神经网络模型在煤碳生物脱硫仿真及优化中的应用  

Application of simulation and optimization in coal biodesulphurization based on the neural network method

作　　者：鲁建江[1] 郭志国[1] 徐小琳[1] 

机构地区：[1]石河子大学化学化工学院新疆兵团化工绿色过程重点实验室,石河子832001

出　　处：《四川大学学报（自然科学版）》2013年第5期1079-1084,共6页Journal of Sichuan University(Natural Science Edition)

基　　金：新疆生产建设兵团国际科技合作资助项目(2010YD38);国家自然科学基金(21206095)

摘　　要：通过单因素实验研究了转速、培养温度、初始pH、脱硫时间、煤浆浓度和煤炭颗粒度对煤炭生物脱硫的影响,建立了煤炭生物脱硫反应过程的BP神经网络模型.研究结果表明,神经网络训练中的模型检验均方误差接近1×10-3,模型检验样本预测输出值和试验值的决定相关系数达到0.9997,表明该模型对煤炭生物脱硫过程仿真及结果预测效果良好;采用遗传算法工具箱对建立的BP神经网络模型进行优化求解,在最优条件下得到的脱硫率为47.6%,该结果经实验验证具有可靠稳定性.Effects of rotation speed, culture temperature, initial pH, desulfurization time, pulp density, coal particle size on coal biodesulphurization reaction process were investigated by single-factor experi ments, and a BP artificial neural network model was established for coal biodesulphurization. The re suits showed that the mean square error of the model test was close to 1 X 10-3 , and the determination coefficient between the output value predicted by the model and the experimental value was worked out as 0. 9997, which indicated that the model was reliable enough to indicating the model established and could well and truly forecast the coal biodesulphurization. Genetic algorithm was applied to optimize the biodesulphurization conditions based on the BP artificial neural network model. The desulfurization rate at the optimized conditions was 47.6 %, the result indicated stability and reliablity by experiment verif- ying.

关 键 词：煤炭 生物脱硫 BP神经网络 模拟与优化 

分 类 号：TQ536.1[化学工程—煤化学工程]