论文部分内容阅读
为了有效地控制液压非线性系统,提出基于RBF神经网络的自适应最优控制系统,应用于机器人液压驱动器.首先,建立了液压系统的动力学模型;然后,输入幅值和频率连续变化的信号,应用卡尔曼滤波器估计液压系统状态,进而计算出模型参数,对模型参数进行分组用于训练RBF神经网络;接着,对不同组参数求平均作为参考点,用RBF神经网络学习最优控制器反馈增益随系统参数的变化规律;最后,训练完成的神经网络根据卡尔曼滤波器参数估计值在线预测并调节控制器增益.经实验验证,该控制系统调节时间和跟踪误差仅为普通线性二次型最优控制器的1/2和1/3左右.
In order to effectively control the hydraulic nonlinear system, an adaptive optimal control system based on RBF neural network is proposed, which is applied to a robotic hydraulic actuator. First, a dynamic model of the hydraulic system is established. Then, a signal with continuously changing amplitude and frequency , The application of Kalman filter to estimate the state of the hydraulic system, and then calculate the model parameters, the model parameters are grouped for training RBF neural network; then, the average of different groups of parameters as a reference point RBF neural network learning optimal controller Finally, the trained neural network predicts and regulates the gain of the controller on the basis of the Kalman filter parameter estimation.The experimental results show that the adjustment time and the tracking error of the control system are only normal linear quadratic Optimal controller type 1/2 and 1/3 or so.