阅读说明：本技术 一种基于无人机遥感的航标巡检系统及巡检方法 (Navigation mark inspection system and inspection method based on unmanned aerial vehicle remote sensing ) 是由 季克淮 霍虎伟 毛建峰 李铁 李金鹏 李栋 于 2021-11-05 设计创作，主要内容包括：本申请提供了一种基于无人机遥感的航标巡检系统及其巡检方法,包括控制系统、巡检主机和待巡检的航标；所述控制系统包括航标历史信息库；所述控制系统还包括路线分配模块、航标失常判断模块、失常原因采集模块和路线调整模块；所述路线分配模块用于分配所述巡检主机的巡检路线；所述航标失常判断模块用于通过巡检主机获得的航标信息判断航标是否失常；所述失常原因采集模块用于通过巡检主机对失常航标的失常信息进行采集；路线调整模块用于当巡检主机采集的失常航标的失常信息不能判断其失常原因时,调整巡检路线,使用其他巡检主机对失常航标的失常信息进行采集。本申请提高了巡检效率。(The application provides a navigation mark inspection system based on unmanned aerial vehicle remote sensing and an inspection method thereof, wherein the navigation mark inspection system comprises a control system, an inspection host and a navigation mark to be inspected; the control system comprises a navigation mark historical information base; the control system also comprises a route distribution module, a navigation mark abnormality judgment module, an abnormality reason acquisition module and a route adjustment module; the route distribution module is used for distributing routing inspection routes of the routing inspection host; the navigation mark abnormality judging module is used for judging whether the navigation mark is abnormal or not according to the navigation mark information obtained by the inspection host; the abnormality reason acquisition module is used for acquiring abnormality information of the abnormality navigation mark through the inspection host; the route adjusting module is used for adjusting the routing inspection route when the abnormality information of the abnormal navigation mark acquired by the routing inspection host can not judge the reason of the abnormality, and other routing inspection hosts are used for acquiring the abnormality information of the abnormal navigation mark. This application has improved and has patrolled and examined efficiency.)

1. A navigation mark inspection system based on unmanned aerial vehicle remote sensing is characterized by comprising a control system, an inspection host and a navigation mark to be inspected;

the control system comprises a navigation mark historical information base, wherein the navigation mark historical information base comprises basic navigation mark information and historical patrol inspection information, and the historical patrol inspection information comprises patrol inspection characteristics of navigation marks obtained in historical patrol inspection and cruise weight of the navigation marks;

the control system also comprises a route distribution module, a navigation mark abnormality judgment module, an abnormality reason acquisition module and a route adjustment module; the route distribution module is used for distributing routing inspection routes of the routing inspection host; the navigation mark abnormality judging module is used for judging whether the navigation mark is abnormal or not through navigation mark information obtained by the inspection host, and when the navigation mark is judged to be abnormal, the navigation mark is set as an abnormal navigation mark; the abnormality reason acquisition module is used for acquiring abnormality information of the abnormality navigation mark through the inspection host; the route adjusting module is used for adjusting the routing inspection route when the abnormality information of the abnormal navigation mark acquired by the routing inspection host can not judge the reason of the abnormality, and acquiring the abnormality information of the abnormal navigation mark by using other routing inspection hosts;

the inspection host comprises an acquisition component for acquiring navigation mark information and abnormal information.

2. The unmanned aerial vehicle remote sensing-based beacon inspection system according to claim 1, wherein the collection component of the inspection host comprises a photoelectric pod and a laser range finder.

3. The unmanned aerial vehicle remote sensing-based beacon inspection system according to claim 1, wherein the beacon includes a beacon body, the beacon body being one of a lighthouse, a buoy, a post, or a lightboat.

4. The drone remote sensing-based beacon inspection system according to claim 3, wherein the beacon further includes a beacon kiosk, the beacon kiosk being disposed on the beacon body for collecting big data information in the aspects of weather, water flow, and ship traffic.

5. A navigation mark inspection method is applied to the navigation mark inspection system based on unmanned aerial vehicle remote sensing according to claims 1-4, and is characterized by comprising the following steps:

s10, according to n navigation marks R = [ R ] to be inspected₁,R₂,R₃,…,R_n]Obtaining patrol weight R theta = [ R theta ] corresponding to navigation mark R in navigation mark historical information base₁,Rθ₂,Rθ₃,…,Rθ_n]Simultaneously establishing a removal beacon list YR that includes a first removal list YR1 and a second removal list YR 2;

s20, setting a patrol task A, wherein the patrol task A comprises m patrol routes AH = [ AH ]₁,AH₂,AH₃,…,AH_m]Each routing inspection route is correspondingly inspected by one routing inspection host, and the routing inspection route AH is correspondingly inspected by the routing inspection host P = [ P ]₁,P₂,P₃,…,P_m]Carrying out routing inspection; wherein, set up the ith platform and patrol inspection host P_iAt inspection route AH_iNavigation mark IR = [ IR ] requiring inspection₁,IR₂,IR₃,…,IR_k]Wherein k is less than m; patrol route AH_iThe patrol weight IR theta = [ IR theta ] corresponding to the navigation mark IR needing patrol₁,IRθ_2,IRθ₃,…,IRθ_k](ii) a Wherein, a routing inspection route AH is set_iGo up j navigation mark IR that needs to patrol and examine_jHas a patrol weight of IR theta_jRouting inspection route AH_iTotal weight of theta_i=；

