A Method for the Detection of Decrease in Power in PV Systems Using Satellite Data

Abstract

In this paper, a method to detect a decrease in the output power of photovoltaic systems is proposed. This method is based on using satellite irradiance data. In addition, fault detection is carried out with only one day's data in this method. Thus, the time elapses since the decrease in output is shorter than with the other methods. In order to mitigate the error in satellite data and improve the accuracy of fault detection, data extraction is carried out, which consists of two steps. In the first step, effective data are extracted by setting a lower irradiance limit. In the second step, "Calculation day" is determined depending on the number of effective data in one day. Fault detection, which is only conducted on the Calculation day, is conducted by comparing the expected power and the measured power. The parameters used in this study were optimized by testing 45 systems that appear normal. Subsequently, 340 systems were analyzed with the proposed method, using optimized parameters. The results showed the effectiveness of our method from the viewpoints of both accuracy and time required. In addition, three data extraction methods were considered to distinguish between the permanent decrease caused by failure, and the temporary decrease caused by partial shade. Fuzzy cluster analysis showed the best result among the three methods used.

Fault detection using acceptance ratio analysis on polycrystalline grid-connected photovoltaics system

“…Where Ypredict is the predicted energy yield (kWh), and kderation_y is derationed factor of energy yield by year (dimensionless) [16] and can be calculated as (6).…”

“…Fault detection using acceptance ratio analysis on polycrystalline … (Nurmalessa Muhammad) 1099 AC side of PV systems, such as hot-spot detection [28], [29] and satellite data [6]. Artificial intelligence (AI) approaches also have been used such as neural networks and the fuzzy classification system [30], [31].…”