Ship path planning and tracking control method based on improved differential evolution algorithm

Abstract

The invention provides a ship path planning and tracking control method based on an improved differential evolution algorithm, and the method comprises the steps: constructing a collision risk degree model, and improving the calculation precision of the collision risk degree through combining with parameters of a quaternary ship domain optimization model; determining a fitness function Fitness to evaluate the quality of the path points according to the constraints of the ship collision risk degree, the voyage, the steering angle, the international marine collision avoidance rule and the optimal collision avoidance distance; a self-adaptive cross factor dynamic adjustment strategy based on individual fitness and population statistical indexes is introduced, candidate path points are screened according to an improved differential evolution algorithm, and collision with obstacles and dynamic ships is avoided; and establishing a ship course motion mathematical model, taking the reference path as an input instruction of a ship motion control subsystem, and tracking a ship planning path by using an adaptive neural network controller. According to the invention, the ship collision avoidance path planning is more in line with the actual navigation, and the safety of the collision avoidance path and the accuracy of tracking control are improved.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and more particularly to a method for ship path planning and tracking control based on an improved differential evolution algorithm. Background Technology

[0002] In the complex and ever-changing maritime navigation environment, ship path planning and tracking control are key technologies for ensuring navigation safety and economy. Path planning aims to find a safe and efficient path for a ship from its starting point to its destination, while tracking control ensures that the ship can navigate precisely along the planned path. In recent years, with the rapid development of emerging technologies such as the Internet of Things, big data, and cloud computing, the field of intelligent navigation has received widespread attention, and ship path planning and tracking control technologies have also made significant progress.

[0003] Currently, various intelligent optimization algorithms are widely used in the field of ship path planning, such as ant colony optimization, genetic algorithms, simulated annealing, and fast random search tree algorithms. Each of these algorithms has its advantages. For example, ant colony optimization optimizes paths by simulating ant foraging behavior; genetic algorithms use natural selection and genetic mutation mechanisms to find the optimal solution; simulated annealing avoids local optima by simulating the physical annealing process; and fast random search tree algorithms find feasible paths by constructing a path tree through random sampling. Furthermore, there are techniques that integrate multiple algorithms, such as fusing ant colony optimization and genetic algorithms to improve the latter's slow convergence speed and premature convergence; using simulated annealing to optimize the repulsion function of the artificial potential field method to address its susceptibility to local minima and unreachable targets; and integrating Dijkstra's algorithm into the fast random search tree algorithm to optimize the initial path. Differential evolutionary algorithms have also received widespread attention in recent years due to their simple structure, high computational efficiency, and strong adaptability. This algorithm possesses a unique memory ability to dynamically track the search state and adjust the strategy in real time, and its adaptive information-based evolutionary mechanism demonstrates good adaptability in the information-scarce and complex marine environment. The differential evolution algorithm exhibits a linear impact of population size on computation time, demonstrating considerable efficiency and robustness even with large populations, while exhibiting low sensitivity to core parameters. Therefore, existing research has explored various strategies to improve it, such as initializing the population with chaotic sequences and combining them with multi-subpopulation competition mechanisms to enhance population diversity, or integrating multiple mutation strategies with local search techniques to improve optimization capabilities in complex environments.

[0004] However, existing technologies still have some limitations and shortcomings. The Fast Random Search Tree algorithm often produces non-smooth paths in gridded environment modeling, and its computation time increases dramatically with increasing accuracy requirements. The artificial potential field method fails to utilize global information adequately, making it difficult to guarantee global optimality and prone to stalling at positions where the net force is zero. Furthermore, existing technologies still have shortcomings in terms of the completeness of ship path planning and control algorithms, the smoothness of generated paths, and the accuracy of tracking control signals. For example, the application effect of existing technologies in complex ship path planning is limited, making it difficult to meet the requirements of high-precision path planning and tracking control. In addition, the A* algorithm combined with the International Maritime Collision Prevention Regulations (IMC) collaborative planning, along with the line-of-sight method-neural network composite control architecture, has also demonstrated engineering application value. However, current technologies still face technical bottlenecks such as insufficient completeness of ship path planning and control algorithms, lack of smoothness in generated paths, and virtualization of tracking control signals. Therefore, a more effective solution is urgently needed to improve the safety, accuracy, and adaptability of ship path planning. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides a ship path planning and tracking control method based on an improved differential evolution algorithm. By constructing a collision hazard model and optimizing the path planning algorithm, combined with an adaptive neural network controller, this invention achieves safe and efficient path planning and accurate path tracking for ships in complex marine environments, significantly improving the safety of collision avoidance paths and the accuracy of tracking control.

