动量改进BP神经网络在湿法脱硫系统中的应用  

Application of BP Neural Network Improved by Momentum in Wet Flue Gas Desulfurization

作　　者：刘锁清[1] 翟奕博 张耀华 董森[1] 彭伟娟 

机构地区：[1]山西大学,山西太原030013 [2]华能西宁热电有限责任公司,青海西宁811600

出　　处：《自动化技术与应用》2016年第8期23-25,36,共4页Techniques of Automation and Applications

摘　　要：以国内某350MW超临界机组为例,充分考虑影响石灰石/石膏湿法烟气脱硫效率的主要因素,基于动量改进BP神经网络建立了脱硫效率预测模型。并随机选取该电厂脱硫系统现场运行的15组数据进行仿真预测。结果表明,所建立的烟气脱硫效率预测模型拥有较高的预测精度,对现场运行工作有一定的指导意义。Taking a 350MW supercritical unit as an example, the main factors that affect the efficiency of limestone-gypsum wet flue gas desulfurization （WFGD） are taken into account. The prediction model of desulfurization efficiency is built, it is based on the BP neural network improved by momentum. It is verified with 15 sets of data from operating desulfurization of this power plant randomly. The results show that the prediction model has a higher prediction precision, it has some directive meaning to operation work.

关 键 词：石灰石/石膏湿法脱硫 脱硫效率 BP神经网络 预测模型 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]