机理混合自适应时延神经网络建模和控制算法

过程机理和数据驱动模型的结合一直是过程控制中的重点和难点问题。提出了一种新颖的机理混合模型结构。该模型基于自适应时延神经网络,并结合了被控对象的机理模型,因此可以对建模的对象进行预测感知,并可大幅提高神经网络的泛化功能。基于这种模型设计了预测控制算法,并分析了该模型的逼近能力、收敛性以及该控制算法的闭环稳定性。在双容水箱液位控制系统的大量实验结果表明,基于该混合模型的预测控制算法比现有算法具有更好的控制效果,从而验证了该混合模型的优越性。

An Algorithm of Modeling and Control Based on Mechanism Hybrid Adaptive Time Delay Neural Network

The integration of process mechanism and data-driven model is an important and difficult problem in the field of process control all along. A novel structure of mechanism hybrid model is presented. This model is based on adaptive time delay neural network (ATDNN) combined with process mechanism model, so it can predict the plant to be modeled and can improve the generalization ability of the neural network greatly. A predictive control algorithm is designed based on this model, and the approach ability and convergence of this model are analyzed. Moreover, the close loop stability of this control algorithm is analyzed as well. A great deal of experimental results, which are done on the double tank water level control system, show that the control performance of the predictive control algorithm based on this hybrid model is superior to the counterpart of the predictive control algorithm based on traditional feedforward neural network model. Therefore, the superiority of this hybrid model is validated.