Research On Unit Commitment Problems In Power System Based On Improved Gravitational Search Algorithm

Unit commitment problem is an engineering problem with practical application value, which can bring significant social benefits and economic benefits and is very important in power system research. It has a very important significance to make a reasonable power generation plan in the required time for the construction of a reliable, safe, economical and intelligent power grid.Unit commitment problem is discrete and nonlinear. So far it still does not exist a method that can not only obtain the ideal accuracy and calculation speed but also can comprehensively consider all constraints. How to improve the speed and accuracy of unit commitment problems of the power system is still very important. The conventional mathematical methods have some limitations in dealing with this kind of problems; however, intelligent optimization methods are suitable for solving the high dispersion and high nonlinear problems. According to the properties of gravitational search algorithm and chaotic local search method, this paper combines the two methods to design a new algorithm to solve unit commitment problems.First of all, this paper presents two kinds of improved gravitational search algorithms for solving unit commitment problems. In order to improve the convergence speed and weaken the defect of falling into local optimum, random variables used in the gravitational search algorithm are instead by chaotic variables generated by chaotic system. What?s more, the ergodicity of chaotic systems can also enhance the global searching ability of gravitational search algorithm. In order to prevent the algorithm from being trapped in local optimum, the second method is to find the optimal solution firstly by gravitational search algorithm. Then making the optimal solution as the center and establish a hypercube to search a better solution in this space. The two methods are used to search the best solutions of six standard test functions. Simulation results show that improved algorithm is better than the basic gravitational search algorithm which verifies that the second improved method is effective and feasible to solve continuous problems.Finally, in order to further enhance global search ability of the algorithm, piece wise linear chaotic map with stronger ergodicity is combined with gravitational search algorithm. Then this combined algorithm is applied to the test system with 6 and 10 generators. Simulation and optimization results are compared with selective pruning method and iterative linear algorithm etc. The total generation cost of power systems get by the improved algorithm is better than the that of others which verify the feasibility to solve unit commitment problems.