A hybrid prediction model based on improved multivariable grey model for long-term electricity consumption
- PDF / 1,190,010 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 15 Downloads / 197 Views
ORIGINAL PAPER
A hybrid prediction model based on improved multivariable grey model for long-term electricity consumption Xiaohong Han1 · Jun Chang1 Received: 8 April 2020 / Accepted: 6 October 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract The accurate and stable prediction of electricity consumption is essential for intelligent power systems in rapidly developing countries. Grey prediction model is one of choices for prediction under the condition of limited historical data. Nonetheless, it seems rather sceptical using single-variable grey prediction model to predict the dynamics of a complex system. This paper presents a novel multivariable grey prediction model based on first-order linear difference equation for long-term electricity consumption prediction. The proposed model solves the problem of parameter estimation and variable prediction deriving from different approaches through rewriting the whitenization equation of multivariable grey model (MGM(1, m)). To validate the effectiveness of the proposed hybrid model, the electricity consumption is estimated and predicted over the data from Shanxi province and Beijing city in China from 1999 to 2018. The results show that the hybrid model provides a better estimation and prediction performance compared with other prediction model for predicting electricity consumption. Keywords Difference equation · Electricity consumption forecasting · Multivariable grey model · Grey relational analysis
List of symbols Grey relational analysis χ0 (k) Reference data at time step k χi (k) Comparative data of the ith influencing factor at time step k χ˜ 0 (k) Normalized reference data at time step k χ˜ i (k) Normalized comparative data of the ith influencing factor at time step k ζ0i (k) Grey relational coefficient between χ˜ 0 (k) and χ˜ i (k) ρ Distinguishing coefficient γ0i Grey relational grade between reference and comparative data sequences
Multivariate grey model X (0) (0) Xi
B 1
Matrix of original data sequence Original data sequence of the variable i
Xiaohong Han [email protected] College of Data Science, Taiyuan University of Technology, Taiyuan 030024, Shanxi, China
(0)
xi ( j) X (1) (1) Xi (1) xi ( j) Z (1) (k) (1) xˆi (k)
Original data of the variable i at time step j Matrix of accumulated data sequence Accumulated data sequence of the variable i Accumulated data of the variable i at time step j Vector of the background values at time step k (1) Estimated sequence of the variable xi at time step k (0) Xˆ (0) (k) Estimated sequence of the variable xi at time step k
1 Introduction 1.1 Background With the increase in population and large-scale industrialization, electricity consumption around the world is rising rapidly. Due to the non-storage of power resources and the uncertainty of coal, hydro, wind, and solar power, electricity consumption forecasting is very important for managing power resources successfully and using energy effectively. Excessive electricity supplying will lead to energy investment waste and energy di
Data Loading...
