welding preheat engineering AI data analysis I spent 1 month reverse-engineering an industrial thermal preheating process with AI. Not because I had a massive budget. Because I wanted to close the gap between guessing and physics. Here's the workflow 🧵 --- 1/ The problem: In heavy manufacturing, preheating thick metal components before robotic welding is critical. Too often, shops rely on tribal knowledge "leave it in for a while" leading to thermal distortion or suboptimal fusion. I wanted a deterministic formula, not vibes. --- 2/ The setup: → Standard 3-phase industrial oven → 12-channel data logger → 50,000+ data points collected over a week → Local Python scripts + Cursor AI for data analysis No enterprise software. Just raw data and LLM-assisted scripting. --- 3/ What the data revealed: The internal thermodynamics were highly non-linear: • The zone nearest the heating elements hit target temp roughly 45% faster than the protected center zone.…