Battery Capacity Estimation based on Convolutional Neural Network

Estimating the battery capacity is an efficient way to monitor battery usage in real time. The data-driven capacity estimation methods proposed in the current literature usually require all the voltage, temperature, current data, etc. during the battery operation. However, obtaining such a large amount of data is unrealistic in practical application scenarios. Therefore, in this paper, a convolutional neural network is proposed, which needs only part of the voltage data during the battery discharge process, to estimate the battery capacity. Our model is tested on the Li(NiMnCo)O 2 / carbon battery, and experimental results show that the proposed convolutional neural network model can accurately estimate the battery capacity with limited voltage data.