Published: 16 October 2020| Version 1 | DOI: 10.17632/g6m3pbh5kw.1
Wanderley Celeste,


The PPGEN-SILVA2020-csv.rar compressed folder contains 20 files in CSV format. Each file contains samples of voltage (column 1, leftmost) and current (column 2) generated by a photovoltaic string (PV) consisting of six solar panels in each of the following operating conditions: - Standard; - Full shading on panel 1; - Full shading on panel 2; - Full shading on panel 3; - Full shading on panel 4; - Full shading on panel 5; - Full shading on panel 6; - Partial shading of panel 1; - Partial shading of panel 2; - Partial shading of panel 3; - Partial shading of panel 4; - Partial shading of panel 5; - Partial shading of panel 6; - Short-circuit of panel 1; - Short-circuit of panel 2; - Short-circuit of panel 3; - Short-circuit of panel 4; - Short-circuit of panel 5; - Short-circuit of panel 6; - Electric arc. The files also contain solar irradiation (column 3) and ambient temperature (column 4) in the PV string's vicinity. Each file line represents a sample of the respective operational condition. Samples were obtained at a sampling rate of 100 kilo-samples per second, in 99.96 ms acquisition windows per PV string operation condition, generating 9,960 samples per string operation condition per collection cycle. The collection cycle implies a data acquisition window for each of the 20 operating conditions. This data collection cycle was repeated 220 times for 21 days, and at different times of the day. According to the processing architecture adopted, up to 9,960 samples can be taken as an example (or case). Initially, 1666 continuous samples were taken as an example of a specific photovoltaic string operating condition, allowing the separation of six examples per collection cycle. The compressed folder PPGEN-SILVA2020-bin.rar contains the same dataset represented in the float32 binary format..



Universidade Federal do Espirito Santo


Renewable Energy, Deep Learning, Smart Grid