Short-term Load Forecasting for Microgrids Based on Discrete Wavelet Transform and BP Neural Network

Electricity is of great vital and indispensable to national economies. A new short-term load forecasting for micro grid is proposed in this paper. After comparing and analyzing all load characteristic in the time domain and frequency domain, we apply wavelet transform to decompose the load signal. After that, the training set and text set are selected in consideration of the effects generated by the temperature and day type. At length, BP natural network is employed you forecast the micro grid load. The final result proves that the forecasting precision of the method we propose is obviously better than the traditional ones. What’s more, our method has Strong adaptability and good generalization ability.