Airport remote control tower inspection system based on unmanned aerial vehicle assistance

Abstract

The invention relates to the technical field of unmanned aerial vehicle inspection, in particular to an unmanned aerial vehicle assistance-based airport remote control tower inspection system, which comprises an unmanned aerial vehicle module, a remote control module and a data analysis module, and is characterized in that the unmanned aerial vehicle module comprises a data acquisition module, a repelling module, a driving module and a first communication module; the remote control module comprises a second communication module, a path planning module, a display module and a storage module, and the data analysis module comprises a control optimization module and a target detection module. According to the invention, the unmanned aerial vehicle is used to improve the inspection efficiency and enhance the inspection safety, the inspection cost is reduced by searching the optimal sequence of traversing the working area, whether an abnormal condition occurs in the image is detected, and the human and animals are repelled in time through the repelling module. And comprehensive inspection, emergency disposal and real-time monitoring of the airport control tower and the surrounding area are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle inspection, and in particular to an airport remote tower inspection system assisted by unmanned aerial vehicles. Background Art

[0002] With the rapid development of the aviation transportation industry, the safe operation of airports is of vital importance; traditional airport tower inspections are mainly carried out manually, which has many limitations.

[0003] On the one hand, manual inspections are inefficient and require a lot of time and manpower; airport towers have complex structures and numerous equipment, making it difficult for manual inspections to fully cover all areas in a short period of time, and omissions are prone to occur; moreover, manual inspections are relatively slow and cannot respond to emergencies in a timely manner; on the other hand, manual inspections involve certain safety risks, as airport towers are usually located at higher locations, and staff may face the danger of working at heights when conducting inspections. In addition, manual inspections are even more difficult to conduct under severe weather conditions, such as strong winds and heavy rains.

[0004] With the continuous advancement of science and technology, drone technology is becoming increasingly mature. Drones are flexible, efficient, and remotely controlled, and can operate in complex environments. Applying drone technology to airport tower inspections can effectively make up for the shortcomings of traditional manual inspections, improve the efficiency and accuracy of inspections, and ensure the safe operation of airports. Summary of the invention

[0005] The purpose of the present invention is to provide an airport remote tower inspection system assisted by a drone to solve the problems raised in the above background technology.

[0006] In order to solve the above technical problems, the purpose of the present invention is to provide an airport remote tower inspection system based on drone assistance, including a drone module, a remote control module and a data analysis module, the drone module includes a data acquisition module, a drive module, a drive module and a first communication module, the remote control module includes a second communication module, a path planning module, a display module and a storage module, and the data analysis module includes a control optimization module and a target detection module;

[0007] The specific steps of the path planning module to plan the path are:

[0008] S1. Determine the cell size based on the size of the drone, divide the airport based on the cells, and establish a grid map;

[0009] S2. Distinguish the working area and the obstacle area on the established grid map;

[0010] S3, dividing the working area of ​​the grid map into a plurality of rectangular sub-areas again, and numbering each rectangular sub-area;

[0011] S4, obtaining the shortest path between two rectangular sub-areas based on the improved A-star algorithm;

[0012] S5. Obtain the optimal order for traversing all rectangular sub-regions based on the improved genetic algorithm.

[0013] As a further improvement of the technical solution, the step of obtaining the shortest path between two rectangular sub-areas in S4 is:

[0014] S4.1, taking the center points of the two rectangular sub-areas as the starting point and the target point respectively, using bidirectional search, searching for the shortest path from the starting point and the target point respectively;

[0015] S4.2. Initialize an open list and a closed list. The open list is used to store nodes to be checked, and the closed list is used to store checked nodes. The open list includes a forward open list and a reverse open list, and the closed list includes a forward closed list and a reverse closed list. The obstacle area belongs to the closed list.

