Wind Turbine Anomaly Dashboard
📊 Wind Turbine Anomaly Dashboard
Turbine.SCOPE is a production-grade interactive dashboard built with Streamlit, designed to monitor and evaluate the operational health of wind turbines.
It combines SCADA data, machine learning residuals, and visual analytics to help engineers and analysts: - Detect anomalies in turbine signals - Rank turbines by urgency (Priority Score) - Explore interactive plots for deeper diagnosis
🧠 Key Features
- Machine Learning Residuals: Uses model predictions vs actuals to compute deviations.
- Priority Scoring: Combines average residual, trend slope, and out-of-bound frequency into a risk score.
- Heatmaps & Boxplots: Visual summaries of turbine-level anomalies.
- Outlier Detection: Identifies and marks out-of-bound values.
- Dynamic Filtering: Select by plant, signal, date range, and turbine.
🛠️ Tech Stack
- Frontend: Streamlit
- Backend: Microsoft SQL Server
- ML Backend: Pre-generated model predictions and bounds
- Visualization: Plotly, Seaborn, Matplotlib
- Data Layer: SCADA time series + model-generated results
🧮 Sample Logic
-
Residual Calculation:
Residual = Actual - Prediction -
Out-of-Bounds Detection:
Flag ifResidual > UpperBound -
Trendline Slope:
Captures performance trend over time via linear fit. -
Priority Score:
🖥️ Screenshots (Optional)
Add screenshots later like: - Heatmap of Residuals - Prediction vs Actual Chart - Priority Table View
🔐 Notes
- Data and connections are anonymized.
- Based on real operational SCADA data from wind turbines.
- Part of an internal monitoring toolkit for predictive maintenance.