一种基于神经网络的自适应预测控制方法

An Adaptively Predictive Control Method Based on Neural Network

强非线性系统预测控制问题当前仍有许多难点需要解决,而神经网络由于具有非线性逼近能力、自学习、容错性等特点,使它在自动控制理论中得到了广泛的应用。针对非线性系统,提出了一种基于神经网络的自适应预测控制方案,该方案结构简单,易于实现。对于一些难于用传统方法控制的复杂非线性对象提供了一种有效的方法。对非线性定常及时变对象,利用仿真表明了该控制策略的有效性。最后讨论了神经网络控制器发展方向与亟待解决的问题。

There are still many difficulties in controlling strong non-liner system now.Neural network（NN） have many characteristics such as ability of approximating to nonlinear system,self-learning and failure-tolerant etc.Therefore it is applied extensively in automatic control theory.According to the non-liner system,an adaptively predictive control strategy is put forward.Its structure is simple and easy to implement.Simulation results have shown the effectiveness of the control strategy for constant and time-varying object. Finally,the research direction and key problem to be solved are also discussed for Neural Network controller.