Realtime Battery State of Charge Estimation on Embedded devices

In this project, realtime Machine Learning based Battery state of charge (SOC) estimation is performed on embedded devices. Typically WSN and IoT devices are focused (mostly Texas Instrument CC13xx 26xx family, as well as ARM cortex M4). Extensive battery discharge data was collected for various load and operational profiles for typical WSN devices for multiple batteries. This unique dataset is one of its kind especially for actual working WSN devices working under real-life conditions. The datasets were then used to develop LSTM based networks that were fine-tuned using Genetic Algorithms for each battery as well as load type. The dataset, along with the code, as well as step-by-step instructions are provided to replicate the experiment and extend by applying other techniques for SOC estimation improvements.