Unmanned aerial vehicle group collaborative inspection system and method based on adaptive algorithm

Abstract

The invention relates to the technical field of unmanned aerial vehicle inspection systems, and provides an unmanned aerial vehicle group collaborative inspection system and method based on an adaptive algorithm, and the system comprises an unmanned aerial vehicle, a collaborative control terminal, an adaptive adjustment terminal, a ground emergency terminal and an inspection data processing terminal. The unmanned aerial vehicle is used for executing inspection tasks and collecting inspection data; the cooperative control terminal is used for managing cooperative work of the unmanned aerial vehicle group; the adaptive adjustment terminal is used for adjusting the task strategy of the unmanned aerial vehicle group in the execution task in real time; the ground emergency terminal is used for monitoring the working state of the unmanned aerial vehicle group and providing a fault emergency processing scheme; and the inspection data processing terminal is used for collecting inspection data from the unmanned aerial vehicle group and carrying out data real-time analysis, storage and detection. According to the method, the unmanned aerial vehicle group coordinated inspection quality is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle inspection systems, and in particular to a unmanned aerial vehicle group collaborative inspection system and method based on an adaptive algorithm. Background Art

[0002] The drone swarm collaborative inspection system is designed to achieve efficient inspection tasks over large areas through the collaborative work of multiple drones. The system achieves efficient inspection, obstacle avoidance and information collection in complex and changing environments, and is particularly suitable for power line inspection, traffic monitoring, environmental monitoring and other fields.

[0003] Now many drone swarm inspection systems have been developed. After a lot of searching and reference, we found that the drone swarm inspection systems in the prior art include the drone swarm inspection systems disclosed in publication numbers CN113093796A, CN105514872A, CN111966124A, EP3695191A1, and US20240092483A1. These drone swarm inspection systems generally include: drones, drone control terminals, and inspection data acquisition terminals; drones are used to inspect target areas; drone control terminals are used to control drone swarms to inspect target areas according to instructions; and inspection data acquisition terminals are used to collect inspection data from drones and analyze them. Since the inspection process of the above drone swarm inspection system is relatively simple and lacks an adaptive control process based on environmental changes and drone state changes, the quality of drone swarm inspections is reduced. Summary of the invention

[0004] The purpose of the present invention is to propose a drone swarm collaborative inspection system and method based on an adaptive algorithm in response to the shortcomings of the above-mentioned drone swarm inspection system.

[0005] The present invention adopts the following technical solution: A drone swarm collaborative inspection system based on an adaptive algorithm comprises a drone, a collaborative control terminal, an adaptive adjustment terminal, a ground emergency terminal and an inspection data processing terminal; the drone is used to perform inspection tasks and collect inspection data; the collaborative control terminal is used to manage the collaborative work of the drone swarm; the adaptive adjustment terminal is used to adjust the task strategy of the drone swarm in real time during the execution of the task; the ground emergency terminal is used to monitor the working status of the drone swarm and provide emergency fault handling solutions; the inspection data processing terminal is used to collect inspection data from the drone swarm and perform real-time data analysis, storage and detection; The collaborative control terminal includes a task allocation module, a path planning module and a collaborative control module; the task allocation module is used to allocate inspection tasks to each drone in the drone swarm; the path planning module is used to plan the flight path for each drone in the drone swarm; the collaborative control module is used to collaboratively control each drone in the drone swarm based on the inspection task allocation results and the flight path planning results.

[0006] Optionally, the task assignment module includes a task matching degree calculation submodule and a task assignment submodule; the task matching degree calculation submodule is used to calculate the task matching degree based on the plan information of the initial task assignment plan and the drone information of each drone in the drone group; the task assignment submodule is used to first perform task assignment according to the initial task assignment plan, and then adjust the task assignment according to the task matching degree.

[0007] Optionally, the path planning module includes a preferred path calculation submodule and a path planning result output submodule; the preferred path calculation submodule is used to calculate the preferred path parameters of the corresponding UAV in real time according to the task information of the inspection task and the flight information of the UAV; the path planning result output submodule is used to output the preferred path parameters as the path planning results of the corresponding UAV.

