Research On Middle-term And Short-term Load Forecasting Of Electric Power System

Short-term load forecasting is one of the most important contents of running and dispatching of power system. It can be economic and reasonable to arrange start and stop of the Generator in wire net, reduce otiose revolve the storage capacity. It can be reasonable to arrange Generator the maintain plan, assurance normal produce and live of society, raise the economic efficiency and social efficiency of electric power enterprise. Middle and long load forecasting is one of important jobs of power planning office. It can help to decide the setup of new generator group and planning, capacity increase, reconstruction of wire net.As for the middle-term and short-term load forecasting of electric power system, the components and characteristics of electric loads and the methods of load forecasting are analyzed in detail. Carry through algorithm study using Matlab and the models for the middle-term and short-term load forecasting are proposed.For the middle-term load forecasting, grey theory forecasting method is adopted. This paper changes clutter history load data list to data list of exponential rule, sets up one order linear equation for new data list, gains forecasting value through resolving equation and carrying out inverse accumulated generating operation to equation value, optimizes grey model during forecasting, takes place of the oldest data using the forecasted data under maintaining the dimension of original number sequence, realizes the progressive reasoning of monthly forecasting. The example result proves that this model can obtain a higher forecasting precision using a few datas.For the short-term load forecasting, genetic neural network is adopted. It chooses net input variables using regression analysis method, takes Radial-Basis Function as the excitation function of hidden layer, trains the connected weights between hidden layer and output layer by adaptive genetic algorithm. This can help build a more reasonable frame of ANN. In the basis of this make weather factors as inputs of ANN. By making a comparison among time series, genetic neural network and support vector machine methods and analyzing the example, the result proves that genetic neural network can increase forecasting precision evidently.On the study result, this paper opens up electric power system middle-term and short-term load forecasting software using Visual Basic 6 and Access 2000 database as plat. This software has some advantages like operation with graphics, simple, intuitionist, good alternation, higher forecasting precision and expansion facility.