Louiza Ait Mouloud, Aissa Kheldoun, Abdelhakim Deboucha, Saad Mekhilef
{"title":"利用时间融合变压器对多时间步长的全球水平辐照度的可解释预报","authors":"Louiza Ait Mouloud, Aissa Kheldoun, Abdelhakim Deboucha, Saad Mekhilef","doi":"10.1063/5.0159899","DOIUrl":null,"url":null,"abstract":"Accurate prediction of solar irradiance is essential for the successful integration of solar power plants into electrical systems. Despite recent advancements in deep learning technology yielding impressive results in solar forecasting, their lack of interpretability has hindered their widespread adoption. In this paper, we propose a novel approach that integrates a Temporal Fusion Transformer (TFT) with a McClear model to achieve accurate and interpretable forecasting performance. The TFT is a deep learning model that provides transparency in its predictions through the use of interpretable self-attention layers for long-term dependencies, recurrent layers for local processing, specialized components for feature selection, and gating layers to suppress extraneous components. The model is capable of learning temporal associations between continuous time-series variables, namely, historical global horizontal irradiance (GHI) and clear sky GHI, accounting for cloud cover variability and clear sky conditions that are often ignored by most machine learning solar forecasters. Additionally, it minimizes a quantile loss during training to produce accurate probabilistic forecasts. In this study, we evaluate the performance of hourly GHI forecasts on eight diverse datasets with varying climates: temperate, cold, arid, and equatorial, for multiple temporal horizons of 2, 3, 6, 12, and 24 h. The model is benchmarked against both climatological persistence for deterministic forecasting and Complete History Persistence Ensemble for probabilistic forecasting. To prove that our model is not location locked, it has been blind tested on four completely different datasets. The results demonstrate that the proposed model outperforms its counterparts across all forecast horizons.","PeriodicalId":16953,"journal":{"name":"Journal of Renewable and Sustainable Energy","volume":"17 1","pages":"0"},"PeriodicalIF":1.9000,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Explainable forecasting of global horizontal irradiance over multiple time steps using temporal fusion transformer\",\"authors\":\"Louiza Ait Mouloud, Aissa Kheldoun, Abdelhakim Deboucha, Saad Mekhilef\",\"doi\":\"10.1063/5.0159899\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Accurate prediction of solar irradiance is essential for the successful integration of solar power plants into electrical systems. Despite recent advancements in deep learning technology yielding impressive results in solar forecasting, their lack of interpretability has hindered their widespread adoption. In this paper, we propose a novel approach that integrates a Temporal Fusion Transformer (TFT) with a McClear model to achieve accurate and interpretable forecasting performance. The TFT is a deep learning model that provides transparency in its predictions through the use of interpretable self-attention layers for long-term dependencies, recurrent layers for local processing, specialized components for feature selection, and gating layers to suppress extraneous components. The model is capable of learning temporal associations between continuous time-series variables, namely, historical global horizontal irradiance (GHI) and clear sky GHI, accounting for cloud cover variability and clear sky conditions that are often ignored by most machine learning solar forecasters. Additionally, it minimizes a quantile loss during training to produce accurate probabilistic forecasts. In this study, we evaluate the performance of hourly GHI forecasts on eight diverse datasets with varying climates: temperate, cold, arid, and equatorial, for multiple temporal horizons of 2, 3, 6, 12, and 24 h. The model is benchmarked against both climatological persistence for deterministic forecasting and Complete History Persistence Ensemble for probabilistic forecasting. To prove that our model is not location locked, it has been blind tested on four completely different datasets. The results demonstrate that the proposed model outperforms its counterparts across all forecast horizons.\",\"PeriodicalId\":16953,\"journal\":{\"name\":\"Journal of Renewable and Sustainable Energy\",\"volume\":\"17 1\",\"pages\":\"0\"},\"PeriodicalIF\":1.9000,\"publicationDate\":\"2023-09-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of Renewable and Sustainable Energy\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.1063/5.0159899\",\"RegionNum\":4,\"RegionCategory\":\"工程技术\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q4\",\"JCRName\":\"ENERGY & FUELS\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Renewable and Sustainable Energy","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1063/5.0159899","RegionNum":4,"RegionCategory":"工程技术","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q4","JCRName":"ENERGY & FUELS","Score":null,"Total":0}
Explainable forecasting of global horizontal irradiance over multiple time steps using temporal fusion transformer
Accurate prediction of solar irradiance is essential for the successful integration of solar power plants into electrical systems. Despite recent advancements in deep learning technology yielding impressive results in solar forecasting, their lack of interpretability has hindered their widespread adoption. In this paper, we propose a novel approach that integrates a Temporal Fusion Transformer (TFT) with a McClear model to achieve accurate and interpretable forecasting performance. The TFT is a deep learning model that provides transparency in its predictions through the use of interpretable self-attention layers for long-term dependencies, recurrent layers for local processing, specialized components for feature selection, and gating layers to suppress extraneous components. The model is capable of learning temporal associations between continuous time-series variables, namely, historical global horizontal irradiance (GHI) and clear sky GHI, accounting for cloud cover variability and clear sky conditions that are often ignored by most machine learning solar forecasters. Additionally, it minimizes a quantile loss during training to produce accurate probabilistic forecasts. In this study, we evaluate the performance of hourly GHI forecasts on eight diverse datasets with varying climates: temperate, cold, arid, and equatorial, for multiple temporal horizons of 2, 3, 6, 12, and 24 h. The model is benchmarked against both climatological persistence for deterministic forecasting and Complete History Persistence Ensemble for probabilistic forecasting. To prove that our model is not location locked, it has been blind tested on four completely different datasets. The results demonstrate that the proposed model outperforms its counterparts across all forecast horizons.
期刊介绍:
The Journal of Renewable and Sustainable Energy (JRSE) is an interdisciplinary, peer-reviewed journal covering all areas of renewable and sustainable energy relevant to the physical science and engineering communities. The interdisciplinary approach of the publication ensures that the editors draw from researchers worldwide in a diverse range of fields.
Topics covered include:
Renewable energy economics and policy
Renewable energy resource assessment
Solar energy: photovoltaics, solar thermal energy, solar energy for fuels
Wind energy: wind farms, rotors and blades, on- and offshore wind conditions, aerodynamics, fluid dynamics
Bioenergy: biofuels, biomass conversion, artificial photosynthesis
Distributed energy generation: rooftop PV, distributed fuel cells, distributed wind, micro-hydrogen power generation
Power distribution & systems modeling: power electronics and controls, smart grid
Energy efficient buildings: smart windows, PV, wind, power management
Energy conversion: flexoelectric, piezoelectric, thermoelectric, other technologies
Energy storage: batteries, supercapacitors, hydrogen storage, other fuels
Fuel cells: proton exchange membrane cells, solid oxide cells, hybrid fuel cells, other
Marine and hydroelectric energy: dams, tides, waves, other
Transportation: alternative vehicle technologies, plug-in technologies, other
Geothermal energy