Particle Swarm Optimization Algorithm In Power System Planning In Applied Research

The electrical power system planning is comprised of power load forecasting, power source planning and electrical network planning. Power load forecasting is the base of electrical power system planning. Its veracity has great impact on planning result. The electrical network planning may further be devided into transmission network planning and distribution network planning. Transformer substation planning is the important part of distribution network planning. The capacity of transformer, the number of transformers and substations are one of the important elements which influence the structure, the reliability and the economy of urban electrical network. Therefore, doing researches in load forecasting and transformer substation optimize planning have practical meanings.Basing on the standard particle swarm optimization (PSO), this thesis brings forward an improved particle swarm optimization (IPSO). IPSO considers the social influence of other particles in the whole swarm to each particle, not only the single particle that in the global best position. Furthermore, the social influence factor is adjusted dynamically alonging with the searching behavior. Through the test of several standard testing functions, IPSO shows better ability of searching global best result and avoiding falling into local best position. It is a banausic algorithm.Aiming at the question of load forecasting, this thesis presents a combined load forecasting model based on IPSO, which uses IPSO to calculate the weight of combined model. The computing result of practical data shows that the combined forecasting model based on IPSO has less predicted errors than PSO and other traditional combined forecasting models.Aiming at the question of transformer substation optimize planning, this thesis presents a mathematical model based on discrete improved particle swarm optimization (DIPSO). This model optimizes the 110 kV-substation number, its capacity and unit number, using the minimum cost of network construction and grid energyloss as object function, the capacity-load ratio of transformer and the radius of power supply as constraint condition. DIPSO is used to find out the best result of this model. It combines mathematic optimization algorithm and engineering practice. The results and the conclution of a computational example show the effective value of this model.