Dynamic vector collaborative unmanned aerial vehicle path planning system based on chaos and differential disturbance

Abstract

The invention relates to the field of unmanned aerial vehicle autonomous navigation and path planning, and particularly discloses a dynamic vector collaborative unmanned aerial vehicle path planning system based on chaos and differential disturbance, and the system consists of a path comprehensive constraint model module, a chaos initialization module and a path optimization module. The path comprehensive constraint model module integrates path length, obstacle threat, height and smoothness indexes to construct a weighted objective function, and establishes a mathematical optimization model. The chaos initialization module generates uniformly distributed initial populations by utilizing piecewise linear traversal mapping. The path optimization module adopts a dynamic vector collaboration method fusing vector differential disturbance, and introduces differential variation, crossover and selection operators to carry out secondary disturbance updating while simulating collaboration behaviors such as rejection, synchronization and condensation among individuals to carry out basic optimization. According to the invention, the problem that a traditional algorithm is easy to fall into local optimum is effectively solved, and efficient autonomous navigation of the unmanned aerial vehicle in a complex three-dimensional environment is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of unmanned aerial vehicle autonomous navigation and path planning, in particular to a dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and differential disturbance. BACKGROUND

[0002] In the technical field of unmanned aerial vehicle autonomous navigation and path planning, with the continuous expansion of unmanned aerial vehicle application scenarios, autonomous flight tasks in complex environments face multiple challenges such as three-dimensional space obstacle avoidance, kinematic constraint satisfaction, and real-time requirements. Traditional path planning methods such as A* algorithm and RRT method have low computational efficiency and poor path smoothness in three-dimensional environments, making it difficult to meet actual flight requirements.

[0003] In order to improve planning quality and real-time obstacle avoidance capability, in recent years, swarm intelligence algorithms based on biological group cooperative behavior (such as particle swarm algorithm, dragonfly algorithm, etc.) have become a research hotspot. However, existing swarm intelligence path planning methods still have significant limitations: Low initialization quality: existing works usually use pseudo-random functions to generate initial populations, resulting in uneven distribution of initial paths in three-dimensional solution space, low coverage, and high search blindness, which seriously affects the convergence speed of the algorithm in the early stage; Easily fall into local optimum: traditional algorithms often only use simple cooperative updating mechanism, when facing high-dimensional complex space, dense obstacles or dynamic obstacles, the population diversity is lost too quickly, easily falling into local extreme value, resulting in the planned path is not the global optimum, even the obstacle avoidance fails.

[0004] In summary, the existing unmanned aerial vehicle path planning technology has problems such as large initialization blindness, easy to fall into local optimum, and difficulty in handling kinematic constraints in complex three-dimensional environments, and there is an urgent need for a path planning system with fast convergence speed, high solution quality, and strict satisfaction of flight constraints to realize efficient autonomous navigation of unmanned aerial vehicles in complex environments. SUMMARY

[0005] The present application provides a dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and differential disturbance to solve the problems of large initialization blindness, easy to fall into local optimum, and difficulty in handling kinematic constraints in complex environments, and improve the accuracy and efficiency of the solution.

[0006] To achieve the above purpose, the present application proposes a dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and differential disturbance, which includes: A path synthesis constraint model module, a chaos initialization module and a path optimization module realize the global optimal planning of unmanned aerial vehicle path in complex three-dimensional environment through the three-dimensional integrated architecture of chaos initialization-constraint modeling-cooperative optimization; The path synthesis constraint model module is configured to construct a global cost evaluation function of path planning, to quantitatively calculate path length indicators, obstacle threat indicators, height constraint indicators and smoothness indicators, and to assign weight coefficients to the indicators respectively, and to construct a multi-constraint objective function by weighted summation. The chaotic initialization module is configured to generate a normalized chaotic variable sequence by using piecewise linear traversal mapping before the start of path optimization iteration, to execute a random disturbance mechanism in the iteration process to avoid the chaotic variable falling into a small cycle of a set of fixed points, and to combine boundary constraints of a three-dimensional search space to convert the chaotic variable sequence into three-dimensional coordinates of a UAV path point by carrier mapping, and to output an initial dynamic vector population with uniform distribution. The path optimization module is connected to the path synthesis constraint model module and the chaotic initialization module respectively, receives the initial dynamic vector population, and guides the iteration and update of the path by using a vector differential disturbance enhanced dynamic vector cooperation method with the minimization of the objective function as the guide, wherein the iteration and update process includes a cooperative search stage and a vector differential disturbance stage, in the cooperative search stage, basic optimization is performed by simulating behaviors such as repulsion, synchronization and condensation among individuals, in the vector differential disturbance stage, mutation, crossover and selection operators of differential evolution are introduced to perform secondary disturbance and update of the dynamic vector, the cooperative search stage and the vector differential disturbance stage of the cooperative iteration and update process are repeated until a termination condition is met to output an optimal path.