[0006] The technical means employed in this invention are as follows: A ship path planning and tracking control method based on an improved differential evolution algorithm includes: S1. Construct a collision hazard model and optimize the model parameters in conjunction with the quaternary ship domain to improve the accuracy of collision hazard calculation; S2. Based on the constraints of ship collision risk, range, turning angle, international maritime collision avoidance rules, and optimal collision avoidance distance, determine the fitness function to evaluate the quality of the waypoints; S3. Introduce an adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators, adjust the differential evolution algorithm to generate an improved differential evolution algorithm, and select candidate path points to plan a reasonable path and avoid collisions with obstacles and dynamic ships. S4. Establish a mathematical model of the ship's heading motion, use the planned reference path as the input command of the ship's motion control subsystem, and use the designed adaptive neural network controller to track the ship's planned path.

[0007] Further, step S1 specifically includes: S11. Using a quaternion-based asymmetric elliptical ship domain model, collision avoidance boundaries are dynamically defined, and the safe radii for the bow, stern, starboard side, and port side are calculated using the following formulas:

[0008] in, Captain; and These are the advance gain and cyclotron diameter gain, respectively:

[0009]

[0010] in, This is the ship's speed; Based on the target ship's hull angle Segmented calculation of ship domain boundary values:

[0011] S12. Calculate the spatial collision hazard degree and define the spatial membership function. :

[0012] Among them, spatial membership degree threshold Adjustments are made in segments based on relative orientation, and ; S13. Calculate the time collision hazard and define the time membership function. :

[0013] Among them, the critical value and By avoidance distance , and relative orientation RB Dynamically determined; S14. Construct a composite assessment model based on the Spatial Collision Risk Ratio (SCR) and the Temporal Collision Risk Ratio (TCR) to determine the Composite Collision Risk Ratio (CRI): .

[0014] Further, step S2 specifically includes: S21. Determine based on collision hazard level CRI objective function ,as follows:

[0015] in, CRI Indicates the degree of risk of a ship collision; S22. Determine the objective function based on the flight distance. ,as follows:

[0016] in, Let the coordinates of a node be... and Its adjacent nodes before and after, The endpoint is [node name], and the total number of nodes is [number]. ; S23. Determine the objective function based on steering amplitude. ,as follows:

[0017] S24. Determine the objective function based on the international maritime collision avoidance rules, according to the classification of ship encounter situations. ,as follows:

[0018] in, Indicates the relative bearing of the target ship; S25. Determine the objective function based on collision avoidance timing. ,as follows:

[0019] S26. Construct the final fitness function Fitness ,as follows: .

[0020] Further, step S3 specifically includes: An adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators is introduced to achieve adaptive parameter adjustment. The crossover factor... The calculation formula is as follows:

[0021] in, Indicates the first The fitness of an individual The average fitness of the population. This represents the minimum fitness value in the population.

[0022] Further, step S4 specifically includes: S41. Establish a systematic processing flow for ship motion characteristics, construct a mathematical model of ship heading motion by extracting the path coordinates planned by the improved differential evolution algorithm, and use the heading angle calculation result as a reference input signal for path tracking. S42. Based on this, an adaptive control framework is constructed using neural networks to achieve multi-dimensional matching between the ship's motion trajectory and the planning command signal, forming a closed-loop control system, and ultimately achieving precise tracking of the desired heading angle. S43. To address the stability problem of discrete closed-loop systems, conduct stability analysis.

[0023] Further, step S41 specifically includes: S411, based on rudder angle With heading Based on the dynamic relationship, a mathematical model for the ship's nonlinear heading control system is constructed as follows:

[0024] in, The ship's turning index. For ship following behavior index; nonlinear function The approximate expansion is as follows:

[0025] in, The coefficient is constant. S412, Define state variables By discretizing the original equations, a nonlinear discrete system model for the ship's heading is derived as follows:

[0026] in, The state of the system; and These are the system's input and output, respectively. For system control gain; function ; S413, Based on planned path coordinates The reference heading angle is calculated using the following formula:

[0027] in, For ideal planning instruction signals.

