Pareto Entropy Based Multi-objective Gravitational Search Algorithm Research And Application

The Gravitational Search Algorithm (GSA) has good global search ability fastconvergence speed and simple operation.Compared with similar algorithms in the standard test functions, it reflects the significant performance advantages. This paper proposes the Pareto Entropy Based Multi-objective Gravitational Search Algorithm (PE-MGSA), which improves the local optimization ability and solves the premature convergence problem. And it is applied to the multi-objective optimal allocation model forwind, photovoltaic and battery combined as a generation unit. The main research workin this dissertation is as follows:Research the basic principle of GSA and the basic steps of the algorithm and realization process. Analyze the influence of the parameters on the performance, its applicable condition and its superiority.ProposesPE-MGSA andimprove the algorithm. Introduce the elite reserved strategy and make updating strategy of Pareto archive. On the basis of the change of the Pareto entropy give definition of three evolutionary status and the corresponding judging conditions, give the PE-MGSA algorithm flow.Apply PE-MGSA to the multi-objective optimal allocation model forwind, photovoltaic and battery combined as a generation unit. Put forward a multi-objective optimal capacity allocation method with comprehensive cost and renewable energy loss rate as the objective function and adopt PE-MGSA to optimize each part capacity proportion of wind turbines, solar panels and battery energy storage. Finally close to the ideal solutions for sorting method from the selection of the optimal solution concentration to compromise.Depend on the algorithm research, design and development the system of the multi-objective optimal allocation model. Calculated on the basis of the measured data in a given area, and compare the optimization results with the traditional optimization algorithms’. The comparing results verify the feasibility and superiority of the algorithm is given in this paper.