and the iteration number, judging whether the iteration termination condition is met, and returning to the substep S0503 if the iteration termination condition is not met.

The scheme is further preferable, transmission data in the existing high-speed train car networking control system is utilized, optimization solution is carried out in a designed adaptive genetic algorithm, and a final optimization result can be obtained, wherein the optimization result comprises the position distribution condition of the optimized terminal equipment nodes in the switching node sub-networks, the total communication traffic among the switching node sub-networks, the maximum difference of the communication loads among the switching node sub-networks and the end-to-end communication delay of data transmission.

In summary, because the invention adopts the above technical scheme, the invention has the following remarkable effects:

(1) the high-speed train car networking topology optimization scheme fully considers a plurality of targets including network load and delay, and adopts a search algorithm based on an improved adaptive genetic algorithm to solve the optimization target according to the actual physical constraint condition of the Ethernet switching node, so that the convergence speed is higher and the obtained distribution result is more accurate in the designed optimization algorithm compared with the common genetic search; finally, the designed topology optimization method can reduce the total communication traffic among the switching node subnets, the maximum difference of the communication loads among the subnets and the communication delay of data transmission through the simulation verification of a mathematical example;

(2) aiming at the problem of poor real-time performance of the high-speed train car networking based on the industrial Ethernet, the communication load of each exchange node subnet in the topology is reduced by optimizing the distribution mode of the vehicle-mounted terminal equipment of the physical topological structure of the high-speed train car networking; the optimization method is applied to the high-speed train car networking technology, and can more efficiently reduce the end-to-end time delay among the vehicle-mounted terminal devices in the high-speed train car networking, thereby reducing the communication delay in the network and achieving the purpose of effectively improving the real-time performance of the high-speed train car networking.

Drawings

FIG. 1 is an optimization flow chart of a high-speed train networking topology optimization method based on an IAGA algorithm of the invention;

FIG. 2 is a topological structure diagram of a high-speed train networking topological optimization method based on an IAGA algorithm of the present invention;

FIG. 3 is a diagram of traffic weight matrices computed in an embodiment of the present invention;

FIG. 4 is a flow chart of the improved adaptive genetic algorithm solution of the present invention;

FIG. 5 is an evolutionary iterative comparison graph of two solution algorithms of the optimization model in the embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail below with reference to the accompanying drawings by way of examples of preferred embodiments. It should be noted, however, that the numerous details set forth in the description are merely for the purpose of providing the reader with a thorough understanding of one or more aspects of the present invention, which may be practiced without these specific details.

With reference to fig. 1, according to the high-speed train-vehicle networking topology optimization method based on the IAGA algorithm, the optimization of a network topology structure is converted into the distribution planning problem of network terminal devices, a two-layer topology structure model of the train-vehicle networking is preliminarily constructed under the constraint condition of network communication reliability, the communication flow of communication between the vehicle-mounted terminal devices in the two-layer topology structure model is calculated, an objective function model with network flow load and communication delay as indexes is established on the basis of the calculation of data communication flow, and the objective function model is optimized and solved by adopting an improved adaptive genetic algorithm; adjusting the cross probability and the variation probability of the adaptive genetic algorithm according to the fitness value of the evolved individual in the adaptive genetic algorithm; finally, optimizing the position of the equipment in the exchange node network by using the communication data of the high-speed train networking to obtain the position distribution result of the network topology equipment nodes in the exchange node network; and finally, solving and analyzing the traffic load index and the time delay index in the network after the position optimization, and comparing and verifying the significance of the improvement on the real-time performance of the network before the optimization. The specific optimization process comprises the following steps:

step S01: the optimization of the network topological structure is converted into a distribution planning process of the vehicle-mounted terminal equipment, various parameters of a vehicle-mounted network of the vehicle-mounted terminal equipment are defined, the parameter definitions comprise the definitions of communication flow, communication time delay, exchange node performance parameters, terminal nodes and link parameters, and a physical topological structure with high reliability is selected according to the network communication reliability. The physical topology comprises a line type, a ring type, a tree type, a star type and the like, the network communication reliability of the structures is compared, finally, the line type topology is selected on a train layer network, and the ring type topology is selected on a vehicle layer network; the parameter definition of the communication flow comprises data length, data period and data priority weight, the definition of the performance parameters of the switching nodes comprises the number of the switching nodes, the number of ports of the switching nodes and port transmission rate, the definition of the terminal nodes and the definition of the link parameters comprise the number of the terminal nodes, the number of links and link communication bandwidth in a car networking network topology structure, and the bandwidth of the Ethernet in the links is 100 Mbps.

