基于遗传神经网络的锅炉入炉煤质软测量研究

STUDY ON SOFT-MEASUREMENT OF QUALITY FOR FURNACEENTERING COAL BASED ON GENETIC NEURAL NETWORK

基于软测量的非机理建模原理,利用遗传算法结合人工神经网络建立了煤质在线软测量模型。确定了BP网络与遗传算法(GA)两者结合的建模方式。分析了原煤从进入制粉系统到完全燃烧及排出的整个过程,得到BP网络模型的输入和输出节点参数集;通过对权参数初值进行实数编码,设计了基于实数编码的GA-BP算法流程,并在Visual Studio2005开发平台上进行了GA-BP算法程序编制及其调试。使用山西某电厂的200 MW机组实时运行数据进行模型训练和检测,结果表明煤质软测量模型可以较为准确地预测煤质参数。

Based on the non-mechanism modelling principle for soft-measurement,an on-line soft-measuring model used for coal quality has been established by using genetic algorithm（GA） combined with artificial neural network.The BP neural network and GA have been presented,and the modelling mode of their combination being determined.The whole process,including the raw coal entering into the coal-pulverizing system to being fully burnt-up and discharged away has been analysed,obtaining a collection of parameters at the input and output nodal points,through actual number programming of initial values in weighing parameters,the process of GA-BP algorithm based on actual number programming has been designed,and the working-out of GA-BP algorithm programming and its debugging are completed on the developing platform Visual Studio 2005.By using real-time operation data of a 200 MW unit in one power plant of Shanxi Province,the model training and detection have been carried out,results show that the said model can more accurately predict the coal quality parameters.