A novel entropy-based fault diagnosis and inconsistency evaluation approach for lithium-ion battery energy storage systems

Detection and diagnosis of faults at the early stage, as well as inconsistency monitoring and control are of extreme importance for operating Li-ion batteries (LIBs) safely and reliably, handling performance degradation and cell unbalancing, and avoiding accidents like thermal runaway (TR). In this work, a general procedure based on multi-level Shannon entropy algorithms is put forward to perform fault diagnosis as well as inconsistency evaluation for LIB-based energy storage systems (ESSs). More specifically, the cell-level Shannon entropy algorithm is used to detect faults by comparing Shannon entropies of different LIB cells in each module while the module-level and cluster-level Shannon entropy algorithms are used to evaluate the overall inconsistency among LIB cells in each module and in each cluster respectively. The proposed approach is then applied in a large-scale LIB-based ESS (1 MW/2 MWh). Through simulated data, the availability of the cell-level Shannon entropy algorithm to detect small changes in gradual faults is testified while the module-level and the cluster-level Shannon entropy algorithms are demonstrated to be effective for assessing inconsistences of LIBs in every module and in every cluster respectively, by comparing results of the normal case with those from two cases each with a different faulty LIB cell at the early stage of internal short circuit (ISC).

Automated summary

This study introduces a general procedure based on Shannon entropy algorithms for fault diagnosis and inconsistency evaluation of LIB-based energy storage systems. Through simulated data, the effectiveness of different level Shannon entropy algorithms in detecting and evaluating LIB cells is demonstrated.