[0016] S4.3, put the starting point and the target point into the forward open list and the reverse open list respectively, traverse the 24 adjacent nodes of the starting point and the target point respectively, give priority to expanding the nodes with lower comprehensive costs, and set the starting point as p(x 1 ,y 1 ) and the target point is q(x 2 ,y 2 ), the calculation formula is:

[0017] f(n)=g(n)+αh(n)

[0018]

[0019] Where n represents the node n(x n ,y n ), f(n) represents the comprehensive cost of node n, g(n) represents the actual cost of node n, h(n) represents the estimated cost of node n, α is the local obstacle rate, and m is the number of nodes in the obstacle area of ​​the 24 nodes around node n;

[0020] When searching forward:

[0021] h(n)=|x 2 -x n ∣+∣y 2 -y n ∣

[0022] At this time, g(n) represents the cost required to move the starting point to node n;

[0023] When searching in reverse:

[0024] h(n)=|x 1 -x n ∣+∣y 1 -y n ∣

[0025] At this time, g(n) represents the cost required to move the target point to node n;

[0026] S4.4. Set a safety buffer zone and treat it as an impassable area to ensure that the drone maintains a certain safety distance from actual obstacles.

[0027] S4.5. If there are duplicate nodes in the forward open list and the reverse open list in the bidirectional search, stop the search.

[0028] As a further improvement of the technical solution, the step of obtaining the order of traversing all rectangular sub-regions based on the improved genetic algorithm in S5 is:

[0029] S5.1. Encode the order of multiple rectangular sub-regions into a chromosome. Suppose there are m rectangular sub-regions, which are represented by an integer sequence of length m, and each integer represents the number of a rectangular sub-region;

[0030] S5.2, randomly generate a certain number of chromosomes as the initial population, each chromosome represents a possible working order of the rectangular sub-region;

[0031] S5.3, the total distance that the UAV traverses the entire working area is taken as the fitness value;

[0032] S5.4, selection, using the elite retention strategy to retain the individuals with the highest fitness in each generation, and using roulette to select individuals as parents to reproduce the next generation;

[0033] S5.5. The selected parent chromosomes are crossovered, mutated and repaired to produce offspring. Sequential crossover is used to generate two random numbers as nodes for the crossover operation. The gene fragments between the two nodes are exchanged. Mutation also generates two random numbers. The genes of the two nodes are exchanged. The probability of crossover and mutation is introduced to adjust the probability of adaptive adjustment. Then:

[0034]

[0035]

[0036] Among them, p c represents the crossover probability, Represents the set initial crossover probability, p m represents the probability of mutation, represents the set initial mutation probability, k 1 , k 2 , k 3 and k 4 are adjustment coefficients, u represents the population diversity measurement value, v represents the population convergence measurement value, and p i represents the probability of the i-th gene appearing in the chromosome, N represents the number of populations, and f e represents the fitness value of the e-th individual in the population, Represents the average fitness value of the population;

[0037] S5.6. Organize the newly generated offspring into a new population, repeat S5.3, S5.4 and S5.5, and stop when the maximum number of iterations is reached.

[0038] As a further improvement of the technical solution, the fitness calculation step in S5.3 is: according to the shortest path between the two rectangular sub-areas obtained in S4, a distance matrix D is established, then

[0039]

[0040] Among them, d 1,1 represents the shortest path distance between the first rectangular sub-region and the first rectangular sub-region, d 1,m represents the shortest path distance between the first rectangular sub-region and the mth rectangular sub-region, d m,1 represents the shortest path distance between the mth rectangular sub-region and the first rectangular sub-region, d m,m represents the shortest path distance between the m-th rectangular sub-region and the m-th rectangular sub-region, d 1,1 =…=d m,m =0.

[0041] As a further improvement of the present technical solution, the data acquisition module includes a camera, an infrared thermal imager and a laser radar.

[0042] As a further improvement of the technical solution, the expulsion module includes a loudspeaker, an audible and visual alarm, and an electromagnetic expulsion device.

[0043] As a further improvement of the present technical solution, the first communication module and the second communication module are connected via a wireless communication signal, and the driving module receives the route data sent by the first communication module and controls the movement of the drone module based on a PID control algorithm.

