Machine Learning for Power Engineers

After participating in this course, you will have an understanding of:

• The basics of machine learning algorithms
• The overall implementation of machine learning from collecting the data to the classification process
• Sources of machine learning errors
• New trends in machine learning
• The use of available machine learning software
• Application of machine learning to different power engineering problems

Description

The application of machine learning in the power system is an integral part of the smart grid. This course introduces the basic knowledge of machine learning for power engineers with an emphasis on applications related to the power industry. The types of machine learning algorithms, such as supervised and unsupervised learning, classification, and regression problems, will be discussed. Also, the practical problems that diminish the efficiency of machine learning, like overfitting and imbalance of data, will be highlighted. The application of deep learning as a new trend in machine learning will be introduced. The main focus of the course is on the practical application of machine learning to the power grid without going deep into the math or the statistics behind the different algorithms. Hands-on tutorials using free software will be utilized to demonstrate the application of machine learning to the power industry.

Who Should Attend

This course is intended as an introduction to machine learning and its application to the power industry for power managers and engineers who are dealing with different parts of the power grid. The course does not require any background in programming, math, or statistics.

Day I (Theory)

Introduction

• Type of machine learning algorithms
• Supervised vs Unsupervised learning
• Classification vs regression problems
•  
• The overall machine-learning process
• Data collection
• Extracting features
• Learning process
• Application of machine learning algorithms

Examples of Machine Learning Applications in Power Engineering

• Predicting the output of renewable energy sources
• Identifying the source of partial discharge in electrical equipment
• Estimating the transformer health index
• Detecting the source of transient in the power grid
• Enhancing the operation of the distribution system using data from smart meters
• Quantifying the damage in outdoor ceramic and non-ceramic insulators

Problems With Machine Learning and Their Remedies

• Overfitting
• Data imbalance
• Curse of dimensionality

New Trends in Machine Learning

• Big data
• Deep learning
• Multi label problems
• Applications on edge devices

Day II (Application)

Introduction To FREE Commercial Machine Learning Software

• Hands-on tutorial on using WEKA to explore the following:
• Application of supervised learning
• Regression
• Classification
• Application of unsupervised learning
• Clustering

Hands-on Study Cases (Using WEKA):

• Detecting different defects in outdoor ceramic insulators:
• Problems with ceramic insulators: crack, pollution discharge, etc.
• Sensor application and data gathering
• Classifying the different insulator defects.
• Predicting transformer oil parameters:
• Importance and types of transformer oil tests
• Predicting transformer oil interfacial tension and furan content.