Causal Network Analysis and Fault Root Point Detection Based on Symbolic Transfer Entropy

Abstract

Transfer entropy (TE) is a model-free method based on data-driven information theory. It can obtain causal relationships between variables. It has been used for modeling, monitoring and fault diagnosis of complex industrial processes. It can detect the causal relationship between variables without the need to assume any underlying model, but its calculation process is complicated and the calculation time is long. In order to overcome this limitation, symbol transfer entropy is proposed. The symbol transfer entropy is robust and fast to calculate. It can also quantify the dominant direction of information flow between time series with identical and non-identical coupling systems, thereby improving the accuracy of causal paths. Sex. Through the symbolic transfer of entropy, a causal network diagram can be obtained, and the root cause of the fault can be found. The effectiveness and accuracy of the method are verified by simulation and actual industrial cases (Tennessee-Eastman process)