基于BP网络的全垫升气垫船运动模型的辨识与仿真

Identification and Simulation of ACV Motion Model Based on BP Network

为了研究全垫升气垫船的操纵性能,建立了一种全垫升气垫船操纵运动的灰箱模型。通过MATLAB神经网络工具箱(nntool)建立BP神经网络,对全垫升气垫船的灰箱模型进行系统辨识,建立了全垫升气垫艇的人工神经网络操纵运动模型(Artificial Neural Networks Maneuvering Model,ANNMM),并对该模型进行Z型舵操纵仿真试验。结果表明:该模型的航向最大相对误差为0.36%,南北向速度最大相对误差为0.31%,东西向速度最大相对误差为0.13%。充分表明基于BP网络的辨识建模方法是有效的、可行的。

In order to study the maneuverability of air cushion vehicle(ACV),a grey box model for maneuvering ACV was established.The BP neural network is established by using MATLAB (nntool)to identify the gray box model of the ACV.An artificial neural network maneuvering model (ANNMM)was established for the ACV ,and Z-shaped rudder manipulation simulation tests were performed on this model.The results show that the maximum relative error of the course of the model is 0.36% the maximum relative error of the north-south velocity is 0.31%,and the maximum relative error of the east-west velocity is 0.13%.Above fully shows that the identification modeling method based on BP network is effective and feasible.