Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model
High-precision, short-term power forecasting for photovoltaic systems not only reduces unnecessary energy consumption but also provides power grid security. To this end, in this paper we propose a photovoltaic short-term power forecasting model based on the division of data of the 24 traditional Chi...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Frontiers Media S.A.
2023-09-01
|
| Series: | Frontiers in Energy Research |
| Subjects: | |
| Online Access: | https://www.frontiersin.org/articles/10.3389/fenrg.2023.1229695/full |
| _version_ | 1797685541196529664 |
|---|---|
| author | Yujun Liu Shutong Duan Xinrui He Hongqing Wang |
| author_facet | Yujun Liu Shutong Duan Xinrui He Hongqing Wang |
| author_sort | Yujun Liu |
| collection | DOAJ |
| description | High-precision, short-term power forecasting for photovoltaic systems not only reduces unnecessary energy consumption but also provides power grid security. To this end, in this paper we propose a photovoltaic short-term power forecasting model based on the division of data of the 24 traditional Chinese solar terms and the Adaboost-GA-BP model. The 24 solar terms were condensed from the laws of meteorology, phenology, and seasonal changes to adapt to agricultural times in ancient China and have become intangible cultural heritage. This article first analyzes the numerical characteristics of meteorological factors and demonstrates their close correlation with the turning points of the 24 solar terms. Second, using Standardized Euclidean Distance and Spearman’s Correlation Coefficients to analyze data similarity between the Gregorian half-months and the 24 solar terms divisions for comparative analysis purposes, it is shown that the intragroup data under the division of the 24 solar terms have a higher similarity, leading to an average decrease of 15.68%, 40.57%, 14.68%, and 14.64% in the MAE, MSE, RMSE, and WMAPE of the predicted results, respectively. Finally, based on the data derived from the 24 solar terms, the combined algorithm was compared with the Adaboost-GA-BP model and then was verified. The genetic algorithm and Adaboost were used to optimize the BP neural network algorithm in initial value assignment and neural network structure, resulting in a 23.42%, 18.12%, and 22.28% reduction in the mean values of the MAE, RMSE, and WMAPE of the predicted results, respectively. Analysis of the results show that using the Adaboost-GA-BP model based on the 24 solar terms for short-term photovoltaic power forecasting can improve the accuracy of photovoltaic power forecasting and significantly improve the predictive performance of the model. |
| first_indexed | 2024-03-12T00:46:38Z |
| format | Article |
| id | doaj.art-f417f03067304bd4b556473e9a563f63 |
| institution | Directory Open Access Journal |
| issn | 2296-598X |
| language | English |
| last_indexed | 2024-03-12T00:46:38Z |
| publishDate | 2023-09-01 |
| publisher | Frontiers Media S.A. |
| record_format | Article |
| series | Frontiers in Energy Research |
| spelling | doaj.art-f417f03067304bd4b556473e9a563f632023-09-14T17:15:19ZengFrontiers Media S.A.Frontiers in Energy Research2296-598X2023-09-011110.3389/fenrg.2023.12296951229695Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP modelYujun Liu0Shutong Duan1Xinrui He2Hongqing Wang3College of Science, Beijing Forestry University, Beijing, ChinaSchool of Economics and Management, Beijing Forestry University, Beijing, ChinaCollege of Science, Beijing Forestry University, Beijing, ChinaCollege of Science, Beijing Forestry University, Beijing, ChinaHigh-precision, short-term power forecasting for photovoltaic systems not only reduces unnecessary energy consumption but also provides power grid security. To this end, in this paper we propose a photovoltaic short-term power forecasting model based on the division of data of the 24 traditional Chinese solar terms and the Adaboost-GA-BP model. The 24 solar terms were condensed from the laws of meteorology, phenology, and seasonal changes to adapt to agricultural times in ancient China and have become intangible cultural heritage. This article first analyzes the numerical characteristics of meteorological factors and demonstrates their close correlation with the turning points of the 24 solar terms. Second, using Standardized Euclidean Distance and Spearman’s Correlation Coefficients to analyze data similarity between the Gregorian half-months and the 24 solar terms divisions for comparative analysis purposes, it is shown that the intragroup data under the division of the 24 solar terms have a higher similarity, leading to an average decrease of 15.68%, 40.57%, 14.68%, and 14.64% in the MAE, MSE, RMSE, and WMAPE of the predicted results, respectively. Finally, based on the data derived from the 24 solar terms, the combined algorithm was compared with the Adaboost-GA-BP model and then was verified. The genetic algorithm and Adaboost were used to optimize the BP neural network algorithm in initial value assignment and neural network structure, resulting in a 23.42%, 18.12%, and 22.28% reduction in the mean values of the MAE, RMSE, and WMAPE of the predicted results, respectively. Analysis of the results show that using the Adaboost-GA-BP model based on the 24 solar terms for short-term photovoltaic power forecasting can improve the accuracy of photovoltaic power forecasting and significantly improve the predictive performance of the model.https://www.frontiersin.org/articles/10.3389/fenrg.2023.1229695/fullphotovoltaic powershort-term forecast24 solar termsback-propagation neural networkgenetic algorithmadaboost algorithm |
| spellingShingle | Yujun Liu Shutong Duan Xinrui He Hongqing Wang Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model Frontiers in Energy Research photovoltaic power short-term forecast 24 solar terms back-propagation neural network genetic algorithm adaboost algorithm |
| title | Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model |
| title_full | Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model |
| title_fullStr | Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model |
| title_full_unstemmed | Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model |
| title_short | Short-term PV power prediction based on the 24 traditional Chinese solar terms and adaboost-GA-BP model |
| title_sort | short term pv power prediction based on the 24 traditional chinese solar terms and adaboost ga bp model |
| topic | photovoltaic power short-term forecast 24 solar terms back-propagation neural network genetic algorithm adaboost algorithm |
| url | https://www.frontiersin.org/articles/10.3389/fenrg.2023.1229695/full |
| work_keys_str_mv | AT yujunliu shorttermpvpowerpredictionbasedonthe24traditionalchinesesolartermsandadaboostgabpmodel AT shutongduan shorttermpvpowerpredictionbasedonthe24traditionalchinesesolartermsandadaboostgabpmodel AT xinruihe shorttermpvpowerpredictionbasedonthe24traditionalchinesesolartermsandadaboostgabpmodel AT hongqingwang shorttermpvpowerpredictionbasedonthe24traditionalchinesesolartermsandadaboostgabpmodel |