[0008] Optionally, the adaptive adjustment terminal includes an adaptive obstacle avoidance adjustment module and a real-time adjustment strategy module; the adaptive obstacle avoidance adjustment module is used to perform state update calculations based on the flight status and flight environment information of the UAV, and generate adaptive obstacle avoidance adjustment parameters; the real-time adjustment strategy module is used to make real-time adjustments to the mission strategy being executed by the UAV according to the adaptive obstacle avoidance adjustment parameters.

[0009] Optionally, the ground emergency terminal includes a drone swarm monitoring information acquisition module, an emergency event analysis module and an emergency strategy execution module; the drone swarm monitoring information acquisition module is used to obtain the aircraft condition information, flight information and flight environment information of each drone in the drone swarm; the emergency event analysis module is used to analyze and identify emergency events based on the aircraft condition information, flight information and flight environment information; the emergency strategy execution module is used to generate and execute corresponding emergency strategies based on the identification results.

[0010] A drone swarm collaborative inspection method based on an adaptive algorithm is applied to the above-mentioned drone swarm collaborative inspection system based on an adaptive algorithm. The drone swarm collaborative inspection method based on an adaptive algorithm includes: S1, perform inspection tasks and collect inspection data; S2, manages the collaborative work of drone swarms; S3, real-time adjustment of the mission strategy of the drone swarm in the mission execution; S4, monitors the working status of the drone swarm and provides emergency response solutions for faults; S5, collects inspection data from the drone swarm and performs real-time data analysis, storage, and detection.

[0011] The beneficial effects achieved by the present invention are: 1. The setting of drones, collaborative control terminals, adaptive adjustment terminals, ground emergency terminals and inspection data processing terminals is conducive to optimizing the system's collaborative control of drone groups during inspection tasks. The adaptive adjustment process is added so that the drone groups can perform adaptive adjustments in time when performing inspection tasks. With the protection of the ground emergency terminal, the work efficiency and quality of the drone groups are greatly improved, which is conducive to improving the inspection quality of the drone groups. 2. Through the setting of task allocation module, path planning module and collaborative control module, task allocation, path planning and collaborative control are independently realized, which is conducive to improving the stability of the basic functions of the system, and also improves the efficiency and stability of drone group inspection, thereby improving the inspection quality of drone group; 3. By setting up the task matching degree calculation submodule and the task allocation submodule in conjunction with the task matching degree algorithm, the task matching degree is calculated according to the scheme information of the initial task allocation scheme and the drone information of each drone in the drone group, which is conducive to improving the calculation accuracy of the task matching degree, thereby optimizing the task allocation process, thereby helping to improve the inspection quality of the drone group; 4. Through the setting of the preferred path calculation submodule and the path planning result output submodule in conjunction with the preferred path algorithm, the preferred path parameters of the corresponding drone are calculated in real time according to the task information of the inspection task and the flight information of the drone, which is conducive to improving the accuracy of the selection of the preferred path, thereby improving the inspection quality of the drone group; 5. By setting up the adaptive obstacle avoidance adjustment module in conjunction with the state parameter prediction algorithm, the state update calculation is performed according to the flight state and flight environment information of the UAV, and the adaptive obstacle avoidance adjustment parameters are generated, which is beneficial to improve the accuracy and timeliness of predicting the state parameters of the UAV at the next moment, thereby improving the obstacle avoidance performance of the UAV during the inspection task, which is beneficial to improving the inspection quality of the UAV swarm.

[0012] To further understand the features and technical contents of the present invention, please refer to the following detailed description and drawings of the present invention. However, the drawings provided are only for reference and description and are not intended to limit the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the overall structure of the present invention; Figure 2 It is a structural diagram of the task allocation module in the present invention; Figure 3 A schematic diagram showing the visualization of the broken line effect of the result of multiple calculations of task matching in the present invention; Figure 4 It is a structural schematic diagram of the path planning module in the present invention; Figure 5 A schematic diagram of a flow chart of the steps of calculating the preferred path parameters in the present invention; Figure 6 A visualization distribution diagram of the calculation effect of performing multiple preferred path calculations on the same drone group in the present invention; Figure 7 A schematic diagram of a method flow of a drone group collaborative inspection method based on an adaptive algorithm in the present invention; Figure 8 This is a rendering of the obstacle avoidance process in which a UAV avoids obstacles through state parameters in another embodiment of the present invention. DETAILED DESCRIPTION