Step S02: preliminarily constructing a two-layer topological structure model of the train-to-train network under the condition of the communication reliability of the train-to-train network, calculating the communication flow of two types of data, namely process data and message data of communication between the train-to-train terminal equipment in the two-layer train-to-train network topological structure model, and constructing a communication traffic weight matrix and an adjacent mapping matrix of terminal equipment nodes according to the communication flow value; wherein the content of the first and second substances,

traffic q_ijThe calculation of (a) satisfies:

where i and j represent two terminal equipment nodes communicating in the network, beta_ijFlow mean, τ, representing periodic real-time process data in a network_ijRefers to communication traffic of aperiodic message data in the network,andweights respectively representing priorities of process data and message data can construct a corresponding traffic weight matrix A according to the calculated traffic flow value, and the traffic weight matrix A meets the requirements;

in which the matrix element q_ijThe method comprises the steps that a directional communication flow weight value of data between a source device i and a target device j is represented, i, j is {1,2, …, m }, m is the total number of nodes of communication devices in a network, the size of a matrix element is defined as the ratio of actual communication traffic of two device nodes to the minimum communication traffic of all communication devices in the network, and if the flow of round-trip communication between the two device nodes is equal and no communication exists between the device nodes, a diagonal element in the matrix is 0;

after constructing the corresponding traffic weight matrix A, constructing a corresponding adjacency mapping matrix, wherein the construction process comprises the following steps: first, defining element X of adjacent matrix according to the relation between equipment node and exchange node word network_ikComprises the following steps:

in the formula, i represents a device node for communication, k represents a subnet number of a switching node at a vehicle layer, k is {1,2,3, …, N }, and N represents the total number of switching nodes in the vehicle layer (the total number of switching nodes in the vehicle layer of the topology network), and the formula represents that when a terminal device node i is in the switching node subnet k, the corresponding matrix element is 1, otherwise, the corresponding matrix element is 0;

secondly, assuming that there are N switching nodes and i terminal device nodes, a device node adjacency mapping matrix B is constructed for the N vehicle layer switching nodes:

the formula of the condition for determining communication between device nodes of two different subnets (referring to the condition for establishing communication connection between devices of different subnets) can be expressed as:

H_ij＝X_ik(1-X_jk)；

in the formula, H_ijIndicating the communication status of a pair of device nodes in different switch node subnets, when the source device i and the target device j are in the same switch node subnet, H_ij0, when not in the same subnet, H_ij＝1；

Step S03: establishing an objective function model of a network traffic load index and a communication delay index on the basis of data communication traffic calculation, and establishing a multi-objective function taking the minimum communication traffic load, the minimum load balance and the minimum transmission delay among network terminal devices as optimization objectives in all vehicle layer subnets in the vehicle networking topology structure according to a traffic weight matrix and an adjacent mapping matrix; the first objective is to minimize the communication load among all the vehicle layer switching node subnets in the network, and the objective function is expressed as:

in the formula, the device node i, j of communication is {1,2, …, M }, M is the total number of device nodes participating in communication in the network, k is the subnet number of the vehicle layer switching node, k is {1,2,3, …, N }, and N represents the total intersection in the vehicle layerNumber of change nodes, only if H_ijWhen 1, the objective function f can be matched₁Can be calculated.

The second objective is to minimize the difference in load traffic transmitted between the switching node subnets, whose objective function is expressed as:

wherein k represents the subnet serial number of the switching node on the vehicle layer, and the total communication load flow ω (k) of any switching node subnet k is defined as:

in the formula, n_kThe expression represents the sum of communication traffic of all device nodes in a certain switching node subnet and all communication nodes except the subnet in the network.

A third objective is to minimize the data transmission delay in the network, whose objective function is expressed as:

in the formula, the objective function f₃Is a qualitative representation of the communication time delay between nodes in the network, and establishes an objective function for the communication time delay of all nodes in the network as a parameter, D_ijThe number of switching nodes through which data streams communicated by the devices i and j pass is M, the number of total nodes of the communication terminal device is M, and D is the total number of switching nodes in the network topology;

step S04: establishing an evaluation function model of multi-objective optimization, and determining constraint conditions of a target function according to the limitation of physical conditions of full-duplex switching nodes; for the objective function f₁、f₂、f₃Adaptive genetic programming using improved techniquesThe method comprises the steps that a distribution mode of vehicle-mounted terminal equipment is planned by a traditional algorithm, a multi-objective problem is converted into a single objective planning problem by a linear weighting method, before a total objective function model is determined, dimension of the three objective functions is subjected to consistency processing, then the influence of difference in relative importance degree and magnitude of each objective function is comprehensively considered, and a weight coefficient is determined; the target function is restricted by the physical condition of the full-duplex switching node, and the target evaluation function and the restriction condition are as follows:

wherein, in the objective function F,andrespectively representing the t-th function f_tT ═ 1,2,3}, η₁、η₂、η₃Representing weight coefficients corresponding to the three objective functions, wherein M is the number of equipment nodes participating in communication in the network topology in constraint condition 1, and N is the total number of exchange nodes in the vehicle layer; constraint 2 indicates that each device node in the network topology is assigned to a sub-network of switching nodes; constraint 3S_kRepresenting the port number of the switching node k, the condition being that the number of node devices connected in the switching node subnet cannot exceed the port number of the switching node; in constraint 4 and constraint 5, q_cThe two conditions restrict that the transmission rate of the uplink and downlink traffic of any switching node subnet cannot exceed the maximum transmission rate of the switching node port in the subnet.

