基于BP网络变量选择与LSSVM结合的锅炉脱硝经济性建模

发布时间：2018-02-28 16:06

本文关键词： 脱硝成本变量选择 BP网络锅炉最小二乘支持向量机(LSSVM)　出处：《中国电机工程学报》2015年24期 　论文类型：期刊论文

【摘要】：大型燃煤电站锅炉在进行烟气脱硝时会产生较高的运行成本,建立有效的脱硝成本预测模型是对其进行经济性优化的基础。从某660MW火电机组的厂级监控信息系统(supervisory information system,SIS)选取历史运行数据,将BP神经网络算法与最小二乘支持向量机(least squares support vector machine,LSSVM)结合,利用BP网络算法对输入变量进行选择以降低模型的维数与复杂度,将筛选出来的变量作为LSSVM模型的输入,建立了脱硝成本预测的BP-LSSVM模型。将该模型与单纯的LSSVM模型进行对比,结果表明通过神经网络变量选择,能有效降低模型的复杂度,提高模型的泛化能力,从而实现火电机组脱硝成本精确快速地预测。
[Abstract]:During flue gas denitrification, large coal-fired power plant boilers will have higher operating costs. To establish an effective denitrification cost prediction model is the basis for economic optimization of a 660MW thermal power unit. The historical operation data are selected from the supervisory information system SISs, a supervisory information system at the plant level of a 660MW thermal power unit. The BP neural network algorithm is combined with the least square support vector machine (LSSVM). The input variables are selected by BP neural network algorithm to reduce the dimension and complexity of the model, and the selected variables are used as the input of the LSSVM model. The BP-LSSVM model of denitrification cost prediction is established and compared with the simple LSSVM model. The results show that the neural network variable selection can effectively reduce the complexity of the model and improve the generalization ability of the model. Thus the cost of denitrification of thermal power units can be predicted accurately and quickly.
【作者单位】：工业过程测控新技术与系统北京市重点实验室(华北电力大学);
【基金】：国家重点基础研究发展计划项目(973计划)(2012CBC215203)~~
【分类号】：X773

【参考文献】