[0014] The following is an explanation of the embodiments of the present invention through specific embodiments. Those skilled in the art can understand the advantages and effects of the present invention from the contents disclosed in this specification. The present invention can be implemented or applied through other different specific embodiments, and the details in this specification can also be modified and changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. In addition, the drawings of the present invention are only simple schematic illustrations and are not depicted according to actual dimensions. It is stated in advance. The following embodiments will further explain the relevant technical contents of the present invention in detail, but the disclosed contents are not intended to limit the scope of protection of the present invention.

[0015] Embodiment 1: This embodiment provides a drone group collaborative inspection system based on an adaptive algorithm. Figure 1 As shown, a drone swarm collaborative inspection system based on an adaptive algorithm includes a drone, a collaborative control terminal, an adaptive adjustment terminal, a ground emergency terminal and an inspection data processing terminal; the drone is used to perform inspection tasks and collect inspection data; the collaborative control terminal is used to manage the collaborative work of the drone swarm; the adaptive adjustment terminal is used to adjust the task strategy of the drone swarm in real time during the execution of the task; the ground emergency terminal is used to monitor the working status of the drone swarm and provide emergency fault handling solutions; the inspection data processing terminal is used to collect inspection data from the drone swarm and perform real-time data analysis, storage and detection; The collaborative control terminal includes a task allocation module, a path planning module and a collaborative control module; the task allocation module is used to allocate inspection tasks to each drone in the drone swarm; the path planning module is used to plan the flight path for each drone in the drone swarm; the collaborative control module is used to collaboratively control each drone in the drone swarm based on the inspection task allocation results and the flight path planning results.

[0016] Optional, combined Figure 2 As shown, the task allocation module includes a task matching degree calculation submodule and a task allocation submodule; the task matching degree calculation submodule is used to calculate the task matching degree according to the plan information of the initial task allocation plan and the drone information of each drone in the drone group; the task allocation submodule is used to first perform task allocation according to the initial task allocation plan, and then adjust the task allocation according to the task matching degree.

[0017] Specifically, when the drone group completes the inspection according to the initial task allocation plan, the task matching degree calculation submodule performs calculations to satisfy the following formula: ; Among them, R total Indicates the current task matching degree of the drone group; w i represents the task weight of the i-th UAV. In order to enhance the flexibility of task allocation, the higher the task priority and the stronger the urgency of the inspection process, the higher the task weight of the corresponding inspection task. The specific value is pre-set by the administrator to enable the UAV to prioritize urgent or high-priority tasks under different task conditions. For example, a UAV with insufficient power will reduce its task weight to ensure endurance. i It represents the instant score of the i-th drone after it performs the current task. The instant score is evaluated by the administrator or system program based on the completion progress and task effect of the drone task, such as whether the task is completed on time and whether anomalies are found. The instant score ranges from 0 to 10 points. The smaller the anomaly in the process of the drone completing the inspection task, the larger the instant score; γ represents the score discount factor, which is a constant with a value range between 0 and 1, indicating the weight of future rewards. The longer the inspection task duration, the larger the discount factor. The specific value is set by the administrator based on experience; It represents the value score of the i-th and j-th drones in the drone group when performing action a' in the inspection task under state s'; it is used to characterize the risk of the i-th and j-th drones working together. The larger the value score, the lower the risk and the better the action completed by the collaboration. State s' represents the current environmental condition or internal state of the drone in the inspection task. State s' can include information in multiple dimensions, such as: drone position, speed, heading, sensor data, power, task progress, etc. Assuming that the drone is currently performing an inspection task, state s' can be: s'=(x=500m, y=200m, z=50m, v=10m / s, residual current=80%); action a' Indicates the next action that the drone can take in state s'. Action a' can be of various types, usually related to the drone's navigation, obstacle avoidance, and mission execution, for example: a'=(accelerate to 12m / s, ascend 5m); the value score can be obtained by scoring according to the rules preset by the administrator, or it can be predicted by the training model based on historical data. For example, when drones i and j are inspecting an area at the same time, state s' can indicate the current risk level of the inspection area, and action a' can be the two drones working together to complete the inspection. If no fault is detected during the inspection, ,on the contrary ; It means selecting the task action with the largest value score among the task actions performed by any two drones when they jointly complete the inspection task, so that each value score in the summation process is the maximum value score obtained by performing the task action in the corresponding state when drones i and j are inspecting a certain area at the same time; If R total ≤R ref , indicating that the task matching degree does not meet the standard, the task allocation submodule adjusts the task allocation; if R total >R ref , indicating that the task matching degree meets the standard, the task allocation submodule allocates tasks to the same drone group at the same time every day in the manner of the initial task plan; R ref Indicates the matching threshold. The older the average age of the drone group, the smaller the matching threshold. The specific value is set by the administrator based on experience. Figure 3 As shown, Figure 3 This is a line graph showing the changes in the results of the system's multiple calculations of task matching, which shows the process, results, and visual analysis of 10 task matching calculations in actual use.