[0007] Preferably, the path length indicator represents the cumulative Euclidean distance of the UAV from the starting point to the ending point, the obstacle threat indicator represents the minimum distance between the path segment and the projection center of the obstacle, the height constraint indicator represents the degree of deviation of the flight height from the safety interval, and the smoothness indicator represents the change rate of the turning angle of the path in the horizontal plane and the climbing angle in the vertical plane.

[0008] Preferably, the path synthesis constraint model module takes the minimization of the path length as the optimization objective, integrates the path length indicator, the obstacle threat indicator, the height constraint indicator and the smoothness indicator, and assigns different weights to the indicators to form a complete objective function, thereby constructing a complete mathematical optimization model and realizing global optimization and dynamic adjustment of the UAV flight path.

[0009] Preferably, the specific execution steps of the path synthesis constraint model module include: calculating the path length cost: the UAV flight path for a sequence of n path points, the three-dimensional coordinates of the path points are the straight-line distance between two points in the three-dimensional space is calculated by using the Euclidean distance formula of adjacent path points to obtain a path length cost function; wherein the Euclidean distance formula of adjacent path points is: ; The path length cost function is: ; Calculate the obstacle threat cost: Let the obstacle projection center be... The radius of the region is Calculate path segments To the obstacle Projection Center Proportional coefficient, shortest distance The threat cost of a single path segment is calculated by dividing the path into segments based on the value of t. The total threat cost function is obtained by segmenting the path segments according to distance thresholds; where, path segments To the obstacle Projection Center The formula for calculating the proportionality coefficient is: ; The total threat cost function is: ; In the formula, The number of path points. This represents the total number of obstacles. Calculating altitude costs: Flight altitude must meet certain requirements. Single path point High cost By height Whether the calculation is performed in segments within the safe zone, the total height cost function is: ; Calculating the smoothness cost: Turning angle and climb angle are two key indicators for measuring path smoothness. The smoothness cost function is obtained by calculating the turning angle and climb angle and combining them; where the turning angle... The calculation formula is: ; In the formula, Path segment arrive The horizontal projection vector, Path segment arrive The horizontal projection vector is given by the formula, where the numerator is the cross product modulus of the horizontal projection vectors of the two path segments, representing the change in direction of the turn, and the denominator is the dot product of the horizontal projection vectors of the two path segments, representing the similarity of the directions of the path segments. Angle of climb The calculation formula is: ; In the formula, and flight height at path point and flight height at path point projection length of path segment on horizontal plane The smoothness cost function calculation formula is: ; In the formula, the first term is the cumulative sum of all turning angles , which measures the smoothness of the overall direction of the path, is the penalty coefficient of the turning angle cost; the second term is the change between consecutive climb angles, which measures the smoothness of the change in path height, is the penalty coefficient of the climb angle cost; Construct the total objective function: combine the multi-dimensional costs including path length, threat cost, flight height cost and smoothness cost to obtain the total objective function, and the calculation formula is: ; In the formula, is the weight coefficient of each cost, which is used to balance the importance of each cost in the overall optimization.

[0010] Preferably, the specific execution steps of the chaotic initialization module include: According to the preset population size and dynamic vector dimension, a normalized chaotic variable sequence is iteratively generated by using a piecewise linear traversal mapping formula; In the iteration, a random disturbance mechanism is executed, and when the chaotic variable falls into a preset small period cycle fixed point set, a random small amount is added to the chaotic variable to jump out of the cycle; The normalized chaotic variable sequence generated by iteration is mapped to a three-dimensional search space composed of the boundaries of the unmanned aerial vehicle flight area and the flight height limit by using a carrier mapping formula, to generate initial path point coordinates.

