Performs prognostic assessment on vibration signals: trend analysis, degradation onset detection, and RUL estimation using predictive-maintenance-mcp server.
How this skill is triggered — by the user, by Claude, or both
Slash command
/predictive-maintenance:prognosticsThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Estimate degradation trends and remaining useful life from vibration signals.
Estimate degradation trends and remaining useful life from vibration signals. Orchestrate MCP tools in a precise prognostic sequence.
Prerequisite: The predictive-maintenance-mcp MCP server must be connected.
Call list_stored_signals() to check cached signals, or list_signals() to
browse the data/ directory. Identify the signal file to analyze.
The prognostics tools segment a single signal file internally, extracting a feature series over time from consecutive segments. This works best on run-to-failure recordings or long monitoring sessions where degradation evolves within the signal.
Call analyze_signal_trend(signal_file=..., feature_name="rms").
The tool automatically segments the signal, extracts the chosen feature per segment, and fits a linear trend.
Parameters:
feature_name: degradation indicator — "rms" (default), "kurtosis",
"crest_factor", "peak_to_peak", etc.sampling_rate: auto-detected from metadata if not providedsegment_duration: segment length in seconds (default: 0.1s)overlap_ratio: overlap between segments (default: 0.5)Good feature candidates:
| Indicator | Use when | Notes |
|---|---|---|
| rms | General degradation | Most common, ISO 20816 aligned |
| kurtosis | Bearing degradation | Peaks early, then may drop |
| crest_factor | Impulsive faults | Sensitive to early damage |
| peak_to_peak | Looseness, imbalance | Good for mechanical looseness |
Ask the user which indicator to track, or default to rms if unsure.
Interpretation:
| R-squared | Trend direction | Meaning |
|---|---|---|
| > 0.7 | Increasing | Strong degradation trend |
| 0.3 - 0.7 | Increasing | Moderate trend, monitor closely |
| < 0.3 | Any | No clear trend (or non-linear) |
| Any | Decreasing | Improving or post-maintenance |
| Any | Stable | Stationary condition |
Decision gate: If trend_direction is "stable" or "decreasing", report that no degradation trend is detected. Ask if the user wants to proceed with RUL estimation anyway.
Call detect_signal_degradation_onset(signal_file=..., feature_name="rms", threshold_sigma=3.0).
If onset is detected, report:
onset_segment_index)Call estimate_rul(signal_file=..., failure_threshold=..., method="linear").
Choosing the failure threshold:
Choosing sampling_interval:
Choosing method:
"linear" — simpler, works for steady degradation (default)"exponential" — for accelerating degradation curves"weibull" — Weibull distribution fit (if available)"kalman" — Kalman filter estimation (if available)If RUL is infinity or confidence is 0, the curve doesn't reach the threshold. Explain why and suggest:
Combine all findings into a concise prognostic summary:
Recommended actions based on RUL:
| RUL | Urgency | Recommendation |
|---|---|---|
| < 1 interval | Critical | Immediate inspection/replacement |
| 1-3 intervals | High | Schedule maintenance soon |
| 3-10 intervals | Moderate | Monitor more frequently |
| > 10 intervals | Low | Continue routine monitoring |
npx claudepluginhub lgdimaggio/predictive-maintenance-mcp --plugin predictive-maintenancePerforms fast vibration health screening for industrial machinery using predictive-maintenance-mcp server. Generates reports with RMS, crest factor, FFT peaks, ISO 20816 zones, and recommendations.
Zero-shot univariate time-series forecasting with Google's TimesFM foundation model. Produces point forecasts and prediction intervals from CSV/DataFrame/array inputs with a preflight system checker.
Monitors deployed model performance, detects data drift, and manages model health using the DataRobot Python SDK. Use for tracking prediction accuracy, feature drift, and prediction anomalies.