针对无刷直流电机(brushless DC motor,BLDCM)非线性的特点,引入了一种基于神经网络的自适应逆控制方法。该方案中,用非线性自回归(NARX)动态网络做为模型辨识器和控制器。辨识器采用了BP(back propagation)算法在线调整参数,并获取被...针对无刷直流电机(brushless DC motor,BLDCM)非线性的特点,引入了一种基于神经网络的自适应逆控制方法。该方案中,用非线性自回归(NARX)动态网络做为模型辨识器和控制器。辨识器采用了BP(back propagation)算法在线调整参数,并获取被控对象精确的Jacobian信息,再由实时递归学习算法(RTRL)实现对控制器的在线整定。仿真结果表明,方法具有响应速度较快、无超调的优点,且具备较强的自适应性和鲁棒性。展开更多
The drawbacks of common nonlinear Filtered-ε adaptive inverse control （AIC） method, such as the unreliability due to the change of delay time and the faultiness existing in its disturbance control loop, are discuss...The drawbacks of common nonlinear Filtered-ε adaptive inverse control （AIC） method, such as the unreliability due to the change of delay time and the faultiness existing in its disturbance control loop, are discussed. Based on it, the diagram of AIC is amended to accommodate with the characteristic of nonlinear object with time delay. The corresponding Filtered-ε adaptive algorithm based on RTRL is presented to identify the parameters and design the controller. The simulation results on a nonlinear ship model of ＂The R.O.V Zeefakker＂ show that compared with the previous scheme and adaptive PID control, the improved method not only keeps the same dynamic response performance, but also owns higher robustness and disturbance rejection ability, and it is suitable for the control of nonlinear objects which have higher requirement to the maneuverability under complex disturbance environment.展开更多
文摘针对无刷直流电机(brushless DC motor,BLDCM)非线性的特点,引入了一种基于神经网络的自适应逆控制方法。该方案中,用非线性自回归(NARX)动态网络做为模型辨识器和控制器。辨识器采用了BP(back propagation)算法在线调整参数,并获取被控对象精确的Jacobian信息,再由实时递归学习算法(RTRL)实现对控制器的在线整定。仿真结果表明,方法具有响应速度较快、无超调的优点,且具备较强的自适应性和鲁棒性。
基金This project was supported by the National Defence Pre-research Foundation of Shipbuilding Industry (01J1.50) and theWeapon & Equipment Pre-research Foundation of General Armament Department (51414030204JW0322).
文摘The drawbacks of common nonlinear Filtered-ε adaptive inverse control （AIC） method, such as the unreliability due to the change of delay time and the faultiness existing in its disturbance control loop, are discussed. Based on it, the diagram of AIC is amended to accommodate with the characteristic of nonlinear object with time delay. The corresponding Filtered-ε adaptive algorithm based on RTRL is presented to identify the parameters and design the controller. The simulation results on a nonlinear ship model of ＂The R.O.V Zeefakker＂ show that compared with the previous scheme and adaptive PID control, the improved method not only keeps the same dynamic response performance, but also owns higher robustness and disturbance rejection ability, and it is suitable for the control of nonlinear objects which have higher requirement to the maneuverability under complex disturbance environment.