[0011] Preferably, the specific execution steps of the path optimization module include: In the cooperative search stage, the repulsion, synchronization, condensation, optimal solution attraction, and worst solution avoidance vectors of each dynamic vector of the population are calculated, the above vectors are weighted and summed to obtain a velocity update increment by combining the preset inertia weight and the behavior coefficients, and the current position of the dynamic vector is updated by using the velocity update increment to generate a preliminary position of the dynamic vector; In the mutation stage in the vector difference disturbance, three mutually different individual vectors are randomly selected from the current population, and a mutation vector is generated by using a vector difference formula to represent a preliminary position vector; In the crossover stage in the vector difference disturbance, a trial vector is generated by performing a binomial crossover operation on the generated mutation vector and the preliminary position vector generated in the first step; In the greedy selection stage of the vector differential perturbation, the path synthesis constraint model module is called to calculate the objective function fitness value of the trial vector and the prepared position vector respectively; if the fitness value of the trial vector is better than that of the prepared position vector, the trial vector is selected as the next generation path position, otherwise the prepared position vector is retained as the next generation path position; The above two stages are repeated until the termination condition is met, and the optimal path is output.

[0012] Preferably, the iteration formula of the piecewise linear traversal mapping is: ; In the formula, is the mapping number, is the th iteration of the chaotic variable of the th path point on the th path; The calculation formula for generating the initial path point coordinates is: ; In the formula, is the th initial position of the th dynamic vector in the th dimension, and are the upper limit and lower limit of the search space in the th dimension, respectively.

[0013] Preferably, the dynamic vector speed and position update in the cooperative search stage has a clear calculation formula, wherein the speed update formula is: ; The position update formula is: ; In the formula, is the speed of the dynamic vector at the path point , is the inertia weight, is the repulsion vector, is the repulsion coefficient, is the synchronization vector, is the synchronization coefficient, is the condensation vector, is the condensation coefficient, is the optimal solution attraction vector, is the optimal solution coefficient, is the worst solution avoidance vector, is the worst solution avoidance coefficient, is the speed of the dynamic vector at the path point coordinate representation.

[0014] Preferably, the mutation, crossover, selection operation of the vector difference perturbation stage has a clear calculation formula, wherein the mutation operation formula is: ; The crossover operation formula is: ; The selection operation formula is: ; In the formula, is a scaling factor, and the value range is , is a crossover probability, is a random number between , is a randomly selected dimension index.

[0015] Preferably, the path optimization module comprises optimal and worst position update rules, and has a complete algorithm flow, and the total algorithm flow is to obtain initial planning information and initialize parameters, call the chaotic initialization module to generate an initial population, calculate the initial fitness and set the initial optimal or worst position, iteratively execute the cooperative search and difference perturbation operation, update the fitness and the optimal or worst position, and output the optimal path after reaching the maximum function evaluation times. Among them, the local optimal position The update rule is: ; The global optimal position is determined by the dynamic vector position with the optimal fitness, and is the path with the optimal fitness among all local optimal positions: ; The local worst position The update rule is: ; The global worst position is determined by the dynamic vector position with the worst fitness, and is the path with the worst fitness among all local worst positions: .

[0016] Therefore, the application proposes a dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and difference perturbation, which has the following beneficial effects: (1) Significantly improve the global search ability and obstacle avoidance robustness, through the two-stage mechanism of cooperative guidance and difference correction, both realize fast convergence by means of cooperative behavior, and maintain population diversity by means of difference perturbation, effectively jump out of local optimum, and improve the path planning success rate and solution quality.

[0017] (2) solve the blindness of traditional random initialization, generate an initial population with uniform distribution by using piecewise linear traversal mapping, supplemented by random disturbance to avoid small cycle, provide high-quality initial solution for subsequent optimization, greatly speed up the convergence speed of the algorithm.

[0018] The technical solutions of the present application will be further described in detail below with the help of the accompanying drawings and examples. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a structural schematic diagram of the dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and differential disturbance of the present application. DETAILED DESCRIPTION

[0020] In order to make the technical solutions, advantages and purposes of the present application clearer, the technical solutions of the embodiments of the present application will be described clearly and completely below. The described embodiments are part of the embodiments of the present application, not all the embodiments. Based on the described embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor belong to the protection scope of the present application.

[0021] Unless otherwise defined, the technical terms or scientific terms used in the present application shall be understood as the usual meaning understood by those skilled in the art in the field to which the present application belongs.

[0022] As shown in Figure 1 The dynamic vector cooperative unmanned aerial vehicle path planning system based on chaos and differential disturbance provided by the present application comprises a path synthesis constraint model module, a chaos initialization module and a path optimization module.

[0023] The path synthesis constraint model module is used to build a global cost evaluation function of path planning, to quantitatively calculate path length indicators, obstacle threat indicators, height constraint indicators and smoothness indicators, and to respectively assign weight coefficients to them, and to build a multi-constraint objective function by weighted summation.

