Short term load forcast using deep learning

Abstract

As the economy increasingly developed, the demand for power systems increasingly large. Should be to improve the reliability of the grid, short term load forecasting is the indispensable factor for improving the fastness of power supply and distribution smart grid (SG) applications. Based on the comparison of applicability and aptitude of Feed-forward Deep Neural Network (FF-DNN) according to the criteria of accuracy and performance applied for short term forecast, and integrated with Recurrent Deep Neural Network (R-DNN). In my article, we study the pertinency and combination two deep neural network architectures Feed-forward Deep Neural Network (FF-DNN) and Recurrent Deep Neural Network (R-DNN) to the Ho Chi Minh City electricity load forecasting task. Thereby, we suggest DNN based electricity load forecasting system to manage in distribution and consumption in the best way. My article used method with activation functions which are different such as Sigmoid, Hyperbolic Tangent (tanh) and Rectifier Linear Unit (ReLU). The propound sheme is administrated to required offline load parameters of Ho Chi Minh City. Load forecasts of weather attribute analysis using DNN. It allows for the capture of these factors affecting the electricity consumption model. Three performance metrics include absolute percentage error (MAPE), normalized root mean square (NRMSE), and correlation coefficients are used to formalize the propound sheme. The final results have demonstrated that FF-DNN and R-DNN are authntic.