VRLA battery state of health estimation using a stress factor-based machine-learning algorithm
Graduate Research Scientist, TU Berlin
State of health estimation is an important practice for battery management systems. Degrading batteries influence system performance and knowledge of their current state of health help to optimize the operating parameters. In this study, a state of health algorithm based on machine learning is developed. The different stress factors that affect lead acid battery degradation are identified. These are for instance average discharge current, depth of discharge, temperature and average discharge voltage per cycle. Based on a real-world application, a generic ageing test cycle is created and ageing tests are performed using this cycle. The measurements are analysed and a dataset containing the different stress factors is created for a specific period. The capacity of the battery is determined periodically during the test to obtain the current state of health of the battery. Each set of stress factors can be allocated to a specific state of health. This data is used to train a neural network. The neural network can be used to predict the state of health of the battery for the given application.
- received his Bachelors and Masters degree in Electrical Engineering from Technische Universität Berlin in 2014 and 2015 respectively.
- working on the modelling of secondary batteries and battery systems with focus on ageing characteristics.