[0024] The path length indicator represents the cumulative Euclidean distance of the unmanned aerial vehicle from the starting point to the end point, the obstacle threat indicator represents the minimum distance between the path segment and the projection center of the obstacle, the height constraint indicator represents the degree of deviation of the flight height from the safety interval, and the smoothness indicator represents the change rate of the turning angle of the path in the horizontal plane and the climbing angle in the vertical plane.

[0025] The path synthesis constraint model module takes the minimization of path length as the optimization objective, calculates the path length indicators, the obstacle threat indicators, the height constraint indicators and the smoothness indicators, and gives them different weights, organically fuses them to form a complete objective function, builds a complete mathematical optimization model, and realizes the global optimization and dynamic adjustment of the unmanned aerial vehicle flight path.

[0026] The specific execution steps of the path synthesis constraint model module include: Calculate path length cost: UAV flight path For a sequence of n path points, the three-dimensional coordinates of the path points are , the straight-line distance between two points in three-dimensional space is calculated using the Euclidean distance formula of adjacent path points to obtain a path length cost function; wherein the Euclidean distance formula of adjacent path points is: ; The path length cost function is: ; Calculate obstacle threat cost: Let the obstacle projection center be , the area radius be , the path segment to the obstacle projection center proportional coefficient, and the shortest distance be calculated by segmenting according to the value of t, and the threat cost of a single path segment be defined by segmenting according to the distance threshold to obtain a total threat cost function; wherein the path segment to the obstacle projection center proportional coefficient calculation formula is: ; The calculation formula of the shortest distance is: ; In the formula, when , it indicates that the projection center is before the starting point of the path segment; when , it indicates that the projection center is inside the path segment; when , it indicates that the projection center is after the end point of the path segment; wherein is the projection point of the obstacle projection center on the path segment, and the calculation formula is: ; Then the threat cost of a single path segment to the obstacle is defined as: ; In the formula, is the obstacle buffer distance, which is determined by the UAV application environment, UAV operation accuracy, GPS signal quality and other factors, is the diameter of the UAV; is the total threat cost function: wherein, is the number of path points, is the total number of obstacles; Calculate the height cost: the flight height needs to meet , the single path point height cost is segmented according to the height whether in the safety interval, the total height cost function is: Calculate the smoothness cost: the turning angle and the climbing angle are two key indicators to measure the smoothness of the path. The turning angle and the climbing angle are calculated, and the smoothness cost function is obtained by comprehensively considering the turning angle and the climbing angle; wherein, the turning angle The calculation formula of the turning angle is: wherein, is the horizontal projection vector of the path segment to , and is the horizontal projection vector of the path segment to , wherein the numerator is the cross product length of the horizontal projection vectors of the two path segments, indicating the direction change amount of the turning, and the denominator is the dot product of the horizontal projection vectors of the two path segments, indicating the similarity of the path segment direction; The calculation formula of the climbing angle is: wherein, and are the flight heights at the path points and respectively; is the projection length of the path segment in the horizontal plane; The calculation formula of the smoothness cost function is: wherein, the first term is the cumulative sum of all turning angles , which measures the smoothness of the overall direction of the path, is the penalty coefficient of the turning angle cost; the second term is the change amount between the consecutive climbing angles, which measures the smoothness of the height change of the path, is the penalty coefficient of the climbing angle cost; ​​​​​The total target function is constructed by fusing the multi-dimensional cost including path length, threat cost, flight height cost and smoothness cost, and the calculation formula of the total target function is: ; In the formula, is a weight coefficient of each cost, which is used to balance the importance of each cost in the overall optimization.

[0027] The chaotic initialization module is configured to, before starting the path optimization iteration, generate a normalized chaotic variable sequence by using piecewise linear ergodic mapping, and combine the boundary constraint of the three-dimensional search space to convert the chaotic variable sequence into three-dimensional coordinates of the UAV path point by carrier mapping, so as to output an initial dynamic vector population.

[0028] The specific execution steps of the chaotic initialization module include: According to the preset population size and dynamic vector dimension, a normalized chaotic variable sequence is generated by using the piecewise linear ergodic mapping formula; the iteration formula of the piecewise linear ergodic mapping is: ; In the formula, is the number of mapping, is the chaotic variable generated by the i th iteration of the j th path point on the i th path, is the chaotic variable generated by the i th iteration of the j th path point on the i th path, is the chaotic variable generated by the i th iteration of the j th path point on the i th path, is the chaotic variable generated by the i th iteration of the j th path point on the i th path; In the iteration, a random disturbance mechanism is executed, that is, when , let , wherein is a very small random number, and when the chaotic variable falls into a preset small periodic cycle fixed point set, a random small amount is added to the chaotic variable to jump out of the cycle. The normalized chaotic variable sequence generated by iteration is mapped to a three-dimensional search space composed of the flight area boundary and the flight height limit by using a carrier mapping formula, to generate initial path point coordinates.

