2 comments

  • lgdimaggio 2 hours ago
    Hi HN, I built an open-source MCP server for vibration-based predictive maintenance.

    The idea: the LLM handles conversation and orchestration; the server runs deterministic vibration analysis and writes an interactive local report.

    Repo: https://github.com/LGDiMaggio/predictive-maintenance-mcp

    Quick try: - git clone https://github.com/LGDiMaggio/predictive-maintenance-mcp.git - cd predictive-maintenance-mcp - python setup_venv.py - python validate_server.py

    What’s included: - real bearing vibration signals (healthy + inner/outer race faults) - spectrum peak detection + envelope analysis - ISO 20816-3 severity assessment - (baseline) anomaly detection like OneClassSVM/LOF - local HTML reports under reports/

    Looking for feedback on: MCP tool design, missing diagnostics to prioritize next, and whether the combo LLM with deterministic tooling is useful in real PdM workflows.

  • JojoFatsani 1 hour ago
    This is probably the AI we need in cars instead of the one that will tell you Taylor Swift’s shoe size.
    • lgdimaggio 1 hour ago
      Thanks, that’s exactly the intent. Automotive would be interesting, but it would need real-time constraints. Happy to get feedback on what signals would be most valuable