[0044] As a further improvement of the technical solution, the control optimization module optimizes the PID parameters used in the driving module based on the improved whale algorithm, and the specific steps are as follows:

[0045] A1. Initialize parameters, set population size, spatial dimension to three, number of iterations, initial population of whales and position and initial fitness value of the optimal whale, the position vector of the whale represents proportional gain, integral time and differential time respectively;

[0046] A2. Boundary condition processing: process the individuals in the population separately and calculate the coefficient vector and Generate uniformly distributed decision random number ρ,

[0047]

[0048] in, is the control vector, and decreases from 2 to 0 as the number of iterations increases during the entire iteration cycle. is a random vector on [0,1];

[0049] A3. Calculate the fitness value according to the fitness function;

[0050] A4. Predation to find the optimal solution:

[0051] Random search for prey: When ρ < 0.5 and hour,

[0052]

[0053] Where t is the current iteration number, is the position vector of a randomly selected individual in the current population, is the position vector of the current whale individual in the tth generation, is the position vector of the current whale individual in the (t+1) generation;

[0054] Contraction to surround the prey: When ρ < 0.5 and hour,

[0055]

[0056] in, is the position vector of the best individual in the current group;

[0057] Spiral update position: When ρ ≥ 0.5,

[0058]

[0059] Where s is a constant used to define the shape of the logarithmic spiral, and l is a uniformly distributed random number between [-1,1];

[0060] A5. Iterate the search until the maximum number of iterations is reached.

[0061] As a further improvement of this technical solution, the fitness function in A3 is:

[0062] J = ∫ 0 ∞ k|e(k)|dk

[0063] Among them, J represents the fitness function, k represents time, and e(k) represents the system error, that is, the difference between the expected output and the actual output.

[0064] As a further improvement of the present technical solution, the data acquisition module takes images and transmits them to the remote control module and the target detection module in sequence. The target detection module uses a trained Faster R-CNN network to detect anomalies. The Faster R-CNN network detects people and animals in the image and determines abnormal situations.

[0065] Compared with the prior art, the beneficial effects of the present invention are as follows: in the drone-assisted airport remote tower inspection system, drones are used to improve inspection efficiency and enhance inspection safety, and the inspection cost is reduced by finding the optimal order of traversing the working area. Whether abnormal conditions appear in the image is detected, and people and animals are promptly driven away through the removal module, thereby realizing comprehensive inspection, emergency response and real-time monitoring of the airport tower and surrounding areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is the overall structure diagram of the present invention;

[0067] Figure 2 This is the overall structure diagram of the drone module in the present invention;

[0068] Figure 3 It is the overall structure diagram of the data acquisition module in the present invention;

[0069] Figure 4 This is the overall structural diagram of the expulsion module in the present invention;

[0070] Figure 5 is the overall structure diagram of the remote control module in the present invention;

[0071] Figure 6 A flow chart of route planning by the path planning module in the present invention;

[0072] Figure 7 It is the overall structure diagram of the data analysis module in the present invention;

[0073] The meaning of each mark in the figure is:

[0074] 1. UAV module; 10. Data acquisition module; 100. Camera; 101. Infrared thermal imager; 102. LiDAR; 11. Expulsion module; 110. Loudspeaker; 111. Sound and light alarm; 112. Electromagnetic expulsion device; 12. Driving module; 13. First communication module;

[0075] 2. Remote control module; 20. Second communication module; 21. Path planning module; 22. Display module; 23. Storage module;

[0076] 3. Data analysis module; 30. Control optimization module; 31. Target detection module. DETAILED DESCRIPTION

[0077] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] like Figure 1-Figure 7 As shown, the purpose of this embodiment is to provide an airport remote tower inspection system based on drone assistance, including a drone module 1, a remote control module 2 and a data analysis module 3, the drone module 1 includes a data acquisition module 10, a drive module 11, a drive module 12 and a first communication module 13, the remote control module 2 includes a second communication module 20, a path planning module 21, a display module 22 and a storage module 23, the display module 22 uses a common LCD display screen, and buttons can be added to realize interactive functions, and the data analysis module 3 includes a control optimization module 30 and a target detection module 31;

[0079] The specific steps of the path planning module 21 to plan the path are:

[0080] S1. Determine the cell size based on the size of the drone, divide the airport based on the cells, and establish a grid map;

[0081] S2. Distinguish the working area and the obstacle area on the established grid map;

[0082] S3, dividing the working area of ​​the grid map into multiple rectangular sub-areas again, and numbering each rectangular sub-area, and dividing it step by step according to a predetermined rule, starting from a corner, for example, from the upper left corner, and expanding to the right or downward in sequence, and stopping when it expands to an obstacle;

[0083] S4, obtaining the shortest path between two rectangular sub-areas based on the improved A-star algorithm;

[0084] S5. Based on the improved genetic algorithm, the optimal order of traversing all rectangular sub-areas is obtained to ensure that the UAV can traverse the entire working area while reducing the flight distance of the UAV.

