Application and research for electricity price forecasting system based on multi-objective optimization and sub-models s
- PDF / 5,172,630 Bytes
- 27 Pages / 595.276 x 790.866 pts Page_size
- 52 Downloads / 153 Views
(0123456789().,-volV)(0123456789(). ,- volV)
METHODOLOGIES AND APPLICATION
Application and research for electricity price forecasting system based on multi-objective optimization and sub-models selection strategy Tonglin Fu1 • Shenghui Zhang2 • Chen Wang3
Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract In general, electricity prices reflect the cost to build, finance, maintain, and operate power plants and the electricity grid. Therefore, the cost-optimized scheduling of industrial loads with accurate price forecasts is very important. As such, recent studies have attempted to combine models to forecast electricity prices more accurately. Earlier combined models have tended to ignore the selection of sub-models and data analyses, leading to poor forecasting performance. In order to select the best forecasting models in a combined model, we propose a hybrid electricity price forecasting system that includes a data analysis module, a sub-model selection strategy module, optimized forecasting processing, and a model evaluation module. As such, the hybrid system fully exploits the advantages of a single model, thus improving the forecasting performance of the combined model. The experimental results show that the proposed system selects optimal sub-models effectively and successfully identifies future trend changes in the electricity price. Thus, the system can be an effective tool in the planning and implementation of smart grids. Keywords Electricity prices Sub-models selection strategy Combined model Hybrid forecasting system
1 Introduction The problem of electricity price forecasting influences various processes in the operation of modern electric power systems. These processes are related to other contemporary scientific and engineering problems, such as optimal power grid scheduling, energy consumption, the exploitation of energy resources, greenhouse gas emissions, simulations of electric power systems, and electricity-load demand modeling. As a result, it is a multi-disciplinary research topic within the power systems community. In addition, although price forecasting is concerned with the operation of the energy market and transactions, market participants (e.g.,
Communicated by V. Loia. & Shenghui Zhang [email protected] 1
School of Mathematics and Statistics, LongDong University, Qingyang 745000, Gansu, China
2
Faculty of Science and Technology, University of Macau, Macao 999078, China
3
School of Information Science and Engineering, Lanzhou University, Lanzhou 730000, Gansu, China
utilities, grid operators, retailers, aggregators, etc.) are interested in short-term electricity price forecasts. A variety of methods have been proposed to forecast electricity prices and can be divided into two general categories: (I) statistical models; and (II) artificial neural networks (ANN) (Wang et al. 2016; Rani and Victoire 2019; Aghaiani et al. 2018). Statistical models are direct random time-series models (Li et al. 2013) and include autoregressive (AR) (Liu and Shi 2013) mo
Data Loading...
