基于PSO的无人艇操纵响应模型参数辨识结果优化

Optimization of Parameter Identification Results of USV Maneuvering Response Model Based on PSO

为解决无人艇操纵性预报问题,基于无人艇二阶非线性响应模型,采用差分离散法设计辨识模型,在Matlab平台上采用四阶龙格库塔法进行20°Z形操纵仿真实验,利用采集的数据基于递推最小二乘算法辨识响应模型参数,并基于此辨识值提出采用粒子群算法做进一步优化,基于正弦操纵运动数据作为适应度函数数据源,优化后的各个参数精度均较以前有所提高,最后基于优化前后的辨识参数结果开展10°,20°及30°Z形、正弦以及回转仿真操纵运动,验证了利用粒子群算法能优化递推最小二乘算法的辨识结果,其优化结果能精确预报无人艇的各种操纵运动.

In order to solve the problem of USV maneuverability prediction,based on the second-order nonlinear response model of USV,the identification model was designed by using the difference discrete method,and the fourth-order Runge Kutta method was used to carry out the 20°Z-shaped maneuvering simulation experiment on the Matlab platform.Using the collected data,the parameters of response model were identified based on recursive least square algorithm,and based on the identified values,particle swarm optimization was proposed for further optimization.Based on the sinusoidal manipulation motion data as the data source of fitness function,the precision of each parameter after optimization has been improved.Finally,10,20 and 30°Z-shaped,sinusoidal and rotary simulation maneuvers were carried out based on the identification parameters before and after optimization,which proves that the identification results of recursive least squares algorithm can be optimized by particle swarm optimization,and the optimized results can accurately predict various maneuvers of USV.