摘要
为优化锅炉配风配粉降低NO x排放浓度,以某1000 MW火电机组现场变工况运行数据为基础,采用正弦算法(Sine algorithm,SA)优化的正则化极限学习机(Regularized extreme learning machine,RELM)建立了NO x排放量的预测模型。为提高预测模型的精度,通过比较不同激活函数对RELM模型性能的影响,选用了新的Swish激活函数;采用奇异值分解法确定RELM的最佳隐含层节点个数,并引入一种自适应调整惯性权重的正弦算法对RELM的输入权值和阈值进行优化。将基于以上策略建立的SA-RELM模型与SA-ELM、RELM及PSO-RELM等模型的预测结果进行对比,表明基于SA-RELM的NO x排放量预测模型具有更高的精度和泛化能力。
In order to optimize boiler air and fuel distribution and reduce its NO x emission concentration.We established a prediction model for boiler NO x emission based on the sine algorithm(SA)optimized regularized extreme learning machine(RELM)with field operation data of a given 1000 MW thermal power unit under varying work conditions.To improve the predictive accuracy of the model,we compared the effects of different activation functions on the accuracy of RELM model and adopted a new Swish activation function.The singular value decomposition method was used to determine the optimal number of hidden layer nodes in RELM,and a sine algorithm with adaptive adjustment of inertia weight was employed to optimize the input weights and thresholds of RELM.Finally,we compared the prediction result of the SA-RELM NO x model based on above strategies with that of SA-ELM,RELM and PSO-RELM models.The result indicates that the prediction model of NO x emission based on SA-RELM has better accuracy and generalization ability.
作者
马良玉
程善珍
王永军
MA Liangyu;CHENG Shanzhen;WANG Yongjun(School of Control and Computer Engineering,North China Electric Power University,Baoding 071003,China;Shandong Electrical Engineering&Equipment Group Co.,Ltd.,Jinan 250002,China)
出处
《华北电力大学学报（自然科学版）》
CAS
北大核心
2022年第3期112-118,共7页
Journal of North China Electric Power University：Natural Science Edition
