Overcoming limited battery data challenges: A coupled neural network approach

Abstract

The electric vehicle (EV) industry has seen extraordinary growth in the last few years. This is primarily due to an ever increasing awareness of the detrimental environmental effects of fossil fuel powered vehicles and availability of inexpensive lithium-ion batteries (LIBs). In order to safely deploy these LIBs in electric vehicles, certain battery states need to be constantly monitored to ensure safe and healthy operation. The use of machine learning to estimate battery states such as state-of-charge and state-of-health have become an extremely active area of research. However, limited availability of open-source diverse datasets has stifled the growth of this field, and is a problem largely ignored in the literature. In this work, we propose a novel method of timeseries battery data augmentation using deep neural networks. We introduce and analyze the method of using two neural networks working together to alternatively produce synthetic charging and discharging battery profiles. One model produces battery charging profiles, and another produces battery discharging profiles. The proposed approach is evaluated using few public battery datasets to illustrate its effectiveness, and our results show the efficacy of this approach to solve the challenges of limited battery data. We also test this approach on dynamic electric vehicle drive cycles as well.

Estimation of electric vehicle battery capacity requirements based on synthetic cycles

“…However, most studies focus on charge profiles, such as (Schäuble et al, 2017), and, to the author's knowledge, only a couple have generated synthetic current profiles from real driving cycles. In one of these studies, two coupled feedforward Neural Networks were employed to model the charge and discharge profiles (Herle et al, 2021). The Markov Chain has also been used to generate synthetic current profiles, which uses transition probability matrixes to predict the current based on the previous state (Pyne et al, 2019).…”

A comprehensive review on the state of charge estimation for lithium‐ion battery based on neural network

“…Herle et al argued that the limited availability of data sets limits the development of the field of SOC estimation. 93 Acceptable estimation results with only a small amount of data could be achieved by coupling two neural networks. Table 6 lists the comparison of the previous studies.…”

Battery Voltage Prediction Technology Using Machine Learning Model with High Extrapolation Accuracy

“…Models that use big data and machine learning (ML) models to predict state of health (SOH) and state of charge (SOC) have been actively studied [14][15][16][17]. As mentioned earlier, battery voltage prediction with high accuracy is necessary to estimate battery performance, and some studies have been conducted using timeseries data and ML models [18][19][20][21].…”