论文部分内容阅读
采用多层前馈网络结构进行动态建模,并用Davidon最小二乘法作为在线学习算法,将辨识后得到的模型进行线性化.基于线性化模型设计广义预测控制器。将其与非线性前馈相结合,建立了一种适合于非线性系统的前馈补偿广义预测自校正控制器.仿真结果验证了本控制器对非线性系统控制的有效性
The multi-layer feedforward network structure is used for dynamic modeling, and the Davidon least squares method is used as an online learning algorithm to linearize the identified model. Design of Generalized Predictive Controller Based on Linear Model. Combining this with nonlinear feedforward, a feedforward compensation generalized predictive self-tuning controller suitable for nonlinear systems is established. The simulation results verify the effectiveness of this controller in the control of nonlinear systems