[0029] The calculation formula for generating the initial path point coordinates is: ; In the formula, is the initial position of the i th dynamic vector in the j th dimension, is the initial position of the i th dynamic vector in the j th dimension, is the upper limit and the lower limit of the search space in the j th dimension, respectively. is the upper limit and the lower limit of the search space in the j th dimension, respectively.

[0030] ​​The path optimization module is connected to the path comprehensive constraint model module and the chaos initialization module, respectively. It receives the initial dynamic vector population and, guided by minimizing the objective function, it iteratively updates the path using a vector difference perturbation-enhanced dynamic vector cooperative method. The iterative update process includes a cooperative search phase and a vector difference perturbation phase. In the cooperative search phase, basic optimization is performed by simulating behaviors such as repulsion, synchronization, and cohesion among individuals. In the vector difference perturbation phase, mutation, crossover, and selection operators of differential evolution are introduced to perform secondary perturbation updates on the dynamic vectors. The cooperative search phase and vector difference perturbation phase of the cooperative iterative update process are repeated until the termination condition is met and the optimal path is output.

[0031] The specific execution steps of the path optimization module include: During the collaborative search phase, the repulsion, synchronization, cohesion, optimal solution attraction, and worst solution avoidance vectors of each dynamic vector in the population are calculated. Combined with preset inertia weights and various behavior coefficients, the above vectors are weighted and summed to obtain the velocity update increment. The current position of the dynamic vector is updated using the velocity update increment to generate the preliminary position of the dynamic vector. In the mutation stage of vector difference perturbation, three distinct individual vectors are randomly selected from the current population, and the vector difference formula is used to generate a preliminary position vector representing the mutation vector. In the crossover phase of vector difference perturbation, the generated mutation vector is subjected to a binomial crossover operation with the pre-position vector generated in the first step to generate the test vector. In the greedy selection stage of vector difference perturbation, the path synthesis constraint model module is called to calculate the objective function fitness values ​​of the test vector and the preliminary position vector respectively; if the fitness value of the test vector is better than that of the preliminary position vector, the test vector is selected as the next generation path position; otherwise, the preliminary position vector is retained as the next generation path position. Repeat the above two stages until the termination condition is met, and output the optimal path.

[0032] The dynamic vector velocity and position updates during the cooperative search phase have explicit calculation formulas, with the velocity update formula being: ; The position update formula is: ; In the formula, For dynamic vectors At the waypoint speed, For inertial weights, For repulsion vectors, The repulsion coefficient, For synchronization vectors, is the synchronization coefficient, is the cohesion vector, is the cohesion coefficient, is the optimal solution attraction vector, is the optimal solution coefficient, is the worst solution repulsion vector, is the worst solution repulsion coefficient, is the dynamic vector is the coordinate representation of the path point .

[0033] where, repulsion vector calculation formula: ; synchronization vector calculation formula: ; cohesion vector calculation formula: ; optimal solution attraction vector calculation formula: ; where is the optimal position of the th path point in all paths.

[0034] worst solution repulsion vector calculation formula: ; where is the worst position of the th path point in all paths.

[0035] The mutation, crossover, and selection operations in the vector difference perturbation stage have explicit calculation formulas, where the mutation operation formula is: ; The crossover operation formula is: ; The selection operation formula is: ; In the formula, is the scaling factor, which takes a value in the range of , is the crossover probability, is a random number between , is a randomly selected dimension index.

[0036] The path optimization module includes best and worst position update rules and has a complete algorithm flow. The overall algorithm flow is as follows: obtain initial planning information and initialize parameters, call the chaos initialization module to generate the initial population, calculate the initial fitness and set the initial best or worst position, iteratively execute cooperative search and differential perturbation operations, update fitness and best or worst position, until the maximum number of function evaluations is reached and the optimal path is output. Among them, the local optimal position The update rules are as follows: ; Global optimal position The path is determined by the fitness-optimal dynamic vector position, and is the fitness-optimal path among all local optima: ; Local worst position The update rules are as follows: ; The global worst position is determined by the position of the dynamic vector with the worst fitness, and it is also the path with the worst fitness among all local worst positions: .

