A nonlinear process fault detection and diagnosis method was presented based on the information divergence in order to utilize the nonGaussian fault characteristic in Kalman filter innovation. An unscented Kalman filter was used to produce the innovation sequence, and the kernel density estimation was applied to obtain the probability distribution of the innovation. Then an information divergence statistic was constructed to monitor the state of the process. Once a fault was detected, a symmetric information divergence was introduced to isolate the fault by measuring the distance between the monitored process and the fault process in fault database. Simulation results on the continuous stirred tank reactor demonstrate that the method can detect the fault in time and correctly distinguish the fault type.
［1］ 周东华,胡艳艳.动态系统的故障诊断技术［J］.自动化学报,2009,35(6): 748758.
ZHOU Donghua, HU Yanyan. Fault diagnosis technigues for dynamic systems ［J］.Acta Automatica Sinica,2009,35(6): 748758.
［2］ 陈国金,梁军,刘育明,等.基于多元统计投影方法的过程监控技术研究［J］.浙江大学学报：工学版,2004,38(12): 15611565.
CHEN Guojin, LIANG Jun, LIU Yuming, et al. Process monitoring based on multivariate statistical projection analysis ［J］. Journal of Zhejiang University: Engineering Science, 2004, 38(12): 15611565.
［3］ PATWARDHAN S, MANUJA S, NARASIMHAN S, et al. From data to diagnosis and control using generalized orthonormal basis filters. Part II: model predictive and fault tolerant control ［J］. Journal of Process Control, 2006, 16(2): 157175.
［4］ BENKOUIDER A, BUVAT J, COSMAO J, et al. Fault detection in semibatch reactor using the EKF and statistical method ［J］. Journal of Loss Prevention in the Process Industries, 2009, 22(2): 153161.
［5］ 张鹏,黄金泉.航空发动机气路故障诊断的平方根UKF方法研究［J］.航空动力学报,2008,23(1): 169173.
ZHANG Peng,HUANG Jinquan. RUKF research on aeroengines for gas path component fault diagnostic ［J］. Journal of Aerospace Power, 2008, 23(1): 169173.［6］ 葛哲学,杨拥民,温熙森.强跟踪UKF方法及其在故障辨识中的应用,仪器仪表学报,2008,29(8): 16701674.
GE Zhexue, YANG Yongmin, WEN Xisen. Strong tracking UKF method and its application in fault identification ［J］. Chinese Journal of Scientific Instrument,2008,29(8): 16701674.
［7］ LIN J. Divergence measures based on the Shannon entropy ［J］. IEEE Transactions on Information Theory, 1991, 37(1): 145151.
［8］ SILVERMAN B. Density estimation for statistics and data analysis ［M］. London: Chapman & Hall, 2000: 7577.
［9］ JULIER S, UHLMANN J. Unscented filtering and nonlinear estimation ［J］. Proceedings of the IEEE, 2004, 92(3): 401422.
［10］ JURICEK B, SEBORG D, LARIMORE W. Predictive monitoring for abnormal situation management ［J］. Journal of Process Control, 2001, 11(2): 111128.
