Performance evaluation method and system for path planning algorithm of unmanned aerial vehicle

Abstract

The invention relates to an unmanned aerial vehicle path planning algorithm performance evaluation method and system, and belongs to the field of algorithm performance evaluation, a hierarchical structure model is constructed based on an analytic hierarchy process (AHP), evaluation indexes are classified and hierarchical, and the hierarchical structure model comprises two evaluation dimensions, an algorithm universality index and a specific task index; constructing a judgment matrix through an analytic hierarchy process (AHP), and calculating a subjective weight vector of each evaluation index; based on a CRITIC weighting method, objective weight vectors of all evaluation indexes under all evaluation dimensions are calculated, and the objective weight vectors are obtained by calculating index variability and index conflict; and combining the subjective weight calculated by the AHP method with the objective weight calculated by the CRITIC method to form a comprehensive weight vector. Through the combination of AHP and CRITIC, the importance of each evaluation index can be reflected more accurately.

Description

Technical Field

[0001] The present invention belongs to the field of algorithm performance evaluation, and in particular relates to a method and system for evaluating the performance of an algorithm for unmanned aerial vehicle path planning. Background Art

[0002] With the rapid development of UAV technology, UAVs are increasingly used in various fields. As one of the key technologies for autonomous flight of UAVs, the performance of path planning algorithms directly affects the efficiency and safety of UAV missions. Therefore, it is particularly important to conduct a scientific and objective evaluation of the performance of UAV path planning algorithms.

[0003] At present, in the field of performance evaluation of UAV path planning algorithms, a variety of evaluation methods have been proposed, such as fuzzy comprehensive evaluation method, extension geometry evaluation method, etc.

[0004] The fuzzy comprehensive evaluation method can show good flexibility in dealing with uncertain information by introducing the concept of fuzzy mathematics. However, since this method is greatly affected by subjective factors in the weight determination process, the objectivity of its evaluation results is limited. On the other hand, the extension geometry evaluation method can handle multi-attribute decision-making problems through geometric methods and reduce the influence of subjectivity to a certain extent. However, its complex calculation process limits the application of this method when facing large-scale problems, which is not conducive to UAV tasks with high real-time requirements.

[0005] Based on the above problems, a performance evaluation method and system for UAV path planning algorithm is proposed. Summary of the invention

[0006] The purpose of the present invention is to provide a method and system for evaluating the performance of an unmanned aerial vehicle path planning algorithm, thereby improving the efficiency of evaluating the performance of an unmanned aerial vehicle path planning algorithm.

[0007] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0008] A method for evaluating the performance of an unmanned aerial vehicle path planning algorithm, the method comprising the following steps:

[0009] Based on the analytic hierarchy process (AHP), a hierarchical model is constructed to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions, an algorithm universality indicator and a specific task indicator. A judgment matrix is ​​constructed through the analytic hierarchy process (AHP) to calculate the subjective weight vector of each evaluation indicator.

[0010] Based on the CRITIC weighting method, the objective weight vector of each evaluation indicator under each evaluation dimension is calculated, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict;

[0011] The subjective weights calculated by the AHP method are combined with the objective weights calculated by the CRITIC method to form a comprehensive weight vector.

[0012] On the other hand, a UAV path planning algorithm performance evaluation system is provided, the system comprising:

[0013] A hierarchical model building module is used to build a hierarchical model based on the analytic hierarchy process (AHP) to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions: algorithm universality indicators and specific task indicators.

[0014] The subjective weight calculation module is used to construct a judgment matrix through the analytic hierarchy process (AHP) and calculate the subjective weight vector of each evaluation index;

[0015] An objective weight calculation module is used to calculate the objective weight vector of each evaluation indicator under each evaluation dimension based on the CRITIC weighting method, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict;

[0016] The comprehensive weight calculation module is used to combine the subjective weight calculated by the AHP method with the objective weight calculated by the CRITIC method to form a comprehensive weight vector.

[0017] Beneficial effects:

[0018] This application combines AHP and CRITIC to more accurately reflect the importance of each evaluation index. Compared with the traditional single weight allocation method, the present invention can flexibly adjust the weight according to the specific performance of different task requirements and evaluation indicators, so that the evaluation results are more in line with the actual application needs;

[0019] This application optimizes the evaluation model and algorithm, and significantly reduces the computational complexity and improves the computational efficiency by reasonably simplifying the calculation steps of the path planning algorithm performance evaluation, making it applicable to large-scale complex scenarios and tasks, with high computational efficiency and practical applicability.

