The Comparison Of Several Nonlinear Control Algorithms And Process Control Experiment Test

At present, the traditional control algorithms represented by the PID control and the advanced control algorithms represented by the MPC control algorithm which develops in recent decades have been very mature both in theory and practical industrial applications. These algorithms are aimed at linear control process from the beginning of their emergence,so they are called linear control algorithm. However, almost all the actual controlled process are nonlinear, and the only difference between them is that the degree of nonlinearity is strong or weak. For those wake nonlinear object, we can deal with them as linear systems. But when the nonlinear is very strong, the measure mentioned above is short of out request. More over, most of the time we have a high control demand. In face of this condition, linear control algorithm has been outstripping.In view of the difficulties of the linear control algorithm, nonlinear control algorithm has become a hot topic in the control field in recent years, and this article will pay much attention to it. And in this article, the focus of the content is about modeling for several nonlinear control algorithms, because a good model that can reflect the characteristics of the control process is the foundation of control algorithm implication .Neural network has good approximation properties for nonlinear system and the characteristics of self-learning, so this paper introduces two most common neural network identification algorithm, and on the basis, this article describes three kinds of application for neural network in the field of nonlinear process control. In third chapter, we introduce a class of experiment model, focusing on the most common two of them—Hammerstein model and Wiener model. Beside, in this part we put forward a kind of nonlinear model predictive control algorithm based on Hammerstein model, and the algorithm will be simulated in a pH process to verify the effectiveness of the algorithm. The last part of the thesis is about experiment studies of process control. So in order to realize that, we must actualize the data communication between MATLAB and configuration software through the OPC technology, we can establish such a platform via this way, in which the control algorithm can realize process control of a A3000 experiment system. On this basis above, we achieve the control experiment studies of the previous several non-linear control algorithm aimed at a nonlinear single-tank, which can verify these algorithms'efficiency in the actual process control system.