Step S05: planning the distribution mode of the vehicle-mounted terminal equipment by using an improved adaptive genetic algorithm according to the objective function model optimized in the step S03, determining a coding scheme, a fitness function, adjusting a genetic operator and designing the whole algorithm flow;

step S06: and (4) carrying out optimization solution in an improved adaptive genetic algorithm by using the traffic of the high-speed train car networking obtained in the step (S02) to obtain the distribution result of the terminal equipment nodes in the optimized network topology.

In the invention, constraint conditions of a multi-objective optimized objective function are determined according to the limitation of physical conditions of full-duplex switching nodes, the multi-objective optimized objective function is established by a communication flow weight matrix and an adjacent mapping matrix, and the distribution mode of vehicle-mounted terminal equipment is planned and solved by utilizing an improved adaptive genetic algorithm according to an optimized objective function model; adjusting the crossing probability and the mutation probability of the adaptive genetic algorithm according to the fitness value of the evolved individual in the adaptive genetic algorithm, adjusting crossing and mutation operators, optimizing the terminal equipment distribution mode in the network topology by referring to a target function, and finally obtaining the position division mode of the optimized equipment nodes in the exchange node network, wherein the method specifically comprises the following substeps:

s0501: an integer coding mode is adopted in a genetic algorithm, and a device allocation mode in a switching node subnet is described by a chromosome in the genetic algorithm; chromosome string [112332231] may indicate that device 1,2, 9 is assigned to switch node subnet 1, device 3, 6, 7 is assigned to switch node subnet 2, and device 4, 5, 8 is assigned to switch node subnet 3;

s0502: the optimization goal of the objective function model (multi-objective function optimization model) is to minimize the transmission flow load and transmission delay in the network, i.e. find the equipment distribution mode with the best real-time performance, directly use the objective function as the fitness function of the algorithm, finally measure the weight relation of each index, and preliminarily determine the weight coefficient eta₁、η₂、η₃0.3, 0.3 and 0.4 respectively;

s0503: calibrating and adjusting the fitness of each individual again, and determining a calibration formula of the individual fitness value according to the maximum fitness value and the minimum fitness value of the individuals in the current population:

wherein F' is the adjusted fitness value, F is the original fitness value, F_maxIs the maximum fitness value in the current population, F_minTaking phi as a regulating factor and taking 0.65 as the minimum fitness value in the current population;

s0504: selecting an individual with a large fitness value before next operation and directly copying the individual into the next generation by adopting a roulette mode;

s0505: in the crossing operation, the crossing operator adopts a double-point crossing method to adjust the value of the self-adaptive crossing probability, and the adjusting formula is as follows:

in the formula, z₁、z₂Is the adjusting coefficient of self-adaptive cross probability, which is respectively taken as 0.5 and 0.3, F_maxIs the maximum fitness value in the current population, F_avgThe average fitness value of all individuals in the current population is obtained;

s0506: the mutation method used in the mutation operation is bit-flipping mutation, the rationality of the mutated individuals is checked, if the constraint conditions cannot be met, the individuals need to be mutated again, and the adaptive mutation probability adjustment meets the following requirements:

in the formula, v₁、v₂Is the adjusting coefficient of the self-adaptive variation probability, and is respectively 0.05 and 0.049. F represents the fitness value of the variant individual;

s0507: setting the number of the initialized population, the number of the generation ditches and the number of the iteration, setting the number of the initialized population of the algorithm to be 40, the number of the generation ditches to be 0.9 and the number of the iteration to be 100, judging whether the iteration termination condition is met, and if the iteration termination condition is not met, returning to the substep S0503;