[0018] The implementation example of the above formula is as follows: Assume that there are 3 drones in the drone swarm to conduct power line inspections, and the 3 drones collaborate to complete the inspection of the initial task plan. Given w1=0.7, w2=0.5, w3=0.4, r1=8, r2=6, r3=5, and discount factor γ=0.8, After selection Q 12 (s',a')=12,Q 23 (s',a')=10,Q 31 (s',a')=6,R ref =20. Then: ; Due to R total =23.24>R ref , so the task matching degree of this inspection task plan meets the requirements.

[0019] The program code for the calculation process of the above formula in the implementation process is as follows: import numpy as np # Parameter settings n = 3 # Number of drones m = 3 # Number of cooperative drones # Assumed weights, instant scores, and discount factors w = np.array([0.7, 0.5, 0.8]) # Weight of each drone r = np.array([10, 8, 12]) # The instant score of each drone gamma = 0.9 # discount factor # Hypothetical joint Q value matrix Q_ij(s', a') # Three drones, each drone corresponds to three collaborative Q values ​​(joint Q value under action a') Q = np.array([ [[8, 12], [6, 9], [5, 10]], # Q value of drone 1 [[7, 10], [4, 11], [6, 8]], # Q value of drone 2 [[10, 5], [11, 7], [8, 6]] # Q value of drone 3 ]) # R_total calculation R_total = 0 for i in range(n): # For each drone, find max_{a'} sum_{j=1}^m Q_ij max_Q_value = 0 for a in range(Q.shape[2]): # Traverse action a' Q_sum = np.sum(Q[i, :, a]) # Accumulate for all j if Q_sum > max_Q_value: max_Q_value = Q_sum # Find the action a' that maximizes Q_sum # Calculate the current drone's score R_i = w[i] * (r[i] + gamma * max_Q_value) R_total += R_i # Add to total score # Output results print(f"Task matching degree R_total: {R_total}").

[0020] Optional, combined Figure 4 As shown, the path planning module includes a preferred path calculation submodule and a path planning result output submodule; the preferred path calculation submodule is used to calculate the preferred path parameters of the corresponding UAV in real time according to the task information of the inspection task and the flight information of the UAV; the path planning result output submodule is used to output the preferred path parameters as the path planning results of the corresponding UAV.

[0021] Specifically, when the UAV is assigned to an inspection task and starts to execute, the preferred path calculation submodule calculates the preferred path parameters for the UAV, combined with Figure 5 As shown, the steps to calculate the preferred path parameters are as follows: A1, initialize the path candidate set and read the path candidate set of the system.

[0022] Among them, each path contains different combinations of inspection points, and the initial path set can be generated by random generation or the nearest neighbor method; A2, calculate the objective function value of each path, and calculate the flight time term, synergy utility term and obstacle avoidance complexity corresponding to each path; A3, calculate the total path cost of each path according to the flight time term, synergy utility term and obstacle avoidance complexity corresponding to each path; A4, select the path with the smallest total path cost as the preferred path.

