A Comparative Study of Recurrent Neural Network Architectures for Battery Voltage Prediction

Electrification is the well-accepted solution to address carbon emissions and modernize vehicle controls. Batteries play a critical in the journey of electrification and modernization with battery voltage prediction as the foundation for safe and efficient operation. Due to its strong dependency on the prior information, the battery voltage was estimated with recurrent neural network methods in the recent literatures exploring variety deep learning techniques to estimate battery behaviors. In these studies, standard recurrent neural networks, gated recurrent units, and long-short term memory are popular neural network architectures under review. However, in most cases, each neural network architecture is individually assessed and therefore the knowledge about comparative study among three neural network architecture is limited. This paper presents a comparative study on the battery voltage predication using all three neural network architectures. In this study, all neural network architectures use common pulse data and cycle data for training and validation. Then the predictions are made in severe battery operation conditions such as high current, low temperature, and long duration is investigated. The results indicate that long-short term memory achieves the highest accuracy among the three architectures.