[0020] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0022] Figure 1This is the overall evaluation framework diagram of the UAV path planning algorithm disclosed in this disclosure;

[0023] Figure 2 This is the AHP operation flow chart disclosed herein;

[0024] Figure 3 This is a diagram of the hierarchical structure model disclosed herein;

[0025] Figure 4 It is a schematic diagram of constructing a judgment matrix of the analytic hierarchy process disclosed in the present invention;

[0026] Figure 5 This is the operation flow chart of the CRITIC objective weighting method disclosed in this disclosure;

[0027] Figure 6 For this disclosure Figure 5 CRITIC objective weight diagram;

[0028] Figure 7 This is a simulated map of mountain terrain disclosed in the present invention;

[0029] Figure 8 It is the no-fly zone map in the disclosed topographic map;

[0030] Fig. 9 This is a schematic diagram of the control points disclosed in the present invention;

[0031] Fig.10 This is a schematic diagram of the yaw angle constraint disclosed in the present invention;

[0032] Fig.11 This is the path diagram after interpolation fitting of the present disclosure. DETAILED DESCRIPTION

[0033] 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.

[0034] Implementation method,The performance evaluation method of the UAV path planning algorithm includes the following steps;

[0035] Based on the analytic hierarchy process (AHP), a hierarchical model is constructed to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions, an algorithm universality indicator and a specific task indicator. A judgment matrix is ​​constructed through the analytic hierarchy process (AHP) to calculate the subjective weight vector of each evaluation indicator.

[0036] Based on the CRITIC weighting method, the objective weight vector of each evaluation indicator under each evaluation dimension is calculated, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict;

[0037] The subjective weights calculated by the AHP method are combined with the objective weights calculated by the CRITIC method to form a comprehensive weight vector.

[0038] The weighting method can be divided into subjective weighting method and objective weighting method according to whether there is a subjective impression of the decision maker. The subjective weighting method can reflect the degree of attention paid by the decision maker to different evaluation indicators, but there is a problem of subjective randomness; while the objective weighting method can highlight the intrinsic attribute value of the statistical law of data, but it is overly dependent on objective data and lacks subjective interpretation.

[0039] This implementation method proposes a comprehensive performance evaluation method for the UAV path planning algorithm, which is a combination of the AHP subjective weighting method and the CRITIC objective weighting method, while taking into account the advantages of both methods. By reasonably assigning the weight of each indicator, an evaluation result with the inherent statistical laws of objective data can be obtained, and a scientific algorithm evaluation result can also be obtained based on the subjective preferences of decision makers.

[0040] For the UAV path planning algorithm, the evaluation problem generally focuses on the performance of the algorithm in four aspects: the effectiveness of the algorithm, the stability of the algorithm, the convergence of the algorithm, and the safety of the path planned for the UAV path planning scenario. The UAV path planning algorithm test evaluation model proposed in this project mainly considers the algorithm performance in these four aspects, and because the constructed intelligent iterative optimization algorithm evaluation model has multiple characteristics such as scalability and free combination. This project proposes a comprehensive performance evaluation method of hierarchical analysis subjective weighting method and CRITIC objective weighting method.

[0041] The evaluation method framework of UAV path planning algorithm based on AHP-CRITIC combined weight is as follows Figure 1 As shown, the specific operation steps of the method are as follows:

[0042] S1 Determine the evaluation indicators: First, determine the various indicators for evaluating the UAV path planning algorithm. The evaluation indicators proposed in this paper include average running time, optimization success rate, number of iterations, path length, and path safety.

[0043] S2. Construct a hierarchical model based on AHP: classify and divide the evaluation indicators into levels to construct a hierarchical model. This paper determines two evaluation dimensions: algorithm universality indicator A1 and specific task indicator A2.

[0044] S3. Give a judgment matrix again based on AHP and calculate the weight vector: Compare each indicator according to its relative importance, give a judgment matrix, and then calculate the weight vector U of each indicator by using the hierarchical analysis method. ij .

[0045] S4. Calculate CRITIC weights: By using the CRITIC objective evaluation method, calculate the internal connections between the indicators under the two evaluation dimensions respectively, and finally obtain the objective weight vector W2.

[0046] S5. Calculate the comprehensive score: Substitute the evaluation results of the two evaluation methods into the weight vector calculation formula to obtain the final comprehensive weight vector W.

[0047] S6. Combine the path planning experimental data and the comprehensive weight vector W to obtain the final evaluation result F.

[0048] The AHP (Analytical Hierarchy Process) proposed in this embodiment is a system analysis method based on the combination of qualitative and quantitative methods. It is a quantitative analysis method that can perform comprehensive evaluation among multiple standards. The core idea of ​​the AHP is to build a multi-level hierarchical model, analyze the relative relationship between indicators at each level, and finally derive a comprehensive weight to make decisions. For the problem of applying this method to the evaluation of drone path planning algorithms, Figure 2-Figure 4 As shown, the specific steps are as follows:

[0049] S21. Establish a hierarchical model;

[0050] When the AHP method is applied to the decision-making problem of evaluating the UAV path planning algorithm, it is necessary to organize and hierarchize the problem and build a hierarchical structure model. The hierarchical structure of the model is divided into three categories:

[0051] The highest layer: This layer contains only one element, which is generally used as the goal or ideal result of the evaluation problem and is called the target layer. In this paper, the final evaluation result of the UAV path planning algorithm is taken as the target layer F.

