基于大时滞特性预估补偿的模型辨识及模糊控制方法

Model Identification and Fuzzy Logic Control Method Based on Time-Lag Prediction

对具有大迟延特性的对象采用径向基神经网络辨识其预测控制模型,在不需要知道被控对象的精确物理解析数学模型,也不需要知道对象的脉冲激励响应模型和阶跃激励响应模型的情况下,可以实现对系统响应特性的在线辨识.采用分级模糊建模的思想,设计一种分级模糊控制器,可以极大地减小模糊规则基的规模,在分级模糊控制器的设计中,采用共生进化遗传算法对参数寻优,提高了进化速度.仿真试验证明,该方法的效果很好.

For plants to be controlled with great time-delay characteristics, RBF based neural network is used to identify their predictive control models. It is not necessary to know the concise analytic physical models or the pulse stimulation response model or the step stimulation response model of the plants. By this method, the stimulus-response characteristics of the system can be identified on-line. Multiple-stage fuzzy logic modeling method is used to design a fuzzy logic controller. This can greatly reduce the scale of the fuzzy logic base. In the design of multiple-stage controller, symbiotic genetic algorithm is used to optimize the parameters. This can greatly speed the evolution process. Simulations show that above methods work well.