误差补偿和时滞辨识预测控制算法  

Predictive Control Algorithm with Error Estimation and Time Delay Identification

作　　者：王家祥[1] 何信[2] 

机构地区：[1]天津大学管理学院,天津300072 [2]北京石油化工学院信管系,北京102617

出　　处：《电子科技大学学报》2005年第6期817-820,共4页Journal of University of Electronic Science and Technology of China

摘　　要：通过神经网络辨别出非线性系统的滞后时间,并采用误差补偿手段对神经网络进行修正,控制系统的反馈取自不带时滞的非线性神经网络的输出,采用预测控制策略改善整个系统的控制性能。仿真实验表明该方法具有较快的响应速度和较强的自适应性与鲁棒性,能有效克服延迟和干扰给控制品质带来的不利影响,取得了良好的控制性能。Competitive advantage of Artificial Neural Network （ANN） compensation control is that it can overcome the time delay, which is of vital importance in process control. ANN identify time lag in the system, and is modified through error compensation method. Feedback of control system comes from the output of nonlinear ANN without time lag. Predictive control strategy is used to improve the performance of overall system. Simulation test shows the algorithm has little response time and strong adaptability and robustness; it can overcome the degradation introduced by time delay and disturbance in system; fin e performance of control is acquired.

关 键 词：时滞 辨识 神经网络 误差补偿 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]