Setting the weight threshold value of the routing inspection route as theta₀Then patrol route AH_iNeed to satisfyWhile satisfying the inspection route AH_iNumber of navigation marks on(ii) a Wherein e is a natural constant, θ₀＜2e；

S30, the inspection host P_iAt inspection route AH_iUpper pair navigation mark IR = [ IR =₁,IR₂,IR₃,…,IR_k]In the process of executing the polling task, when the polling host P_iAt inspection route AH_iGo up jth fairway buoy IR who patrols and examines_jIf the navigation mark does not belong to the removed navigation mark list YR, the step is switched to step S31;

when patrolling and examining host P_iAt inspection route AH_iGo up jth fairway buoy IR who patrols and examines_jIf the navigation mark belongs to the removed navigation mark list YR, the step is switched to step S32;

s31, according to the IR to the navigation mark_jThe inspection information obtained during inspection is used for judging the navigation mark IR_jWhether the navigation mark is abnormal or not, and if not, continuing the routing inspection task; if yes, the process proceeds to step S32;

s32, patrol and examine host P_iFor abnormal navigation mark IR_jCarrying out a secondary collection task for judging the reason of the abnormality;

when according to the patrol main machine P_iThe inspection information obtained by the secondary acquisition task can judge the abnormal navigation mark IR_jWhen the abnormality is caused, the inspection host P_iContinuing the polling task;

when according to the patrol main machine P_iThe inspection information obtained by the secondary acquisition task cannot judge the abnormal navigation mark IR_jIf the cause of the malfunction is (3), the process proceeds to step S40;

s40, the abnormal navigation mark IR_jSlave patrol route AH_iTransfer to the first removal List YR₁；

Updating patrol inspection host P_iTo obtain a patrol main machine P_iNew routing inspection route AH_i', patrol and examine the host computer P_iAccording to the new routing inspectionRoute AH_i' continuing the inspection task;

s50, obtain a first removal list YR1= [ YR1₁,YR1₂,YR1₃,…,YR1_c]C is less than n; setting the distance threshold D simultaneously₀；

When a polling host P exists_jPatrol route AH_jSuch that the z-th disarranged navigation mark YR in the first removal list YR1_zTo patrol route AH_jDistance D of_z≤D₀If so, wherein j ≠ i, proceed to step S51;

when there is not a patrol inspection host P_jPatrol route AH_jSuch that the z-th disarranged navigation mark YR in the first removal list YR1_zTo patrol route AH_jDistance D of_z≤D₀If so, wherein j ≠ i, proceed to step S52;

s51, the abnormal navigation mark YR_zTransferring the first removal list into the routing inspection route AH_jIn which j ≠ i, updating patrol route AH_jObtain a new routing inspection route AH_j', patrol and examine the host computer P_jAccording to the updated patrol route AH_j' go to patrol, go to step S30;

s52, the abnormal navigation mark YR_zThe second removal list YR2= [ YR2 ] is obtained by shifting from the first removal list YR1 to the second removal list YR2₁,YR2₂,YR2₃,…,YR2_b]，b＜c；

S60, when all the inspection tasks of the inspection host are finished, the abnormal navigation mark in the second removal list YR2 is used as the navigation mark to be inspected, and the step S10 is carried out.

6. The beacon inspection method according to claim 5, wherein in step S50, it is set that the abnormal beacon IR is performed_jThe cruise host of the secondary collection task is abnormal navigation mark IR_jThe filtering host set without executing the abnormal navigation mark IR_jThe cruise host of the secondary collection task is abnormal navigation mark IR_jThe non-filtering host;

first remove list of pixilated fairway signs IR_jOnly abnormal navigation mark IR_jIs in the cruise route in which the non-filtered host is located.

7. The beacon inspection method according to claim 6, wherein the IR is applied to the malfunctioning beacon_jThe distance does not exceed a distance threshold D₀When there is more than one route for polling non-filtering host computer, it can be used for making navigation mark IR with abnormality_jThe shortest distance routing inspection route will be abnormal navigation mark IR_jAnd (6) turning in.

8. The beacon inspection method according to claim 5, wherein the inspection host AP is set_iHas a flying speed v_iInspection host AP_jHas a flying speed v_jThen there is a distance threshold D₀=(ii) a Wherein D is_maxThe total inspection distance is used for inspecting all n navigation marks to be inspected by using a single machine.

9. The beacon inspection method according to claim 5, wherein the inspection weight IR θ = [ IR θ = ] corresponding to the beacon IR₁,IRθ_2,IRθ₃,…,IRθ_k](ii) a Setting inspection route AH_iGo up j navigation mark IR that needs to patrol and examine_jPatrol weight IR θ = epsilon_jφ_jWherein phi is_jAs navigation marks IR_jDegree of historical risk of epsilon_jAs navigation marks IR_jThe position importance of.

