基于BP神经网络的生物质固定床热解气化过程模拟

为研究生物质的热解气化规律,基于BP人工神经网络原理,利用Matlab神经网络工具箱,以试验得到的两种生物质54组试验数据作为样本,建立了以停留时间、水分、温度、催化剂种类和催化剂用量为输入变量,气、液、固产物产率和不同种类气体组成为输出变量的生物质固定床热解气化过程模型。模型输出的7个变量的预测结果与试验数据吻合较好,证明该模型对生物质热解气化过程模拟的可行性和有效性。

BP neural network simulation of biomass pyrolysis gasification in a fixed-bed reactor

A BP neural network was built to simulate the pyrolysis gasification process of biomass in a fixed-bed reactor by using Matlab neural network toolbox.Five input variables,i.e.residence time,initial moisture of biomass,pyrolysis gasification temperature,the kinds of catalyst and the catalyst weight/biomass weight ratios,and seven output variables,i.e.gas yield,liquid yield,solid yield and four kinds of gas component were selected.54 groups experimental data were taken as training and checking samples.The results show that model-predicted results of seven output variables are in sound agreement with the experimental data.Thereby the neural network model is considered to properly reflect the real pyrolysis gasification process of a biomass.The feasibility and effectiveness of the BP based model are also presented.