一种自适应神经网络控制器的研究

THE RESEARCH ON AN ADAPTIVE NEURAL CONTROLLER

在非线性系统的神经网络模型基础上，以 Ｃｌａｒｔｅ一阶条件为目标函数，运用线性化方法提出了一种自适应预测控制算法。由于利用了神经网络良好的非线性逼近能力和学习特性，所设计的控制器具有较强的自适应能力，能够实现对非线性系统的有效控制。通过仿真证明了算法的有效性。

In this paper an adaptive neural network predicative control algorithm is presented on the basis of the BP neural network model of nonlinear. In the algorithm, Clarke one - step condition is chosen to be the optimal function and linearization method is used. As the neural network has a good property of nonlinear approach and self - learning, our controller is well self- adaptive and able to coned a nonlinear system Simulation results show the effectiveness of our algorithm.