10. The beacon inspection method according to claim 5, wherein a neural network model is established according to historical inspection characteristics of the beacon historical information base, and the inspection characteristics acquired by the inspection host and the judgment result of whether the beacon is abnormal are used as input and output data samples (x, y) to judge whether the beacon is an abnormal beacon;

setting and inputting D routing inspection characteristics x = [ x ]₁; x₂ ; …; x_D ]Corresponding to weight w = [ w = [ w ]₁; w₂;…; w _D]Setting a bias b epsilon R; then we can get the weighted sum z of the input features, the specific formula is:

using the ReLU function as the activation function, then

In a multi-layer feedforward neural network, letThen, the feedforward neural network continuously iterates to carry out the propagation formula layer by layer as follows:；

the composite function is:

where W and b represent the connection weights and offsets for all layers in the network,is the number of layers of the neural network,is as followsThe number of layer neurons;is as followsLayer to layerA weight matrix of the layer;is as followsLayer to layerBiasing of the layers;is as followsOutput of layer neurons;

using a cross-entropy loss function, for sample (x, y) the loss function is:

wherein the content of the first and second substances,representing by a one-hot vector corresponding to y;

given a training set ofEach sample is sampledInput to the pre-neural network to obtain the network output ofThe risk function on the data set is:

wherein the content of the first and second substances,is a regularization term; λ is a long parameter, the larger λ the closer W is to 0,

in each iteration of the gradient descent method, a learning rate alpha is set to obtain an updating mode of the parameters W and b,

calculating the gradient of the l-th layer weight and bias, δ^(l)Error term for layer i:

obtaining an iterative formula:

。

Technical Field

The application relates to the field of navigation mark inspection, in particular to a navigation mark inspection system and an inspection method based on unmanned aerial vehicle remote sensing.

Background

In recent years, on the background of the continuous development of shipping economy in China, the number of ports and the number of ships are continuously increased, and marine sudden accidents and traffic accidents also occur at times, so that the demand on the navigation mark is gradually increased, the workload of a navigation mark department is increased day by day, and the responsibility for ensuring safety is more important. Therefore, the maintenance and management of the navigation mark become key points, and in order to ensure the safety of the navigation path, the navigation mark needs to be inspected at regular time so as to grasp and process abnormal conditions such as damage, displacement and the like of the navigation mark in time. At present, technical means such as field inspection or ship inspection, near-shore monitoring and the like are mainly utilized for the navigation mark guarantee work, the advantages of a field inspection mode are obvious, but the defects are also obvious, such as the influence and the restriction of meteorological conditions, the visual navigation mark needs to be simulated by a mark-climbing and light-covering lamp, the consumed time is long, operators are easy to fatigue, and the danger coefficient is increased; the modes of ship inspection and the like have the problems of low response speed, high cost, large limitation by meteorological conditions, limited operation capability and range, occasional false alarm and missing report, incapability of displaying main body information such as the appearance of a navigation mark body and the like. The development of the remote sensing technology provides a remote sensing inspection mode for the navigation mark inspection, the remote sensing inspection mode makes up for many defects of field inspection, and the maintenance level of the navigation mark is obviously enhanced along with the continuous development of the remote sensing measurement and control technology. On this basis, the design idea that the unmanned aerial vehicle remote sensing is applied to the navigation mark field is provided, the navigation mark condition is shown through the mode of video images, on this basis, the unmanned aerial vehicle remote sensing mode is further provided, the unmanned remote sensing patrol can provide powerful management basis for navigation mark departments, the current condition of the navigation mark is mastered in real time, and targeted management and maintenance are carried out on the navigation mark according to actual conditions, so that the goal of improving the maintenance efficiency of the navigation mark is achieved. An Unmanned Aerial Vehicle (UAV) is an unmanned plane, which is an unmanned plane operated by a radio remote control device or a self-contained program control device, and generally includes an unmanned helicopter or a fixed-wing drone. Be applied to the navigation guarantee field with unmanned aerial vehicle, can give full play to its advantage such as with low costs, transportation convenience, easy operation, the quick high flexibility of reaction and can independently fly, compensate the not enough of present technical means, provide fine technical support for the navigation guarantee, improve navigation guarantee service level comprehensively.

