The Modeling Of Fuzzy Neural Network And Generalized Predictive Control Of 330MW Circulating Fluidized Bed Boiler

As is well known, the coal resource is unevenly distributed in china which the low-quality coal makes a higher proportion more than 50%. Circulating Fluidized Bed Boiler (CFBB) is developed rapidly in recent years for its advantages of high combustion efficiency, low pollution and fuel flexibility. It can burn a variety of low-quality coal and achieve furnace desulfurization, so the CFBB plays an important role in modern power plants. However, CFBB is a complex thermal system with multiple-input-multiple-output, strong coupling, nonlinearity, and large time delay, so it is difficult to get its accurate analytical model. Besides, complex analytical models are not suitable for most control strategies. Therefore, how to get accurate and practical models of a CFBB, and how to control a CFBB are always research concerns.For the 330MW CFBB, this paper studies the characteristics and the modeling of circulating fluidized bed boiler combustion, the coupled between multi variables, and the model predictive control strategy. The thesis investigates modeling and controlling in the following aspects:(1) Simplified CFBB’s model appropriately after analyzing its characteristics, and make coal feed quantity, primary air flow,turbine valve position as the inputs, while bed temperature, main steam pressure, output power as the outputs. This dynamic model is based on data-driven modeling approach, which using the T-S fuzzy reasoning and triangle membership function, is better than the models built from recursive least squares method and the RBF neural network method. The simulation results demonstrate that the proposed T-S fuzzy neural network identification approach is of high accuracy and compactness, and suitable for on-line modeling and prediction even under large range of change in the load.(2) This paper proposed an advanced generalized predictive control (GPC) strategy under the T-S fuzzy neural network model, and applied it to CFBB’s main steam pressure control, bed temperature control, power control. Simulation results show that the effectiveness of GPC. Besides, the power can also ensure good track performance which better than traditional PI controller.In the end, the paper analyzes deficiency and questions in research, and the research directions in the future.