Data-based Adaptive Fault Prediction Method and Its Application

摘要：

Multi-level recursive method is an adaptive and datadriven fault prediction process. In terms of inputoutput equivalence, a nonlinear model can be modified into a multi-level linearized model using the multilevel recursive method. The time-varying characteristics of model parameters are accounted for at the same time. Therefore, the proposed approach obtained satisfied results when utilized in prediction issues. The fault prediction for CSTR (Continuous Stirred Tank Reactor) system has been studied based on the integrated multi-level recursive forecasting method which considering the CSTR systems dynamic & time-variable characteristics. The optimal match of models and the algorithm of multilevel recursive method have been investigated through simulation. Through used in digital simulation experiments, the proposed method which is specific for CSTR system fault prediction has been validated and proved to be effective. This method can be used to predict the faults in such a class of nonlinear time-varying systems. Hence applying the proposed method in engineering and industry is proved to be feasible.

关键词： data-driven integrated multi-level recursive forecasting method CSTR system fault prediction technique.

作者: Jie Ma Di Li Shaohong Wang Xiaoli Xu

作者单位: Automation College,Beijing Information Science and Technology University,Beijing 100192,China School of Astronautics,Harbin Institute of Technology,Harbin,Heilongjiang 150001,China

会议类型: 国际会议

会议名称: 2009 9th International Conference on Electronic Measurement & Instruments(第九届电子测量与仪器国际会议 ICEMI2009)

会议地点: 北京

会议语种:英文

页码: 4239-4244

在线出版日期: 2009-08-16（万方平台首次上网日期，不代表论文的发表时间）