Leo Zhang
Back to Projects

Device Maintenance Prediction

Chunxiao Technology · 2023

Role: AI Algorithm Developer

Device Maintenance Prediction

Predictive maintenance AI: SparkNet + Random Forest models analyze equipment aging trends and predict next maintenance windows. Feature engineering on intervals, fault frequency, and usage intensity with multi-model comparison.

Predictive maintenance recommendations replacing fixed-cycle scheduling, reusable model pipeline

Problem

Equipment maintenance is typically reactive or fixed-cycle, leading to either premature servicing or unexpected failures.

Solution

ML pipeline: historical maintenance data → feature engineering (intervals, fault frequency, usage intensity) → SparkNet + Random Forest prediction → maintenance window recommendations with risk-priority scoring.

Key Highlights

  • Designed predictive maintenance workflow from historical service records
  • Implemented SparkNet + Random Forest for next-maintenance timing prediction
  • Built feature engineering pipeline for intervals, fault frequency, and usage intensity
  • Multi-model comparison with regression and time-series baselines

Tech Stack

PythonSparkNetRandom ForestRegression ModelsTime-series ForecastingEnsemble ModelingFeature Engineering

What I Learned

Feature engineering on maintenance intervals and usage patterns is more impactful than complex models; ensemble approaches provide more stable predictions than single models.