Electric load forecasting using virtual instrument based on dynamic recurrent Elman neural network

In order to improve accuracy of load forecasting for power grid, since the load characteristics of Yichang power grid is sensitive to climate impact, an Elman neural network (NN)-based short-term load forecasting model under comprehensive consideration of various meteorological factors is established. Elman NN has a dynamic recurrent performance which is able to enhance the adaptability of forecasting model. Actual historical hourly loads and weather data of Yichang city are used to build training sample set for NN. The simulation results indicate that the model based on Elman NN has a higher accuracy. Using the method of LabVIEW calling MATLAB, the NN load forecasting model was implanted in and a Virtual Instrument (VI) for load forecasting has been designed. Inputting meteorological factors such as temperature, precipitation, the VI can output load curve, error curve, maximum, minimum and average load. The VI is easy to implement and intuitive. The result shows the effectiveness of this load forecasting method which can be used in practical application.