SS04 Objective-based Machine Learning for Low-carbon Power Systems

ABSTRACT

With the trend of digitalization, machine-learning techniques will be widely applied in the planning and operation of power systems. Traditionally, machine learning and decision-making models are independent. The objective of machine learning is usually based on statistical metrics while decision-making models usually aim at lower cost and higher reliability.
An underlying assumption here is that a more statistically accurate result generated by machine learning will guarantee more effective decision-making. However, recent research shows different results. For example, the forecasting error might have an asymmetric impact on the system operational cost. The clustering error may have a distributional influence on network expansion decisions. Therefore a series of objective-based machine learning models have been proposed to address specific objectives of power system decision making such as the cost-oriented forecasting model, the closed-loop clustering model and the reliability-based expansion model.
Supported by authors of these recent emerging papers, this panel will further explore how objective-based machine learning could support more effective planning and operation of low-carbon power systems. These are essential ingredients to smart decisions which will increase the operating efficiency of the system.
The panel will discuss the following topics: 1). Theory: What is the fundamental science of objective-based machine learning models? How are they compared to other decision-making approaches such as probabilistic models and uncertainty optimisations? 2). Techniques: What algorithm and data innovations are required to implement objective-based machine learning models? Could the models truly reflect the objectives of the system? In reality, when the data is limited, what is the practical value of this approach? Is there a practical approach to manage the data in real-time or is there an alternative rule-based/AI-based solution? 3). Applications: What are the potential applications of this approach to different sectors and tasks in generators, networks, customers and markets? How flexible is the approach when the system is of high uncertainty? How transferrable is the approach to different systems and markets?

CHAIR

Ran Li, Shanghai Jiao Tong University

SPEAKERS

Dr. Fei Teng, Imperial College

Prof. Ran Li, Shanghai Jiao Tong University/University of Bath

Prof. Salvador Pineda Morente, University of Málaga

SHORT BIO

	Fei Teng (Senior Member, IEEE) received the B.Eng. degree in electrical engineering from Beihang University, China, in 2009, and the M.Sc. and Ph.D. degrees in electrical engineering from the Imperial College London, U.K., in 2010 and 2015, respectively. He is currently a Senior Lecturer with the Department of Electrical and Electronic Engineering, Imperial College London. His research focuses on the power system operation with high penetration of inverter-based resources (IBRs) and the cyber-resilient and privacy-preserving cyber-physical power grid.
	Ran Li received the B.Eng. degrees in electrical power engineering from the University of Bath, Bath, U.K., and North China Electric Power University, Beijing, China, in 2011, and the Ph.D. degree in electrical engineering from the University of Bath in 2014. He is currently an Associate Professor with the Department of Power Electrical Engineering, Shanghai Jiao Tong University, Shanghai, China. His major interests include big data in power system and power economics.
	Salvador Pineda received the Ingeniero Industrial degree from the University of Malaga, Malaga, Spain, in 2006, and the Ph.D. degree in electrical engineering from the University of Castilla-La Mancha, Ciudad Real, Spain, in 2011. He is currently an Associate Professor with the University of Malaga. His research interests are in the fields of power system operation and planning, decision-making under uncertainty, bilevel programming, machine learning and statistics