On the basis of the BP modle of multi-layer feedforward neural network ,this paper presents a short-term load forecast method for power system. According to the variation characteristics of short-term system load, a day load model which not only reflects the continuity ,periodicity and variation trend of system load but also includes the effects of weather on the load is set up as the vector sample set for training the BP neural network. Practical examples indicate that the application of ANN to short-term load forecast is feasible and effective ,and can produce more accurate results than conventional method.
The fault locating in distribution networks based on unified matrix algorithm may result in false judge under the information distortion mode; the construction of the fitness function is a main bottleneck for the application of genetic algorithm.The method of fault section diagnosis of distribution networks based on new nerve network is presented in this paper. The method has high fault-tolerance performance and the result of simulation show that this method is effective.