[0028] Further, step S42 specifically includes: S421. Regarding the nonlinear terms in the system An approximation function is constructed using an RBF neural network, as follows:

[0029] in, For weight vectors, The Gaussian function is typically used. ,in, As the center vector, For width parameters; S422. Define the neural network approximation error and set the optimal weights. ,satisfy The neural network approximation error is then expressed as:

[0030] in, ; S423, combined with tracking error The adaptive control law is designed as follows:

[0031] in, To adjust the parameters; S424. Substitute the control law into the system model and establish the error dynamic equation as follows:

[0032] S425. To enhance robustness, a filtering error function is introduced as follows:

[0033] in, ; S426. After simplification, the recurrence relation can be obtained: .

[0034] Further, step S43 specifically includes: S431. Regarding the stability problem of discrete closed-loop systems, an analysis is conducted based on Lyapunov theory, defining the discrete-time Lyapunov function as follows:

[0035] in, The design parameters are strictly positive definite. S432. Derive the Lyapunov function difference equation and calculate the first-order difference:

[0036] S433. To verify stability, it is necessary to simplify the equation by considering the dynamic characteristics of the error. Substituting the error recursive formula into the difference equation and performing algebraic operations, we obtain:

[0037] in, To adjust the parameters; S434. Design parameter constraints to ensure parameters Satisfies nonnegativity:

[0038] S435. Construct a piecewise parameter update adaptive control law as follows:

[0039] in, It is the gain constant. ; S436. Substituting the adaptive law into the Lyapunov function difference expression simplifies the analysis into two cases:

[0040] S437. Decompose the auxiliary control signal and design the disturbance compensation term as follows:

[0041] in, , ; S438. Establish the explicit expression for the error and derive the dynamics of the closed-loop error:

[0042] S439. Verify stability under different scenarios. The verification process is as follows: for In this case, combined with the upper bound of the approximation error, the following is derived: This leads to the inequality relationship. This proves that at this time Thus ensuring ; When the system approaches steady state, the gain effect of the denominator term in step S436 makes... Combined with error propagation relationship And the parameters satisfy Ultimately, the global convergence of the tracking error was achieved, verifying the asymptotic stability of the closed-loop system.

[0043] Compared with the prior art, the present invention has the following advantages: 1. This invention, by constructing a collision risk model and combining it with the parameters of a four-dimensional ship domain optimization model, can more accurately assess the collision risk of ships in complex marine environments, thereby improving the safety of path planning.

[0044] 2. This invention improves the differential evolution algorithm by introducing an adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators. This solves the problems of insufficient convergence speed and premature convergence of the traditional differential evolution algorithm, realizes global optimal path planning, and improves the efficiency and quality of path planning.

[0045] 3. By utilizing an adaptive neural network controller for path tracking control, this invention can adjust the control strategy in real time to adapt to the dynamic characteristics of the ship, achieving precise control of complex nonlinear systems and enhancing the stability and robustness of path tracking control.

[0046] 4. This invention uses the planned route of the improved differential evolution algorithm as a reference tracking path and combines it with an adaptive neural network controller to achieve multi-dimensional matching between the ship's motion trajectory and the planning command signal, forming a closed-loop control system that ensures that the ship can accurately track the planned path.

[0047] 5. By considering the characteristics of the ship's course discrete nonlinear control system and introducing a neural network to approximate the unknown nonlinear uncertainties, this invention designs an adaptive neural network controller with low complexity and low computational burden, realizing compensation for system disturbances and nonlinear terms, and solving the inherent defect of the traditional differential evolution method in ignoring the dynamic disturbances of the navigation environment during path planning.

[0048] 6. The present invention verifies the effectiveness of the improved differential evolution algorithm path planning and the feasibility of the ship's path tracking of the desired course through simulation experiments, proving the feasibility and superiority of the present invention in practical applications, and providing an efficient and accurate technical solution for unmanned ships to track and control paths in complex sea areas.

