Project 02

Neural Network Based Battery Management System for Electric Vehicles

A battery management system based on LSTM neural networks, taking real-time voltage, current, and temperature as inputs to predict battery State of Charge. Includes a custom embedded device and mobile monitoring app.

System Overview

BMS system architecture showing electric vehicle, embedded system, upper computer, and mobile APP

System architecture — embedded device connected to EV battery pack with mobile app monitoring

Hardware Design

Custom PCB board with digital and analog circuits for BMS

Custom PCB — digital and analog circuits

Hardware architecture with TI chip, STM32, power electronics

Hardware architecture — TI chip, STM32, power electronics

Technical Details

Collected charging and discharging data of lithium iron phosphate battery for electric vehicles
Selected voltage, current, and temperature as inputs and State of Charge as output to train an LSTM network
Designed and programmed an embedded device based on TI chip and STM32 microprocessor
Developed a mobile phone app to monitor the battery pack in real time
Led a six-member team
Applied for one invention patent and one utility model patent as the first inventor

Special Prize — 2019 "Challenge Cup" Shaanxi Province (over 7,000 groups) · Patent No. ZL201910261973.3

← Back to all projects