[0052] Middle layer: This layer contains the criteria, indicators, and sub-criteria that need to be considered to achieve the goal, so it is also called the criterion layer. This layer is composed of several levels and plays the role of the intermediate link in achieving the goal. This article divides the middle criterion layer into two layers, one is the evaluation dimension, and the other is the evaluation indicator, which includes different indicators under each dimension.

[0053] The lowest level: This level contains various measures, decision-making plans and other options to choose from, also known as the measure layer or plan layer. The elements of this level form the basis for the final decision. This paper regards different UAV path planning algorithms as different plan layers.

[0054] According to the evaluation index system of the UAV path planning algorithm designed in this embodiment, taking the overall evaluation framework as an example, a hierarchical structure model of the hierarchical analysis method is constructed as follows: Figure 3 :

[0055] S22. Construct a judgment matrix;

[0056] A quantitative scale is required when comparing the relative importance of two indicators. Therefore, the consistent matrix method is usually used to determine the relationship and weight of N factors. By comparing N indicators pairwise, an N×N judgment matrix A can be obtained, as shown in the following formula:

[0057]

[0058] Among them, a ij It represents the relative importance ratio of the i-th algorithm evaluation index to the j-th one. At the same time, the judgment matrix A is a positive reciprocal matrix, so it satisfies a ij ·a ji = 1. When comparing the importance of evaluation indicators, the importance scale value is usually used to determine the relative importance between two indicators, as shown in Table 1:

[0059] Table 1 Relative importance scale of evaluation indicators

[0060]

[0061]

[0062] like Figure 4 As shown, the present disclosure designs two criterion layers, and judgment matrices need to be constructed separately for different criterion layers.

[0063] The construction method of the judgment matrix is ​​as follows: for the first-level evaluation dimensions, take A1 as the unit and judge the importance of evaluation dimension A2 relative to A1 according to Table 4.2; for the second-level evaluation indicators, they need to be grouped according to the evaluation dimensions first, and then the construction method is the same as the first level.

[0064] S23. Calculate the eigenvector;

[0065] By calculating the maximum eigenvalue λ of the judgment matrix max , we can find the maximum eigenvector W corresponding to each judgment matrix max =(w1, w2, ..., w n ) T For example, the weight vector of the judgment matrix U1 is W max =(w1,w2,w3) T , the corresponding three evaluation index weights are w1, w2, w3, then w1+w2+w3=1.

[0066] S24.Consistency check;

[0067] In order to ensure the reliability of the judgment matrix, a consistency test is required. The consistency degree is tested by calculating the consistency ratio CR of the matrix. The calculation formula is as follows:

[0068]

[0069] When using the AHP method for evaluation, if the calculation result shows that the CR value is less than 0.1, it is considered that the consistency requirements have been met, the judgment matrix is ​​reasonable, and the weight vector can be used for performance evaluation. If the CR value is greater than 0.1, it is necessary to re-evaluate the relative importance of each evaluation indicator, adjust the judgment matrix, and perform consistency test again until the weight vector is reasonable.

[0070] In order to check whether the judgment matrix meets the consistency, the consistency index CI (Consistence Index) is used. This index is used to judge whether the consistency of the judgment matrix is ​​reasonable. The calculation formula is as follows: Where RI represents the consistent random index of the n-order matrix, which is obtained by taking the random consistency table with n ranging from 1 to 9, as shown in Table 2:

[0071] Table 2 Randomness index of n-order matrix

[0072] n 1 2 3 4 5 6 7 8 9 10 RI 0 0 0.58 0.9 1.12 1.24 1.32 1.41 1.45 1.49

[0073] S24..calculate the comprehensive weight;

[0074] The subjective weights of the evaluation dimensions of the first-level algorithm are: A1+A2=1;

[0075] The subjective weights of the evaluation indicators of the algorithm under the second dimension are: U 11 +U 12 +U 13 =1,U 21 +U 22 =1.

[0076] Multiply the weight of the first layer by the specific indicator under the evaluation dimension to obtain the total weight vector of AHP: T AGP ={T1, T2, T3, T4, T5}. Where T n =A i ·U ij , represents the subjective weight value of the first-layer weight and the j-th evaluation index of the second layer under the i-th evaluation dimension.

