Root Cause Diagnosis with Error Correction Model Based Granger Causality

With the rise of data science, data-based fault root diagnosis methods have attracted widespread attention. Among these methods, the Granger causalitytest is one of the most common methods which can infer causal associations between signals based on temporal precedence. However, there are some strong constraints when using this method. First, the time series analyzed should be stationary. Besides, the GC is based on the linear model. In the actual process, the system is often nonlinear, and the time series caused by the fault are mostly nonstationary. In this paper, error correction model is introduced into the root cause diagnosis to solve the problem that Granger causality can't be applied to non-stationary time series analysis directly. The effectiveness of the proposed method is illustrated by two cases of TE process.