A Machine Learning-based overload detection and electrical condition monitoring system developed using experimentally generated electrical parameter datasets and a Random Forest Classifier.
The project focuses on identifying overload and abnormal operating conditions in electrical systems using electrical signal characteristics such as voltage, current, power, and power factor.
The objective of this project is to develop an intelligent overload detection model capable of classifying electrical operating conditions using machine learning techniques.
The model was later conceptually integrated into the Smart Energy Monitoring ecosystem for appliance condition analysis and monitoring demonstrations.
- Random Forest-based overload classification
- Electrical parameter analysis
- Condition monitoring
- Relay action analysis
- Feature importance visualization
- Correlation analysis
- ROC curve evaluation
- Confusion matrix visualization
- Signal trend visualization
The model uses electrical parameters such as:
- RMS Voltage (Vrms)
- RMS Current (Irms)
- Real Power
- Apparent Power
- Power Factor (PF)
The Random Forest model was trained for overload detection and condition classification using experimentally generated datasets.
| Metric | Value |
|---|---|
| Accuracy | 100% |
| Test Samples | 6000 |
| Model Type | Random Forest Classifier |
The dataset primarily classifies operating conditions into:
- Normal Condition
- Overload Condition
The project contains several visualization and analysis plots including:
- Confusion Matrix
- ROC Curve
- Feature Importance
- Correlation Heatmap
- Pairplot Analysis
- Electrical Signal Trends
- Voltage vs Current Scatter Plot
- Real Power vs Apparent Power Plot
- Relay Action Frequency Analysis
data/ -> datasets used for training/testing
models/ -> saved ML models
notebooks/ -> Jupyter notebooks
plots/ -> generated plots and visualizations
- Python
- Scikit-learn
- Pandas
- NumPy
- Matplotlib
- Seaborn
- Jupyter Notebook
This model can be integrated with:
- ESP32-based real-time monitoring systems
- IoT-enabled smart energy systems
- Real-time mobile monitoring applications
- Predictive maintenance systems
This repository focuses on the Machine Learning and analysis component of the project. It does not currently include real-time hardware deployment but is fully extendable for live IoT integration.
Department of Electrical Engineering Indian Institute of Engineering Science and Technology (IIEST), Shibpur
- Predictive Energy Intelligence System (Machine Learning & Predictive Analysis)
- ESP32 True RMS Energy Monitor (Embedded System)
- Smart Energy Monitoring Mobile Application (Flutter App)