基于GWO-BP神经网络算法的WFGD系统在线优化  

Online Optimization of a WFGD System Based on GWO-BP Neural Network Algorithm

作　　者：王涛[1] 任少君[1] 司风琪[1] 马利君 王力光 Wang Tao;Ren Shaojun;Si Fengqi;Ma Lijun;Wang Liguang(Key Laboratory of Energy Thermal Conversion and Control,Ministry of Education,Southeast University,Nanjing 210096,China;Franchise Branch of Datang Environmental Industry Group Co.,Ltd.,Nanjing 210096,China)

机构地区：[1]东南大学能源热转换及其过程测控教育部重点实验室,南京210096 [2]大唐环境产业集团股份有限公司特许经营分公司,南京210096

出　　处：《发电设备》2021年第2期122-130,共9页Power Equipment

摘　　要：以浆液循环泵运行情况作为工况划分条件,通过提出的灰狼优化(GWO)-BP神经网络(GWO-BP神经网络)算法建立了针对湿法烟气脱硫(WFGD)系统多模态在线优化模型组,分析了机组负荷、入口SO_(2)质量浓度对出口SO2质量浓度变化量的影响,并利用某660 MW机组切换试验对该模型组性能进行验证。结果表明:随着机组负荷和入口SO_(2)质量浓度增大,切换后出口SO_(2)质量浓度变化量增大,该模型组具有较好的预测精度和泛化能力。Based on operation conditions of slurry circulation pumps,a group of multi-modality models were established for online optimization of a wet flue gas desulfurization(WFGD)system by the grey wolf-optimized BP(GWO-BP)neural network algorithm proposed,and subsequently the effects of the following factors on the variation of outlet SO_(2) concentration were analyzed,such as the unit load and inlet SO_(2) concentration,etc.In addition,the performance of the group of models was verified through the switching test of a 660 MW unit.Results show that with the increase of unit load and inlet SO_(2) concentration,the change of outlet SO_(2) concentration increases after switching the pump,indicating high prediction accuracy and strong generalization ability of the group of models.

关 键 词：湿法烟气脱硫 浆液循环泵 BP神经网络 GWO算法 

分 类 号：TM621[电气工程—电力系统及自动化] TP183[自动化与计算机技术—控制理论与控制工程]