Research Of Improved Fuzzy Generalized Predictive Control Method For Glass Furnace System

Due to the affection of large number of uncertain factors, horseshoe flame glass furnace thermal system is a complex parameters uncertain multi-variable coupled nonlinear system, of which effective mechanism model cannot be established. The existing control algorithms which require high precision model tend to cause a poor control performance with large overshoot and frequent fluctuations, which was difficult to meet the needs of the production process. Fuzzy Generalized Predictive Control is a powerful tool for solving nonlinear system modeling and control problem, however when dealing with the furnace object, it appears to be slightly weak. By analyzing the characteristics of the furnace, this paper targeted in-depth study on the basis of the fuzzy generalized predictive control in order to build a suitable furnace modeling and control algorithms, the main contents are as follows:1. In view of T-S model’modeling unstable characteristic and slow online learning speed, a fuzzy model based on simplify T-S model with a sub-cluster method was developed. Using the sub-cluster method to ensure the optimal model structure and the initial value of the model to solve the problem of instability modeling; through the combination of sub-cluster a simplified T-S model (STS), we designed an improved STS fuzzy model (ISTS), to speed up the fuzzy model’s online learning rate.2. Because of the large amount of computation and the control overshoot problem of traditional generalized predictive control, a small-overshoot quick generalized predictive control (SOQGPC) strategy was designed. SOQGPC is a kind of multiple-step prediction and one-step control method due to the combination of weighted rolling optimization function and soften control law.3. In view of Furnace pressure parameters uncertain nonlinear system control problem, a kind of fuzzy generalized predictive control methods based on ISTS model (ISTS-QGPC) were studied. The ISTS model was used as the predictive model of generalized predictive control algorithm to enhance capa-bility of tracking the controlled object characteristics’hopping.4. In order to solve the multivariable control problem of glass furnace parameters uncertain nonlinear thermal system, we formulated a weighted quick multivariable fuzzy generalized predictive control based on ISTS model (ISTS-WQMGPC). The multivariable ISTS model was used to enhance capa-bility of tracking the controlled object characteristics’hopping; the WQMGPC was a strategy to simplify the computation process of traditional MGPC.