[0085] The steps for obtaining the shortest path between two rectangular sub-regions in S4 are:

[0086] S4.1, taking the center points of the two rectangular sub-areas as the starting point and the target point respectively, using bidirectional search, searching for the shortest path from the starting point and the target point respectively;

[0087] S4.2. Initialize an open list and a closed list. The open list is used to store nodes to be checked, and the closed list is used to store checked nodes. The open list includes a forward open list and a reverse open list, and the closed list includes a forward closed list and a reverse closed list. The obstacle area belongs to the closed list.

[0088] S4.3, put the starting point and the target point into the forward open list and the reverse open list respectively, traverse the 24 adjacent nodes of the starting point and the target point respectively, give priority to expanding the nodes with lower comprehensive costs, and set the starting point as p(x 1 ,y 1 ) and the target point is q(x 2 ,y 2 ), the calculation formula is:

[0089] f(n)=g(n)+αh(n)

[0090]

[0091] Where n represents the node n(x n ,y n ), f(n) represents the comprehensive cost of node n, g(n) represents the actual cost of node n, h(n) represents the estimated cost of node n, α is the local obstacle rate, and m is the number of nodes in the obstacle area of ​​the 24 nodes around node n. When there are fewer obstacles nearby, the local obstacle rate is smaller, which reduces the impact of the estimated cost and expands to the nodes that need to be reached faster;

[0092] When searching forward:

[0093] h(n)=|x 2 -x n ∣+∣y 2 -y n ∣

[0094] At this time, g(n) represents the cost required to move the starting point to node n;

[0095] When searching in reverse:

[0096] h(n)=|x1 -x n ∣+∣y 1 -y n ∣

[0097] At this time, g(n) represents the cost required to move the target point to node n;

[0098] S4.4. Set up a safety buffer zone and treat it as an impassable area to ensure that the drone maintains a certain safety distance from actual obstacles to prevent the drone from being damaged by touching obstacles during flight;

[0099] S4.5. If there are duplicate nodes in the forward open list and the reverse open list in the bidirectional search, stop the search and find paths from the initial point and the target point respectively. Then choose the shorter path as the shortest path. The two paths may also meet in the middle, reducing the time required to find the path.

[0100] The steps in S5 to obtain the order of traversing all rectangular sub-regions based on the improved genetic algorithm are:

[0101] S5.1. Encode the order of multiple rectangular sub-regions into a chromosome. Suppose there are m rectangular sub-regions, which are represented by an integer sequence of length m, and each integer represents the number of a rectangular sub-region;

[0102] S5.2. Randomly generate a certain number of chromosomes as the initial population, each chromosome representing a possible working order of the rectangular sub-region;

[0103] S5.3, the total distance that the UAV traverses the entire working area is taken as the fitness value;

[0104] S5.4, selection, using the elite retention strategy to retain the individuals with the highest fitness in each generation, using roulette to select individuals as parents to reproduce the next generation, and selecting the probability of genetic operations based on the fitness value of the individuals;

[0105] S5.5. The selected parent chromosomes are crossovered, mutated and repaired to produce offspring. When performing genetic operations, random numbers are generated, and the values ​​of the random numbers are between 0 and 1. The random numbers are compared with the adaptive crossover probability and the adaptive mutation probability. When the random numbers are less than the adaptive crossover probability and the adaptive mutation probability, the corresponding genetic operations are performed respectively. Sequential crossover is used to generate two random numbers as the nodes of the crossover operation, and the gene fragments between the two nodes are exchanged. Mutation also generates two random numbers, and the genes of the two nodes are exchanged. The probability of adaptive adjustment of crossover and mutation is introduced, and then:

[0106]

[0107]

