Abstract
As objects of load prediction are becoming increasingly diversified and complicated, it is extremely important to improve the accuracy of load forecasting under complex systems. When using the group method of data handling (GMDH), it is easy for the load forecasting to suffer from overfitting and be unable to deal with multicollinearity under complex systems. To solve this problem, this paper proposes a GMDH algorithm based on elastic net regression, that is, group method of data handling based on elastic net (EN-GMDH), as a short-term load forecasting model. The algorithm uses an elastic net to compress and punish the coefficients of the Kolmogorov–Gabor (K–G) polynomial and select variables. Meanwhile, based on the difference degree of historical data, this paper carries out variable weight processing on the input data of load forecasting, so as to solve the impact brought by the abrupt change of load law. Ten characteristic variables, including meteorological factors, meteorological accumulation factors, and holiday factors, are taken as input variables. Then, EN-GMDH is used to establish the relationship between the characteristic variables and the load, and a short-term load forecasting model is established. The results demonstrate that, compared with other algorithms, the evaluation index of EN-GMDH is significantly better than that of the rest algorithm models in short-term load forecasting, and the accuracy of prediction is obviously improved.
Highlights
Load forecasting plays an important role in power systems
This paper proposes a short-term load forecasting model based on elastic net regression, which is EN-group method of data handling (GMDH)
A GMDH algorithm based on elastic net regression is proposed, named EN-GMDH
Summary
Load forecasting plays an important role in power systems. Power grid planning, power scheduling, equipment maintenance, and so on all need to be carried out according to load forecasting. The authors of a past paper [19] proposed a method where factor analysis and similar-day thinking were combined into a prediction model for short-term load forecasting, which is performing well. Through an actual measurement, we found that the model output is unstable and that the prediction accuracy is quite different when GMDH is used for short-term load forecasting This is because the GMDH algorithm uses least square estimates to fit polynomials. The GMDH algorithm, based on the elastic net regression, is proposed as a short-term load forecasting model to improve the accuracy of load forecasting and ensure the smooth operation of power systems. This paper proposes a short-term load forecasting model based on elastic net regression, which is EN-GMDH.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.