Economic distribution (ED) is a typical optimization problem of power grid and it is also very important for power system energy saving. However, the optimization method based on the traditional particle swarm algorithm only considers speed and position parameters, which easily leads to local optimization. Under this background, taking into account the valvepoint effect of generators, an improved particle swarm algorithm is proposed for the active power distribution of power systems. The perspective parameter is introduced, and then the moving state of particle is decided by the new highdimensional parameters. The proposed algorithm can avoid the local optimal, reduce the search randomness and improve the optimization accuracy. Simulation results shows that the improved particle swarm algorithm with viewing angle parameters has a more efficient global search capability and a more reliable optimal solution, which provides an effective new algorithm for the power system economic distribution problem.