CAJ | 학술논문

准确检测空气预热器(以下简称空预器)热点对火力发电机组的安全运行具有重要意义.综合利用热电偶和红外传感器的温度信息,通过Dempster-Shafer(DS)证据理论对这两类温度信息进行融合推理,决策当前空预器内部的火情状态.为准确计算各证据体的基本概率,首先利用最小二乘支持向量机(LS-SVM)和Sigmoid函数建立多元分类器,实现对温度测点较精确的类别判断,然后根据类别票数情况计算各证据体的基本概率.实验结果表明:由最小二乘支持向量机(LS-SVM)和Sigmoid函数建立的多元分类器具有较高的分类准确率,所提出的空预器热点检测方法具有较高的判警准确率.
준학검측공기예열기(이하간칭공예기)열점대화력발전궤조적안전운행구유중요의의.종합이용열전우화홍외전감기적온도신식,통과Dempster-Shafer(DS)증거이론대저량류온도신식진행융합추리,결책당전공예기내부적화정상태.위준학계산각증거체적기본개솔,수선이용최소이승지지향량궤(LS-SVM)화Sigmoid함수건립다원분류기,실현대온도측점교정학적유별판단,연후근거유별표수정황계산각증거체적기본개솔.실험결과표명:유최소이승지지향량궤(LS-SVM)화Sigmoid함수건립적다원분류기구유교고적분류준학솔,소제출적공예기열점검측방법구유교고적판경준학솔.
It is important to accurately detect air preheater spots for power plant units. The temperature information of thermocouples and infrared sensors has been comprehensively utilized in this paper. DempsterShafer(DS) vidential theory has been applied to determine present fire status of air preheater by fusing the two kinds of temperature information. To accurately acquire the basic probability of all evidences two steps as follows are necessary. The multivariate classifier has been built based on LS-SVM and Sigmoid function,which can accurately classify the temperature points. The basic probability can be obtained from the votes of classifications. Experimental results show that this multivariate classifier has higher accuracy rate and the detection method of air preheater hot spots is sensible.