Relative Discharge Stage (RDS) Classification: A Practical Indicator of Battery Discharge Progress

TL;DR

This paper introduces Relative Discharge Stage (RDS), a five-class indicator that assesses battery discharge progress without needing future load data. It combines physics-based state estimation with a lightweight temporal CNN to classify RDS accurately under dynamic conditions.

Why It Matters

RDS provides a practical, interpretable way to monitor battery health during discharge, improving real-time decision-making in applications like electric vehicles and grid storage where predicting exact remaining time is unreliable.

Key Findings

• RDS classification achieves over 80% accuracy across diverse load and temperature conditions.
• The framework integrates second-order equivalent circuit modeling with adaptive parameter estimation and neural residual correction.
• A sliding window of current, voltage, temperature, and SOC fed into a lightweight temporal CNN enables robust RDS inference.