An Improved STS Model Based Nonlinear GPC Method Orienting with Furnace Pressure Control

Furnace pressure is one of the important variable to be controlled in the process of glass furnace operation. Owing to the influence of a large number of uncertainties, existent control algorithms tend to produce overshooting and frequent fluctuation in follow-up. Considering that simplified T-S model (STS) has a strong self-adaptive learning capability and subtractive clustering is able to handle the unsteadiness in the optimal model structure, this paper proposes, for the purpose of furnace pressure control, a generalized fuzzy nonlinear prediction approach by combining subtractive clustering and STS. This approach allows a quick approximation of the object following jumps of furnace pressure and may eliminate system unsteadiness caused by jumps. Additionally, in respect of excessive overshooting in furnace pressure step following-up, a control method is designed based on the improved STS model and integrating global optimal rolling optimization function with optimal control increment algorithm so that overshooting is restrained essentially. This approach involves little computation work and is easy to implement.