A novel EFSM-based ELM double-faults identification approach and its application to non-linear processes

Fault identification is a meaningful field that meets the demands of industrial safe and stable. There are lots of researches on single fault issues. However, multi-fault really exists in non-linear process and it has a higher risk than single fault. Aiming to solve double-faults identification problem, an extended finite state machine (EFSM) based extreme learning machine (ELM) double-fault identification approach has been proposed. First, EFSM model is introduced to analyze the correlation between process variables and process states by establishing the variable dependence diagram and state dependence diagram. Second, according to the EFSM-based time series analysis, an ELM prediction model is constructed to predict the process variables. Third, based on the prediction results, the state transition rule in EFSM is used to recognize the faults. Through the verification on Tennessee Eastman (TE) benchmark process, the results indicate that the new approach has higher reliability and practicability in multi-faults identification.

[1]  Yuan Xu,et al.  A New Feedback DE‐ELM with Time Delay‐Based EFSM Approach for Fault Prediction of Non‐Linear Processes , 2015 .

[2]  Yan-Lin He,et al.  A Monte Carlo and PSO based virtual sample generation method for enhancing the energy prediction and energy optimization on small data problem: An empirical study of petrochemical industries , 2017 .

[3]  Chee Kheong Siew,et al.  Extreme learning machine: Theory and applications , 2006, Neurocomputing.

[4]  Jing-Rong Li,et al.  A Rough Set Approach to the Ordering of Basic Events in a Fault Tree for Fault Diagnosis , 2001 .

[5]  Sigurd Skogestad,et al.  Self-optimizing control of a large-scale plant: The Tennessee Eastman process , 2001 .

[6]  M. Hallett,et al.  Linear and nonlinear information flow based on time-delayed mutual information method and its application to corticomuscular interaction , 2010, Clinical Neurophysiology.

[7]  Yan-Lin He,et al.  Soft-sensing model development using PLSR-based dynamic extreme learning machine with an enhanced hidden layer , 2016 .

[8]  Seoung Bum Kim,et al.  Time series forecasting based on wavelet filtering , 2015, Expert Syst. Appl..

[9]  S.M. Virk,et al.  Fault Prediction Using Artificial Neural Network and Fuzzy Logic , 2008, 2008 Seventh Mexican International Conference on Artificial Intelligence.

[10]  Jin Li,et al.  Fault prediction of the nonlinear systems with uncertainty , 2008, Simul. Model. Pract. Theory.

[11]  Chia-Hung Lin,et al.  Dissolved gases forecast to enhance oil‐immersed transformer fault diagnosis with grey prediction–clustering analysis , 2011, Expert Syst. J. Knowl. Eng..

[12]  Robert M. Hierons,et al.  An integrated search-based approach for automatic testing from extended finite state machine (EFSM) models , 2011, Inf. Softw. Technol..

[13]  Yanlin He,et al.  A PSO based virtual sample generation method for small sample sets: Applications to regression datasets , 2017, Eng. Appl. Artif. Intell..