| Complex marine environment and diverse detection engineering requirements for underwater vehicle performance requirements are also getting higher and higher,the establishment of dynamic modeling for simulation,research on its dynamic characteristics and control algorithm is the commonly used underwater vehicle research and development technology roadmap.Dynamic model of traditional dynamics modeling on the nonlinear term is obtained by simplifying the model of six degrees of freedom vehicle linear space motion,but can not effectively solve the coupling problems of strong nonlinearity,only approximate motion of space is divided into horizontal and vertical calculation,modeling and simulation precision is high,serious depends on the accuracy of the hydrodynamic parameters.Vector propulsion underwater vehicle has stronger maneuverability and more complex dynamic characteristics.Therefore,a more effective and simple modeling method is needed.In this paper,the neural network is used to replace the traditional dynamics model to build the underwater vehicle dynamics system,and then the neural network parameters are optimized by a large number of navigation experimental data,and the practicability of the model is verified.The research results are as follows:(1)Combining neural network and dynamic system identification method based on vector propulsion underwater vehicle for modeling target,establish the dynamic model and trajectory prediction model,and based on the navigation of a large number of experimental data,through the reverse derivation algorithm to optimize the parameters,by neural network model.(2)According to the trajectory prediction model obtained by supervised learning,the kinematics simulation analysis of the vehicle is carried out,and the results are in good agreement with the experimental results.The method is proved to be correct and effective.Moreover,the simulation of turning process and the consistency of nonlinear process simulation are very high.And the operation is simple and easy to implement.(3)By the commercial simulation software Matlab,we carry out the joint simulation of neural network trajectory prediction model and control algorithm,and predict the PID control parameters of underwater vehicle in the process of steering and submergence deep navigation,which is similar to the experimental results. |
