Efficient Computation for Sparse Load Shifting in Demand Side Management

• Published in: IEEE transactions on smart grid Vol. 8; no. 1; pp. 250 - 261
• Main Authors: Li, Chaojie, Yu, Xinghuo, Yu, Wenwu, Chen, Guo, Wang, Jianhui
• Format: Journal Article
• Language: English
• Published: Piscataway IEEE 01.01.2017; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Algorithms; Computational efficiency; Customers; Demand side management; Discomfort; Energy consumption; Energy conversion efficiency; Energy costs; fast gradient; Game theory; Home appliances; Household appliances; Load modeling; Load shifting; Nash equilibrium; Newton method; Newton methods; Peak load; Peak to average power ratio; smart appliances scheduling; Smart grids; Sparse load shifting; Stress concentration
• ISSN: 1949-3053; 1949-3061
• DOI: 10.1109/TSG.2016.2521377

This paper introduces a distributed algorithm for sparse load shifting in demand-side management with a focus on the scheduling problem of residential smart appliances. By the sparse load shifting strategy, customers' discomfort is reduced. Although there are many game theoretic models for the demand-side management problem, the computational efficiency of finding Nash equilibrium that globally minimizes the total energy consumption cost and the peak-to-average ratio is still an outstanding issue. We develop a bidirectional framework for solving the demand-side management problem in a distributed way to substantially improve the search efficiency. A Newton method is employed to accelerate the centralized coordination of demand side management strategies that superlinearly converge to a better Nash equilibrium minimizing the peak-to-average ratio. Furthermore, dual fast gradient and convex relaxation are applied to tackle the sub-problem for customers' best response, which is able to relieve customers' discomfort from load shifting or interrupting. Detailed results from illustrative case studies are presented and discussed, which shows the costs of energy consumption and daily peak demand by our algorithm are reduced. Finally, some conclusions are drawn.