[0049] In summary, the technical solution of this invention not only improves the safety and accuracy of ship path planning, but also enhances the stability and robustness of path tracking control, providing an innovative and practical technical solution for the field of intelligent navigation. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0051] Figure 1 This is a flowchart of the method of the present invention.

[0052] Figure 2 A flowchart illustrating the combination of the collision hazard model and the improved differential evolution algorithm provided in this embodiment of the invention.

[0053] Figure 3 A diagram of an adaptive control system based on neural network approximation provided in an embodiment of the present invention.

[0054] Figure 4 A comparison diagram of traditional and improved differential evolution path planning provided for embodiments of the present invention.

[0055] Figure 5 The diagram shows the results of the USV's avoidance process under two-ship encounters and multiple obstacles, as provided in the embodiments of the present invention.

[0056] Figure 6 The tracking curve of the heading angle provided for the embodiments of the present invention.

[0057] Figure 7 The input signal variation curve is provided for an embodiment of the present invention.

[0058] Figure 8 This refers to the USV heading angle tracking error provided in the embodiments of the present invention. Detailed Implementation

[0059] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0060] It should be noted that the terms "comprising" and "having" and any variations thereof in the specification, claims and accompanying drawings of this invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or units that are not explicitly listed or that are inherent to such processes, methods, products or devices.

[0061] like Figure 1 As shown, this invention provides a ship path planning and tracking control method based on an improved differential evolution algorithm, comprising: S1. Construct a collision hazard model and optimize the model parameters in conjunction with the quaternary ship domain to improve the accuracy of collision hazard calculation; S2. Based on the constraints of ship collision risk, range, turning angle, international maritime collision avoidance rules, and optimal collision avoidance distance, determine the fitness function to evaluate the quality of the waypoints; S3. Introduce an adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators, adjust the differential evolution algorithm to generate an improved differential evolution algorithm, and select candidate path points to plan a reasonable path and avoid collisions with obstacles and dynamic ships. S4. Establish a mathematical model of the ship's heading motion, use the planned reference path as the input command of the ship's motion control subsystem, and use the designed adaptive neural network controller to track the ship's planned path.

[0062] In a specific implementation, as a preferred embodiment of the present invention, step S1 specifically includes: S11. Using a quaternion-based asymmetric elliptical ship domain model, collision avoidance boundaries are dynamically defined, and the safe radii for the bow, stern, starboard side, and port side are calculated using the following formulas:

[0063] in, Captain; and These are the advance gain and cyclotron diameter gain, respectively:

[0064]

[0065] in, This is the ship's speed; Based on the target ship's hull angle Segmented calculation of ship domain boundary values:

[0066] S12. Calculate the spatial collision hazard degree and define the spatial membership function. :

[0067] Among them, spatial membership degree threshold Adjustments are made in segments based on relative orientation, and ; S13. Calculate the time collision hazard and define the time membership function. :

[0068] Among them, the critical value and By avoidance distance , and relative orientation RB Dynamically determined; S14. Construct a composite assessment model based on the Spatial Collision Risk Ratio (SCR) and the Temporal Collision Risk Ratio (TCR) to determine the Composite Collision Risk Ratio (CRI): .

[0069] In a specific implementation, as a preferred embodiment of the present invention, step S2 specifically includes: S21. Determine based on collision hazard level CRI objective function ,as follows:

[0070] in, CRI Indicates the degree of risk of a ship collision; S22. Determine the objective function based on the flight distance. ,as follows:

[0071] in, Let the coordinates of a node be... and Its adjacent nodes before and after, The endpoint is [node name], and the total number of nodes is [number]. ; S23. Determine the objective function based on steering amplitude. ,as follows:

[0072] S24. Determine the objective function based on the international maritime collision avoidance rules, according to the classification of ship encounter situations. ,as follows:

[0073] in, Indicates the relative bearing of the target ship; S25. Determine the objective function based on collision avoidance timing. ,as follows:

[0074] S26. Construct the final fitness function Fitness ,as follows: .

[0075] In a specific implementation, as a preferred embodiment of the present invention, step S3 specifically includes: An adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators is introduced to achieve adaptive parameter adjustment. The crossover factor... The calculation formula is as follows:

[0076] in, Indicates the first The fitness of an individual The average fitness of the population. This represents the minimum fitness value in the population. Figure 2 The flowchart shows the combination of the collision hazard model and the improved differential evolution algorithm.

