Multi-objective Optimization of a Wind-solar Microgrid System Based on the Improved NSGA-II Method

Abstract

Distributed energy planning is a complex issue, and the non-dominated sorting genetic algorithm (NSGA-II) is widely employed for the system multi-objective optimization. This algorithm screens the intermediate population based on the fixed crowding distance principle, it does not consider the dynamic crowding change and cannot satisfy the diverse search requirements of solution space in different evolutionary periods. In this paper, an improved NSGA-II method based on dynamic crowding distance and information entropy is proposed. Then a case study of a wind-solar integrated microgrid system is implemented, by refereeing the local meteorological conditions and power loads in Yunnan of China, the renewable energy system under study is optimized in terms of the system energy efficiency, energy volatility and net present cost. Results indicate that the energy system achieves a lower cost after optimization by the improved NSGA-II method, and the matching degree of renewable energy generation and electricity demand is evidently enhanced which means a better system operation stability. Comparing to the general NSGA-II, the improved algorithm also has superiority in convergence speed, and the research findings provide an alternative method for optimizing the renewable microgrid system.

Analysis of the current status of research on optimal scheduling and demand-side response for active distribution networks

“…The alternative is to employ artificial intelligence algorithms to solve the issue following self-learning. For instance, literature [8] employed the particle swarm algorithm for distributed cooperative optimization; literature [9] employed the simulated annealing algorithm with branch exchange and literature [10] employed the enhanced NSGA-II algorithm. Algorithms for artificial intelligence can be easily learned, adjusted, self-iterative, and very effective if tailored to the specific circumstances.…”