Selbstgesteuertes Projekt
Predictive Maintenance: Industrial ML for Fault Detection
Condition-Monitoring Engineer · KI / ML · August 2026
Eine ProoV-Fallstudie · Lernprojekt, kein Arbeitsverhältnis
Built an end-to-end rolling-bearing fault detector using vibration snapshots, physics-based time and frequency features, and a compact decision tree sized for embedded use. The work includes a held-out evaluation and a maintenance memo that justifies the alarm threshold with explicit cost trade-offs, catch-rate, and false-alarm figures, while identifying the outer-race fault from the dominant defect frequency.
Bewertet nach
- Feature engineering (signal to honest numbers)25%
- Held-out evaluation (honest measurement)25%
- Cost-based threshold decision + memo25%
- Compact model + embedded-budget justification15%
- Plain-language explanation10%
Bestanden · Bestehensgrenze 60/100
Was herausstach6
- Used a depth-capped decision tree and explicitly stated depth 2 with 5 nodes against an embedded budget.
- Selected kurtosis as the strongest time-domain fault indicator, which aligns with impulsive bearing damage.
- Wrote a maintenance memo that ties the threshold to concrete euro costs and names the outer race from the BPFO peak.
- Included a plain-language analogy to explain why damaged bearings produce distinctive vibration patterns.
- Leakage-Aware Model Building
- Physics-Informed Feature Selection
Meine eingereichte Arbeit17 Aufgaben
9 Aufgaben · 5.6k Zeichen · Python, JavaScript · 8 Textantworten · 452 Wörter
- Initial Prediction62 Zeichen
- Load Plot Signal CodePython · 1.8k Zeichen
- Example Rms WorkedPython · 697 Zeichen
- Time Features143 Zeichen
- Freq FeaturesPython · 1.1k Zeichen
- Train Classifier363 Zeichen
- Evaluation Heldout283 Zeichen
- Threshold Decision335 Zeichen
- Decision MemoJavaScript · 835 Zeichen
- Initial Prediction9 Wörter
- Load Plot Signal Code163 Wörter
- Example Rms Worked72 Wörter
- Time Features10 Wörter
- Freq Features117 Wörter
- Train Classifier31 Wörter
- Evaluation Heldout24 Wörter
- Threshold Decision26 Wörter
Bewusst zusammengefasst — der eingereichte Code und die Texte bleiben privat, damit diese Seite nicht als Lösungsvorlage dient. Die vollständige Einreichung steht hinter dem verifizierten Zertifikat.