Underwater mechanical arm control method for optimizing RBF (Radial Basis Function) neural network through improved Harris eagle algorithm

Abstract

The invention relates to an underwater mechanical arm control method for optimizing an RBF (Radial Basis Function) neural network through an improved Harris Hawk algorithm (PTHHO). The method comprises the following steps: firstly, constructing an RBF (Radial Basis Function) network which comprises an input layer, a hidden layer and an output layer, and then generating a group of initial solutions (weight and bias) by using a PTHHO algorithm; and then, by simulating hunting behaviors of eagles and iteratively updating and optimizing the parameters, the loss function is minimized, and the approximation precision is improved. The combination utilizes the global search capability of the PTHHO algorithm and the fitting capability of the RBF neural network, so that the performance and generalization capability of the neural network can be improved to a certain extent. Compared with a traditional control method, the method has the advantages that the robustness of a control system is improved, hydrodynamic interference is effectively compensated, and the purpose of accurately controlling the underwater mechanical arm is achieved.

Description

Technical Field

[0001] The invention relates to the technical field of automatic control, and in particular to an underwater manipulator control method using an improved Harris Hawk optimization algorithm to optimize a RBF neural network. Background Art

[0002] In underwater operations, robotic arms are widely used in salvage, maintenance, scientific exploration and other tasks. The underwater environment is complex and changeable, with unfavorable conditions such as unstable water flow, buoyancy changes and pressure changes, which puts high demands on the control accuracy and stability of the robotic arm. Traditional control algorithms often show the shortcomings of low tracking accuracy and slow response speed when dealing with complex underwater environments. Therefore, developing a high-precision and fast-response control method is crucial to improving the performance of underwater robotic arms.

[0003] Radial basis function (RBF) neural network is a nonlinear neural network that has the ability to approximate functions and can process nonlinear systems through learning and modeling. It has broad application potential in complex systems such as controlling robotic arms. However, the performance of RBF neural networks is highly dependent on the selection of parameters such as weights, biases, and basis function centers. If the parameters are inappropriate, the network may reach a local optimal state and fail to perceive effectively changing environments.

[0004] The Harris Hawk Algorithm is a new intelligent optimization algorithm based on the hunting behavior of eagles. It can quickly and globally search for the optimal solution to the problem by simulating the collaborative hunting process of eagle clusters. Derived from traditional optimization algorithms, the Harris Hawk Algorithm has the characteristics of fast convergence speed and strong global search ability when dealing with complex multi-peak problems, so it is very suitable for optimizing complex nonlinear systems.

[0005] Based on this, the Harris Eagle algorithm will optimize the RBF neural network to improve the control performance of the network in underwater manipulator path tracking. Through this optimization method, the path tracking accuracy and real-time control capability of the underwater manipulator in a dynamic environment can be effectively improved, so as to better cope with the operation requirements in complex underwater environments. Summary of the invention

[0006] The present invention aims to solve the technical problems arising in the above-mentioned underwater manipulator control, and provides an underwater manipulator control method by optimizing RBF neural network with improved Harris Hawk optimization algorithm.

[0007] The technical solution of the present invention is as follows:

[0008] An underwater manipulator control method using an improved Harris Hawk optimization algorithm to optimize RBF neural network, the method comprising:

[0009] S1: Improvement of Harris Hawk Algorithm: Improve the global search capability of the algorithm by changing the Harris Hawk search mechanism and search path;

[0010] S2: Use the improved Harris Eagle algorithm to optimize the RBF neural network: iteratively optimize the network parameters through the algorithm's optimization performance to find the best weight combination;

[0011] S3: Achieve precise control of underwater robot arm trajectory tracking: The neural network after algorithm optimization has better tracking accuracy, enabling the underwater robot arm to achieve precise control in complex environments;

[0012] S1 specifically includes:

[0013] S11: Improving population quality and diversity using chaotic map initialization;

[0014] S12: Use adaptive t-distribution strategy to improve convergence speed and accuracy;

[0015] S13: Use dynamic reverse learning strategy to enhance the ability to escape from local optimal solutions;

[0016] S2 uses the improved Harris Eagle algorithm to optimize the RBF neural network. S2 specifically includes:

[0017] S21: The initial settings for establishing the Harris Hawk algorithm are as follows: the population size is set to 100; the maximum number of iterations is set to 30;

[0018] S22: Initialize network parameters: randomly generate weights and biases, and set the RBF network structure;