[0077] The essence of the CRITIC objective weighting method is to determine the objective weight according to the information content of the data, use the standard deviation to represent the comparison intensity, and use the correlation coefficient to measure the conflict between indicators. Finally, the information content of each indicator is obtained by multiplying the comparison intensity and the conflict.

[0078] Suppose that when performing a performance test on an algorithm to be evaluated, n evaluation indicators and p data samples are considered. ij It represents the performance test value of the algorithm for the jth sample under the i-th evaluation index, and X represents the test data matrix of the algorithm on all indicators and samples.

[0079]

[0080] like Figure 5-Figure 6 As shown in Figure 1, the basic steps of the CRITIC objective weighting method include dimensionless processing of different evaluation indicators through forward and reverse transformation, so as to unify the numerical ranges of different evaluation indicators to avoid excessive or insufficient influence of certain indicators in the evaluation system.

[0081] S31. Dimensionless processing;

[0082] The purpose of the forward and reverse normalization in the CRITIC objective weighting method is to unify the numerical ranges of different variables, thereby avoiding the problem of too large or too small impact on some variables when calculating similarity. Dimensionless normalization can make the weights of different indicators equal, avoiding inaccurate evaluation results caused by different normalization levels of indicators.

[0083] For example, in the evaluation index set in this paper, the magnitude of path length is usually 10 2 ~10 3 The maximum success rate will not exceed 100%. If dimensionless processing is not used, the path length index will have a greater impact on the entire index system, while the success rate will have a smaller impact on the entire index system, resulting in inaccurate evaluation results.

[0084] The meaning of positive conversion is that the larger the value of the evaluation index, the better the evaluation result. It is calculated using the following formula:

[0085]

[0086] The meaning of inversion is that the smaller the value of the evaluation index, the better the evaluation result is. It is calculated using the following formula:

[0087]

[0088] In the formula, x′ ij is the dimensionless processed data, x ij represents the original data, x imax represents the maximum value of the algorithm's i-th evaluation index in p data samples, x imin Represents the minimum value of the algorithm's i-th evaluation indicator among p data samples.

[0089] S32.Indicator variability;

[0090] The significance of calculating the variability of indicators is to evaluate the degree of dispersion of indicator values ​​within a certain period of time or in different samples, that is, the stability and reliability of the indicator values. Therefore, the purpose of using the standard deviation to calculate the variability of indicators by the CRITIC weighting method is to measure the degree of deviation of the indicator value from the mean. The smaller the standard deviation, the more stable the indicator value. The calculation formula is as follows:

[0091]

[0092] In the formula, represents the average value of the test data under the i-th evaluation index, S i Represents the standard deviation of the algorithm's test data under this indicator.

[0093] S33. Conflict of indicators;

[0094] There are mutual influences between different evaluation indicators. In order to avoid contradictions when using multiple indicators for comprehensive evaluation, it is necessary to calculate the conflict of indicators. The principle of calculating the conflict of indicators is to measure the linear relationship between different indicators by calculating the correlation coefficient between them, so as to evaluate their mutual influence.

[0095] In the CRITIC weighting method, the method of calculating indicator conflict is to measure the degree of conflict by calculating the correlation coefficient between different indicators. The larger the correlation coefficient, the greater the degree of conflict between different indicators. The calculation formula is as follows:

[0096]

[0097] In the formula, the index conflict coefficient R x Represents the sum of the correlation coefficients between indicator x and other evaluation indicators.

[0098] S34..CRITIC objective weight

[0099] Combining conflict and variability, a comprehensive objective weight value is obtained.

[0100] First, we need to obtain the information content of the evaluation index. The larger the information content, the more important the index is compared with other indexes in the objective evaluation system, and more weight needs to be allocated. The calculation method is as follows:

[0101] C i =S i ×R i

[0102] In the formula, C i Represents the amount of information contained in the i-th indicator.

[0103] Finally, C i Perform normalization to obtain the final objective weight W i , the calculation formula is as follows:

[0104]

[0105] According to the evaluation indicators set in this paper, the objective weight of CRITIC under each evaluation dimension can be obtained.

[0106] The subjective weights calculated by the AHP method are combined with the objective weights calculated by the CRITIC method to form a comprehensive weight vector. The specific steps are as follows;

[0107] Firstly, the subjective weight vector W1 and objective weight vector W2 of the evaluation index are obtained by using the AHP analytic hierarchy process and the CRITIC weighting method. The traditional comprehensive weight method only uses a simple arithmetic average or weighted average method, lacks an effective fusion and correction mechanism, and cannot give full play to the advantages of subjective and objective weights.

