"An intelligent desktop ecosystem for metro maintenance, integrating ML.NET for predictive fault forecasting, SQLite for resilient data management, and automated risk scoring."
Infrastructure reliability is mission-critical. This project is a Predictive Maintenance System designed to transform reactive fault logs into proactive operational insights. By implementing an ML.NET FastTree regression/binary classification model, the system analyzes historical patterns to forecast future fault probabilities. This bridge between traditional desktop CRUD operations and Machine Learning allows for data-driven scheduling in high-stakes metro environments.
This application demonstrates proficiency in Industrial Software Architecture:
- Predictive Modeling (ML.NET): Implementing FastTree algorithms with One-Hot Encoding to process categorical station data and temporal features, enabling millisecond-speed fault probability forecasting.
- Resilient Data Layer: Utilizing SQLite for lightweight yet robust local persistence, featuring an automated one-click backup mechanism (
metro_backup.db) to ensure zero data loss. - Algorithmic Risk Scoring: Developing a dynamic calculation engine that monitors 7-day fault rolling windows to assign real-time RiskScores to individual stations.
- Professional Reporting Pipeline: Integrating iText7 for programmatic PDF generation and custom CSV exporters to facilitate formal maintenance documentation and inter-departmental data sharing.
- Binary Model Management: Designing a workflow for on-the-fly model training (
faultModel.zip) and immediate deployment within the application lifecycle.
- 🤖 AI-Powered Forecasting: Predict fault probability (%) for specific stations based on date/time matrices.
⚠️ Dynamic Risk Mapping: Instant visual feedback on station health using weighted fault frequency analysis.- 📤 Enterprise Export: Generate high-fidelity PDF reports for maintenance audits and CSVs for external analysis.
- 💾 Integrity Management: Single-click database mirroring and maintenance history tracking.
- Core: .NET Framework / WinForms.
- Intelligence: ML.NET (FastTree), One-Hot Encoding.
- Database: System.Data.SQLite.
- Reporting: iText7 (PDF), CSV Serializers.
- Clone the repository:
git clone [https://github.com/emineugurlu/MetroBakimTakip.git](https://github.com/emineugurlu/MetroBakimTakip.git)
2.Setup: Open .sln in Visual Studio and restore NuGet packages (ML.NET, iText7, SQLite).
3.Database: Ensure metro.db is in the project root.
- Run: Press F5 to launch the Maintenance Dashboard.
Developed by Emine Uğurlu - Computer Engineer. Inspired by industrial field operations.