Multivariable Modeling Of Bed Temperature For 300MW Circulating Fluidized Bed Boiler

Circulating fluidized bed technology relative to pulverized coal furnace combustion with high efficiency,wide range of material selection,and low sulfide production,the development of thermal power plants in the future,can improve the plant production efficiency,reduce the coal burning loss and cost,reduce the corresponding emissions of pollutants,the application will be more extensive.However,in the application of circulating fluidized bed,it is found that the control of bed temperature is a difficult problem,so how to choose more reasonable influencing factors to establish a high-quality model,is very important for circulating fluidized bed boiler.In this thesis,the structure and working principle of circulating fluidized bed boiler are briefly introduced,and the influencing factors of bed temperature are analyzed in detail.Combining theory and practice,this thesis applies a method called principal component analysis,which can find out the main reasons influencing the change of process monitoring parameters or the decrease of indicators in the change of a large number of process variables,and is used to reduce the dimension of high-dimensional data space.It is divided into subspaces by combining it with model accuracy verification.The number of pivot elements in the subspace is determined according to the contribution rate accumulation and percentage score of pivot elements,and it is used as the input of mathematical model in the subspace.In addition,due to the uncertainty of quantum particle swarm optimization,random spatial search and randomness,ergodicity and regularity of chaos in optimization design,chaos quantum particle swarm optimization algorithm is obtained to establish the model of bed temperature.The main purpose of writing this thesis is based on the bed temperature characteristics of circulating fluidized bed,mainly focus on how to more reasonable delimit scope of sub-window,using the method of principal component analysis as a basis for the sub-model partitioning,find a dynamic can draw different stage of the main factors influencing degree of different bed temperature,achieve the purpose of improve the model generalization ability,and then through specific examples of the analysis process,described the process of how to build a child model,have a certain application value.