رسالة جامعية
Multi-horizon forecasting of global horizontal irradiance for the smart management of the power distribution grid in Occitania ; Prévision multi-horizon de l'éclairement global horizontal pour la gestion intelligente du réseau électrique de distribution en région Occitanie
| العنوان: | Multi-horizon forecasting of global horizontal irradiance for the smart management of the power distribution grid in Occitania ; Prévision multi-horizon de l'éclairement global horizontal pour la gestion intelligente du réseau électrique de distribution en région Occitanie |
|---|---|
| المؤلفون: | Gbemou, Shab |
| المساهمون: | Procédés, Matériaux et Energie Solaire (PROMES), Université de Perpignan Via Domitia (UPVD)-Centre National de la Recherche Scientifique (CNRS), Université de Perpignan, Stéphane Grieu |
| المصدر: | https://theses.hal.science/tel-03641953Test ; Autre. Université de Perpignan, 2021. Français. ⟨NNT : 2021PERP0050⟩. |
| بيانات النشر: | HAL CCSD |
| سنة النشر: | 2021 |
| المجموعة: | Université de Perpignan: HAL |
| مصطلحات موضوعية: | Global horizontal irradiance, Gaussian process regression, Kernel, Photovoltaic power generation, Energy transition, Forecasting, Machine learning, Éclairement global horizontal, Régression non paramétrique par processus gaussien, Noyau, Production photovoltaïque, Transition énergétique, Prévision, Apprentissage automatique, [SPI.OTHER]Engineering Sciences [physics]/Other |
| الوصف: | The penetration of intermittent renewable energies, in particular solar photovoltaic (PV), in the power distribution grid is constantly increasing due to the dynamics instilled by the energy transition. A change of the power distribution grid has thus become necessary. This mutation must allow to increase the capacity of the network without reinforcing the infrastructures and to limit the constraints, in particular voltage constraints, which start to appear on the low voltage power distribution grid.This thesis is part of the project Smart Occitania, which deals with the development of a control strategy with controlled computing cost for the predictive management of a low voltage power distribution grid. The control strategy involves intrahour and intraday forecasts of stochastic quantities, such as solar photovoltaic generation. This thesis deals with the temporal and spatio-temporal forecasting of global horizontal irradiance (GHI), in order to forecast PV power generation.First, a comparative study of machine learning tools was conducted in order to choose the most suitable method for GHI forecasting. The results showed that non-parametric Gaussian process regression (GPR) is the most suitable tool for our application. Then, the choice of the input of the GPR models has been studied. Indeed, for GPR, time or historical observations can be used as input. A comparative study was conducted, highlighting that, with respect to our application, time as an adequate input. Furthermore, the known inductive biases on the GHI dynamics have been used to build a suitable kernel. Finally, a preliminary study has been conducted, so as to determine the spatial kernel to be associated with the temporal kernel for spatio-temporal forecasting of GHI. ; La pénétration des énergies renouvelables intermittentes, en particulier du solaire photovoltaïque, dans le réseau électrique de distribution ne cesse d'augmenter du fait de la dynamique insufflée par la transition énergétique. Une mutation du réseau électrique de distribution ... |
| نوع الوثيقة: | doctoral or postdoctoral thesis |
| اللغة: | French |
| العلاقة: | NNT: 2021PERP0050; tel-03641953; https://theses.hal.science/tel-03641953Test; https://theses.hal.science/tel-03641953/documentTest; https://theses.hal.science/tel-03641953/file/These_Gbemou_Shab_2021.pdfTest |
| الإتاحة: | https://theses.hal.science/tel-03641953Test https://theses.hal.science/tel-03641953/documentTest https://theses.hal.science/tel-03641953/file/These_Gbemou_Shab_2021.pdfTest |
| حقوق: | info:eu-repo/semantics/OpenAccess |
| رقم الانضمام: | edsbas.42788BDC |
| قاعدة البيانات: | BASE |
| الوصف غير متاح. |