Heinrich, FelixFelixHeinrichLehmann, ThomasThomasLehmannJonas, KonstantinKonstantinJonasPruckner, MarcoMarcoPruckner2022-03-142022-03-142021https://publica.fraunhofer.de/handle/publica/411383To assure safety and performance for battery electric vehicles it is important to monitor the battery state of health. This state is influenced by many different conditions such as state of charge (SOC), time in service, and electrical/ thermal/ mechanical stress. Recent approaches for battery state estimation base on physical models and are subject to lab data under certain testing conditions. However, the stet of health is rarely considered in battery state modeling. In this work, we present a data driven method to continuously estimate the battery state of health (SOH) by applying neural networks on real life in-vehicle time series data. We thereby aim to automatize the complex and time-consuming modeling process. Special adjustments regarding design, composition, and aging of the cell can be omitted.enbattery state estimationneural networksstate of healthbattery modelingbattery monitoring629004conference paper