[0037] Example 1 This embodiment takes a UAV power line inspection task in a city airspace as an application scenario, and details the specific implementation process of a dynamic vector cooperative UAV path planning system based on chaos and differential perturbation, and verifies the path planning effect of the system in a complex three-dimensional environment.

[0038] I. Setting Parameters for Implementation Scenarios: Airspace range: The three-dimensional airspace for UAV flight is defined as a horizontal range of [0,1000]m × [0,1000]m, and a safe altitude range of h. min =50m, h max =150m, the starting coordinates of the drone are P0=(0,0,80)m, and the ending coordinates are P n =(1000,1000,80)m, total number of path points n=10.

[0039] Obstacle parameters: Three cylindrical obstacles are set in the airspace. The projection center of obstacle 1 is (300, 300), and the area radius is r. o =50m, obstacle 2 projection center (600,600), area radius r o =60m, obstacle 3 projection center (800,200), area radius r o =40m; UAV diameter R=1m, obstacle buffer distance d b =5m.

[0040] Algorithm parameters: Chaos initialization module population size N = 30, piecewise linear ergodic mapping parameters a = 0.4, iteration number k = 100; Path optimization module inertia weight 0.7, repulsion coefficient 0.2, synchronization coefficient 0.3, condensation coefficient 0.2, optimal solution coefficient 0.5, worst solution avoidance coefficient 0.4; differential perturbation scaling factor F ∈ [0.5, 0.8], crossover probability CR is 0.6; The path length weight coefficient of the objective function is 0.3, the threat cost weight coefficient is 0.4, the flight height cost weight coefficient is 0.15, and the smoothness cost weight coefficient is 0.15, the turning angle penalty coefficient is 0.8, and the climbing angle penalty coefficient is 0.6.

[0041] II. System module execution steps: Path synthesis constraint model module execution: (1) Path length cost calculation: the to-be-planned path is expressed as 11 three-dimensional path point sequence P = {P0, P1, …, P 10}, the Euclidean distance between adjacent path points is calculated according to the formula: ; Then the path length cost is obtained through

[0042] (2) Obstacle threat cost calculation: calculate the proportional coefficient t of each path segment to the obstacle projection center, determine the projection point coordinates to get the shortest distance, then calculate the single segment threat cost, and finally accumulate the total threat cost.

[0043] (3) Height cost calculation: calculate the height cost of a single path point, and accumulate to get the total height cost.

[0044] (4) Smoothness cost calculation: calculate the turning angle and climbing angle of each path segment respectively, and then get the smoothness cost.

[0045] (5) Total objective function construction: construct the global cost evaluation function.

[0046] Chaos initialization module execution: (1) According to the piecewise linear ergodic mapping formula, the normalized chaotic variable sequence is generated by iteration. When it is monitored that the chaotic variable falls into the small cycle fixed point set, the perturbation operation is performed.

[0047] (2) Through the carrier mapping formula, the chaotic variable is mapped to the three-dimensional search space to generate 30 initial dynamic vector populations (i.e. 30 initial paths), and is passed to the path synthesis constraint model module to complete the initial fitness calculation.

[0048] Path optimization module execution: (1) Cooperative search stage: calculate the velocity update increment according to the formula, and then update the initial position of the dynamic vector.​

[0049] (2) Vector difference perturbation stage: first generate mutation vector, then complete crossover operation, and finally determine the next generation path through greedy selection.

[0050] (3) Repeat the above two-stage iteration until the maximum function evaluation number is reached, and output the global optimal path.

[0051] III. Implementation result verification: After iteration optimization, the optimal path output by the system meets the high safety interval constraint, and the shortest distance between all path segments and obstacles is greater than , the path turning angle and climbing angle change smoothly, the total objective function value is lower than that of the traditional particle swarm algorithm, the algorithm convergence speed is improved, and the safe and efficient planning of the UAV path in the complex airspace is realized.

[0052] Therefore, the present application provides a dynamic vector cooperative UAV path planning system based on chaos and differential perturbation, which is composed of three modules of path synthesis constraint model, chaos initialization and path optimization, and can realize global optimal planning of UAV path in complex three-dimensional environment, solve the blindness problem of traditional initialization, realize preliminary optimization, jump out of local optimum, and finally output efficient autonomous navigation path meeting safety and task efficiency.

[0053] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application and not to limit them, although the present application has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that: it can still modify or equivalently replace the technical solutions of the present application, and these modifications or equivalent replacements cannot make the modified technical solutions deviate from the spirit and scope of the technical solutions of the present application.