[0019] S23: Fitness function design: Select mean square error (MSE) and mean absolute error (MAE) to evaluate the difference between the network's output under a given input and the target output;

[0020] S24: Generate initial eagle group: create a set of candidate solutions representing different weight and bias combinations;

[0021] S25: Simulate hunting behavior: Calculate the fitness of each candidate solution. Update the position and speed of the solution according to the fitness, and simulate the hunting strategy of the eagle. During this period, emphasize the attraction of the better solution and allow a certain degree of randomness to avoid local optimality;

[0022] S26: Iterative optimization: Continue to iterate and update weights and biases until convergence conditions are met;

[0023] S3 specifically includes:

[0024] S31: Use the optimized RBF neural network to estimate the state of the robot arm in real time, including position, velocity and acceleration;

[0025] S32: Combine the state estimation value output by the optimized RBF neural network, the dynamic characteristics of the system and the environmental interference that need to be considered to design a closed-loop control law to ensure the stability and accuracy of the robot arm;

[0026] S33: Design the motion trajectory of the robot arm according to the task requirements, and use the optimized RBF neural network to predict the state of each node of the robot arm in the execution trajectory to ensure the accuracy of trajectory planning;

[0027] S34: During the execution of the robot arm, the state information of the robot arm is monitored in real time, the error between the actual motion trajectory and the planned motion trajectory is analyzed, and the parameters are adjusted and the error is compensated through the optimized RBF neural network to improve the robustness and adaptability of the system;

[0028] The present invention is a control method for improving the poor control effect of an underwater manipulator, and uses the optimization advantage of an improved Harris Eagle algorithm to optimize the network parameters of a RBF neural network to achieve a better control effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the implementation steps of the method of improving the Harris Eagle algorithm to optimize the RBF neural network to control the underwater manipulator;

[0030] Figure 2 It is a schematic diagram of underwater manipulator control with improved Harris Eagle algorithm added on the basis of the original RBF neural network controller.

Claims

1. A method for controlling an underwater manipulator based on an improved Harris Hawk algorithm to optimize RBF neural network, characterized in that: The method comprises: S1: Improvement of Harris Hawk algorithm; S2: Optimizing RBF neural network using improved Harris Eagle algorithm; S3: Achieve precise control of underwater robotic arm trajectory tracking.

2. The underwater manipulator control method based on the improved Harris Hawk optimization algorithm to optimize the RBF neural network according to claim 1 is characterized in that Improved the fitting ability of the neural network and improved the control accuracy. S1 specifically includes: S11: Improving population quality and diversity using chaotic map initialization; S12: Use adaptive t-distribution strategy to improve convergence speed and accuracy; S13: Use dynamic reverse learning strategy to enhance the ability to escape from local optimal solutions.

3. The underwater manipulator control method based on the improved Harris Hawk optimization algorithm to optimize the RBF neural network according to claim 1 is characterized in that Optimize the weights and biases of the RBF neural network, the specific steps of S2: S21: Establish the starting settings of the Harris Hawk algorithm; S22: Initialize network parameters: randomly generate weights and biases, and set the RBF network structure; S23: Fitness function design: Select mean square error (MSE) and mean absolute error (MAE) to evaluate the difference between the network's output under a given input and the target output; S24: Generate initial eagle group: create a set of candidate solutions representing different weight and bias combinations; S25: Simulate hunting behavior: calculate the fitness of each candidate solution. The position and speed of the solution are updated according to the fitness, simulating the hunting strategy of the eagle. During this period, the attraction of the better solution is emphasized, and a certain degree of randomness is allowed to avoid the local optimum; S26: Iterative optimization: Continue to iterate and update weights and biases until convergence conditions are met.

4. The underwater manipulator control method based on the improved Harris Hawk optimization algorithm to optimize the RBF neural network according to claim 1 is characterized in that: The robot arm has multiple joints. In S3, the action value output by the control method is the angle and angular velocity of the joint movement, and the control algorithm can control the robot arm to move within a predetermined time. S3 specifically includes: S31: Use the optimized RBF neural network to estimate the state of the robot arm in real time, including position, velocity and acceleration; S32: Combine the state estimation value output by the optimized RBF neural network, the dynamic characteristics of the system and the environmental interference that need to be considered to design a closed-loop control law to ensure the stability and accuracy of the robot arm; S33: Design the motion trajectory of the robot arm according to the task requirements, and use the optimized RBF neural network to predict the state of each node of the robot arm in the execution trajectory to ensure the accuracy of trajectory planning.

Images