Short Term Load Forecasting by Adaptive Neural Network

2018 2nd International Conference on Aerospace Technology, Communications and Energy Systems (ATCES 2018)

航空航天工程;无线电电子学;能源学

Li, Hong^1

Computer Systems Technology, New York City College of Technology, City University of New York, Brooklyn

NY

11201, United States^1

关键词: Adaptive learning algorithm;    Adaptive neural networks;    Convergence conditions;    Economic scheduling;    Energy forecasting;    Network structures;    Short term load forecasting;    Short-term forecasting;   

Others  :  https://iopscience.iop.org/article/10.1088/1757-899X/449/1/012028/pdf
DOI  :  10.1088/1757-899X/449/1/012028

学科分类：航空航天科学

来源: IOP

【 摘 要 】

The Generation and load balance is required in economic scheduling of the generating units and in electricity market trades. Energy forecasting became very important to mitigate some of the challenges that arise from the uncertainty in the resource. The paper presents a structure of artificial neural network with an adaptive learning algorithm used for a short-term forecasting of hourly electric power load. Historical data are sourced from Global Energy Forecasting Competition 2017 (GEFCom2017) including forecasting in the domains of electric load, weather, wind power, solar power, and electricity prices. An adaptive learning algorithm is derived from analysis of system stability to ensure convergence of training process. A simplified condition of learning factor is driven for use of computer simulation. An upper bound of learning factors is derived from the theory of convergence. At iteration of network training, a learning factor is defined to satisfy the convergence condition. The simulations with different initial state of network structure demonstrate that training error steadily decrease with an adaptive learning factor starting at different initial values whereas errors behave volatile with constant learning factors. The comparison demonstrated that a learning factor arbitrarily chosen out of the predefined stability domain leads to an unstable identification of the considered system; however, an adaptive learning factor satisfying the conditions chosen for this study ensures the stability of the identification system.