Claims

1. A dynamic vector cooperative UAV path planning system based on chaos and differential perturbation, characterized in that, The system comprises a path comprehensive constraint model module, a chaos initialization module and a path optimization module, and realizes global optimal planning of a UAV path in a complex three-dimensional environment through a three-dimensional integrated architecture of chaos initialization-constraint modeling-collaborative optimization. The path comprehensive constraint model module is configured to construct a global cost evaluation function of path planning, to quantitatively calculate path length indexes, obstacle threat indexes, height constraint indexes and smoothness indexes, and to assign weight coefficients to the indexes respectively, and to construct a multi-constraint objective function through weighted summation. The chaos initialization module is configured to generate a normalized chaos variable sequence through piecewise linear traversal mapping before the start of path optimization iteration, to execute a random disturbance mechanism in the iteration process to avoid the chaos variable from falling into a small cycle of a fixed point set, and to combine boundary constraints of a three-dimensional search space to convert the chaos variable sequence into three-dimensional coordinates of UAV path points through carrier mapping, and to output an initial dynamic vector population with uniform distribution. The path optimization module is connected to the path comprehensive constraint model module and the chaos initialization module respectively, receives the initial dynamic vector population, and guides the iteration and update of the path by using a vector differential disturbance enhanced dynamic vector collaborative method with the minimization of the objective function as the guide, wherein the iteration and update process comprises a collaborative search stage and a vector differential disturbance stage, in the collaborative search stage, basic optimization is performed through simulation of behaviors such as repulsion, synchronization and condensation among individuals, in the vector differential disturbance stage, mutation, crossover and selection operators of differential evolution are introduced to perform secondary disturbance and update of the dynamic vector, the collaborative search stage and the vector differential disturbance stage of the collaborative iteration and update process are repeated until a termination condition is met, and an optimal path is output.

2. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 1, wherein, The path length index represents the cumulative Euclidean distance of the UAV from the starting point to the ending point, the obstacle threat index represents the minimum distance between the path segment and the projection center of the obstacle, the height constraint index represents the degree of deviation of the flight height from the safety interval, and the smoothness index represents the change rate of the turning angle of the path in the horizontal plane and the climbing angle in the vertical plane.

3. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 1, wherein, The path comprehensive constraint model module takes the minimization of the path length as the optimization objective, comprehensively calculates the path length index, the obstacle threat index, the height constraint index and the smoothness index, assigns different weights to the indexes, organically integrates the indexes to form a complete objective function, constructs a complete mathematical optimization model, and realizes global optimization and dynamic adjustment of the UAV flight path.

4. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 3, wherein, The specific execution steps of the path comprehensive constraint model module include: Computing path length cost: UAV flight path For a sequence of n path points, the three-dimensional coordinates of the path points are The straight-line distance between two points in three-dimensional space is calculated using the Euclidean distance formula of adjacent path points to obtain a path length cost function; wherein the Euclidean distance formula of adjacent path points is: ； The path length cost function is: ； Calculate the obstacle threat cost: Let the obstacle projection center be... The radius of the region is Calculate path segments To the obstacle Projection Center Proportional coefficient, shortest distance The threat cost of a single path segment is calculated by dividing the path into segments based on the value of t. The total threat cost function is obtained by segmenting the path segments according to distance thresholds; where, path segments To the obstacle Projection Center The formula for calculating the proportionality coefficient is: ； The total threat cost function is: ； In the formula, is the number of path points, is the total number of obstacles; Compute height cost: flight height needs to satisfy , single path point height cost by height whether in the safety zone section, the total height cost function is: ； Calculate the smoothness cost: the turning angle and the climbing angle are two key indicators for measuring the path smoothness, the turning angle and the climbing angle are calculated, and the smoothness cost function is obtained by comprehensively considering the turning angle and the climbing angle; wherein, the calculation formula of the turning angle is: ； wherein is the horizontal projection vector of path segment to , is the horizontal projection vector of path segment to , wherein the numerator is the magnitude of the cross product of the two horizontal projection vectors of path segments, indicating the amount of change in the direction of the turn, and the denominator is the dot product of the two horizontal projection vectors of path segments, indicating the similarity of the direction of the path segments; Climb angle The formula for calculating the climb angle is: ； wherein with are the flight altitudes at the path points and respectively; is the projection length of the path segment on the horizontal plane; The smoothness cost function calculation formula is: ； where the first term is the cumulative sum of all turning angles measuring the smoothness of the overall direction of the path, is the penalty coefficient of the turning angle cost; the second term is the variation between consecutive climb angles, measuring the smoothness of the height variation of the path, is the penalty coefficient of the climb angle cost. The total objective function is constructed by integrating the multi-dimensional costs of the path length, the threat cost, the flight height cost and the smoothness cost, and the calculation formula of the total objective function is: ； wherein are weight coefficients for each cost, used to balance the importance of each cost in the overall optimization.

