计算机与应用化学
計算機與應用化學
계산궤여응용화학
COMPUTERS AND APPLIED CHEMISTRY
2014年
8期
1015-1018
,共4页
非负矩阵分解%支持向量机%故障诊断
非負矩陣分解%支持嚮量機%故障診斷
비부구진분해%지지향량궤%고장진단
Non-negative matrix factorization%support vector machines%fault diagnosis
针对复杂化工过程中存在故障数据维数大与识别率低的问题,提出了1种非负矩阵分解与支持向量机相结合的故障诊断方法。该方法首先对原始特征数据进行非负矩阵分解,得到基向量矩阵与系数矩阵,用基向量矩阵作为输入训练SVM分类器,针对NMF结果的不稳定性,采用PCA模型确定NMF的初始值;然后通过系数矩阵构造超定线性方程组,并将其最小二乘解作为样本特征矩阵输入分类器进行故障类型的识别。通过对Tennessee Eastman (TE)过程数据的仿真研究,实验结果验证了该方法的有效性。
針對複雜化工過程中存在故障數據維數大與識彆率低的問題,提齣瞭1種非負矩陣分解與支持嚮量機相結閤的故障診斷方法。該方法首先對原始特徵數據進行非負矩陣分解,得到基嚮量矩陣與繫數矩陣,用基嚮量矩陣作為輸入訓練SVM分類器,針對NMF結果的不穩定性,採用PCA模型確定NMF的初始值;然後通過繫數矩陣構造超定線性方程組,併將其最小二乘解作為樣本特徵矩陣輸入分類器進行故障類型的識彆。通過對Tennessee Eastman (TE)過程數據的倣真研究,實驗結果驗證瞭該方法的有效性。
침대복잡화공과정중존재고장수거유수대여식별솔저적문제,제출료1충비부구진분해여지지향량궤상결합적고장진단방법。해방법수선대원시특정수거진행비부구진분해,득도기향량구진여계수구진,용기향량구진작위수입훈련SVM분류기,침대NMF결과적불은정성,채용PCA모형학정NMF적초시치;연후통과계수구진구조초정선성방정조,병장기최소이승해작위양본특정구진수입분류기진행고장류형적식별。통과대Tennessee Eastman (TE)과정수거적방진연구,실험결과험증료해방법적유효성。
A fault diagnosis method based on non-negative matrix factorization(NMF) and support vector machine(SVM) is proposed to resolve the problems of high dimensional data and the low efficiency of fault classification in complex chemical process. Firstly, the original feature data are decomposed by NMF, and get base vector matrix and coefficient matrix, and using base vector matrix as input to train the SVM classifier. PCA method is used for getting initial value to avoid the NMF instability. Secondly, constructing an over determined linear equations by coefficient matrix, and its least-squares as feature matrix inputted into support vector machine to identify the faults. The proposed fault diagnosis method is proved to be effective by simulation with the data from Tennessee Eastman process.