[0108] This application introduces the idea of ​​game theory and complements the shortcomings of various weighting methods on the premise of finding a reasonable weight ratio to achieve the purpose of minimizing errors. The innovation of this method is that it is oriented to the different focuses of users in evaluating different task plans. AHP can meet the focus of users, and the CRITIC method can supplement the shortcomings of subjective weighting methods from the perspective of data, making full use of the advantages of subjective and objective weighting methods, coordinating the relationship between subjective and objective weights, and improving the accuracy and reliability of evaluation results.

[0109] First, there are two weighting methods. Each method calculates the subjective weight vector W1 and the objective weight vector W2, with weight ratios of λ1 and λ2 respectively, and obtains the comprehensive weight vector W0. The calculation formula is as follows:

[0110]

[0111] The value of the weight ratio λ is solved and normalized, and the comprehensive weight vector W0 of the performance evaluation index is:

[0112]

[0113] In the present disclosure, the performance is evaluated, and the indicators include: algorithm universality indicators and algorithm specific task indicators;

[0114] The algorithm universality indicators include:

[0115] 1. Algorithm accuracy

[0116] The formula for quantifying the accuracy of the UAV path planning algorithm is: In a certain test function, assuming that the current best search value of the tested algorithm is fiBEST , then the accuracy score of the algorithm in this test function is:

[0117]

[0118] 2. Algorithm time complexity

[0119] The formula for quantifying the time complexity of the UAV path planning algorithm is: In a certain test function, assuming that the current best search value of the random search algorithm is f RMEAN , the running time of the random search algorithm is T RTIME , the random search algorithm calls the objective function N times RNCF , the average running time of the tested algorithm is T iTIME ,, the average accuracy of the tested algorithm is f iMEAN , the average number of times the tested function calls the target function is N iNCF , then the time complexity score of the algorithm in this test function is

[0120]

[0121] 3. Algorithm stability

[0122] The formula for quantifying the stability of the intelligent iterative optimization algorithm is: in a certain test function, assuming that the current best search value of the random search algorithm is f RMEAN , the standard deviation of the accuracy of the tested algorithm is σ iSTD , then the stability score of the algorithm in this test function is

[0123]

[0124] The specific task indicators of the algorithm include:

[0125] 1. Algorithm path security

[0126] The path safety metric mainly refers to the path safety of the path planned by the UAV path planning algorithm, in combination with the characteristics of the UAV path planning, for typical UAV path planning scenarios. In the UAV mountain reconnaissance scenario, we set up an objective function by integrating the turning angle constraint, maximum path constraint, extreme flight altitude constraint, and no-fly zone constraint of the UAV maneuver constraints.

[0127] Turning angle constraint means that although the rotor UAV can change the flight direction freely, due to its inertia, it takes a certain amount of time to change the direction. Under the requirements of the UAV path planning scenario in this article, the UAV is required not to make too large a turn and can only be limited to a certain angle. The turning angle can be calculated as follows.

[0128]

[0129] In the formula, l i represents the projection of the i-th path vector on the horizontal plane, l i =(x i -x i-1 ,y i -y i-1 ) T .

[0130] The maximum path constraint means that the rotorcraft cannot fly a long distance due to its small size. If the planned flight path exceeds this distance, it will exceed the range that the equipment can bear, resulting in mission failure. This constraint can be written as:

[0131] W l =|P o P n |≤L max

[0132] |P0P n |Here refers to the distance between the starting point and the end point, L max Indicates the maximum flight distance.

[0133] The extreme flight altitude constraint means that if the rotorcraft UAV flies too high, it will lose the signal. This constraint limits the maximum flight altitude of the UAV. The constraint can be written as:

[0134] W h =H i ≤H max ,i=1,2,...,n

[0135] In the formula, H i represents the flight height of the UAV in the i-th path segment, H max Indicates the maximum flight altitude.

[0136] The no-fly zone constraint means that the path planned for the reconnaissance drone should be avoided as much as possible in the no-fly zone for safety. The constraint can be written as:

[0137]

[0138] Among them, x i ,i=1,2,...,n represents the planned path, and s represents the vector between the radar area and the origin.

[0139] The objective function for UAV path planning in this project mainly considers the above four constraints. The expression of the objective function is as follows:

[0140] s8=α1W a +α2W l +α3W h +α4Wr

[0141] Where W a is the turning angle, W l is the total path length, W h is the flight altitude, W r is the minimum distance between the path and the radar area, and α1, α2, α3, and α4 are the weight coefficients of each item. After obtaining the objective function value of the UAV path planning algorithm, it is normalized to obtain the path safety measurement index of the UAV path planning algorithm in the typical UAV path planning scenario.

[0142] When the result of the algorithm iteration optimization is lower than the set threshold, it is judged that the optimization result of this UAV path planning algorithm is converged, otherwise the convergence fails. The final UAV path planning algorithm convergence efficiency metric is obtained by counting the proportion of successfully optimized problems in all runs of the algorithm to the total number of run problems.

