Grant: Discovery Grants Program - Individual
McGill University
$32,000.00
Data Efficient Load Forecasting and Control Methods for Smart Grid
EF 2024-25
Résumé
Smart grids use advanced meters, energy storage, communication, and computation tools to improve power grid reliability, energy efficiency, and operational security. Modern power grids are now faced with fundamental changes on both the energy generation side as well as the energy demand side. The penetration of electric vehicles (EVs) has also increased quickly in recent years. EV sales have increased very quickly in North America in the last two years, and the Canadian government has recently announced a plan to ban the sale of new fuel-powered cars and light trucks from 2035. The increase in EV adoption will result in additional high power demand on the grid and may jeopardize the power grid infrastructures in the absence of proper charging schedules. Meanwhile, renewable energy generation, e.g., solar and wind power generation, has also increased in recent years. Unfortunately, the output of renewable energy generation is intermittent and sensitive to weather conditions. Thus, modern power grids are now faced with more and more complex challenges. Energy management and load forecasting are two core issues in the smart grid. Electric load forecasting can facilitate operations planning, revenue projection, electricity price design, and real-time energy dispatching. Efficient energy management benefits both end-user and utility companies. Efficient management can help reduce operation costs and greenhouse gas emissions. The solutions of data-efficient forecasting and control methods developed in this project will also be suitable for other real-world problems such as transportation systems, and modern communication systems.
Energy management and electric load forecasting are two core problems for modern power grids. Different types of methods have been developed to deal with these two problems. Machine learning-based methods, e.g., deep learning and reinforcement learning (RL) [1-6], have shown impressive performance in several real-world applications, including electric load forecasting [9-12] and energy management [7, 8]. However, most of these works assume that a large amount of training data is available. This can be very challenging or even impossible for several real-world scenarios, e.g., newly built houses and buildings. Without enough data, it will be hard to learn reliable forecasting and control models via machine learning. Transfer learning [13-14] has shown its capability to improve the performance of machine learning models for a target domain by reusing knowledge from other related domains [15]. In this research program, we aim to 1) present a comprehensive study on understanding the impact of the amount of data on the effectiveness of different machine learning methods and 2) leverage transfer learning and meta learning to design novel data-efficient forecasting and control methods for the power grid. The outputs of this project can serve as good references for both researchers as well as domain engineers.
Programme
Discovery Grants Program - Individual
Bénéficiaire
Emplacement
Quebec, Canada
Type de bénéficiaire
Academia
Type
Grant
Source
NSERC's Awards Data
Tel que publié
Application id
RGPIN-2023-04883
Name nom
Wu, Di
Department département
Elecctrical and Computer Engineering
Organization id
61
Institution établissement
McGill University
Province en
Québec
Province fr
Québec
Country en
CANADA
Country fr
CANADA
Fiscal year exercice financier
2024
Competition year année de concours
2023
Award amount
32000
Program id
RGPIN
Program name en
Discovery Grants Program - Individual
Program name fr
Programme de subventions à la découverte - individuelles
Group en
DISCOVERY RESEARCH
Group fr
RECHERCHE AXÉE SUR LA DÉCOUVERTE
Committee code
1510
Committee name en
Electrical and Computer Engineering
Committee name fr
Génie électrique et informatique
Area of application code
300
Area of application group en
Energy resources
Area of application group fr
Ressources d'énergie
Area of application en
Energy resources (including production, exploration, processing, distribution and use)
Area of application fr
Ressources d'énergie (production, exploration, traitement, distribution et utilisation de l'énergie)
Research subject code
2500
Research subject group en
Electrical and electronic engineering
Research subject group fr
Génie électrique et électronique
Research subject en
Electrical and electronic engineering
Research subject fr
Génie électrique et électronique
Field of research list names en
Not available
Field of research list names fr
Non disponible
Socioeconomic objective list names en
Not available
Socioeconomic objective list names fr
Non disponible
Application title
Data Efficient Load Forecasting and Control Methods for Smart Grid
Keywords
smart grid; electric load forecasting; energy management
Application summary
Smart grids use advanced meters, energy storage, communication, and computation tools to improve power grid reliability, energy efficiency, and operational security. Modern power grids are now faced with fundamental changes on both the energy generation side as well as the energy demand side. The penetration of electric vehicles (EVs) has also increased quickly in recent years. EV sales have increased very quickly in North America in the last two years, and the Canadian government has recently announced a plan to ban the sale of new fuel-powered cars and light trucks from 2035. The increase in EV adoption will result in additional high power demand on the grid and may jeopardize the power grid infrastructures in the absence of proper charging schedules. Meanwhile, renewable energy generation, e.g., solar and wind power generation, has also increased in recent years. Unfortunately, the output of renewable energy generation is intermittent and sensitive to weather conditions. Thus, modern power grids are now faced with more and more complex challenges. Energy management and load forecasting are two core issues in the smart grid. Electric load forecasting can facilitate operations planning, revenue projection, electricity price design, and real-time energy dispatching. Efficient energy management benefits both end-user and utility companies. Efficient management can help reduce operation costs and greenhouse gas emissions. The solutions of data-efficient forecasting and control methods developed in this project will also be suitable for other real-world problems such as transportation systems, and modern communication systems.
Energy management and electric load forecasting are two core problems for modern power grids. Different types of methods have been developed to deal with these two problems. Machine learning-based methods, e.g., deep learning and reinforcement learning (RL) [1-6], have shown impressive performance in several real-world applications, including electric load forecasting [9-12] and energy management [7, 8]. However, most of these works assume that a large amount of training data is available. This can be very challenging or even impossible for several real-world scenarios, e.g., newly built houses and buildings. Without enough data, it will be hard to learn reliable forecasting and control models via machine learning. Transfer learning [13-14] has shown its capability to improve the performance of machine learning models for a target domain by reusing knowledge from other related domains [15]. In this research program, we aim to 1) present a comprehensive study on understanding the impact of the amount of data on the effectiveness of different machine learning methods and 2) leverage transfer learning and meta learning to design novel data-efficient forecasting and control methods for the power grid. The outputs of this project can serve as good references for both researchers as well as domain engineers.