5. The chaos and differential perturbation based dynamic vectorial cooperative UAV path planning system according to claim 1, wherein, The specific execution steps of the chaos initialization module include: According to the preset population size and the dynamic vector dimension, a normalized chaos variable sequence is iteratively generated through a piecewise linear traversal mapping formula; A random perturbation mechanism is performed in the iteration, and when a chaotic variable falls into a preset small periodic cycle fixed point set, a random tiny amount is superimposed on the chaotic variable to jump out of the cycle; The normalized chaotic variable sequence generated by iteration is mapped to a three-dimensional search space constituted by the flight area boundary and flight height limit of the unmanned aerial vehicle through a carrier mapping formula to generate initial path point coordinates.

6. The chaos and differential perturbation based dynamic vectorial cooperative UAV path planning system according to claim 1, wherein, The specific execution steps of the path optimization module include: In the cooperative search stage, the repulsion, synchronization, condensation, optimal solution attraction, and worst solution avoidance vectors of each dynamic vector of the population are calculated, the preset inertia weight and each behavior coefficient are combined, the vectors are weighted and summed to obtain a speed update increment, and the current position of the dynamic vector is updated using the speed update increment to generate a preliminary position vector representing the preliminary position of the dynamic vector; In the mutation stage of vector difference perturbation, three different individual vectors are randomly selected from the current population, and a mutation vector is generated using a vector difference formula; In the crossover stage of vector difference perturbation, the generated mutation vector and the preliminary position vector generated in the first step are subjected to a binomial crossover operation to generate a test vector; In the greedy selection stage of vector difference perturbation, the path synthesis constraint model module is called to calculate the fitness values of the test vector and the preliminary position vector, respectively; if the fitness value of the test vector is better than that of the preliminary position vector, the test vector is selected as the next generation path position, otherwise the preliminary position vector is retained as the next generation path position; The above two stages are repeated until the termination condition is met, and the optimal path is output.

7. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 5, wherein, The iteration formula of the piecewise linear traversal mapping is: ； In the formula, is the number of mappings, is the first path point on the first path on the first iteration of the chaotic variable; The calculation formula for generating initial path point coordinates is: ； wherein is the first dimensional initial position of the th dynamic vector, and are the upper and lower bounds of the search space in the th dimension, respectively.

8. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 6, wherein, The dynamic vector speed and position update in the cooperative search stage has a clear calculation formula, wherein the speed update formula is: ； The position update formula is: ； wherein is a dynamic vector is a dynamic vector is a velocity of the path point is an inertia weight is a repulsion vector is a repulsion coefficient is a synchronization vector is a synchronization coefficient is a cohesion vector is a cohesion coefficient is an optimal solution attraction vector is an optimal solution attraction coefficient is a worst solution repulsion vector is a worst solution repulsion coefficient is a dynamic vector is a dynamic vector is a coordinate representation of the path point 9. The dynamic vectorial cooperative UAV path planning system based on chaos and differential perturbation of claim 6, wherein, The mutation, crossover, and selection operations in the vector difference perturbation stage have clear calculation formulas, wherein the mutation operation formula is: ； The crossover operation formula is: ； The selection operation formula is: ； wherein is a scaling factor, taking values in the range , is a cross probability, is a random number between , is a randomly selected one-dimensional index.

10. The chaos and differential perturbation based dynamic vectorial cooperative UAV path planning system according to claim 1, wherein, The path optimization module includes optimal and worst position update rules, and has a complete algorithm flow, and the total algorithm flow is to obtain initial planning information and initialize parameters, call the chaotic initialization module to generate an initial population, calculate the initial fitness and set the initial optimal or worst position, iteratively perform cooperative search and difference perturbation operations, update the fitness and optimal or worst position, and output the optimal path after reaching the maximum function evaluation number; Wherein, the local optimal position The update rule is: ； globally optimal position determined from the fitness-optimal dynamic vector position and is the path with the best fitness among all local optimal positions: ； local worst position The update rule is: ； The global worst position is determined by the dynamic vector position with the worst fitness, and is the path with the worst fitness among all local worst positions: 。

Images