猎雷声纳基阵姿态稳定系统自适应反步法控制

Adaptive backstepping control for attitude stabilization system of mine-hunting sonar

针对带非线性摩擦力矩和负载扰动的高精度猎雷声纳基阵姿态稳定系统,提出了一种基于神经网络的自适应反步法控制方法。其中神经网络用于估计未知非线性摩擦力矩,进而设计反步法控制器和参数自适应律来对神经网络估计误差和负载扰动进行补偿。最后应用Lyapunov方法证明了所提出的自适应控制器能保证闭环系统的稳定性,并且可以通过选择适当的控制器参数来调整收敛率。仿真结果表明,基于神经网络的自适应反步法控制方法与PID控制相比,系统的动、静态性能指标及鲁棒性得到了全面的改善,与双闭环PID控制相比,跟踪精度提高了3倍多。

A neural network-based backstepping adaptive control method is proposed for high-precision attitude stabilization systems of mine-hunting sonar with nonlinear friction torque and the load disturbance.The neural network is used to estimate the unknown nonlinear friction torque on line,and then a backstepping adaptive controller is designed to compensate the estimation error of neural network and the load disturbance.Finally,the Lyapunov method is applied which proves that the proposed adaptive controller could guarantees the stability of the closed-loop system,and the convergence rates can be adjusted by selecting the appropriate parameter values.Simulation results show that,the dynamic and static performances and the robustness of the system are improved under the proposed control method.Compared with the dual-loop PID control,the tracking accuracy is increased more than three times.