[0077] In a specific implementation, as a preferred embodiment of the present invention, step S4 specifically includes: S41. Establish a systematic processing flow for ship motion characteristics, construct a mathematical model of ship heading motion by extracting the path coordinates planned by the improved differential evolution algorithm, and use the heading angle calculation result as a reference input signal for path tracking. S42. Based on this, an adaptive control framework is constructed using neural networks to achieve multi-dimensional matching between the ship's motion trajectory and the planning command signal, forming a closed-loop control system, and ultimately achieving precise tracking of the desired heading angle. S43. To address the stability problem of discrete closed-loop systems, conduct stability analysis.

[0078] In a specific implementation, as a preferred embodiment of the present invention, step S41 specifically includes: S411, based on rudder angle With heading Based on the dynamic relationship, a mathematical model for the ship's nonlinear heading control system is constructed as follows:

[0079] in, The ship's turning index. For ship following behavior index; nonlinear function The approximate expansion is as follows:

[0080] in, The coefficient is constant. S412, Define state variables By discretizing the original equations, a nonlinear discrete system model for the ship's heading is derived as follows:

[0081] in, The state of the system; and These are the system's input and output, respectively. For system control gain; function ; S413, Based on planned path coordinates The reference heading angle is calculated using the following formula:

[0082] in, For ideal planning instruction signals.

[0083] In a specific implementation, as a preferred embodiment of the present invention, step S42 specifically includes: S421. Regarding the nonlinear terms in the system An approximation function is constructed using an RBF neural network, as follows:

[0084] in, For weight vectors, The Gaussian function is typically used. ,in, As the center vector, For width parameters; S422. Define the neural network approximation error and set the optimal weights. ,satisfy The neural network approximation error is then expressed as:

[0085] in, ; S423, combined with tracking error The adaptive control law is designed as follows:

[0086] in, To adjust the parameters; Figure 3 This is a diagram of an adaptive control system based on neural network approximation.

[0087] S424. Substitute the control law into the system model and establish the error dynamic equation as follows:

[0088] S425. To enhance robustness, a filtering error function is introduced as follows:

[0089] in, ; S426. After simplification, the recurrence relation can be obtained: .

[0090] In a specific implementation, as a preferred embodiment of the present invention, step S43 specifically includes: S431. Regarding the stability problem of discrete closed-loop systems, an analysis is conducted based on Lyapunov theory, defining the discrete-time Lyapunov function as follows:

[0091] in, The design parameters are strictly positive definite. S432. Derive the Lyapunov function difference equation and calculate the first-order difference:

[0092] S433. To verify stability, it is necessary to simplify the equation by considering the dynamic characteristics of the error. Substituting the error recursive formula into the difference equation and performing algebraic operations, we obtain:

[0093] in, To adjust the parameters; S434. Design parameter constraints to ensure parameters Satisfies nonnegativity:

[0094] S435. Construct a piecewise parameter update adaptive control law as follows:

[0095] in, It is the gain constant. ; S436. Substituting the adaptive law into the Lyapunov function difference expression simplifies the analysis into two cases:

[0096] S437. Decompose the auxiliary control signal and design the disturbance compensation term as follows:

[0097] in, , ; S438. Establish the explicit expression for the error and derive the dynamics of the closed-loop error:

[0098] S439. Verify stability under different scenarios. The verification process is as follows: for In this case, combined with the upper bound of the approximation error, the following is derived: This leads to the inequality relationship. This proves that at this time Thus ensuring ; When the system approaches steady state, the gain effect of the denominator term in step S436 makes... Combined with error propagation relationship And the parameters satisfy Ultimately, the global convergence of the tracking error was achieved, verifying the asymptotic stability of the closed-loop system.

[0099] Example To verify the effectiveness of the present invention, a detailed simulation study was conducted on the MATLAB platform. The experimental details are as follows: The simulation experiments included two parts: path planning and tracking control simulation. A square simulation scene with a length and width of [0,10] was constructed, and nine circular obstacles with a radius of 0.2 units were randomly placed in this environment. A Euclidean coordinate system was established, and the starting point of the ship was set to (2,8), and the target point was set to (7,3.5).

