Abstract: The generation expansion planning is a complex non-linear and combinatorial optimization problem.With the plan time lengthening and the factors considered increasing,the traditional optimization methods can not make the satisfactory result.And also so,it's applications in the practice are limited.Firstly,the model of the generation expansion was set up.Then,the paper improved the iteration tactic of the particle swarm optimization.It proposed a hybrid algorithm which combined genetic algorithm with particle swarm optimizaion(GPHA).Considering the various factors concerned,the paper introduced fictitious variables to briefly describe generation expansion planning problem in the optimization process.For the fitness function,the paper used penalty function to enhance the effect.In the end of this paper,it chose actual load forecast and the system load actual data of some province,quoted in the similar question regarding correlation parameter supposition,and compared the algorithm with the ordinary genetic algorithms,particle swarm optimization algorithm and the traditional dynamic programming algorithm.The result proves that the new algorithm has done well in the aspects of optimization and speed.

Key words: genetic algorithm, particle swarm optimization algorithm, generation expansion planning, penalty function, fictitious variable