[0023] The formula for selecting the total cost of the path satisfies the following formula: ; Among them, P opt It represents the preferred path of the i-th UAV when performing the inspection task, that is, the selected path with the minimum total cost; arg min() means finding an input that minimizes the objective function in the brackets of arg min(). In this formula, it means finding an optimal path as the preferred path parameter so that the sum of the accumulated flight time and obstacle avoidance complexity of all path points is minimized, that is, selecting an optimal path to minimize the total flight time of the UAV, while considering the difficulty of obstacle avoidance and the collaborative benefit.

[0024] represents the flight time term; d b,b+1 represents the distance between the bth and b+1th nodes. In the UAV path planning, the node can be a checkpoint or an intermediate path point; v i represents the flight speed of the i-th UAV when it passes the b-th node in the historical data; B represents the total number of nodes, n represents the total number of UAVs in the UAV swarm; U ij represents the collaborative utility term, which represents the maximum collaborative utility value of the i-th UAV and the j-th UAV during calculation; the collaborative utility value of two UAVs in a UAV group can be determined by the task overlap, task complementarity and relative distance of the aircraft. As an example, the following is a calculation formula for determining the collaborative utility value based on the task overlap and relative distance of the aircraft: ; Among them, S cd represents the mission overlap area between the i-th UAV and the j-th UAV; S all represents the total task area of ​​all inspection tasks of the drone group (i.e., the area corresponding to the area covered by all tasks); β represents the distance attenuation coefficient, which is generally 0.5; D ij It represents the relative distance between the i-th UAV and the j-th UAV during calculation.

[0025] represents the obstacle avoidance complexity term; Indicates the complexity of obstacles, indicating the total complexity of obstacles for the i-th UAV and the j-th UAV, which can be but is not limited to: the total number of obstacles in the environment of the i-th UAV and the j-th UAV during calculation. Figure 6 As shown, Figure 6 A visualization of the distribution of computational results from multiple preferred path calculations for the same drone swarm.

[0026] The implementation example of the preferred path selection process is as follows: Assume that the initial candidate path set has three paths: P1, P2, and P3. The distance, flight speed, cooperative utility, and obstacle avoidance complexity between the paths are: P1:d 12 =1000m,v1=20m / s,U 12 =0.8, C 12 ob =2; P2:d 12 =1500m, v1=18m / s, U 12 =0.7, C 12 ob =3; P3:d 12 =1200m, v1=22m / s, U 12 =0.9, C 12 ob =1; Then: the total path cost of path P1 is 34, the total path cost of path P2 is 61.33, and the total path cost of path P3 is 49.1, so P opt = P1, where P1 is the preferred path.

[0027] The program code for the calculation process of the above formula in the implementation process is as follows: import numpy as np # Parameter settings B = 3 # Number of path segments n = 3 # Number of other cooperative drones # Distance of assumed path segment (in meters) d = np.array([1000, 1500, 1200]) # The distance of each path d_{b,b+1} # Assumed flight speed for each path segment (in meters per second) v = np.array([10, 12, 9]) # The speed v_i corresponding to each path segment # Assume the collaborative utility U_{ij} U = np.array([0.8, 0.7, 0.9]) # The collaborative utility of each path # Assume the obstacle complexity matrix C_{ob}^{ij}, the obstacle complexity of each drone corresponding to each path segment C_ob = np.array([ [3, 4, 2], # Obstacle avoidance complexity of drone 1 [2, 3, 5], # Obstacle avoidance complexity of drone 2 [1, 2, 4] # Obstacle avoidance complexity of drone 3 ]) # Calculate the total cost of each path total_costs = [] for b in range(B): # Calculate the time cost of path segment b (path distance / speed * collaboration utility) time_cost = (d[b] / v[b]) * U[b] # Calculate the obstacle avoidance complexity of path segment b obstacle_cost = np.sum(C_ob[b, :]) # Accumulation of obstacle avoidance complexity # The total cost is time cost + obstacle avoidance complexity total_cost = time_cost + obstacle_cost total_costs.append(total_cost) # Find the minimum cost path P_opt = np.argmin(total_costs) + 1 # Because the path number starts from 1 P_opt_value = total_costs[P_opt - 1] # Output results print(f"The optimal path is path {P_opt}, and the total cost is: {P_opt_value:.2f}").

