State of health prediction of lithium-ion batteries using a temporal convolutional network

In 2016, over 190 countries signed the Paris Agreement, committing to reducing greenhouse gases. To achieve this goal, we need to transition to sustainable energies, but without reliable and safe energy storage, the switch will not work. Lithium-ion batteries, with their high conversion efficiency, are being used to store energy, but without accurate methods for predicting the state of health (SOH), dangerous situations can arise during operation. Therefore, in this bachelor thesis we investigated the suitability of temporal convolutional networks (TCNs) for SOH prediction. We found that TCNs are as accurate as recurrent neural networks and are even less computationally intensive. We concluded that TCNs are a simple and ecological method for SOH prediction.