Network Expansion Planning Based On Improved Particle Swarm Optimization

Power industry is a very important economic sector, its level of development not only for other industry sectors will have a huge impact, but also to be the primary energy consumption and a large investment. Reasonably Power System Planning will be great economic benefits and social benefits. In contrast, errors in power system planning nation-building would cause irreparable damage. Which the network planning due to its nonlinear, multi-objective and multi-binding characteristics of the electricity planning occupy an important position, so their planning can greatly improve the overall power system reliability, economy and security.In this paper, studying of the traditional network planning model and solution method, and on this basis the corresponding improvements. Multi-objective model for network planning and uncertainty, this paper establishes the corresponding multi-objective optimization model of satisfaction, the model including a minimum investment, minimum loss, maximum reliability and minimal impact on the environment including four goals, it also considered to the maximum load to the future load growth targets and the uncertainty of the impact of the implementation plan. In order to better integrate the actual situation on the satisfaction function is analyzed, chosen better reflect the satisfaction of all the real objective function change of the model; Simplified mathematical model of the process for the errors caused by AHP ranking of each program, the integration of expert experience to the computer in order to achieve more realistic to reflect the actual purpose of the optimal solution.PSO of more suitable for network planning was selected based on the analysis of the traditional intelligent algorithms. And for defects of its easy to fall into local optimal value this paper have quoted the adaptive crossover and mutation probability into PSO. And according to the characteristics of network planning this paper have made corresponding improvements of the crossover. Test function on the grid computing and solving examples show that the proposed model is practical and the improved algorithm solution is faster, better convergence characteristics.