In recent years, as ships develop towards large-scale and high-speed directions, higher requirements are provided for the advancement, accuracy and timeliness of fault recovery of a navigation mark, the mode of polling and flying by using a single unmanned aerial vehicle is not only time-consuming, but also faults easily occur in the polling process, the service life of the unmanned aerial vehicle is also influenced, and particularly, in areas with complex navigation channels or large navigation distance, if a single unmanned aerial vehicle is used for cruising, a plurality of performance efficiency problems can occur, therefore, under the condition, a plurality of unmanned aerial vehicles are often used for carrying out regional polling, how to effectively distribute the polling efficiency, how to save the electric power of the unmanned aerial vehicle, how to improve the service life of the unmanned aerial vehicle and the like needs are continuously provided by users, and meanwhile, routes for polling a plurality of unmanned aerial vehicles are planned and the like, The navigation mark system has the advantages that the abnormality condition of the navigation mark is judged, the reason of the abnormality is obtained according to the abnormality condition, when the obtained information is insufficient, problems such as proper adjustment of a route and the like also occur simultaneously, even on the navigation mark at some special positions, a device for obtaining the information of the navigation mark and the conditions of a water area around the navigation mark is further arranged to be used as an information station of the navigation mark, a large data system of the navigation mark can be further established by obtaining the information, and the safe operation of the large data of the navigation mark is also required to be ensured in the inspection process. Therefore, how to increase the working efficiency of the navigation mark maintenance, enhance the emergency response capability of the navigation mark, comprehensively improve the navigation mark maintenance and service quality, and reduce the inspection cost is a problem to be solved urgently in the current navigation mark management.

Disclosure of Invention

In order to solve the problems, the navigation mark inspection system comprises a control system, an inspection host and a navigation mark to be inspected;

the control system comprises a navigation mark historical information base, wherein the navigation mark historical information base comprises basic navigation mark information and historical patrol inspection information, and the historical patrol inspection information comprises patrol inspection characteristics of navigation marks obtained in historical patrol inspection and cruise weight of the navigation marks;

the control system also comprises a route distribution module, a navigation mark abnormality judgment module, an abnormality reason acquisition module and a route adjustment module; the route distribution module is used for distributing routing inspection routes of the routing inspection host; the navigation mark abnormality judging module is used for judging whether the navigation mark is abnormal or not through navigation mark information obtained by the inspection host, and when the navigation mark is judged to be abnormal, the navigation mark is set as an abnormal navigation mark; the abnormality reason acquisition module is used for acquiring abnormality information of the abnormality navigation mark through the inspection host; the route adjusting module is used for adjusting the routing inspection route when the abnormality information of the abnormal navigation mark acquired by the routing inspection host can not judge the reason of the abnormality, and acquiring the abnormality information of the abnormal navigation mark by using other routing inspection hosts;

the inspection host comprises an acquisition component for acquiring navigation mark information and abnormal information.

The collection component of the inspection main machine comprises a photoelectric pod and a laser range finder.

The navigation mark comprises a navigation mark main body, wherein the navigation mark main body is one of a lighthouse, a buoy, a stand column or a lightboat.

The navigation mark further comprises a navigation mark information station, and the navigation mark information station is arranged on the navigation mark main body and used for collecting big data information in the aspects of weather, water flow and ship traffic.

The application also provides a method for polling the navigation mark polling system based on the unmanned aerial vehicle remote sensing, which comprises the following steps:

s10, according to n navigation marks R = [ R ] to be inspected₁,R₂,R₃,…,R_n]Obtaining patrol weight R theta = [ R theta ] corresponding to navigation mark R in navigation mark historical information base₁,Rθ₂,Rθ₃,…,Rθ_n]Simultaneously establishing a removal beacon list YR that includes a first removal list YR1 and a second removal list YR 2;

s20, setting a patrol task A, wherein the patrol task A comprises m patrol routes AH = [ AH ]₁,AH₂,AH₃,…,AH_m]Each routing inspection route is correspondingly inspected by one routing inspection host, and the routing inspection route AH is correspondingly inspected by the routing inspection host P = [ P ]₁,P₂,P₃,…,P_m]Carrying out routing inspection; wherein, set up the ith platform and patrol inspection host P_iAt inspection route AH_iNavigation mark IR = [ IR ] requiring inspection₁,IR₂,IR₃,…,IR_k](k < m); patrol route AH_iThe patrol weight IR theta = [ IR theta ] corresponding to the navigation mark IR needing patrol₁,IRθ_2,IRθ₃,…,IRθ_k](ii) a Wherein, a routing inspection route AH is set_iGo up j navigation mark IR that needs to patrol and examine_jHas a patrol weight of IR theta_jRouting inspection route AH_iTotal weight of theta_i=；

Setting the weight threshold value of the routing inspection route as theta₀Then patrol route AH_iNeed to satisfyWhile satisfying the inspection route AH_iNumber of navigation marks on(ii) a Wherein e is a natural constant, θ₀＜2e；

S30, the inspection host P_iAt inspection route AH_iUpper pair navigation mark IR = [ IR =₁,IR₂,IR₃,…,IR_k]In the process of executing the polling task, when the polling host P_iAt inspection route AH_iGo up jth fairway buoy IR who patrols and examines_jIf the navigation mark does not belong to the removed navigation mark list YR, the process proceeds to step S31；

