Skip to Main content Skip to Navigation

A study on the minimum duration of training data to provide a high accuracy forecast for PV generation between two different climatic zones

Abstract : This study focus on the minimum duration of training data required for PV generation forecast. In order to investigate this issue, the study is implemented on 2 PV installations: the first one in Guadeloupe represented for tropical climate, the second in Lille represented for temperate climate; using 3 different forecast models: the Scaled Persistence Model, the Artificial Neural Network and the Multivariate Polynomial Model. The usual statistical forecasting error indicators: NMBE, NMAE and NRMSE are computed in order to compare the accuracy of forecasts. The results show that with the temperate climate such as Lille, a longer training duration is needed. However, once the model is trained, the performance is better.
Complete list of metadatas

Cited literature [20 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01823260
Contributor : Ted Soubdhan <>
Submitted on : Monday, June 3, 2019 - 6:20:42 PM
Last modification on : Thursday, March 26, 2020 - 8:59:56 AM
Document(s) archivé(s) le : Monday, September 30, 2019 - 4:59:21 PM

File

article_rene.pdf
Files produced by the author(s)

Identifiers

Citation

Ted Soubdhan, Minh-Thang Do, Benoît Robyns. A study on the minimum duration of training data to provide a high accuracy forecast for PV generation between two different climatic zones. Renewable Energy, Elsevier, 2016, 85, pp.959-964. ⟨10.1016/j.renene.2015.07.057⟩. ⟨hal-01823260⟩

Share

Metrics

Record views

73

Files downloads

86