锅炉飞灰BP-WA算法优化控制策略  

Optimal control strategy of BP-WA algorithm for fly ash of generator set

作　　者：章家岩 陈雨薇 宋澜波 冯旭刚 ZHANG Jiayan;CHEN Yuwei;SONG Lanbo;FENG Xugang(College of Electrical Engineering and Information,Anhui University of Technology,Maanshan 243032,China;Energy center of Hunan Valin Lianyuan Steel Co.,Ltd.,Loudi 417009,China)

机构地区：[1]安徽工业大学电气与信息工程学院,安徽马鞍山243032 [2]湖南华菱涟源钢铁有限公司能源总厂,湖南娄底417009

出　　处：《湖北大学学报（自然科学版）》2022年第5期514-521,共8页Journal of Hubei University：Natural Science

基　　金：安徽省自然科学基金(1908085ME134)资助。

摘　　要：针对锅炉飞灰含碳量在线测量参数多变、惯性大等问题,设计一种改进型BP神经网络飞灰含碳量预测模型.通过主元分析法分析各燃烧工况与飞灰含碳量的关系,利用信息熵将标准BP神经网络中的误差函数进行改进,以抑制输入样本中的干扰噪声,并采用主元分析法筛选模型中输入参数,精简网络模型.结合所提出的改进型BP-WA(BP神经网络-狼群算法)优化控制策略对锅炉燃烧运行工况进行优化控制仿真研究,结果表明:采用改进型BP-WA优化控制策略优化飞灰含碳量前后,锅炉飞灰含碳量预测与标准BP网络模型方法相比,均方误差降低0.012 1;飞灰含碳量降低3.50%,提升了锅炉运行的稳定性.Aiming at the problems of variable online measurement parameters and large inertia of fly ash carbon content in coal-fired power units, A modified BP neural network fly ash carbon content prediction model was designed. According to the field process, the relationship between each combustion condition and fly ash carbon content was analyzed by principal element analysis, the error function in the standard BP neural network was improved by using information entropy to suppress the interference noise in the input samples, and the input parameters in the model were screened by principal element analysis to streamline the network model.Combined with the proposed improved BP-WA optimal control strategy, the optimal control simulation of the combustion operation condition of the generator set was carried out.The results show that the mean square error of the generator set fly ash carbon content prediction is reduced by 0.012 1 compared with the standard BP network model method before and after optimizing the fly ash carbon content using the improved BP-WA optimal control strategy. The carbon content in fly ash was reduced by 3.48%,and the operation stability of the generator set is improved.

关 键 词：飞灰含碳量 BP神经网络 主元分析 燃烧优化 狼群算法 

分 类 号：TP318[自动化与计算机技术—计算机软件与理论]