When patrolling and examining host P_iAt inspection route AH_iGo up jth fairway buoy IR who patrols and examines_jIf the navigation mark belongs to the removed navigation mark list YR, the step is switched to step S32;

s31, according to the IR to the navigation mark_jThe inspection information obtained during inspection is used for judging the navigation mark IR_jWhether the navigation mark is abnormal or not, and if not, continuing the routing inspection task; if yes, the process proceeds to step S32;

s32, patrol and examine host P_iFor abnormal navigation mark IR_jCarrying out a secondary collection task for judging the reason of the abnormality;

when according to the patrol main machine P_iThe inspection information obtained by the secondary acquisition task can judge the abnormal navigation mark IR_jWhen the abnormality is caused, the inspection host P_iContinuing the polling task;

when according to the patrol main machine P_iThe inspection information obtained by the secondary acquisition task cannot judge the abnormal navigation mark IR_jIf the cause of the malfunction is (3), the process proceeds to step S40;

s40, the abnormal navigation mark IR_jSlave patrol route AH_iTransfer to the first removal List YR₁；

Updating patrol inspection host P_iTo obtain a patrol main machine P_iNew routing inspection route AH_i', patrol and examine the host computer P_iAccording to the new routing inspection route AH_i' continuing the inspection task;

s50, obtain a first removal list YR1= [ YR1₁,YR1₂,YR1₃,…,YR1_c]C is less than n; setting the distance threshold D simultaneously₀；

When a polling host P exists_jPatrol route AH_jSuch that the z-th disarranged navigation mark YR in the first removal list YR1_zTo patrol route AH_j(j ≠ i) distance D_z≤D₀If yes, the process proceeds to step S51;

when there is not a patrol inspection host P_jPatrol route AH_j(j ≠ i) causes the z-th arrhythmic navigation mark YR in the first removal list YR1_zTo patrol route AH_jDistance D of_z≤D₀At the moment, turn over toStep S52;

s51, the abnormal navigation mark YR_zTransferring the first removal list into the routing inspection route AH_j(j ≠ i), the patrol route AH is updated_jObtain a new routing inspection route AH_j', patrol and examine the host computer P_jAccording to the updated patrol route AH_j' go to patrol, go to step S30;

s52, the abnormal navigation mark YR_zThe second removal list YR2= [ YR2 ] is obtained by shifting from the first removal list YR1 to the second removal list YR2₁,YR2₂,YR2₃,…,YR2_b]，b＜c；

S60, when all the inspection tasks of the inspection host are finished, the abnormal navigation mark in the second removal list YR2 is used as the navigation mark to be inspected, and the step S10 is carried out.

Wherein, in step S50, the execution of the abnormal navigation mark IR is set_jThe cruise host of the secondary collection task is abnormal navigation mark IR_jThe filtering host set without executing the abnormal navigation mark IR_jThe cruise host of the secondary collection task is abnormal navigation mark IR_jThe non-filtering host;

first remove list of pixilated fairway signs IR_jOnly abnormal navigation mark IR_jIs in the cruise route in which the non-filtered host is located.

Wherein, when and the abnormal navigation mark IR_jThe distance does not exceed a distance threshold D₀When the routing inspection route of the non-filtering host computer has more than one route, the routing inspection route is preferably selected from the IR of the abnormal navigation mark_jThe shortest distance routing inspection route will be abnormal navigation mark IR_jTransferring in;

wherein, the inspection host AP is arranged_iHas a flying speed v_iInspection host AP_jHas a flying speed v_jThen there is a distance threshold D₀=(ii) a Wherein D is_maxThe total inspection distance is used for inspecting all n navigation marks to be inspected by using a single machine.

Wherein, the patrol inspection weight IR theta = [ IR theta ] corresponding to the navigation mark IR₁,IRθ_2,IRθ₃,…,IRθ_k](ii) a Setting inspection route AH_iGo up j navigation mark IR that needs to patrol and examine_jPatrol weight IR θ = epsilon_jφ_jWherein phi is_jAs navigation marks IR_jDegree of historical risk of epsilon_jAs navigation marks IR_jThe position importance of.

Establishing a neural network model according to historical inspection characteristics of a navigation mark historical information base, taking inspection characteristics acquired by an inspection host and a judgment result of whether the navigation mark is abnormal as input and output data samples (x, y), and judging whether the navigation mark is an abnormal navigation mark;

setting and inputting D routing inspection characteristics x = [ x ]₁; x₂ ; …; x_D ]Corresponding to weight w = [ w = [ w ]₁; w₂;…; w _D]Setting a bias b epsilon R; then we can get the weighted sum z of the input features, the specific formula is:

using the ReLU function as the activation function, then

In a multi-layer feedforward neural network, letThen, the feedforward neural network continuously iterates to carry out the propagation formula layer by layer as follows:；

the composite function is:

where W and b represent the connection weights and offsets for all layers in the network,is the nerveThe number of layers of the network,is as followsThe number of layer neurons;is as followsLayer to layerA weight matrix of the layer;is as followsLayer to layerBiasing of the layers;is as followsOutput of layer neurons;

using a cross-entropy loss function, for sample (x, y) the loss function is:

wherein the content of the first and second substances,representing by a one-hot vector corresponding to y;

given a training set ofEach sample is sampledInput to the pre-neural network to obtain the network output ofThe risk function on the data set is:

wherein the content of the first and second substances,is a regularization term; λ is a long parameter, the larger λ the closer W is to 0,

in each iteration of the gradient descent method, a learning rate alpha is set to obtain an updating mode of the parameters W and b,

calculating the gradient of the l-th layer weight and bias, δ^(l)Error term for layer i:

obtaining an iterative formula:

