Enhancement of Stress Cycle-counting Algorithms for Li-ion Batteries by means of Fuzzy Logic

Abstract

The rainflow algorithm is one of the most commonly used tools for studying stress conditions of a wide variety of systems, including power electronics devices and electrochemical batteries. One of the main drawbacks of the algorithm is the trade-off between data compression and the loss of information when classifying the stress cycles into a finite amount of histogram bins. This paper proposes a novel approach for classifying the stress cycles by using fuzzy logic in order to reduce the quantization error of the traditional histogram-based analysis. The method is tested by comparing the accumulated damage estimations of two support-vector regression algorithms when trained with each type of cycle-counting procedure. NASA’s randomized battery usage data set is used as source of stress data. A 50% error reduction was observed when using the fuzzy logic-based approach compared to the traditional one. Thus, the proposed method can effectively improve the accuracy of diagnosis algorithms without penalizing their performance and memory-saving features.