Abstract

Root cause analysis is an important step in process monitoring. In the related research field, the popular causality analysis technique, transfer entropy (TE), has been adopted for its capability to handle process nonlinearity. Its improved version, conditional TE (CTE), is better suited to multivariate cases, because of its capability of neglecting indirect causal relations. Nevertheless, the conventional CTE is often sensitive to noise, which hampers its performance. This problem can be solved by using the concept of symbolization. Statistical process control (SPC) is an industry-standard methodology for determining process performance. Therefore, it is reasonable to symbolize process measurements based on the SPC information. In this work, a control chart-based symbolic CTE method is proposed, which conducts CTE to analyze the symbolized process data and reveal the causality among variables. Then, the identified causality information is visualized on a causal map. The method performance is quantified with several indices.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call