。

the beneficial effect that this application realized is as follows:

the route of the inspection of the unmanned aerial vehicles is planned, various cruise conditions are considered, the possible abnormal conditions of the navigation mark are judged, and the reason of the abnormal conditions is directly obtained through the inspection host according to the abnormal conditions, so that information can be directly obtained in a long distance without repeated round trip; meanwhile, the route is properly adjusted when the acquired information is insufficient, other inspection hosts acquire the reasons of the abnormality at staggered time intervals, so that the acquired information can be more comprehensive, and meanwhile, the inspection hosts have cruise tasks, so that electric power and manpower are saved. Meanwhile, a navigation mark big data system is further established by acquiring meteorological information, ship information and the like of the water areas, and the safe operation of the navigation mark big data is ensured.

Drawings

In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art according to the drawings.

FIG. 1 is a flow chart of the navigation mark inspection method based on unmanned aerial vehicle remote sensing.

Detailed Description

The technical solutions in the embodiments of the present application are clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.

The application provides a system is patrolled and examined to fairway buoy based on unmanned aerial vehicle remote sensing, the system is patrolled and examined to the fairway buoy that the fairway buoy patrolled and examined includes waiting to patrol and examine, ground control system and patrol and examine the host computer, wherein, ground control system is used for distributing the route of patrolling and examining the host computer, and can judge whether the fairway buoy is unusual through the fairway buoy image information of patrolling and examining that the host computer shoots, when the fairway buoy is judged to be out of order, set up the fairway buoy as the abnormal fairway buoy, the control patrols and examines the host computer and gathers the image of the abnormal reason to the abnormal fairway buoy; when the routing inspection host of the routing inspection route to which the abnormal navigation mark belongs can not acquire the abnormal information of the abnormal navigation mark, adjusting the routing inspection route, and acquiring the abnormal information of the abnormal navigation mark by using other routing inspection hosts;

the navigation mark inspection system further comprises a navigation mark historical information base, and the navigation mark historical information base stores basic navigation mark information and historical inspection information of navigation marks obtained in historical inspection. The basic navigation mark information comprises the coordinates, the serial number, the navigation mark type, the initial navigation mark characteristic, the use start time and the like of the navigation mark, and the historical patrol inspection information of the navigation mark comprises the historical hidden danger degree phi and the position importance degree epsilon of the navigation mark.

The navigation mark comprises a navigation mark main body and a navigation mark information station, wherein the navigation mark main body comprises a lighthouse, a buoy, a stand column, a lightboat and other devices which are fixed on a navigation route or fixed in water and used for prompting directions for a navigation ship; the navigation mark information station comprises a monitoring device arranged on a navigation mark main body and is used for collecting hydrological meteorological information, flow velocity and direction information, ship flow information and the like, and the information can be used for constructing big data information in the aspects of meteorology, water flow and ship traffic.

The patrol inspection host machine is an unmanned aircraft which is operated by a radio remote control device or a self-contained program control device, generally comprises an unmanned helicopter or a fixed-wing unmanned aerial vehicle, and comprises a collecting component for collecting navigation mark information and abnormal information, wherein the collecting component comprises and is not limited to image collecting equipment such as a panoramic camera, a photoelectric pod, a laser range finder, an infrared measuring instrument, a surveying instrument and the like. The inspection host can acquire the information of the navigation mark by using a photoelectric pod or other shooting measuring equipment along a preset inspection route in an inspection task planned and designed by the ground control system, and transmits the acquired information to the ground control system in real time, so that the system makes a judgment.

In a specific inspection process, as shown in fig. 1, the ground control system determines n navigation marks R = [ R ] to be inspected according to n navigation marks R to be inspected₁,R₂,R₃,…,R_n]The basic navigation mark information can obtain the patrol weight R theta = [ R theta ] corresponding to the navigation mark R in the navigation mark historical information base₁,Rθ₂,Rθ₃,…,Rθ_n](ii) a Wherein, the ith navigation mark R_iPatrol weight R theta_i=ε_iφ_i，φ_iAs navigation mark R_iDegree of historical risk of epsilon_iAs navigation mark R_iThe location importance of (a); while creating a removal list YR that includes a first removal list YR1 and a second removal list YR 2.