[0028] Optionally, the adaptive adjustment terminal includes an adaptive obstacle avoidance adjustment module and a real-time adjustment strategy module; the adaptive obstacle avoidance adjustment module is used to perform state update calculations based on the flight status and flight environment information of the UAV, and generate adaptive obstacle avoidance adjustment parameters; the real-time adjustment strategy module is used to make real-time adjustments to the mission strategy being executed by the UAV according to the adaptive obstacle avoidance adjustment parameters.

[0029] Optionally, the ground emergency terminal includes a drone swarm monitoring information acquisition module, an emergency event analysis module and an emergency strategy execution module; the drone swarm monitoring information acquisition module is used to obtain the aircraft condition information, flight information and flight environment information of each drone in the drone swarm; the emergency event analysis module is used to analyze and identify emergency events based on the aircraft condition information, flight information and flight environment information; the emergency strategy execution module is used to generate and execute corresponding emergency strategies based on the identification results.

[0030] In summary, the setting of drones, collaborative control terminals, adaptive adjustment terminals, ground emergency terminals and inspection data processing terminals is conducive to optimizing the collaborative control of the system when the drone group performs inspection tasks. The adaptive adjustment process is added so that the drone group can perform adaptive adjustment in time when performing inspection tasks, and cooperate with the protection of the ground emergency terminal, which greatly improves the work efficiency and work quality of the drone group; through the setting of the task allocation module, the path planning module and the collaborative control module, the task allocation, path planning and collaborative control are independently realized, which is conducive to improving the stability of the basic functions of the system and the efficiency and stability of the drone group inspection; through the setting of the task matching degree calculation submodule and the task allocation submodule in conjunction with the task matching degree algorithm, the task matching degree is calculated according to the scheme information of the initial task allocation scheme and the drone information of each drone in the drone group, which is conducive to improving the calculation accuracy of the task matching degree, and then optimizing the task allocation process; through the setting of the preferred path calculation submodule and the path planning result output submodule in conjunction with the preferred path algorithm, the preferred path parameters of the corresponding drone are calculated in real time according to the task information of the inspection task and the flight information of the drone, which is conducive to improving the selection accuracy of the preferred path, thereby improving the inspection quality of the drone group.

[0031] A drone swarm collaborative inspection method based on an adaptive algorithm is applied to the above-mentioned drone swarm collaborative inspection system based on an adaptive algorithm, combined with Figure 7 As shown, the UAV swarm collaborative inspection method based on the adaptive algorithm includes: S1, perform inspection tasks and collect inspection data; S2, manages the collaborative work of drone swarms; S3, real-time adjustment of the mission strategy of the drone swarm in the mission execution; S4, monitors the working status of the drone swarm and provides emergency response solutions for faults; S5, collects inspection data from the drone swarm and performs real-time data analysis, storage, and detection.

[0032] Embodiment 2: This embodiment includes all the contents of Embodiment 1, and provides a drone group collaborative inspection system based on an adaptive algorithm. When the adaptive obstacle avoidance adjustment module performs state update calculation according to the flight state and flight environment information of the drone and generates an adaptive obstacle avoidance adjustment parameter, the following formula is satisfied: ; Among them, X k+1 represents the state parameter of the drone at the next moment, which may be, but is not limited to, position and speed. In this embodiment, the state parameter is position as an example for explanation; G represents the state transfer matrix, which is used to describe the change of the drone from the current state to the next state, such as the change of the position and speed of the drone during the process of flying from the current position to the next position; X k Indicates the current state parameters of the drone, i.e., the current position; H represents the control input matrix, which includes the power output and steering control of the drone, etc., obtained through sensors or flight control systems; U k Represents the control input of the drone at the current moment; W k represents process noise, which means uncontrollable random disturbances during flight, such as wind speed changes, etc., and is obtained by noise modeling of external environmental data; n represents the total number of drones, m represents the total number of filters in the drone group; K ij It represents the fusion coefficient of the jth filter in the i-th drone, which is used to fuse the data of different sensors. The larger the data history processing volume of the filter, the larger the corresponding fusion coefficient. The specific value is set and adjusted by the administrator based on experience. N represents the noise matrix. The noise is the sensor noise, such as the laser radar ranging error, camera recognition error, etc., which can be modeled according to the accuracy of the sensor. When X is calculated k+1 After that, if there is a moving obstacle in front, the system will adjust the filter weight according to the obstacle avoidance algorithm, for example, increasing the weight of the lidar data and reducing the weight of the camera data, and then the drone will continue to complete the inspection task while avoiding the obstacle. Figure 8 As shown, Figure 8 This is a rendering of the obstacle avoidance process in which the drone avoids obstacles through state parameters.

