Review Work on Machine Learning Approaches for Predicting the Remaining Lifespan of Lithium-Ion Batteries

2. Application Of Machine Learning For Forecasting The Remaining Service Life (Rsl)

In the past few years, the domain of Machine Learning (ML) has witnessed substantial expansion, presenting numerous techniques relevant to predicting the Remaining Service Life (RSL) in batteries, as detailed in references (Khumprom & Yodo, 2019; Liu et al., 2017). To employ ML methods, it is imperative to extract crucial raw data from batteries, encompassing variables like current (I), voltage (V), and temperature (T), which demonstrate fluctuations throughout the battery's aging progression.

Figure 1.

Data processing flow (Jin et al., 2021)

ML models perform various tasks, including data preprocessing, network training, and Analysing the dataset is a crucial step in predicting Remaining Service Life (RSL) of the battery. The connection between the tested data values and the Remaining Service Life (RSL) is visualized in the the flow of data processing, as illustrated in Figure 1 (Jin et al., 2021). Achieving high accuracy and fostering independent learning capabilities in the desired output data are essential objectives. After data extraction, the trained ML models establish a simulated relationship between the retrieved data and Remaining Service Life (RSL), facilitating precise predictions for battery RSL. Figure 2 enumerates different types of ML techniques employed context.

Figure 2.

Various ML techniques