Status (updated Oct. 29, 2026): this page is being built during the 2026 Fall semester. Selected lecture recordings are released below as the semester progresses. The complete course materials, including the editable PowerPoint slides of all lectures, will be posted on this page after the semester ends.
Machine Learning with Applications in Power Systems (ML4PS)
ECE6378 Machine Learning with Applications in Power Systems is a graduate-level course offered at the University of Houston (UH) in 2026 Fall.
Course Description:
This course introduces fundamental machine learning (ML) methods and their applications to electric power systems. It begins with selected electrical engineering, transmission grid, and distribution grid foundations so that students from different backgrounds can develop a common vocabulary. It then progresses from grid measurements and data preparation, to classical supervised learning, deep learning, graph-based learning, time-series learning, and unsupervised learning, and finally to ML-assisted and ML-embedded power system optimization.
The course emphasizes not only how an ML model is formulated and trained, but also:
- why a particular model is suitable for a particular task;
- how data resolution and measurement sources affect model design;
- how training, validation, and testing should be separated;
- how physical laws and operational constraints can be used to check predictions;
- how ML can accelerate or be embedded within optimization; and
- when a simpler or physics-based method may be preferable to a more complex ML model.
Course Scope and Specific Topics:
- Lecture 1: Course Introduction: Power Grid & Machine Learning
- Lecture 2: Electrical Engineering Fundamentals
- Lecture 3: Transmission Grid Fundamentals
- Lecture 4: Distribution Grid Fundamentals
- Lecture 5: Python Demo: Research Environment, Jupyter Notebook, and a Case Study
- Lecture 6: Grid Meters & Data
- Lecture 7: Data Preprocessing
- Lecture 8: Feature Engineering & PCA
- Lecture 9: Linear Regression
- Lecture 10: AC Power Flow & N-1 Analysis
- Lecture 11: DC Power Flow, LAC & DistFlow
- Lecture 12: Logistic Regression
- Lecture 13: Support Vector Machine (SVM)
- Lecture 14: Evaluation Metrics, Cross Validation, and End-to-End ML Workflow
- Lecture 15: Short Circuit Analysis & Fault Location
- Lecture 16: DNN Model: MLP & Forward Propagation
- Lecture 17: DNN Training & Backpropagation
- Lecture 18: DNN Tuning: Learning Rates, Optimizers & Regularization
- Lecture 19: CNN, GNN, and Network Science
- Lecture 20: Grid Operations: SCUC and OPF
- Lecture 21: ML-Accelerated Optimization for Power Systems
- Lecture 22: Grid Dynamics & Frequency Regulation
- Lecture 23: ML-Embedded Optimization for Power Systems
- Lectures 24 - 25: Time-Series ML: RNN, LSTM, and GRU
- Lectures 26 - 27: KNN, Decision Tree, Random Forest, Boosting, and Ensemble Learning
Lectures 19 - 27 are tentative as of Oct. 29, 2026, and may change as the semester progresses.
Lecture Recordings (2026 Fall):
Selected lectures are recorded in class and released on the RPG Lab YouTube channel. All released recordings are collected in one playlist:
Lecture 16 - DNN Model: MLP & Forward Propagation: universal approximation, neurons and activation functions, MLP architecture and forward propagation, loss functions (MSE, cross-entropy, weighted loss), and a frequency stability prediction example for frequency-constrained OPF.
Lecture 17 - DNN Training & Backpropagation: gradient descent, chain rule, computation graph and backpropagation, and a complete hand-calculated training step (forward pass, backward pass, and parameter update) on a small MLP, followed by training across multiple epochs.
Lecture 18 - DNN Tuning: learning rate schedules, vanishing and exploding gradients, batch / stochastic / mini-batch gradient descent, adaptive optimizers (AdaGrad, RMSprop, Adam), and regularization and normalization (dropout, weight decay, early stopping, BatchNorm, LayerNorm).
More recordings will be added during the semester, including lectures on ML-accelerated and ML-embedded optimization for power systems.
Planned Shared Contents (after the semester ends):
- Lecture source PowerPoint slides (editable) and converted PDF copies for all lectures.
- Python demo materials: environment setup instructions and Jupyter notebook examples.
- Homework assignments (Python/Jupyter notebook based), with the associated data where applicable, e.g., ERCOT load and weather data and TX-123BT-based power system data.
- Syllabus as a reference.
Assignment solutions, grading scripts, quizzes, and tests/exams are not included. If you are an instructor and plan to use these materials, I would be happy to provide such for-instructor-only materials; please feel free to send me an email with your institutional email account. My contact information is listed below.
Prerequisites:
Basic understanding of electrical circuits, AC and DC systems, basic math, and linear algebra. Prior coursework in power systems is helpful but not required; lectures 2 - 4 provide the power system fundamentals needed for this course.
Software Requirements:
This course uses Python 3.12 with Jupyter Notebook. A class Conda environment file is provided with the course materials.
Related Open-Source Data & Code from the RPG Lab:
- TX-123BT: Synthetic Texas Power System with Time-Series Weather-Dependent Spatiotemporal Profiles
- ML-based ERCOT Net-Load Forecasting in Python Notebook
- ML-based ERCOT Load Forecasting in Python Notebook
- List of Papers with Data/Codes
Some power system lectures in this course draw on materials from my other two courses:
Contact:
Dr. Xingpeng Li
University of Houston
Email: xli83@central.uh.edu
Website: https://rpglab.github.io/
Acknowledgment:
To be added…
License:
This work is licensed under the terms of the Creative Commons Attribution 4.0 (CC BY 4.0) license.
Disclaimer:
The author doesn’t make any warranty for the accuracy, completeness, or usefulness of any information disclosed and the author assumes no liability or responsibility for any errors or omissions for the information (data/code/results etc) disclosed.