【作者】 姜子安； 易辉； 江艳； 冒泽慧；

【Author】 Zian Jiang;Hui Yi;Yan Jiang;Zehui Mao;College of Automation and Electronic Engineering,Nanjing University of Technology;Nanjing University of Aeronautics and Astronautics;

【机构】 南京工业大学电气工程与控制科学学院； 南京航空航天大学；

【摘要】 为更准确的预测选择性催化还原(SCR)反应器入口NOx浓度,提出了基于主成分分析法(PCA)和天牛须算法(BAS)优化的支持向量机(SVM)模型。基于某600 MW电厂采集到的数据,由于影响反应器入口处NOx浓度的辅助变量之间存在相关性,首先利用SPSS软件进行主成分分析对13个辅助变量进行降维处理,以降低预测模型的维数及消除变量间的相关性。将主成分作为支持向量机的输入向量,通过天牛须算法寻优获得最优支持向量机模型参数,最终建立基于PCA-BAS-SVM的NOx浓度预测模型,并将此模型性能与BAS-SVM预测模型以及BAS-BP预测模型进行对比分析。结果表明:使用PCA-BAS-SVM模型的平均绝对百分比误差、均方根误差、迭代时间分别为0.2%、1.541、4.62 s,其平均绝对百分比误差相比BAS-BP、BAS-SVM模型分别降低0.23%、0.75%,提高了预测精度;同时由于PCA-BAS-SVM模型进行了降维处理,迭代也是用时最少的,能够明显提高运行效率;证明该模型应用在NOx浓度预测中具有一定的有效性。更多还原

【Abstract】 In order to predict the NOx concentration at the inlet of the SCR reactor more accurately,a support vector machine(SVM) model based on the principal component analysis(PCA) and the beetle antennae algorithm(BAS) was proposed.Based on data collected by a 600 MW power plant,because there is a correlation between the auxiliary variables that affect the NOx concentration at the reactor inlet,firstly,SPSS software was used to perform principal component analysis to reduce the dimensions of 13 auxiliary variables to reduce the dimensionality of the prediction model and eliminate the correlation between the variables.The principal component is used as the input vector of the support vector machine,and the optimal support vector machine model parameters are obtained through the BAS algorithm,and finally the NOx concentration prediction model based on PCA-BAS-SVM is established,and the performance of this model is compared with BAS-SVM prediction model and the BAS-BP prediction model.The results show that the average absolute percentage error,root mean square error,and iteration time of the PCA-BAS-SVM model are 0.2%,1.541,and 4.62 s,the average absolute percentage error is 0.23% and 0.75% lower than that of BAS-BP model and BAS-SVM model respectively.the prediction accuracy is improved;at the same time,the dimension of PCA-BAS-SVM model is reduced,the iteration is also takes the least time,which can significantly improve the operating efficiency of the model;it proves that the application of the model is effective in the prediction of NOx concentration.更多还原