In allusion to the distributed power planning of enterprises power supply and distribution system, this paper establishes distributed power optimal allocation model which maximizes energy saving benefits. Improved particle swarm optimization algorithm is used to solve the problem, in which integrates parameter adaptive adjustment, particle cross, simulated annealing algorithm into particle swarm algorithm to improve its optimization efficiency. The IEEE33 node distribution system is simulated and analyzed. Simulation results show that using the proposed model to optimize configuration of distributed power supply, distribution network loss is reduced, voltage quality is significantly improved, and energy-saving economic benefits of the enterprises are maximized. Case analysis effectively verifies the feasibility of optimal allocation model and improved particle swarm optimization algorithm.