Reactive power optimization for oil field distribution power system based on hopfield neural network
This paper presents a new mathematical model of reactive power optimization for distribution system. The new model is to minimize the sum of running expense and of investing expense. An associate matrix for describing the topology of the radical distribution system is introduced. Based on the associate matrix, Hopfield neural network is used to solve the mathematical model. The energy function can optimize simultaneously the running expense and the investing expense of a distribution system. The problem of overcompensation is also taken into consideration in the energy function. The dynamic function of the neural network is derived from the energy function. The characteristics of the dynamic function are analyzed, too. By calculating the dynamic function, the overall capacity and the compensation scheme of each load node can be decided simultaneously. A reactive power optimization scheme for a distribution system in a western large oil field is obtained by using Hopfield neural network, and the calculated scheme is better than the one obtained from existing documentations. © 2008 Binary Information Press.