and carrying out optimization solution in a designed adaptive genetic algorithm by utilizing the transmission data information in the existing high-speed train car networking control system to obtain a final optimization result, wherein the optimization result comprises the position distribution condition of the optimized terminal equipment nodes in the switching node sub-network, the total communication traffic among the switching node sub-networks, the maximum difference of the communication loads among the switching node sub-networks and the end-to-end communication delay of data transmission. In the embodiment of the invention, which is verified by the high-speed train-to-train networking topology optimization method based on the IAGA algorithm, a network topology structure comprising three parts of a switching node, a vehicle-mounted equipment node and a link is initially established as shown in figure 2, in the topology, a train backbone layer network adopts a linear topology structure, a vehicle layer marshalling network adopts a ring topology structure, the whole topology structure domain can be divided into a train layer, a vehicle layer and a device layer, in fig. 2, ETBN, ECNN and ED respectively represent the switching node and the terminal equipment layer node of the train backbone layer and the vehicle grouping layer, due to the limitations of the switch node ports, the switch nodes are divided into two layers, and the vehicle layer switch nodes are divided into a considerable number of sub-networks, the nodes of the equipment layer are partitioned in the network, and for convenience of description, 4 train backbones are randomly selected by the topological model, wherein the 4 train backbones comprise 12 vehicle layer switching nodes and 20 equipment nodes. The objective optimization function may be established as per steps S01 through S06 in the function model.

In the invention, a case of optimizing the distribution of 25 vehicle-mounted terminal devices in a network to 5 vehicle layer switching node subnetworks is made by referring to an actual communication scene of a high-speed train car networking, and considering that the transmission types of data in different systems of a train are different, the specification of two types of typical data service parameter values of train Ethernet communication process data and message data in a standard protocol IEC61375-3-4 of train Ethernet communication is referred, and the two types of data parameter information of the process data and the message data extracted in the embodiment is shown in a table 1.

TABLE 1

According to the communication flow calculation step in the invention, two types of data can be integrated to calculate the communication flow between the devices in the network topology, and the communication flow calculation formula is as follows:

wherein the process data and message data priorities are weightedAndtake 0.7 and 0.3, respectively. The traffic weight matrix among 25 terminal equipments can be obtained by calculation as shown in fig. 3, the left and upper digits in fig. 3 represent the serial numbers of 25 terminal equipment nodes, and the middle digit represents the traffic weight. Assuming equal traffic to and from the two nodes and no communication between the node devices themselves, the diagonal element in the matrix is 0. Fig. 3 shows traffic weight of unidirectional communication, the size of the weight is determined according to the ratio of actual traffic to minimum traffic among all communication devices in the network, and the value range is divided into 1-10. Meanwhile, an adjacency matrix between node devices is obtained, so that programming of the objective function is facilitated.

In the invention, the established target optimization model is solved according to the adaptive genetic algorithm, the algorithm flow for solving the target function optimization model is shown in fig. 4, and the steps of the whole algorithm are as follows:

firstly, randomly setting a position distribution mode of terminal equipment in a switching node subnet, adopting integer coding, and expressing chromosome codes corresponding to 25 terminal equipment nodes in the distribution mode as {2353134325424254411251513 };

secondly, setting the objective function as the fitness function of the algorithm, and initially settingThe cross probability is 0.7, the initial variation probability is 0.04, the adjusting factor phi calibrated by the fitness value is 0.65, and the adjusting coefficient z of the self-adaptive cross probability₁、z₂Respectively taking 0.5 and 0.3 as the adjusting coefficient v of the self-adaptive variation probability₁、v₂Take 0.05 and 0.049, respectively. The improved adaptive genetic algorithm is programmed and solved by using MATLAB software, and the obtained distribution mode of the network topology node equipment and the communication load on each switching node subnet are shown in table 2. The numbers in the assignment column in table 2 indicate the serial numbers of the terminal device nodes participating in the assignment.

TABLE 2

In order to verify the superiority of the designed algorithm, a standard genetic algorithm is added for comparison. The number of the initialized population of the two algorithms is set to 40, the generation ditch is set to 0.9, and the number of iteration is 100. The evolution iteration graphs of the two algorithms obtained by solving are shown in fig. 5, and compared with the standard genetic algorithm, the improved adaptive genetic algorithm has the advantages that the fitness value during convergence is reduced by 13.5%, the iteration algebra for achieving convergence is reduced by 10 generations, and the calculation time is reduced by 10.2 s.

Table 3 shows three index values and the total objective function value before and after the topology optimization using the two algorithms. It can be seen from the table that the total traffic between the subnets, the maximum difference amount of the communication loads between the subnets, and the communication delay amount of data transmission, which are solved by the improved adaptive genetic algorithm, are reduced by 26.5%, 87.27%, and 16.92% respectively compared with those before optimization.

TABLE 3

Table 3 fully shows that the real-time performance of the train networking can be significantly improved by using the location allocation result of the device node obtained by the high-speed train networking topology optimization method of the present invention. The invention provides an embodiment result based on an IAGA algorithm, verifies that the convergence speed of the solving algorithm is high, the global optimal solution can be obtained, and each state parameter is well improved compared with the standard genetic algorithm.

The foregoing is only a preferred embodiment of the present invention, and it should be noted that those skilled in the art can make various improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be construed as the protection scope of the present invention.