[0033] When the status parameter is a single parameter, a specific implementation example is as follows: Assume the current state of the drone is X k =(100,150,50), the state transfer matrix G=I, that is, the identity matrix, the control input matrix item HU k =(2,1,0), process noise W k =(0.5,−0.2,0), filter weight K 12 =0.6, K 13 =0.4, noise N=(0.1,0.05). Then: ; X k+1 The calculation results include the obstacle avoidance results obtained through multimodal data fusion. This result will be directly used for the obstacle avoidance behavior of the drone.

[0034] When the status parameter is multiple parameters, the specific implementation example 2 is as follows: Assume that the current state is known to be: ; This means that the current position of the drone is (100, 200, 50) meters, and the speed is (5, 3, 1) meters per second respectively; Assume that the drone flies at a constant speed and the state transfer matrix G is known: ; Assume that the control input is acceleration, and the control input matrix H is known: ; Control input U k : ; This means that the acceleration in the x-axis, y-axis and z-axis directions is 1m / s respectively. 2 , 0.5m / s 2 and 0.2m / s 2 ; Process noise W k : ; Assume that the fusion coefficients of the three sensors, lidar, camera, and GPS, are 0.7, 0.2, and 0.1, respectively, indicating that the lidar has a larger weight and the filter fusion coefficient K is ij : ; Assuming that the measurement noise of the known sensor is: 0.05m, 0.1m, 0.2m, then the noise matrix N is: ; So the comprehensive calculation process is: ; That is, the state prediction of the drone at the next moment is: Position: (105.775, 203.225, 51.225) meters; Speed: (6.1, 3.55, 1.22) m / s.

[0035] The program code for the calculation process of the above formula in the implementation process is as follows: import numpy as np # Parameter settings # Initial state: position (x, y, z) and velocity (v_x, v_y, v_z) X_k = np.array([100, 200, 50, # Current position (x, y, z) 5, 3, 1]) # Current speed (v_x, v_y, v_z) # State transition matrix A A = np.array([[1, 0, 0, 1, 0, 0], # position update = position + speed [0, 1, 0, 0, 1, 0], [0, 0, 1, 0, 0, 1], [0, 0, 0, 1, 0, 0], # the speed remains the same [0, 0, 0, 0, 1, 0], [0, 0, 0, 0, 0, 1]]) # Control input matrix B (assuming the control input affects position and velocity) B = np.array([[0.5, 0, 0], # Control input affects position change [0, 0.5, 0], [0, 0, 0.5], [1, 0, 0], # Control input affects speed change [0, 1, 0], [0, 0, 1]]) # Control input U_k (assuming acceleration is the control input) U_k = np.array([1, 0.5, 0.2]) # Control input, acceleration (a_x, a_y, a_z) # Process noise W_k (external disturbances, such as position and velocity disturbances caused by wind speed) W_k = np.array([0.2, -0.1, 0.05, # position perturbation 0.1, 0.05, 0.02]) # Speed ​​perturbation # Filter fusion coefficient K_{ij} (lidar, camera, GPS) K_ij = np.array([0.7, 0.2, 0.1]) # LiDAR has the highest weight # Sensor noise N (LiDAR, Camera, GPS) N = np.array([0.05, 0.1, 0.2]) # Noise of each sensor # Sensor Noise Fusion sensor_noise = np.dot(K_ij, N) # Noise after multimodal data fusion # Calculate the next state X_{k+1} X_k1 = A @ X_k + B @ U_k + W_k + sensor_noise # Output results print("Position and speed at the next moment:") print(f"Position: ({X_k1[0]:.2f}, {X_k1[1]:.2f}, {X_k1[2]:.2f})") print(f"Speed: ({X_k1[3]:.2f}, {X_k1[4]:.2f}, {X_k1[5]:.2f})").

