基于BP神经网络的污泥酸性发酵过程预测  

Prediction of Sludge Acido-Genic Fermentation Process Based on BP Neural Network

作　　者：石杰[1] 宋秀兰[1] 李亚新[1] 

机构地区：[1]太原理工大学环境科学与工程学院,山西太原030024

出　　处：《中北大学学报（自然科学版）》2015年第4期467-474,共8页Journal of North University of China(Natural Science Edition)

基　　金：国家自然科学基金资助项目(50278060);山西省科技攻关项目(20140313003-1)

摘　　要：为了预测污泥产酸能力及产酸组成,对生物脱氮除磷过程所需的碳源投加量的计算有意义.用两种BP神经网络模型对采用半连续运行方式的污泥酸性发酵过程进行预测.结果表明,8-23-4BP网络除乙酸产率以外,丙酸产率、容积产酸能力、丁酸产率预测值与实测值相关系数都大于0.9,各目标变量预测值与实测值平均相对误差都小于15%;7-22-4BP网络乙酸产率、丙酸产率、丁酸产率、容积产酸能力预测值与实测值相关系数都大于0.9,各目标变量预测值与实测值平均相对误差都小于13%.7-22-4BP模型网络预测性能优于8-23-4BP网络模型.8-23-4BP模型对污泥丙酸产率预测更精确.两种BP神经网络模型可以实现对污泥酸性发酵过程产酸量的预测.实验结果可为污水处理厂污泥酸性发酵的在线控制提供参考.Predicting sludge acid yield and fermentation product composition is of importance to calculate carbon source dosage of biological nutrient removal process.Two kinds of BP neural network models were used to predict sludge acid fermentation process at the condition of semi-continuous operation.In addition to acetic acid yield rate,the correlation coefficients for propionic,butyric acids yield rates and total acid productivity are higher than 0.9in 8-23-4BP network;these average relative errors between predicted value and actual value are less than 15%.These correlation coefficients for acetic,propionic,butyric acids yield rates and total acid productivity are higher than 0.9in 7-22-4BP network;these average relative errors between predicted value and actual value are less than 13%.7-22-4BP network is superior to 8-23-4BP network.8-23-4BP network is more accurate in predicting propionic acid yield rate.Two kinds of BP neural network models can realize acid yield prediction in sludge acidogenic fermentation process.It provides reference for on-line control of sludge acid fermentation.

关 键 词：污水处理厂污泥 酸性发酵 神经网络 

分 类 号：X703.1[环境科学与工程—环境工程] TP183[自动化与计算机技术—控制理论与控制工程]