ML regression models have been used to monitor the operation of a PV system by estimating the expected output, being it either power, current, or voltage, and identifying as anomalies all instances where the measured output deviates from the predicted one. A variety of statistical methods have been used for failure diagnostics, mostly involving machine learning (ML) regression models. īesides gradual degradation, the performance of a PV plant can undergo sudden changes caused by localized failures in the system. For cases of nonlinear degradation rates, change point analysis has been performed to detect the changes in the degradation slopes, after which linear degradation rates are calculated between every two consecutive points of change. In another approach, the degradation ratio is extracted from the distribution of the year-on-year degradation calculated as the rate of change of the PR between the same days in two subsequent years. A simple linear regression model fits a linear model to the raw PR time-series, or to the trend component extracted after seasonal decomposition. When a linear degradation over time is assumed, methods based on linear regression models and seasonal decomposition have been mostly used. PR can be calculated on a yearly, monthly, or daily basis, after which an analysis of the PR time-series is done to evaluate the degradation. Variants of the standard PR include a corrected PR that uses a corrected measured power to compensate for the differences in measured irradiance and module temperature, with respect to the Standard Test Conditions (STC). In many cases the degradation rates are calculated based on a metric called the performance ratio (PR), which is the ratio of the measured and nominal power. Ī PV performance analysis involves the estimation of the long-term degradation rates, that quantify the gradual reduction of performance of a PV system over time. Nowadays increasingly more research is being done on diagnosing a specific set of failures. Because of the many different types of failures, identifying one type of failure in a PV system is a challenging task. Typical failures located in PV modules include cracks, potential induced degradation (PID), burned marks and hail damage of the cells, soiling or physical damage as failure of the front glass, delamination as failure of the encapsulant, and others. Failures in a PV plant can be located in the PV modules, inverters, cables and interconnectors, mounting, or other components. While shading is difficult to measure and quantify, the other parameters can be measured within the PV plant monitoring system. A power loss in a PV plant can be correlated to the values of current, voltage, temperature, irradiance, thermal cycling, shading, and others. Efficient and reliable methods, appropriate for online monitoring, should be used to detect any failure that causes power losses. Considering the cost-effectiveness of the different techniques for failure identification (visual inspection, thermography, electroluminescence, etc.) an efficient procedure for a plant evaluation is to first check for any power loss recorded by the monitoring system, followed, if needed, by other on-site techniques for identifying the plant failure. Operation and maintenance (O & M) companies aim at detecting any failure in a PV system and taking suitable countermeasures. Our work has led us to conclude that the introduced approaches can contribute to the prompt and accurate identification of both gradual degradation and sudden anomalies in PV plants.Įvaluating the status of a PV plant is an important task in maintaining a high output performance and low operating costs. The performance of the introduced methods is demonstrated on data from three different PV plants located in Slovenia and Italy monitored for several years. Depending on the data available in the PV plant monitoring system, the appropriate method for each degradation class can be selected. In this paper, we introduce different approaches for both gradual degradation assessment and anomaly detection. This has motivated our work to develop and implement statistical methods that can reliably and accurately detect the performance issues in a cost-effective manner. Two main classes of degradation exist, being it either gradual or a sudden anomaly in the PV system. Operation and maintenance systems aim at increasing the efficiency and profitability of PV plants by analyzing the monitoring data and by applying data-driven methods for assessing the causes of such performance degradation. Photovoltaic (PV) plants typically suffer from a significant degradation in performance over time due to multiple factors.
0 Comments
Leave a Reply. |