[0036] In summary, by setting up the adaptive obstacle avoidance adjustment module in conjunction with the state parameter prediction algorithm, the state update calculation is performed according to the flight status and flight environment information of the UAV, and the adaptive obstacle avoidance adjustment parameters are generated, which is beneficial to improve the accuracy and timeliness of predicting the state parameters of the UAV at the next moment, thereby improving the obstacle avoidance performance of the UAV during the inspection task, which is beneficial to improving the inspection quality of the UAV swarm.

[0037] The contents disclosed above are only preferred feasible embodiments of the present invention, and do not limit the protection scope of the present invention. Therefore, all equivalent technical changes made using the contents of the present invention description and drawings are included in the protection scope of the present invention. In addition, the elements therein can be updated as technology develops.

Claims

1. A drone swarm collaborative inspection system based on an adaptive algorithm, characterized in that: It includes a drone, a collaborative control terminal, an adaptive adjustment terminal, a ground emergency terminal and an inspection data processing terminal; the drone is used to perform inspection tasks and collect inspection data; The collaborative control terminal is used to manage the collaborative work of the drone swarm; the adaptive adjustment terminal is used to adjust the task strategy of the drone swarm in real time; the ground emergency terminal is used to monitor the working status of the drone swarm and provide emergency fault handling solutions; the inspection data processing terminal is used to collect inspection data from the drone swarm and perform real-time data analysis, storage and detection; The collaborative control terminal includes a task allocation module, a path planning module and a collaborative control module; the task allocation module is used to allocate inspection tasks to each drone in the drone swarm; the path planning module is used to plan the flight path for each drone in the drone swarm; the collaborative control module is used to collaboratively control each drone in the drone swarm based on the inspection task allocation results and the flight path planning results.

2. The UAV swarm collaborative inspection system based on adaptive algorithm as claimed in claim 1, characterized in that: The task assignment module includes a task matching degree calculation submodule and a task assignment submodule; the task matching degree calculation submodule is used to calculate the task matching degree according to the plan information of the initial task assignment plan and the drone information of each drone in the drone group; the task assignment submodule is used to first assign tasks according to the initial task assignment plan, and then adjust the task assignment according to the task matching degree.

3. The UAV swarm collaborative inspection system based on adaptive algorithm as claimed in claim 2, characterized in that: The path planning module includes a preferred path calculation submodule and a path planning result output submodule; the preferred path calculation submodule is used to calculate the preferred path parameters of the corresponding drone in real time according to the task information of the inspection task and the flight information of the drone; The path planning result output submodule is used to output the preferred path parameters as the path planning result of the corresponding drone.

4. The UAV swarm collaborative inspection system based on adaptive algorithm as claimed in claim 3, characterized in that: The adaptive adjustment terminal includes an adaptive obstacle avoidance adjustment module and a real-time adjustment strategy module; the adaptive obstacle avoidance adjustment module is used to perform state update calculation according to the flight state and flight environment information of the UAV to generate adaptive obstacle avoidance adjustment parameters; The real-time adjustment strategy module is used to adjust the mission strategy being executed by the drone in real time according to the adaptive obstacle avoidance adjustment parameters.

5. The UAV swarm collaborative inspection system based on adaptive algorithm as claimed in claim 4, characterized in that: The ground emergency terminal includes a drone swarm monitoring information acquisition module, an emergency event analysis module and an emergency strategy execution module; the drone swarm monitoring information acquisition module is used to obtain the condition information, flight information and flight environment information of each drone in the drone swarm; the emergency event analysis module is used to analyze and identify emergency events based on the condition information, flight information and flight environment information; the emergency strategy execution module is used to generate and execute corresponding emergency strategies based on the identification results.

6. A drone swarm collaborative inspection method based on an adaptive algorithm, applied to a drone swarm collaborative inspection system based on an adaptive algorithm, characterized in that: The drone group collaborative inspection method based on the adaptive algorithm includes: S1, perform inspection tasks and collect inspection data; S2, manages the collaborative work of drone swarms; S3, real-time adjustment of the mission strategy of the drone swarm in the mission execution; S4, monitors the working status of the drone swarm and provides emergency response solutions for faults; S5, collects inspection data from the drone swarm and performs real-time data analysis, storage, and detection.

Images