State of Health Diagnosis and Remaining Useful Life Prediction of Lithium-Ion Batteries Based on Multi-Feature Data and Mechanism Fusion

Abstract

State of Health (SOH) Diagnosis and Remaining Useful Life (RUL) Prediction of lithium-ion batteries (LIBs) are subject to low accuracy due to the complicated aging mechanism of LIBs. This paper investigates a SOH diagnosis and RUL prediction method to improve prediction accuracy by combining multi-feature data with mechanism fusion. With the approach of the normal particle swarm optimization, a support vector regression (SVR)-based SOH diagnosis model is developed. Compared with existing works, more comprehensive features are utilized as the feature variables, including three aspects: the representative feature during a constant-voltage protocol; the capacity; internal resistance. Further, the optimized regularized particle filter (ORPF) model with uncertainty expression is integrated to obtain more accurate RUL prediction and SOH diagnosis. Experiments validate the effectiveness of the proposed method. Results show that the proposed SOH diagnosis and RUL prediction method has higher accuracy and better stability compared with the traditional methods, which help to realize the decision of the maintenance process.

A hybrid remaining useful life prediction method for lithium-ion batteries based on transfer learning with CDRSN-BiGRU-AM

“…Figure 5 displays the capacity degradation curves of these cells. The capacity of the B7 battery did not reach the EOL threshold, therefore, based on previous experience [36,37], 72% of its rated capacity (1.44 Ah) was set as the EOL threshold.…”

Machine Learning-based Remaining Useful Life Prediction Techniques for Lithium-ion Battery Management Systems: A Comprehensive Review

“…Despite their superior performance in RUL prediction, integration of multiple methods is difficult and often increases the computational burden. Xu et al [48] introduced a normal particle swarm optimization method (NPSO) to balance the global search and local search capabilities. However, the improvement in prediction accuracy is minimal.…”

A Novel Remaining Useful Life Prediction Method for Capacity Diving Lithium-Ion Batteries

“…Relative to the data-driven methods, the mining of information contained in degraded data can be flexibly implemented through numerous data analysis methods without obtaining any information about the internal mechanism of the LIBs. This method includes signal processing (empirical modal decomposition (EMD), 19 and wavelet transform (WT) 20 ), machine learning (support vector machines(SVM), 21 − 23 related vector machines (RVM), 24 artificial neural networks (ANN), 25 long–short-term memory network (LSTM) 26 ), time series (autoregression (AR) 27 and autoregressive integrated moving average 28 ), and statistical analysis methods (Gauss process regression) 29 etc. Nuhic et al 1 used support vector machine embedding to diagnose and predict RUL.…”

“…It uses inner product kernel functions for nonlinear mapping of high-dimensional spaces, superior generalization, and computational speed. A large number of studies have recognized the effective performance and applicability of SVR in hybrid methods. − ,, PF is a probabilistic statistical algorithm that estimates the discriminant parameters by calculating the sample mean of a collection of particles. It does not require knowledge of the noise model of the system to estimate data disturbed by task from noise; it can also handle linear or nonlinear systems.…”