Data analysis and tools applied to modeling and simulation of a PV system in Ecuador

Abstract

<jats:p>This paper presents a research was carried out for the management of a photovoltaic system in a Microgrid, with applications and the use of tools applied to modeling and computational simulation in the Microgrid laboratory implanted in the facilities of the University of Cuenca (Ecuador). Additionally, through the use of automatic learning techniques, the behavior of the photovoltaic system has been modeled in the study area based on radiation and temperature with very good results. In addition, several applications can be made in real engineering studies such as feasibility, performance analysis, energy estimation, educational models, etc.</jats:p>

Method of monitoring and detection of failures in PV system based on machine learning

“…The laboratory is made up of a 35kW photovoltaic system with a connection to the electricity grid and a weather station that records data [17]. An example of data analysis and tools applied to the modeling and simulation of a photovoltaic system of this study is presented in [18], Its structure consists of 4 stages, data acquisition, modeling and simulation, validation and finally its applications. This study analyzes its application for the monitoring and detection of PV system failures through the implementation of equations resulting from an analysis of databases in the SCADA system, considering the databases of the PV system and the weather station.…”

“…It should be noted that these values were calculated considering 24 hours a day, so that the zero values generated during the night have been suppressed in the graphs since there is no radiation and consequently no power. In the database corresponding to the weather station, solar radiation and temperature are defined as the most influential variables on the photovoltaic prediction models, as detailed in a preliminary study [18].…”

“…Now, considering the effects of cloudiness in this study, it can be seen that it is not possible to define a model only with the variable "time", it is necessary to find patterns with respect to the meteorological variables described above. For this reason, linear regression models are defined below using temperature and solar radiation as input variables of the functions [18]. In Equation (6), the power generated from the PVS1 is presented as a function of the variables of solar radiation ( / 2 ) and temperature (°).…”

“…On the other hand, the variables of wind speed, wind gust, wind direction and rainfall do not show a pattern of interest in the model. This analysis can also be observed when constructing the correlation matrix between the aforementioned variables, as established in a preliminary study in[18].…”