Energy efficiency is a big concern in industrial sectors. Finding the root cause of anomaly state of energy efficiency can help to improve energy efficiency of industrial systems and therefore save energy cost. In this research, we propose to use transfer entropy (TE) for root cause analysis on energy efficiency of industrial systems. A method, called TE flow, is proposed in that a TE flow from physical measurements of each subsystem to the energy efficiency indicator along timeline is considered as causal strength for diagnosing root cause of anomaly states of energy efficiency of a system. The copula entropy-based nonparametric TE estimator is used in the proposed method. We conducted experiments on real data collected from a compressing air system to verify the proposed method. Experimental results show that the TE flow method successfully identified the root cause of the energy (in)efficiency of the system.

Root Cause Analysis on Energy Efficiency with Transfer Entropy Flow