Simulation of Anaerobic Fermentation Biodegrading Corn Straw by Using Back Propagation-neural Network (BP-NN)
Anaerobic fermentation was of great significance for the treatment of wastewater/wastes,and production of bio energy.Due to the complexity of anaerobic processes with straw-like substrate,the utilization process and biogas production behavior always presents complex nonlinear relation,leading to the difficulty of the simulation of the process.In this study,a novel integrated bioreactor (IBR) was operated by using corn straw and then reached the maximum biogas production rate of 10.08 L/d and the highest COD removal rate of 91.3%.A back propagation neural network (BPNN) model was built to simulate and predict the reactor performance,which consisted of input layer (4 nodes: influent COD.corn straw content,ORP and pH),hidden layer (6 nodes) and output layer (2 nodes: COD removal rate and biogas production rate).After 1831 epochs,the MSE of the BPNN model was 0.0099983,which can meet the setting accuracy standard of 0.01.While comparing the predictive values with real values,the maximum and minimum deviation were 5.31% and 0.13% respectively for COD removal rate,and 8.71% and 0.30% respectively for biogas production rate.The results indicated that the BPNN model can well describe the anaerobic fermentation process.By using weight separation method,the relative importance of input parameters follows: influent COD > pH > ORP > corn straw content with the values of 27.79%,26.04%,25.55%,and 20.62%,respectively.The similar importance for influent COD,ORP,and pH indicated that the anaerobic fermentation might be multiply influenced by various important parameters.
anaerobic fermentation BP neural network corn straw weight separation method
Zhang Guolei Li Zhuo Shi Yue Dai Linchong
College of Power and Energy Engineering, Harbin Engineering University, Harbin 150001, China
国际会议
2013青岛国际脱盐大会(Qingdao International Conference 2013 on Desalination and Water Reuse)
青岛
英文
335-340
2013-06-25（万方平台首次上网日期，不代表论文的发表时间）