Research On Intelligent Algorithms In Model Identification Of 300MW Circulating Fluidized Bed Boiler

At present, circulating fluidized Bed (CFB) Technology is the considerably spread and widely used in the thermal power plants in China, as it is in the advantage that the CFB can realize the combustion with low quality fuels and direct desulfurization in boiler, which is corresponding to the development of clean coal combustion technique advised by our government. Therefore, the modeling and automatic control issues have become the general interest in the thermal control researching area. In this paper, the thermal characteristic of combustion system and bed pressure system are analyzed theoretically, and the occurrence of bed-overturn phenomenon in CFB boiler with the features of underpants-leg structure has been presented firstly. Secondly, identification modeling research is based on data mining. Mining data from the historical database try to avoid any effects in running equipment, according to the actual operation. Then the principles of cross-correlation function and the experience of data selection are determined by utilized in actual production analysis in order to acquire appropriate operating data of each subsystem. The improved genetic algorithm which is based on real-coding, multi-operators cooperation and rank-based selection of crossover and mutation probability is applied to 300MW ALSTOM CFB boiler modeling establishment. The linear transfer function models of combustion system and bed pressure system in different load operating point are identified. Finally, the characteristic of bed material imbalance 300MW ALSTOM CFB boiler with underpants-leg structure have also been identified.300MW ALSTOM CFB boiler modeling is foundation in controller design and optimization, while it is developing to gain the object function method via data mining from historical database in industrial process. This method has significance guide for actual operation, with attempting to solve the practical engineering problem existing in industrial process modeling technology.