[0100] Path planning simulation experiment: A comparative experiment is conducted between the traditional differential evolution algorithm and the improved differential evolution algorithm in the simulation experiment. To ensure consistency of experimental conditions, both algorithms use the exact same parameters: the number of iterations for both algorithms is set. M =100, population size NP =50, population dimension Q =10, , , .

[0101] Table 1 Comparison of simulation results in the scenario of two ships encountering each other.

[0102] Experimental results are as follows Figure 4-5 As shown, under the premise of two ships meeting and avoiding collisions with obstacles, both the traditional differential evolution method and the improved differential evolution method can plan an effective path from the starting point to the ending point. However, the path planned by the improved differential evolution method is shorter and better than that planned by the traditional differential evolution method. Although the traditional differential evolution method has a shorter running time, its total yaw distance is larger. Therefore, the improved differential evolution method is acceptable in terms of time and has a better collision avoidance effect than the traditional differential evolution method, which verifies the superiority of the algorithm.

[0103] Tracking control simulation experiment: In the simulation experiment, the radial basis function neural network structure was selected as 2-9-1, and the parameters of the Gaussian function were... and Take them respectively and The initial weights of the neural network are random numbers between (0, 1). The initial value of the controlled object is 0, and the ideal tracking signal is... Control parameters .

[0104] like Figure 6-8As shown, the planned optimal path is processed and used as input signal for path tracking. It can be seen that when the ship is tracking the path, the heading angle does not deviate significantly from the planned route. Under the control of the adaptive neural network controller, the ship can track the path relatively accurately and complete the tracking task from the starting point to the target point; and In the interval It has a good tracking effect.

[0105] Simulation results show that the improved differential evolution algorithm based on a collision hazard model for ship path planning overcomes the problems of insufficient convergence speed and premature convergence in the traditional differential evolution method, achieving globally optimal path planning. The planned route of the improved differential evolution method is used as a reference tracking path, and through actual calculations, it serves as the reference input signal for the ship's adaptive neural network controller. Considering the characteristics of the ship's discrete nonlinear control system, a low-complexity, low-computation adaptive neural network controller is designed by introducing a neural network to approximate the unknown nonlinear uncertainties of the ship's discrete course. This achieves compensation for system disturbances and nonlinear terms, while also overcoming the inherent defect of the traditional differential evolution method in ignoring dynamic disturbances in the navigation environment during path planning, realizing high-precision path planning and tracking control for unmanned surface vessels. Simulation experiments verify the effectiveness of the improved differential evolution method for path planning and the feasibility of ship path tracking to the desired course.

[0106] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A ship path planning and tracking control method based on an improved differential evolution algorithm, characterized in that, Comprise: S1, construct a collision risk model, and combine the four ship field optimization model parameters to improve the calculation accuracy of the collision risk; S2, according to the ship collision risk, voyage, turning angle, international maritime collision regulations and the constraint of the best collision avoidance distance, determine the fitness function Fitness to evaluate the quality of the path point; S3, introduce an adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indicators, adjust the differential evolution algorithm to generate an improved differential evolution algorithm, and screen the candidate path points to plan a reasonable path and avoid collision with obstacles and dynamic ships; S4, establish a ship heading motion mathematical model, take the reference path generated by planning as the input instruction of the ship motion control subsystem, and use the designed adaptive neural network controller to track the ship planning path.

2. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 1, characterized in that, Step S1, specifically includes: S11, adopt a four-element asymmetric elliptical ship field model, dynamically define the collision avoidance boundary, calculate the safety radius of the bow, stern, starboard and port, and the formula is as follows: wherein is the length of the ship; and are the advance gain and the turn diameter gain, respectively: wherein, V is the ship's speed; According to target ship side angle Segmented calculation of ship field boundary values: S12, calculating the spatial collision risk degree, defining the spatial membership function : wherein the spatial membership threshold value according to the relative orientation segment adjustment, and ; S13, calculate the time collision danger degree, define the time membership function : Wherein, the critical value With By the avoidance distance , And the relative position RB Dynamic decision; S14, based on the space collision risk SCR and the time collision risk TCR, construct a composite evaluation model to determine the composite collision risk CRI: 。 3. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 1, characterized in that, Step S2, specifically includes: S21, determining a target function based on the collision danger degree CRI as follows:​ wherein, CRI represents the degree of danger of ship collision; S22, determining a voyage-based objective function As follows: wherein, is the coordinate of a node, and its preceding and succeeding nodes, is the end point, and the total number of nodes is ; S23, determining a target function based on the steering amplitude As follows: S24, according to the classification of the ship encounter situation, determine the target function based on the international maritime collision regulations As follows: wherein, represents the relative bearing of the target ship; S25, determining a target function based on the collision avoidance opportunity As follows: S26, constructing a final fitness function Fitness As follows: 。 4. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 1, characterized in that, Step S3, specifically includes: An adaptive crossover factor dynamic adjustment strategy based on individual fitness and population statistical indexes is introduced to realize adaptive adjustment of parameters, wherein the crossover factor The calculation formula is as follows: wherein, represents the fitness of the i-th individual, is the average fitness of the population, is the minimum fitness value in the population.​ 5. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 1, characterized in that, Step S4, specifically includes: S41, establish a systematic processing flow for ship motion characteristics, construct a ship heading motion mathematical model by extracting the path coordinates planned by the improved differential evolution algorithm, and take the heading angle calculation result as the reference input signal for path tracking; S42, on this basis, construct an adaptive control framework based on neural network to realize multi-dimensional matching of ship motion trajectory and planning instruction signal, form a closed-loop control system, and finally realize accurate tracking of the expected heading angle; S43, for the stability problem of discrete closed-loop system, stability analysis is carried out.

6. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 5, characterized in that, Step S41, specifically includes: S411、Based on the rudder angle with the dynamic relationship of the heading The mathematical model of the ship heading nonlinear control system is constructed as follows: wherein is the turning quality index of the ship, is the following quality index of the ship; nonlinear function An approximate expansion of the function f is: wherein are constants; S412, define state variable The nonlinear discrete system model of ship course is derived by discretizing the original equation as follows: wherein, is the state of the system; and are the input and output of the system, respectively; is the system control gain; function ; S413、based on the planning path coordinates The reference heading angle is calculated according to the following formula: wherein, is the ideal planning command signal.

7. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 5, characterized in that, Step S42, specifically includes: S421、For the nonlinear term in the system An RBF neural network is used to construct the approximation function as follows: wherein, is a weight vector, a Gaussian kernel is typically employed, wherein, is a center vector, is a width parameter; S422, define a neural network approximation error, set the optimal weight , satisfies The neural network approximation error is expressed as: wherein ; S423, combine tracking error , design adaptive control law as follows: wherein is a tuning parameter; S424, substitute the control law into the system model to establish the error dynamic equation, as follows: S425, in order to enhance the robustness, introduce the filtering error function, as follows: wherein ; S426, after sorting, the recursive relationship is obtained: 。 8. The ship path planning and tracking control method based on the improved differential evolution algorithm according to claim 5, characterized in that, Step S43, specifically includes: S431, for the stability problem of discrete closed-loop system, based on Lyapunov theory, define the discrete time Lyapunov function, as follows: wherein are strictly positive design parameters; S432, derive the difference equation of Lyapunov function, calculate the first-order difference: S433, to verify the stability, the error recursive relationship is substituted into the difference equation, and after algebraic operation: wherein is a tuning parameter; S434、design parameter constraints, ensure parameters non-negativity S435, construct a segmented parameter updating adaptive control law, as follows: wherein is a gain constant, ; S436, after substituting the adaptive law into the difference expression of Lyapunov function, simplify it into the analysis form under two conditions: S437, decompose the auxiliary control signal, design the disturbance compensation term, as follows: wherein , ; S438, establish the explicit expression of error, and derive the closed-loop error dynamics: S439, verify the stability according to the situation, and the verification process is as follows: For the case of , combining the upper bound of approximation error, we have , and then the inequality is obtained. Thus, we prove that , and thus ; When the system tends to be stable, the gain of the denominator in step S436 makes , combined with the error transmission relationship , and the parameters satisfy , the global convergence of the tracking error is finally realized, and the asymptotic stability of the closed-loop system is verified.

Images