A research on family flexible load scheduling based on improved non-dominated sorting genetic algorithm

Abstract

In demand response scheduling, residential electricity load has huge scheduling potential. However, the diversification of characteristics of residential electricity load, the diversity of user demands, and the poor convergence and susceptibility to local optimal solutions of scheduling optimization algorithms significantly increase the difficulty of demand response scheduling. To address the heterogeneity of the load, this paper proposes a new classification method for household electricity load, which is divided into rigid load and flexible load based on different electricity characteristics, and further divides flexible load into interruptible load, non-interruptible load, and reducible load, and establishes mathematical models for each type of load. To address the diversity of user demands, this paper proposes three optimization objectives: minimum electricity cost, minimum carbon emissions, and maximum user satisfaction, which can be comprehensively optimized based on user electricity demands. To address the slow convergence and susceptibility to local optimal solutions of the traditional NSGA-III algorithm, this paper improves the traditional NSGA-III algorithm by using adaptive crossover and mutation to improve algorithm convergence speed and selecting individuals based on crowding distance to avoid local optimal solutions. The results show that compared with the user's original electricity consumption behavior, this method can save up to 21.1% of electricity costs, reduce carbon emissions by up to 10.2%, improve user electricity satisfaction by up to 41.2%, and reduce peak-to-valley difference by up to 42.3%. This provides a powerful basis for residential buildings to participate in demand response scheduling and provides a reference path for achieving low-carbon development goals in the building field.