Based On Fuzzy Nerve Recursion Network Generalized Predictive Control Algorithm Research

Generalized predictive control algorithm (GPC) has robustness strongly, with the superior characteristic to be able to overcome the system lag effectively, therefore it has obtained many success applications in the commercial run control, but the existing generalized predictive control algorithm mostly is aims at the linear system, but the complex industrial system often contains the nonlinear response. Regarding the nonlinear system, many methods used is the misalignment model operating point nearby through the partial linearization. This kind of main existence’s question cannot the very good solution when online computation load is big. if the system has the very strong nonlinear,it often not to be able to achieve satisfaction the control effect. Recent year, research and the progress in the fuzzy neural network has provided a new way to the commercial run control and the modelling. The fuzzy neural network both may carry on the fuzzy reasoning and to be possible to look like the neural network to carry on the study equally, has the very strong misalignment to approach ability, this enables the fuzzy model to be possible the very good description object nonlinear dynamic characteristic. Paper including the following content.First, the predictive control developing process is introduced. And generalized predictive control law is introduced with its characteristic, as well as generalized predictive control domestic and foreign development and research present situation and main application domain, as well as present existence question and limitation. Thus draws out this article to discuss and the research issue.Second, the generalized predictive control algorithm is introduced, how to establishment forecast model as well as the algorithm inferential reasoning and the solution process. Then the generalized predictive control algorithm’s key parameter which is the significance and how to choose and includes is introduced in. And the generalized predictive control algorithm stability and robustness has been discussed.Third, introduced the neural network and fuzzy control’s type and the characteristic, how did study have used the fuzzy nerve recursion network to carry on the model identification to the nonlinear system.Finally, A kind of recurrent fuzzy neural network(RFNN) is constructed,in which,the bility of the Input information handling is enhanced by adding the vector adjustment layer. Based on recursion fuzzy neural network which designs, nonlinear system’s discrete mathematics steps fuzzy forecast model is established.According to this model which is used to forecast the system’s output, the corresponding forecast control law is obtained by the existing predictive control algorithm. Then discussed this method robustness and noise-immune ability. In this process improved in the generalized predictive control algorithm control weighting factor, the use dynamic control weighting factor has enabled the control to achieve a better effect.