Short-term Load Forecasting Model of GRU Network Based on Deep Learning Framework

Along with the development of power system, power load forecasting plays a more and more important role. It is not only an important part of power system dispatching, operation and planning, but also the foundation of economic operation and safe operation of power system. However, power load data generally have the characteristics of complexity and nonlinearity. At present, most forecasting methods, which are based on the traditional linear statistical theory, can not consider the time series and the nonlinear characteristics of load data at the same time. In this paper, a short term load forecasting model, GRU (gated recurrent unit) network is proposed, which is the variant of LSTM(long short term memory) being based on long and short memory neural network, and the simulation prediction is carried out by using the power load data of a region in Heilongjiang province. Comparing the prediction results with the traditional ones, it is proved that the error of GRU model is lower and the prediction effect is better.