Data and code for Synthetic Data Pretraining and Real-world Fine-tuning for Sensor-free Internal Thermal Monitoring of Lithium-ion Batteries

Overview This repository contains datasets and code for internal temperature estimation of lithium-ion batteries by combining physics-based synthetic data with machine learning. Dataset The dataset includes measurements from two cylindrical cells: A123 ANR26650M1B LG INR21700-M50T For each cell, experiments were conducted at -15 °C, 5 °C, and 25 °C, including: Hybrid Pulse Power Characterization (HPPC) test Dynamic current tests: Federal Urban Driving Schedule (FUDS) and the Highway Fuel Economy Test (HWFET) Additionally, the dataset provides: Parameters of the first-order equivalent circuit model (ECM) Thermal parameters of a reduced-order thermal model Code Synthetic data analysis.py → Generates synthetic data and compares it with experimental results. main.py → Pre-trains machine learning models on synthetic data and transfers them to real battery systems via transfer learning.