Syllabus

Data Acquisition and Preprocessing for Power Engineering Datasets:
• Import power system datasets (load demand, voltage, current, renewable energy data).
• Missing value handling, normalization, filtering, and visualization.
• Feature extraction from time-series data.

Electrical Load Forecasting using Linear Regression
• Build a simple load forecasting model using historical demand data.
• Apply simple and multiple linear regression.
• Compare predicted and actual load profiles.

Transformer or Motor Fault Classification using Logistic Regression
• Use simple labeled datasets containing normal and faulty conditions.
• Implement logistic regression classifier.
• Evaluate classification accuracy.

Power Quality Disturbance Identification using Decision Trees
• Detect disturbances such as voltage sag, swell, interruption, and harmonics.
• Train decision tree classifier using extracted signal features.
• Visualize decision boundaries.

Clustering of Load Profiles using K-Means Algorithm
• Group consumers based on load demand patterns.
• Visualize cluster formation.
• Interpret customer categories or operating conditions.

Dimensionality Reduction of Power System Data using PCA
• Apply Principal Component Analysis (PCA) on multi-variable electrical datasets.
• Reduce data dimensions while retaining dominant features.
• Visualize principal components.

Solar Power Prediction using Neural Networks
• Predict solar power output using irradiance and temperature data.
• Implement a small multilayer perceptron (MLP).
• Evaluate prediction performance.

Intelligent MPPT Estimation using Machine Learning
• Use historical photovoltaic operating data.
• Predict optimal duty cycle or voltage for MPPT.
• Compare with conventional MPPT methods.

Reinforcement Learning for Converter Switching Decision (Simulation-Based)
• Implement a simple reinforcement learning framework.
• Learn optimal switching state selection for converter operation.
• Use discrete state-action environment.

Raspberry Pi Based Smart Energy Monitoring and Prediction
• Acquire sensor-based power measurements.
• Transfer data to Python environment.
• Perform lightweight prediction or anomaly detection.

To provide hands-on exposure to machine learning tools and computational methods relevant to power engineering.

To enable students to preprocess, visualize, and analyze electrical engineering datasets.

To develop practical skills in implementing supervised, unsupervised, and neural-network-based models.

To familiarize students with lightweight ML deployment using standard computers and embedded platforms such as Raspberry Pi.

To apply machine learning techniques for forecasting, fault diagnosis, intelligent monitoring, and control in power electronics and energy systems.

Perform preprocessing and visualization of electrical engineering datasets.

Develop simple ML models for prediction and classification.

Apply clustering and dimensionality reduction to power engineering data.

Use neural-network-based methods for renewable energy prediction.

Implement lightweight ML algorithms on Raspberry Pi or standard PCs.

Interpret ML outcomes for diagnostics, forecasting, and intelligent control.

Christopher M. Bishop, Pattern Recognition and Machine Learning, Springer , 2011

Ethem Alpaydin, Machine Learning, MIT Press , 2010

Tom M. Mitchell, Machine Learning, McGraw Hill , 1997

Giuseppe Bonaccorso, Machine Learning Algorithms, Packt Publishing , 2018