A group of Spanish researchers has developed two novel methodologies for the detection of abnormal photovoltaic systems' operation based on minimal data requirements. The first is designed to identify sudden efficiency losses in the short term, while the second is a long-term technique designed to monitor a plant's degradation.
“The manuscript introduces two distinct methodologies employing supervised machine learning (ML) techniques,” said the academics. “Both methodologies are characterized by their reliance on minimal input variables, a deliberate design choice to facilitate their application in facilities with limited data availability. This pragmatic feature allows the proposed approaches to be implemented in pre-existing solar energy plants without installing additional equipment.”
In both approaches, the scientists implemented the recursive least squares (RLS) algorithm for anomaly detection. RLS is an adaptive filter method used in signal processing and control systems to minimize the error between a desired signal and an estimated signal.
“The radiation data at the PV Plane of Array (POA) is necessary for both models, as the PV generation directly correlates with it,” added the group. “Irradiance components are gathered from the Copernicus Atmosphere Monitoring Service (CAMS) using the PV location. However, there is no data provided from the PV installation apart from the energy production, location, and nominal power, so the orientation and elevation of the PV panels are estimated.”
Following the design of the two fault detection algorithms, the research team tested them on a number of PV plants across several locations in Spain. They were located in distinct climatic areas, with peak power generation ranging between 33 kW and 295 kW. Some were brand-new, and some were more than ten years old.
The short-term detection was tested on 22 PV plants and showed that the evolution of the coefficients resulting from using the RLS algorithm can be used for fault detection, according to the scientists. “Using only one coefficient and the radiation incident on the plane of the PV panels as an input makes the fitted coefficient evolve directly proportional to the efficiency of the PV plant,” they explained.
“The long-term model selects the most representative model of the PV plant from each month of operation and compares them by predicting the PV generation on a base year,” the researchers further noted. “This algorithm has been used with the 5 PV plants with higher historical data in this study. The results tracked the expected average 1.5% annual degradation in 4 of these installations and detected an abrupt degradation pattern in one installation.”
Their findings were presented in “Detection of abnormal photovoltaic systems' operation with minimum data requirements based on Recursive Least Squares algorithms,” published in Solar Energy. The group comprised academics from Spain's International Centre for Numerical Methods in Engineering (CIMNE) and the Technical University of Catalonia.
This content is protected by copyright and may not be reused. If you want to cooperate with us and would like to reuse some of our content, please contact: editors@pv-magazine.com.
By submitting this form you agree to pv magazine using your data for the purposes of publishing your comment.
Your personal data will only be disclosed or otherwise transmitted to third parties for the purposes of spam filtering or if this is necessary for technical maintenance of the website. Any other transfer to third parties will not take place unless this is justified on the basis of applicable data protection regulations or if pv magazine is legally obliged to do so.
You may revoke this consent at any time with effect for the future, in which case your personal data will be deleted immediately. Otherwise, your data will be deleted if pv magazine has processed your request or the purpose of data storage is fulfilled.
Further information on data privacy can be found in our Data Protection Policy.