[0143] 2. Robustness

[0144] Robustness metrics are used to measure the stability of drone path planning algorithms in the face of extreme input conditions. Under the construction of typical drone path planning scenarios, extreme test cases are constructed in three ways: by constructing extreme size maps, extreme obstacle maps, and extreme starting and ending positions. By measuring whether the objective function of the drone path planning algorithm under these extreme test cases is lower than the set threshold, it is judged whether the results have successfully converged. The final robustness metric of the drone path planning algorithm is obtained by counting the proportion of problems that are successfully optimized in all calculation test cases. The calculation formula is as follows, where N l It indicates the number of successful optimization under extreme test cases, and N indicates the total number of extreme tests.

[0145]

[0146] 3. Algorithm convergence efficiency

[0147] The formula for quantifying the convergence efficiency of the UAV path planning algorithm is: In a certain test function, assuming that the current best search value of the random search algorithm is f RMEAN , the random search algorithm calls the objective function N times RNCF The average accuracy of the tested algorithm is f iMEAN , the average number of times the tested function calls the target function is N iNCF , then the convergence efficiency score of the algorithm in this test function is

[0148]

[0149] On the other hand, an embodiment of the present application provides a UAV path planning algorithm performance evaluation system, the system comprising:

[0150] A hierarchical model building module is used to build a hierarchical model based on the analytic hierarchy process (AHP) to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions: algorithm universality indicators and specific task indicators.

[0151] The subjective weight calculation module is used to construct a judgment matrix through the analytic hierarchy process (AHP) and calculate the subjective weight vector of each evaluation index;

[0152] An objective weight calculation module is used to calculate the objective weight vector of each evaluation indicator under each evaluation dimension based on the CRITIC weighting method, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict;

[0153] The comprehensive weight calculation module is used to combine the subjective weight calculated by the AHP method with the objective weight calculated by the CRITIC method to form a comprehensive weight vector.

[0154] The specific execution steps and methods of the above module are the same as those recorded in the above-mentioned UAV path planning algorithm performance evaluation method.

[0155] Furthermore, the system includes a test simulation environment establishment unit; the specific steps of the test simulation environment establishment method are as follows:

[0156] 1. Path planning problem definition:

[0157] 1.1 UAV flight environment;

[0158] (1) Terrain threats

[0159] like Figure 7 As shown in the figure, when a drone performs a mission, it often needs to consider many factors, including obstacle avoidance. The mission requires the drone to fly in a mountainous area, and the probability of hitting the ground becomes one of the important factors. Different terrain features also have different effects on the path planning of the drone, among which mountains are the most significant obstacles. Therefore, in the process of drone path planning, factors such as the height and position of the mountain must be fully considered to ensure that the flight path will not be interfered with by terrain obstacles, so as to successfully complete the mission.

[0160] For the original digital terrain of terrain and objects, the following analytical model is used for simulation, as shown in formula (1.1):

[0161]

[0162] Among them, x and y are the coordinates of the point on the horizontal projection plane; z is the terrain height corresponding to the horizontal plane coordinate point; a, b, c, d, e, and f are constant coefficients. By selecting the values ​​of the constant coefficients, the actual terrain of various rugged terrains can be simulated, which can be used as the terrain and geomorphic information of the mission area that has been explored on the ground.

[0163] In addition, the topography of the mountain can also be expressed by formula (1.2):

[0164]

[0165] Where: n represents the total number of peaks, which is 7; x i ,y i Indicates the center coordinates of the peak; h i is a terrain parameter used to control the height of the mountain; x si ,y si Control the attenuation in both directions and control the slope. See Table 3 for the data of mountain terrain simulation.

[0166] In practice, high-precision DEM terrain modeling is often used to obtain a real terrain model. In order to facilitate simulation implementation, this paper mainly uses simulation methods to generate terrain.

[0168] Table 3 Partial simulation data of topographic peaks

[0169] serial number x(m) y(m) Altitude(m) <![CDATA[x si ,]]> <![CDATA[y si ]]> 1 60 60 50 20 20 2 100 100 50 60 35 3 180 160 80 20 20 4 50 140 70 20 20 5 150 45 65 15 15 6 110 150 54 19 19

[0170] (2) Radar Threat

[0171] This article discusses the problem of radar threats in drone route planning, mainly focusing on the threat of no-fly zone detection equipment deployed in a key part of a mountainous area to drones. When a drone flies along a predetermined path, it may detect radar and other detection equipment in front of it. At this time, the path planning system should make timely adjustments and re-plan a safe path to avoid the detection range and ensure the completion of the mission.