[0108] Among them, p c represents the crossover probability, Represents the set initial crossover probability, p m represents the probability of mutation, represents the set initial mutation probability, k 1 , k 2 , k 3 and k 4 are adjustment coefficients, u represents the population diversity measurement value, using information entropy to comprehensively consider the probability of occurrence of various gene combinations in the population, thereby fully reflecting the diversity of the population, v represents the population convergence measurement value, and calculates the variance of the population fitness value. The variance reflects the degree of discreteness of the data. When the variance gradually decreases and approaches zero, it means that the fitness values ​​of individuals in the population are getting closer and closer, and the population tends to converge. i represents the probability of the i-th gene appearing in the chromosome, N represents the number of populations, and f e represents the fitness value of the e-th individual in the population, It represents the average fitness value of the population, taking into account the population diversity and population convergence. When the population diversity measurement value is large, the probability of crossover and mutation becomes larger, which helps to further explore different solutions. When the population convergence measurement value is large, the probability of crossover and mutation becomes smaller, which helps the algorithm to develop in the direction of convergence.

[0109] S5.6. Organize the newly generated offspring into a new population, repeat S5.3, S5.4 and S5.5, and stop when the maximum number of iterations is reached.

[0110] The fitness calculation steps in S5.3 are as follows: According to the shortest path between two rectangular sub-areas obtained in S4, a distance matrix D is established, then

[0111]

[0112] Among them, d 1,1 represents the shortest path distance between the first rectangular sub-region and the first rectangular sub-region, d 1,m represents the shortest path distance between the first rectangular sub-region and the mth rectangular sub-region, d m,1 represents the shortest path distance between the mth rectangular sub-region and the first rectangular sub-region, d m,m represents the shortest path distance between the m-th rectangular sub-region and the m-th rectangular sub-region, d 1,1 =…=d m,m =0.

[0113] The data acquisition module 10 includes a camera 100, an infrared thermal imager 101 and a laser radar 102. The camera 100, the infrared thermal imager 101 and the laser radar 102 are all installed on the drone to meet different inspection requirements.

[0114] The expelling module 11 includes a loudspeaker 110, an audible and visual alarm 111 and an electromagnetic expelling device 112 to expel intruding persons and animals.

[0115] The first communication module 13 and the second communication module 20 are connected via wireless communication signals. The driving module 12 receives the route data sent by the first communication module 13 and controls the movement of the drone module 1 based on the PID control algorithm. The PID control algorithm is based on three basic control functions: proportional, integral, and differential. Its principle is relatively simple and intuitive. Proportional control directly adjusts the control amount according to the size of the deviation, integral control is used to eliminate steady-state errors, and differential control responds to the rate of change of the deviation to predict the changing trend of the system in advance.

[0116] The control optimization module 30 optimizes the PID parameters used in the driving module 12 based on the improved whale algorithm, and the specific steps are as follows:

[0117] A1. Initialize parameters, set population size, spatial dimension to three, number of iterations, initial population of whales and position and initial fitness value of the optimal whale, the position vector of the whale represents proportional gain, integral time and differential time respectively;

[0118] A2. Boundary condition processing: process the individuals in the population separately and calculate the coefficient vector and Generate uniformly distributed decision random number ρ,

[0119]

[0120] in, is the control vector, and decreases from 2 to 0 as the number of iterations increases during the entire iteration cycle. is a random vector on [0,1];

[0121] A3. Calculate the fitness value according to the fitness function;

[0122] A4. Predation to find the optimal solution:

[0123] Random search for prey: When ρ < 0.5 and hour,

[0124]

[0125] Where t is the current iteration number, is the position vector of a randomly selected individual in the current population, is the position vector of the current whale individual in the tth generation, is the position vector of the current whale individual in the (t+1) generation;

[0126] Contraction to surround the prey: When ρ < 0.5 and hour,

[0127]

[0128] in, is the position vector of the best individual in the current group;

[0129] Spiral update position: When ρ ≥ 0.5,

[0130]

[0131] Where s is a constant used to define the shape of the logarithmic spiral, and l is a uniformly distributed random number between [-1,1];

[0132] A5. Iterate the search until the maximum number of iterations is reached.

