Unit Description and General Aims

The objective of this unit is to impart to students a detailed knowledge of the use of machine learning and data analytics in different applications related to the electricity grid. The unit provides a mathematical background and a description of the steps required to build and evaluate a machine learning system. Information presented in this unit also includes: an introduction to different algorithms used in machine learning, a description of typical applications, an overview of software tools commonly used in machine learning and different case studies. Students will complete a project covering the design of a machine learning system to solve a realistic problem related to the electricity grid.

Learning Outcomes

On successful completion of this Unit, students are expected to be able to:

1. Explain the need/importance of machine learning and its applications to data analytics in electricity grid.
   Bloom's Level 2
2. Apply the basic mathematical concepts used in machine learning.
   Bloom's Level 3
3. Evaluate and compare different machine learning systems according to an application.
   Bloom's Level 5
4. Apply machine learning algorithms to solve supervised and unsupervised learning problems using a variety of tools.
   Bloom's Level 3
5. Design machine learning solutions to efficiently manage various electricity grid data.
   Bloom's Level 6

Student Assessment

Overall Requirements: Students must achieve a result of 50% or above in the exam itself to pass the exam and must pass the exam to be able to pass the unit. An overall final unit score of 50% or above must be achieved to pass the unit once all assessment, including the exam, has been completed.

Prescribed and Recommended Readings

Required textbook(s)

• Rodolfo Bonnin. (2017). Machine Learning for Developers. Packt Publishing. ISBN: 978-1786469878
  Available on Knovel
• J. P. Han, H. Jian Tong, Data Mining Concepts and Techniques, 4th Edition. Elsevier, 2023

Reference Materials

• References from authentic websites on the Internet:
• Weka Wiki Documentation
• R Introduction Manual
• IEEE Transaction on Smart Grids journals

Unit Content

Topic 1

Introduction to Data Analytics and Data in Electricity Grids

1. Basics of data analytics
2. Machine Learning vs Artificial Intelligence
3. Supervised, Unsupervised, Reinforcement Learning
4. Types of data in electricity grids: electrical and non-electrical data
5. Introduction of data analytics in electricity grids: event analytics, state analytics, customer analytics and operational analytics

Topic 2

Python in Data Analytics

1. Essential libraries of Python for big data analytics: Pandas, Numpy, Matplotlib, Scikit-learn, Statsmodel, Tensorflow
2. Examples of using the essential libraries to analyse electricity grids data

Topic 3

Data Flow and Feature Engineering

1. Data sources in electricity grids
2. Data preprocessing
3. Data visualization
4. Features and feature vectors
5. Dimensionality reduction
6. Data mining
7. Application of feature engineering to big datasets in electricity grids

Topic 4

Commonly Used Supervised Learning Algorithms

1. K-Nearest Neighbors
2. Decision trees
3. Linear regression
4. Naïve Bayes
5. Implementation of classification algorithms in electricity grids
6. Implementation of regression algorithms in electricity grids

Topic 5

Unsupervised Machine Learning and Reinforcement Learning Algorithms

1. Clustering: K-means, DBSCAN, Hierarchal clustering
2. Association: Apriori and other algorithms
3. Learning models of reinforcement: Markov decision process and Q learning
4. Implementation of clustering algorithms using electricity grids data
5. Implementation of association rules using electricity grids data
6. Example of reinforcement learning in smart grids, e.g., energy trading

Topic 6

Other Algorithms (III)

1. Genetic algorithms and other optimization algorithms
2. Artificial Neural Networks:
   • Feedforward Neural Networks
   • Convolutional Neural Networks
   • Recurrent Neural Networks
3. Artificial neural networks in electricity demand forecasting

Topic 7

Introduction of other Tools in Data Analytics

1. R
2. MATLAB
3. WEKA
4. Cloud-based Solutions
5. Examples of using R and other tools in analyzing electricity grids data

Topic 8

Event Analytics in Smart Grid s

1. Fault and failure diagnosis
2. Fraud detection
3. Predictive outage management
4. Statistical Process Control for event/anomaly detection

Topic 9

State Analytics in Smart Grids

1. Equipment health and condition monitoring
2. System and state identification
3. Predictive control
4. Stability enhancement
5. Asset management

Topic 10

Customer Analytics in Smart Grids

1. Demand response
2. Consumer modelling and segmentation
3. Customer behavior
4. Peer-to-peer energy trading

Topic 11

Operational Analytics in Smart Grids

1. Generation and load forecasting
2. Real-time energy management
3. Price forecasting
4. Economic load dispatch
5. Network constraint forecasting & dynamic network operating envelopes
6. Risk analysis

Topic 12

Unit Review

In the final week, students will have an opportunity to review the contents covered so far. Opportunity will be provided for a review of student work and to clarify any outstanding issues. Instructors/facilitators may choose to cover a specialized topic if applicable to that cohort.

Software/Hardware Used

Software

• WEKA: It is an open-source software suite for machine learning and data mining, offering tools for data preprocessing, classification, regression, clustering, and visualization. It provides students with hands-on experience in applying machine learning algorithms and analyzing data, essential for data-driven decision-making in various fields.
• R: It is a statistical programming language designed for data analysis, statistical computing, and graphical representation. It is extensively used in industries for data mining, predictive modeling, and advanced statistical analysis. By studying R, students gain a strong foundation in statistical modeling and data visualization, which are essential for data-driven decision-making across various industry sectors.
• MATLAB: It is a high-level programming language and computing environment used for numerical computation, data visualization, and mathematical modeling. It is widely applied in industries such as engineering, finance, and data science for solving complex problems, performing simulations, and developing algorithms. For students, it helps build practical skills to analyze data efficiently, model real-world scenarios, and solve engineering problems.
• Python: It is a high-level, open-source programming language widely used for data analysis, machine learning, scientific computing, and automation. By learning Python, students gain practical skills in developing machine learning models, automating data workflows, and supporting data-driven decision-making in real-world applications across engineering, data science, artificial intelligence, and beyond.

Hardware

• N/A

Unit Changes Based on Student Feedback

• The assessment structure is updated based on the action items of the Learning and Teaching Committee (Dec 2024)