基于神经模糊系统的自适应前馈-反馈控制系统设计

在前馈控制器设计思想的启发下,提出了一种基于神经模糊系统的自适应前馈-反馈控制系统。该控制系统首先把非线性过程近似为一个线性的ARX模型和一个基于神经模糊系统的线性化误差模型(FNNM)组成的合成模型,把线性化误差模型的输出看作可测量的“扰动”,然后再引入前馈控制器,利用被控制过程的输入、误差模型的输出、线性ARX模型输出和系统输出值之间的误差以及被控制过程的合成模型的梯度信息对控制器参数进行在线调节,从而获得较好的控制结果。将提出的基于线性化误差模型的自适应控制系统用于简单不可逆放热反应的连续搅拌型化学反应器CSTR中,并与传统的PID控制器进行比较。仿真结果表明：这种基于神经模糊系统的自适应前馈反馈控制器和PID控制器相比,能得到更快、更好的控制效果。

Adaptive Feedback-Feedforward Control Scheme Design Based on Neuro-Fuzzy System

Motivated by the designing idea of conventional feedforward control, an adaptive feedback-feedforward control scheme based on neuro-fuzzy system is proposed. Firstly, a class of SISO nonlinear processes is approximated by a composite model consisting of a linear ARX model and a FNN-based linea-rization error model (FNNM). The output of FNNM is taken as the measurable "disturbance". And then, a feedforward controller is employed, whose parameters are adjusted by using the input of the controlled process, the output of the error model, the output error between the linear ARX model and the system, and the gradient information of the composite model. Finally, both the proposed controller and the conventional PID are used in CSTR. Simulation results show that the proposed controller has better performance than the conventional PID.