The ground control system establishes an inspection task A, wherein the inspection task A comprises m inspection routes AH = [ AH ]₁,AH₂,AH₃,…,AH_m]Each routing inspection route is correspondingly subjected to flight inspection by one routing inspection host P, namely, the m routing inspection routes AH are correspondingly subjected to flight inspection by m routing inspection hosts P = [ P ]₁,P₂,P₃,…,P_m]Performing flight inspection; wherein, set up the ith platform and patrol inspection host P_iAt inspection route AH_iNavigation mark IR = [ IR ] requiring inspection₁,IR₂,IR₃,…,IR_k](ii) a Patrol inspection weight IR theta = [ IR theta ] corresponding to navigation mark IR₁,IRθ_2,IRθ₃,…,IRθ_k](ii) a Setting inspection route AH_iGo up j navigation mark IR that needs to patrol and examine_jHas a patrol weight of IR theta_j(ii) a Patrol route AH_iTotal weight of theta_i=(ii) a Wherein, the weight threshold value of the routing inspection route is set to be theta₀Then patrol route AH_iNeed to satisfy(ii) a At the same time, route AH of patrol_iNumber of navigation marks to be inspected(ii) a Wherein e is a natural constant, θ₀＜2e。

Specifically, in one embodiment, the number of the navigation marks to be inspected is 60, 4 inspection hosts are used for inspection, the ground control system sets the total weight threshold of the inspection route to be 5 according to requirements, and the inspection route AH is_iNumber threshold of navigation mark required to be patrolled=16；

Can obtain and patrol task A: comprises 5 routing inspection routes AH₁、AH₂、AH₃、AH₄、AH₅Corresponding to 5 polling host machines P₁、P₂、P₃、P₄、P₅Carrying out flight inspection setting;

set up 3 rd platform and patrol inspection host P₃At inspection route AH₃The number of the navigation marks needing to be inspected is 15, and the navigation marks are respectively IR₁、IR₂、IR₃、IR₄、IR₅ 、IR₆ 、IR₇、IR₈、IR₉、IR₁₀、IR₁₁、IR₁₂、IR₁₃、IR₁₄、IR₁₅The corresponding inspection weights are 0.2, 0.5, 0.1, 0.5, 0.3, 0.2, 0.5, 0.4, 0.3, 0.2, 0.5, 0.1, 0.2 and 0.6 respectively, and the inspection route AH₃Total weight of theta_i=4.8＜5；

According to the requirement of the polling task, the polling host P_iAt inspection route AH_iUpper pair navigation mark IR = [ IR =₁,IR₂,IR₃,…,IR_k]The inspection is carried out to obtain inspection information such as inspection images of the navigation mark IR, and when the control system inspects the inspection route AH according to the inspection_iGo up j navigation mark IR that needs to patrol and examine_jPatrol information judgment IR_jWhen the navigation system is abnormal, the ground control system controls the inspection host P_iFor abnormal navigation mark IR_jCollecting the aberration information;

wherein, through the information image of patrolling and examining of gathering, ground control system can obtain the fairway buoyThe method comprises the following steps of (1) acquiring polling characteristics of a navigation mark image, establishing a neural network model by acquiring the polling characteristics in the navigation mark image and by using navigation mark characteristic data in a historical navigation mark information base, taking the acquired polling characteristics as input, and judging whether the navigation mark corresponding to the polling image is abnormal according to output, wherein the specific method comprises the following steps: setting and inputting D routing inspection characteristics x = [ x ]₁; x₂ ; …; x_D ]Corresponding to weight w = [ w = [ w ]₁; w₂;…; w _D]Setting a bias b epsilon R; then we can get the weighted sum z of the input features, the specific formula is:

using the ReLU function as the activation function, then

In a multi-layer feedforward neural network, letThen, the feedforward neural network continuously iterates to carry out the propagation formula layer by layer as follows:；

the composite function is:

where W and b represent the connection weights and offsets for all layers in the network,is the number of layers of the neural network,is as followsThe number of layer neurons;is as followsLayer to layerA weight matrix of the layer;is as followsLayer to layerBiasing of the layers;is as followsOutput of layer neurons;

using a cross-entropy loss function, for sample (x, y) the loss function is:

wherein the content of the first and second substances,representing by a one-hot vector corresponding to y;

given a training set ofEach sample is sampledInput to a pre-neural network to obtainNetwork output isThe risk function on the data set is:

wherein the content of the first and second substances,is a regularization term; λ is a long parameter, the larger λ the closer W is to 0,

in each iteration of the gradient descent method, a learning rate alpha is set to obtain an updating mode of the parameters W and b,

calculating the gradient of the l-th layer weight and bias, δ^(l)Error term for layer i:

obtaining an iterative formula:

。

through the neural network model, the patrol characteristic in the patrol information image obtained by the patrol host is input into the network neural model to obtain output, when the output is 0, the corresponding navigation mark is represented to be abnormal, and when the output is 1, the corresponding navigation mark is represented to be normal.

When the navigation mark is abnormal, the ground control system sends an instruction to the inspection host, the inspection host acquires the abnormal reason of the abnormal navigation mark, specifically, the appearance details of the navigation mark can be shot, the water area condition can be shot, the geological structure can be shot, the humidity and the temperature can be acquired, for example, when the navigation mark is displaced, the land or the ship where the navigation mark is located can be shot, and whether the soil slope is loose or the pontoon has a fault can be judged for the control system.

