The Battery Management System with DNN Deployed in STM32

Accurate prediction of the state of charge (SOC) is crucial for extending the lifespan of batteries. While traditional ampere-hour integration methods are simple and practical, they suffer from large cumulative errors and are not well-suited for complex application scenarios. This paper presents a smart battery management system based on the STM32F407VET6 microcontroller, which incorporates a deep neural network (DNN) to effectively model the nonlinear relationship between key parameters such as current, voltage, and SOC. The system features autonomous operation without requiring a host computer, enabling precise measurement of remaining battery capacity and real-time data display on an OLED screen. Furthermore, it supports data transmission in industrial environments via an RS485 module using the Modbus RTU protocol, as well as communication through CAN bus and ESP8266-MQTT for IoT applications. Experimental results demonstrate that this system can accurately predict battery SOC while facilitating multiplatform data communication.