Show HN: EV-QA-Framework – ML-powered QA for electric vehicle battery systems

EV-QA-Framework, an open-source machine learning tool for electric vehicle battery quality assurance, has been released on GitHub. The framework combines rule-based validation with Isolation Forest anomaly detection and LSTM-based State of Health prediction, and includes a CAN bus emulator for testing without physical hardware. The tool aims to help QA engineers and battery researchers catch anomalies, predict degradation, and validate battery performance without expensive test rigs.

Electric vehicle battery systems produce a lot of telemetry data - thousands of readings per minute. EV-QA-Framework helps QA engineers and battery researchers catch anomalies, predict degradation, and validate battery performance without needing expensive test rigs. The framework combines rule-based validation with machine learning Isolation Forest for anomaly detection, LSTM for State of Health prediction , plus a CAN bus emulator so you can test without physical hardware. - ML anomaly detection via Isolation Forest with adjustable sensitivity - State of Health prediction using an LSTM neural network - CAN bus emulation CAN 2.0B and J1939 protocols - Interactive real-time dashboard FastAPI + WebSocket + Chart.js - Configurable safety thresholds per vehicle profile - Save/load trained models as JSON or joblib - 100+ pytest tests with ML validation edge cases pip install -r requirements.txt Run analysis on sample data python -m ev qa framework.cli analyze examples/tesla model s defective.csv Launch dashboard python -m ev qa framework.cli dashboard Generate synthetic CAN data python -m ev qa framework.cli emulate --duration 60 With Docker: docker compose -f docker-compose.prod.yml up -d then open http://localhost:8080 The framework has four main components: Core QA Engine - Pydantic-based validation with configurable safety thresholds ML Analyzer - Isolation Forest for anomaly detection, LSTM for SOH prediction CAN Emulator - Generates CAN 2.0B and J1939 data streams for offline testing Dashboard - FastAPI server with WebSocket-powered real-time charts Data flows: CSV/API input - Pydantic validation - ML analysis - dashboard visualization. python -m ev qa framework.cli analyze examples/tesla model s defective.csv python -m ev qa framework.cli dashboard python -m ev qa framework.cli emulate --duration 120 --protocol j1939 python -m ev qa framework.cli analyze examples/tesla model s defective.csv \ --config config/tesla config.json \ --output report.json \ --save-model docker compose -f docker-compose.prod.yml up -d With custom configuration cp .env.example .env docker compose -f docker-compose.prod.yml --env-file .env up -d Images are published to GitHub Container Registry: ghcr.io/remontsuri/ev-qa-framework:latest pip install -r requirements-dev.txt pytest -v --cov=ev qa framework Specific test suites pytest tests/test ml analysis.py -v pytest tests/test integration.py -v EV-QA-Framework/ ev qa framework/ Core package analysis.py ML anomaly detection cli.py CLI entry point config.py Thresholds and logging setup framework.py Main QA engine models.py Pydantic validation models soh predictor.py LSTM SOH predictor can bus.py CAN bus simulator dashboard/ Web dashboard app.py FastAPI + WebSocket server templates/ Jinja2 frontend api/ REST API routes.py API endpoints config/ Configuration profiles examples/ Usage examples tests/ Test suite - Core QA engine - ML anomaly detection - SOH prediction - CAN bus emulation - Interactive dashboard - Docker deployment - PyPI package - Automated release pipeline - Cell imbalance detection - Thermal runaway prediction - Grafana datasource plugin Bug reports, feature requests, and pull requests are welcome. See CONTRIBUTING.md for the workflow. MIT. See LICENSE for details.