Research On Improvement Of NSGA2 Genetic Algorithm And Its Application In Micro-grid Configuration

Evolutionary algorithm is a highly applicable and advantaged multi-objective global optimization method,which is derived from survival of the fittest of the natural evolutionary process.Optimization of practical engineering applications is often characterized by multiple scenes,multiple period,multiple influencing factors with various constraints.The involved various constraints usually contribute difficulties to solving the multi-objective optimization problems.The widely concerned method of constrained multi-objective optimization problems is penalty function method.This method has one inherent defect,that is,the setting of penalty factor.Non-dominated Sorting Genetic Algorithm 2(NSGA2)is a typical multi-objective genetic algorithm.Based on the classical NSGA2 algorithm,this paper proposes an improved INSGA2 algorithm(Improved Non-dominated Sorting Genetic Algorithm 2)which includes some improved strategies to improve the effectiveness of the algorithm in optimization of multi-objective problem with constraints.The improved algorithm deals with the constraints of the multi-objective optimization problems as object,constraint conditions are transformed into one of the objects to be optimized.Because NSGA2 algorithm could not deal with the problems with more than three targets well,therefore,this paper studies only two goals optimization problems with constraints.In the INSGA2 algorithm,the individuals with better performance in the infeasible domain are used to carry out the genetic operation with the feasible solutions so as to accelerate to move to the feasible direction,at the same time evolutionary algebra of the genetic operation will be changed adaptively to reduce the redundancy and inefficient genetic operation of the late evolutionary.In the process of the evolution,the feasible solution individuals who could remain must meet certain conditional restrictions.This design can strengthen the selective pressure in the evolution,preventing evolutionary stagnation or even degradation.In the latter part of the evolution of the population,the marginal variation operation is performed to relieve the excessive overlapping of the similar individuals,this measure could eliminate the possibility of local convergence in the late stage of population evolution.In the case study,the experimental results show that the improved algorithm has some certain advantages.Traditional power supply and distribution networks have wide range of long-distance interconnections.Power distribution,centralized operation and control are simultaneous.The defects of the operating mode are increasingly valued.People gradually pay attention to the distributed power supply and micro-grid applications,to a large extent making up for the lack of large-scale centralized power supply,to improving power supply reliability and expanding its applicability.However,the improper distribution of distributed power will affect the early planning of power grid which is based on line loss,power quality,economic factors,environmental factors.Therefore,it is of practical significance to analyze the distribution of distributed power supply of the configuration of micro-grid,which do goods to more safe,reliable and efficient operation from the pointview of power supply quality,economic cost and environmental benefit.In order to optimize the configuration problem of microgrid,the mathematical models are established based on the line loss,voltage deviation,initial investment,carbon emission,at the same time two models are selected as the optimization goals,while the constraints of the normal operation of the system are considered.The two contrastive algorithms and the two-objective models are simulated and validated by IEEE 33 nodes power distribution system.Experimental results show that the improved algorithm and models are reasonable and effective.