[0172] In the actual environment, radar threats and other terrains similar to the radar range are processed, and the establishment of the mathematical model of radar requires parameters related to the radar range and effectiveness. Since this paper mainly discusses the mechanism of genetic algorithm to solve the path planning problem, in order to simplify the calculation, the radar threat is simplified into a cylindrical no-fly zone, such as Figure 8 As shown:

[0173] 1.2 UAV dynamics modeling:

[0174] (1) Turning angle constraint

[0175] Compared with the flight requirements of fixed-wing UAVs, rotary-wing UAVs can adapt to more terrains due to their vertical take-off and landing characteristics, and have lower requirements for flight paths. Therefore, this article considers the model of rotary-wing UAVs, and the dynamic constraints of the UAV only consider the constraints of the maximum yaw angle.

[0176] like Fig. 9 As shown, although the rotor UAV can change the flight direction freely, due to its inertia, it takes a certain amount of time to change the direction. Under the military scenario requirements of this article, the UAV is required not to make too large a turn and can only be limited to a certain angle. The turning angle can be calculated by the following formula 1.4.

[0177]

[0178] In the formula, l i represents the projection of the i-th path vector on the horizontal plane, l i =(x i -x i-1 ,y i -y i-1 ) T

[0179] (2) Maximum path constraint

[0180] This constraint limits the maximum flight distance of the drone. If the planned flight path exceeds this distance, it will exceed the range that the device can withstand, resulting in mission failure. This constraint can be written as shown in formula 1.5:

[0181] W l =|P0P n |≤L max (1.5)

[0182] In the formula, |P0P n |Here refers to the distance between the starting point and the end point, L max Indicates the maximum flight distance.

[0183] (3) Limit flight altitude constraints

[0184] If the drone flies too high, it will lose the signal. This constraint determines the maximum altitude that the drone can fly. The constraint can be written as shown in formula 1.6:

[0185] W h =H i ≤H max , i=1,2...n (1.6)

[0186] In the formula, H i represents the flight height of the UAV in the i-th path segment, H max Indicates the maximum flight altitude.

[0187] (4) Objective function

[0188] The objective function of the UAV path planning in this paper mainly considers the above three constraints. The expression of the objective function is as follows:

[0189] W=w1W a +w2W l +w3W h (1.7)

[0190] Where W a is the turning angle, W l is the total path length, W h is the flight altitude, w1, w2, w3 are the weight coefficients of each item, which are taken as 0.2, 0.4, and 0.4 respectively in this paper.

[0191] 2. Algorithm testing:

[0192] 2.1 Control Points

[0193] In the path planning problem defined in this paper, the starting point and the end point of the drone are given. A path can be obtained by solving several points between the starting point and the end point and connecting them into a continuous and smooth curve through fitting methods. These points are called control points.

[0194] This paper regards the individuals to be solved in the algorithm as control points in three-dimensional space, that is, one chromosome corresponds to one control point. Obviously, the more chromosomes there are and the more control points there are, the more likely the three-dimensional path generated will be close to the theoretical optimal solution.

[0195] A set of coordinate points in space contains three coordinate values ​​(x, y, z). In order to reduce the dimension of calculation, each chromosome only considers two values, y and z, and x is evenly distributed between the starting point and the end point. Assuming that there are N control points between the starting point and the end point, the dimension of the chromosome is 2N.

[0196] like Fig.10 As shown, a portion of the control points solved by the algorithm are listed.

[0197] 2.2 Path Solution

[0198] After the path planning algorithm solves the control points, it uses cubic spline interpolation to obtain a smooth path curve based on the given starting and ending points of the problem, such as Fig.11 A simulation schematic diagram of a path curve is shown.

[0199] After obtaining the path curve, it is necessary to determine whether the curve passes through a mountain or a no-fly zone. By comparing the z coordinate of the point on the curve with the mountain area, if it is greater than the z coordinate, it proves that the mountain has not been passed; calculate the distance between the point on the curve and the center of the no-fly zone at the same height. If the calculated distance is greater than the radius of the no-fly zone, it proves that the no-fly zone has not been entered.

[0200] Through the above method system according to the UAV path planning algorithm testing process, seven common intelligent iterative optimization algorithms, including atomic search algorithm, artificial ecosystem optimization algorithm, particle swarm optimization algorithm, genetic algorithm, gray wolf algorithm, black widow optimization algorithm, and whale optimization algorithm, are tested according to the evaluation indicators established above, and good results can be obtained.

[0201] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0202] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for evaluating the performance of an unmanned aerial vehicle path planning algorithm, characterized in that: The method comprises the following steps: Based on the analytic hierarchy process (AHP), a hierarchical model is constructed to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions, an algorithm universality indicator and a specific task indicator. A judgment matrix is ​​constructed through the analytic hierarchy process (AHP) to calculate the subjective weight vector of each evaluation indicator. Based on the CRITIC weighting method, the objective weight vector of each evaluation indicator under each evaluation dimension is calculated, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict; The subjective weights calculated by the AHP method are combined with the objective weights calculated by the CRITIC method to form a comprehensive weight vector.

2. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The method comprises determining evaluation indicators, wherein the evaluation indicators comprise average running time of the algorithm, optimization success rate, number of iterations, path length, and path security.

3. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The hierarchical model includes three layers: The highest layer is the target layer, which is used for the final evaluation results of the UAV path planning algorithm; The middle layer is the criteria layer, which is used to divide the evaluation dimensions; The bottom layer is the measure layer, which is used to include different UAV path planning algorithm solutions.

4. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The judgment matrix in the AHP method is obtained by constructing a consistent matrix, and two indicators are compared in pairs according to an importance scale, where the importance scale includes: equally important, slightly important, obviously important, very important, and extremely important.

5. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The calculation of the subjective weight vector includes performing consistency check on the judgment matrix, calculating consistency index and consistency ratio, and ensuring the rationality of the judgment matrix; If the consistency ratio is less than 0.1, the judgment matrix is ​​considered to meet the consistency requirements, otherwise the judgment matrix needs to be readjusted and tested again.

6. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The CRITIC method performs dimensionless processing on evaluation indicators, and performs dimensionless processing on different evaluation indicators through forward and reverse processing, so as to unify the numerical ranges of different evaluation indicators, so as to avoid that some indicators have too large or too small influence in the evaluation system.

7. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The steps of forming the comprehensive weight vector are as follows: The AHP and CRITIC weight vectors are obtained. The subjective weight vector W1 of each evaluation index is determined by AHP, and the objective weight vector W2 of each evaluation index is determined by CRITIC weighting method. The calculation of weight ratio, introduces the idea of ​​Bo Lun, combines the subjective weight of the hierarchical analysis method and the objective weight of the critics, and determines the weight ratios λ1 and λ2, where λ1 is the ratio of subjective weight, λ2 is the ratio of objective weight, and the comprehensive weight vector W0 of performance evaluation index is; 8. The method for evaluating the performance of a UAV path planning algorithm according to claim 1, characterized in that: The method evaluates the performance of the UAV path planning algorithm through multiple indicators, including: Algorithm universality index; Precision, measured by the difference between the current best search value of the algorithm under test and the reference search value; Time complexity, measured based on the algorithm's running time and the number of target function calls; Stability, evaluated based on the standard deviation of the results of the tested algorithm in different test functions; Algorithm specific task indicators; Path safety: Combined with the turning angle constraint, maximum path constraint, flight altitude constraint and no-fly zone constraint of the UAV, the objective function constructed by the weight coefficient is used to evaluate the safety of the planned path; Robustness, which measures the performance of the algorithm under extreme conditions through extreme test cases, such as extreme map scales, obstacle configurations, and start and end locations; Convergence efficiency is measured based on the ratio of the convergence accuracy of the algorithm being tested to the number of times the objective function is called.

9. A UAV path planning algorithm performance evaluation system, characterized in that: The system comprises: A hierarchical model building module is used to build a hierarchical model based on the analytic hierarchy process (AHP) to classify and hierarchize the evaluation indicators. The hierarchical model includes two evaluation dimensions: algorithm universality indicators and specific task indicators. The subjective weight calculation module is used to construct a judgment matrix through the analytic hierarchy process (AHP) and calculate the subjective weight vector of each evaluation index; An objective weight calculation module is used to calculate the objective weight vector of each evaluation indicator under each evaluation dimension based on the CRITIC weighting method, and the objective weight vector is obtained by calculating the indicator variability and indicator conflict; The comprehensive weight calculation module is used to combine the subjective weight calculated by the AHP method with the objective weight calculated by the CRITIC method to form a comprehensive weight vector.

10. The UAV path planning algorithm performance evaluation system according to claim 1, characterized in that: The system includes a test simulation environment establishment unit, and the test simulation environment establishment unit includes: A terrain threat simulation module, used to simulate terrain obstacles in UAV flight missions, wherein the terrain threat simulation is represented by an analytical model, and the analytical model is calculated based on the point coordinates of the horizontal projection plane and the terrain height to simulate the mountainous terrain; The radar threat simulation module is used to simulate the radar no-fly zone in the UAV flight path. The radar no-fly zone is modeled as a cylindrical area, which represents the threat range of the radar detection equipment. Multiple no-fly zones are set to simulate the distribution and detection radius of the radar equipment. UAV dynamics constraint module, including: Turning angle constraints, maximum path length constraints, and flight altitude constraints; The objective function evaluation module is used to comprehensively consider the turning angle, path length and flight altitude of the UAV, generate the objective function according to the weight coefficient of each indicator, and evaluate the path planning results.

Images