[0133] The fitness function in A3 is:

[0134] J = ∫ 0 ∞ k|e(k)|dk

[0135] Among them, J represents the fitness function, which emphasizes the cumulative effect of error over time and weights time. This means that in the optimization process, not only the size of the error is considered, but also the continuous impact of the error at different time points is considered. k represents time, and e(k) represents the system error, that is, the difference between the expected output and the actual output.

[0136] After taking the image, the data acquisition module 10 transmits it to the remote control module 2 and the target detection module 31 in sequence. The target detection module 31 uses the trained Faster R-CNN network to detect anomalies. The Faster R-CNN network detects people and animals in the image and determines abnormal situations. By extracting features of people and animals, recognizing patterns, and comparing them with preset normal states, the Faster R-CNN network can accurately identify possible abnormal phenomena, providing an important basis for subsequent processing and decision-making.

[0137] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An airport remote tower inspection system based on drone assistance, characterized in that: The invention comprises an unmanned aerial vehicle module (1), a remote control module (2) and a data analysis module (3), wherein the unmanned aerial vehicle module (1) comprises a data acquisition module (10), a driving module (11), a driving module (12) and a first communication module (13), the remote control module (2) comprises a second communication module (20), a path planning module (21), a display module (22) and a storage module (23), and the data analysis module (3) comprises a control optimization module (30) and a target detection module (31); The specific steps of the path planning module (21) for planning a path are: S1. Determine the cell size based on the size of the drone, divide the airport based on the cells, and establish a grid map; S2. Distinguish the working area and the obstacle area on the established grid map; S3, dividing the working area of ​​the grid map into a plurality of rectangular sub-areas again, and numbering each rectangular sub-area; S4, obtaining the shortest path between two rectangular sub-areas based on the improved A-star algorithm; S5. Obtain the optimal order for traversing all rectangular sub-regions based on the improved genetic algorithm.

2. The drone-assisted airport remote tower inspection system according to claim 1 is characterized in that: The step of obtaining the shortest path between two rectangular sub-areas in S4 is: S4.1, taking the center points of the two rectangular sub-areas as the starting point and the target point respectively, using bidirectional search, searching for the shortest path from the starting point and the target point respectively; S4.

2. Initialize an open list and a closed list. The open list is used to store nodes to be checked, and the closed list is used to store checked nodes. The open list includes a forward open list and a reverse open list, and the closed list includes a forward closed list and a reverse closed list. The obstacle area belongs to the closed list. S4.

3. Put the starting point and the target point into the forward open list and the reverse open list respectively, traverse the 24 adjacent nodes of the starting point and the target point respectively, and give priority to expanding the nodes with lower comprehensive costs. Let the starting point be p(x1,y1) and the target point be q(x2,y2), and the calculation formula is: f(n)=g(n)+αh(n) Where n represents the node n(x n ,y n ), f(n) represents the comprehensive cost of node n, g(n) represents the actual cost of node n, h(n) represents the estimated cost of node n, α is the local obstacle rate, and m is the number of nodes in the obstacle area of ​​the 24 nodes around node n; When searching forward: h(n)=∣x2-x n ∣+∣y2-y n ∣ At this time, g(n) represents the cost required to move the starting point to node n; When searching in reverse: h(n)=∣x1-x n ∣+∣y1-y n ∣ At this time, g(n) represents the cost required to move the target point to node n; S4.

4. Set a safety buffer zone and treat it as an impassable area to ensure that the drone maintains a certain safety distance from actual obstacles. S4.

5. If there are duplicate nodes in the forward open list and the reverse open list in the bidirectional search, stop the search.

3. The drone-assisted airport remote tower inspection system according to claim 2 is characterized in that: The step of obtaining the order of traversing all rectangular sub-regions based on the improved genetic algorithm in S5 is: S5.

1. Encode the order of multiple rectangular sub-regions into a chromosome. Suppose there are m rectangular sub-regions, which are represented by an integer sequence of length m, and each integer represents the number of a rectangular sub-region; S5.

2. Randomly generate a certain number of chromosomes as the initial population, each chromosome representing a possible working order of the rectangular sub-region; S5.3, the total distance that the UAV traverses the entire working area is taken as the fitness value; S5.4, selection, using the elite retention strategy to retain the individuals with the highest fitness in each generation, and using roulette to select individuals as parents to reproduce the next generation; S5.

