Systematic Comparison of Deep Learning Approaches for Vibration-Based Bolt Looseness Detection: Why Attention Matters More Than Features
Five deep-learning approaches compared on the same shaft-vibration dataset. An SE-attention CNN hits 93% accuracy and 87% faulty recall; removing attention collapses recall to 15%.
Register to read the full white paper
Enter your details to unlock the complete report — free, no spam.
Get maintenance insights in your inbox
Join operators getting practical CMMS tips, case studies, and product updates. No spam.
Related Articles
FFT & Vibration Monitoring: The Foundation of Predictive Maintenance in Industry 4.0
Machine failures rarely happen without warning. FFT (Fast Fourier Transform) and vibration monitoring enable maintenance teams to identify developing faults before they become costly breakdowns. Learn how frequency analysis helps detect imbalance, misalignment, bearing defects, gear wear, and other machine issues while supporting predictive maintenance initiatives through OpexMX.
White PaperA Cheap, Safe, High-Precision RAG Pipeline: Architecture Notes | OpexMX
How we wired a copilot + retrieval pipeline for a maintenance CMMS that is cheap per request, safe against prompt injection, and precise at retrieval — without a frontier model on every turn. Six stages, cheapest-first, one generation call.
White PaperClean-Room PDF Extraction at the Edge: How OpexMX Runs pdftotext Inside a Cloudflare Worker for RAG | OpexMX Blog
Industrial manuals use subset CID fonts with no ToUnicode map, so every JavaScript PDF library emits glyph garbage — and RAG over garbage text is garbage retrieval. Here is how OpexMX wrote its own clean-room PDF extractor in Rust, compiled it to WASM, and runs it inside the edge Worker that also embeds and indexes the chunks.