According to the patrol inspection host P by the ground control system_iThe obtained image of the cause of the malfunction is used to determine the host P_iWhen the collection of the abnormal reasons is completed, the host P_iAt inspection route AH_iContinuously inspecting;

when the control system is according to the inspection host P_iThe obtained image of the cause of the malfunction is used to determine the host P_iWhen acquisition of the cause of the malfunction is not completed, e.g. due to tide, the patrol inspection master P_iWhen shooting is carried out, the navigation mark under the water surface can not be shot, and at the moment, the navigation mark IR is shot_jSlave patrol route AH_iRemoving and updating the patrol inspection host P_iObtaining a new routing inspection route, and repeating the steps to obtain a removed navigation mark list YR = [ YR ]₁,YR₂,YR₃,…,YR_c]C is less than k; setting a distance threshold D₀(ii) a Arrange and patrol and examine host computer AP_iHas a flying speed v_iInspection host AP_jHas a flying speed v_jThen there is a distance threshold D₀=. (ii) a Wherein D is_maxThe inspection distance of the single machine during inspection of all n pilots is used.

When the z-th abnormal navigation mark YR in the navigation mark list YR is removed_zTo patrol route AH_jDistance D of_z≤D₀When, YR is not present_zJ-th routing inspection route AH added with AH_j(j ≠ i), the patrol route AH is updated_jObtain a new routing inspection route AH_j'; corresponding inspection host P_jCarrying out routing inspection according to the updated routing inspection route;

when there is not an inspection route AH_jSo that YR_zGo to patrolInspection line AH_jDistance D of_z≤D₀And when the navigation mark which is not patrolled in the navigation mark R and the navigation mark in the removed navigation mark list YR are used as the navigation marks to be patrolled, the patrolling task is reestablished, and the steps are repeated according to the method for establishing the patrolling task at first.

For example, in an implementation, patrol host P₃At inspection route AH₃Go up and patrol and examine fairway buoy IR₇Then, the ground control system obtains the navigation mark IR₇Patrol information, judge navigation mark IR₇If the navigation mark is abnormal, the ground control system commands the inspection host P₃To navigation mark IR₇The shooting of the abnormality information is carried out, and the specific shooting content is, for example, the shooting of a navigation mark carrier, such as the ground, or the shooting of the details of the navigation mark body.

When the obtained malfunction information enables the ground control system to judge IR₇When the abnormality is caused, the inspection host P₃Continuously inspecting the next navigation mark;

when the obtained malfunction information enables the ground control system to judge IR₇When the reason is abnormal, the navigation mark is IR₇Patrol route AH₃While removing IR₇Adding the first removal list;

patrol and examine host computer P₃And patrol inspection host P₄The flying speed is the same, the total inspection distance is 55km when a single machine is used for inspecting all 60 to-be-inspected sails, and the current time has 5 cruise host machines to execute 5 routing inspection tasks

Then get the inspection host P₄Location routing inspection route and IR₇The distance threshold is 2.2.2 km;

when IR is measured₇And another patrol route AH₄Within 2.2.2km, the navigation mark IR₇Transferring the first removal list into the routing inspection route AH₄In (1), updating AH simultaneously₄The routing inspection route; when patrolling route AH₄Patrol and examine host computer P₄Patrol and examine fairway buoy IR₇Due to the navigation mark IR₇Transferred from the first removal list, illustrating the navigation mark IR₇The navigation mark information is acquired, so that the patrol main unit P₄Without reacquiring the navigationStandard IR₇Navigation mark information of, directly to, navigation mark IR₇Shooting for acquiring the abnormality information; when P is present₄Obtained navigation mark IR₇Enables the ground control system to determine IR₇When the abnormality is caused, the inspection host P₄Continuously inspecting the next navigation mark; when P is present₄Obtained navigation mark IR₇The malfunction information makes the ground control system still unable to judge IR₇When the reason is abnormal, the navigation mark is IR₇Slave inspection host P₄Patrol route AH₄The first removal list is converted back, and the inspection host P is updated simultaneously₄The routing inspection route;

setting of an execution-error beacon IR₇The cruise host of the abnormal information acquisition task is an abnormal navigation mark IR₇The filtering host set without executing the abnormal navigation mark IR₇The cruise host of the abnormal information acquisition task is the abnormal navigation mark IR₇The non-filtering host; the malformed fairway signs in the first removal list can only be switched into the cruise route in which the non-filtering host computer is located. In addition, when it is reacted with IR₇When more than one routing inspection route of the non-filtering host computer within 2.2km is arranged, the routing inspection route with the shortest distance is preferably used for identifying the navigation mark IR₇Adding;

when navigation mark IR₇When the distance between the navigation mark and the cruising route of any non-filtering host computer is more than 2.2km, the navigation mark IR₇And (4) switching to a second removal list from the first removal list, taking the second removal list as a navigation mark to be patrolled after the patrolling tasks of all the cruise hosts are finished, and distributing the cruise hosts by the ground control system again according to information such as cruise weight and the like, and returning to the initial operation step of the embodiment.

While the preferred embodiments of the present application have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all alterations and modifications as fall within the scope of the application. It will be apparent to those skilled in the art that various changes and modifications may be made in the present application without departing from the spirit and scope of the application. Thus, if such modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is intended to include such modifications and variations as well.