基于神经网络的非线性预测控制的进一步研究

Further Study of Nonlinear PredictiveControl Based on Neural Network

介绍了适于具有纯滞后对象的基于人工神经网络的非线性多步预测控制，提出了具有自补偿功能的非线性预测反馈校正方法。仿真实验证明此方法大大增强了非线性预测控制的鲁棒性，改善了控制系统的动态和静态性能。

Neural network based nonlinear multi-step predictive control for processes withtime-delay is introduced. Then, a self-compensating correction method for non-linear multi-step prediction is presented. The simulation experiments have shown that the algorithm is able to increase greatly the robustness of nonli-near predictive control and improves both the static and dynamic performan-ces of the controlled system.