Affiliations: [a] School of Economics and Management, North China Electric Power University, Baoding, Hebei, China | [b] State Grid Jining Power Supply Company, Jining, Shandong, China
Correspondence:
[*]
Corresponding author: Wei Liu, School of Economics and Management, North China Electric Power University, Baoding 071003, Hebei, China. Tel.: +86 18331131891; E-mail:[email protected]
Abstract: The power load influenced by many random factors shows
the nonlinear chaotic characteristics. At the present, the research on the
load forecasting by using the chaos theory mainly focuses on the short term
local load forecasting and the certification of chaotic characteristics is
insufficient. In this study, the chaotic characteristics of Chinese medium
term power load will be made a full proof and the chaotic neural network
improved by the theory of Euclidean distance will be used to make a precise
forecasting. Firstly, the data of the total electricity consumption from 70
months were processed by the first order difference after taking logarithm.
Secondly, the Wolf algorithm and G_P algorithm were adopted to obtain
maximal Lyapunov exponent and the fractal dimension, which proved that there
are chaotic characteristics of Chinese medium term power load. Finally, the
chaotic neural network based on the Euclidean distance was used to forecast.
The results show an excellent effect of the method. There is a premising
application of the chaotic neural network based on the Euclidean distance in
the study of load forecasting.