5. The selected parent chromosomes are crossovered, mutated and repaired to produce offspring. Sequential crossover is used to generate two random numbers as nodes for the crossover operation. The gene fragments between the two nodes are exchanged. Mutation also generates two random numbers. The genes of the two nodes are exchanged. The probability of crossover and mutation is introduced to adjust the probability of adaptive adjustment. Then: Among them, p c represents the crossover probability, Represents the set initial crossover probability, p m represents the probability of mutation, represents the set initial mutation probability, k1, k2, k3 and k4 are adjustment coefficients, u represents the population diversity measurement value, v represents the population convergence measurement value, p i represents the probability of the i-th gene appearing in the chromosome, N represents the number of populations, and f e represents the fitness value of the e-th individual in the population, Represents the average fitness value of the population; S5.

6. Organize the newly generated offspring into a new population, repeat S5.3, S5.4 and S5.5, and stop when the maximum number of iterations is reached.

4. The drone-assisted airport remote tower inspection system according to claim 3 is characterized in that: The fitness calculation step in S5.3 is: according to the shortest path between two rectangular sub-areas obtained in S4, a distance matrix D is established, then Among them, d 1,1 represents the shortest path distance between the first rectangular sub-region and the first rectangular sub-region, d 1,m represents the shortest path distance between the first rectangular sub-region and the mth rectangular sub-region, d m,1 represents the shortest path distance between the mth rectangular sub-region and the first rectangular sub-region, d m,m represents the shortest path distance between the mth rectangular sub-region and the mth rectangular sub-region, d 1,1 =…=d m,m =0.

5. The drone-assisted airport remote tower inspection system according to claim 1 is characterized in that: The data acquisition module (10) comprises a camera (100), an infrared thermal imager (101) and a laser radar (102).

6. The drone-assisted airport remote tower inspection system according to claim 1 is characterized in that: The driving away module (11) comprises a loudspeaker (110), an audible and visual alarm (111) and an electromagnetic driving away device (112).

7. The drone-assisted airport remote tower inspection system according to claim 1 is characterized in that: The first communication module (13) and the second communication module (20) are connected via a wireless communication signal, and the driving module (12) receives the route data sent by the first communication module (13) and controls the movement of the drone module (1) based on a PID control algorithm.

8. The drone-assisted airport remote tower inspection system according to claim 7 is characterized in that: The control optimization module (30) optimizes the PID parameters used in the driving module (12) based on the improved whale algorithm, and the specific steps are: A1. Initialize parameters, set population size, spatial dimension to three, number of iterations, initial population of whales and position and initial fitness value of the optimal whale, the position vector of the whale represents proportional gain, integral time and differential time respectively; A2. Boundary condition processing: process the individuals in the population separately and calculate the coefficient vector and Generate uniformly distributed decision random number ρ, in, is the control vector, and decreases from 2 to 0 as the number of iterations increases during the entire iteration cycle. is a random vector on [0,1]; A3. Calculate the fitness value according to the fitness function; A4. Predation to find the optimal solution: Random search for prey: When ρ < 0.5 and hour, Where t is the current iteration number, is the position vector of a randomly selected individual in the current population, is the position vector of the current whale individual in the tth generation, is the position vector of the current whale individual in the (t+1) generation; Contraction to surround the prey: When ρ < 0.5 and hour, in, is the position vector of the best individual in the current group; Spiral update position: When ρ ≥ 0.5, Where s is a constant used to define the shape of the logarithmic spiral, and l is a uniformly distributed random number between [-1,1]; A5. Iterate the search until the maximum number of iterations is reached.

9. The drone-assisted airport remote tower inspection system according to claim 8 is characterized in that: The fitness function in A3 is: Among them, J represents the fitness function, k represents time, and e(k) represents the system error, that is, the difference between the expected output and the actual output.

10. The drone-assisted airport remote tower inspection system according to claim 1 is characterized in that: The data acquisition module (10) captures an image and transmits it to the remote control module (2) and the target detection module (31) in sequence. The target detection module (31) uses a trained Faster R-CNN network to detect anomalies. The Faster R-CNN network detects people and animals in the image and determines abnormal conditions.

Images