基于PCA_SVDD的故障检测和自学习辨识精选.pdfVIP

44 4 浙 江 大 学 学 报() Vol.44 No.4 2010 4 Journal of Zhejiang University ( Engineering Science) pr. 2010 DOI: 10.3785/j.issn. 1008973X.2010. 04.005 PCASVDD 1 2 1 祝志博, 王培良 , 宋执环 ( 1. ,, 310027; 2. , 313000) : , .( PC ) PC ,3 ( PC SVDD): , 2 , . Tennessee Eastman(T E) , . : ( PC SVDD) ; ;; : T P 277 : : 1008973X( 2010) PCASVDD based fault detection and selflearning identification 1 2 1 ZHU Zhibo , W NG Peiliang , SONG Zhihuan (1. State Key Laboratory of Industrial Control Technology , I nstitute of Industrial Process Control, Zhej iang University , angz hou 310027, China; 2. School of I nf ormation Engineering, uzhou Teachers College, uz hou 313000, China) Abstract: In order to utilize the advantage of fault detection and to overcome the shortage of fault identifi cation and diagnosis in the multivariable statistical process control, a new algorithm for chemical process fault detection and selflearning identification was proposed. Principal component analysis (PC ) was ap plied for fault detection and feature extraction, threepattern discriminant methods based on principal com ponent analysissupport vector data description ( PC SVDD) for fault selflearning identification were proposed. Considering the faults distribution overlapping problem which happened possibly in fault identi fication, two methods based on Euclid distance and unitary radius were analyzed and compared. method based on weighted unitary radius for novel fault identification was also proposed. case study of T ennes see Eastman (TE) process showed the feasibility and effectiveness of the proposed fault