Remaining useful life prediction and state of health diagnosis for lithium-ion batteries based on improved grey wolf optimization algorithm-deep extreme learning machine algorithm.

The prediction of SOH for Lithium-ion battery systems determines the safety of Electric vehicles and stationary energy storage devices powered by LIBs. State of health diagnosis and remaining useful life prediction also rely significantly on excellent algorithms and effective indicators extraction. Since the data obtained from the aging experiment of Lithium-ion batteries is rich in electrochemical and dynamic information, useful health indicators can be extracted for SOH and RUL prediction of machine learning. This paper presents a method for predicting SOH and RUL based on a data-driven model of deep extreme learning machine based on improved Grey Wolf optimization algorithm. Firstly, GWO algorithm is improved by piecewise chaotic distribution and sine-cosine algorithm, and then multi-layer superposition is performed on an extreme learning machine to form DELM. Additionally, the experimental data of the Center for Advanced Life Cycle Engineering data set was extracted and analyzed, the aging state of batteries was analyzed and verified from multiple scales, and the strong correlation of aging characteristics was extracted and verified. After that, the model was driven by the extracted health indicators, and the accuracy and robustness of the results were checked