Visual Vocabulary Based Photovoltaic Health Monitoring System Using Infrared Thermography

Abstract

Photovoltaic (PV) systems have gained global acceptance in terms of green, replenishable energy resources to meet energy demand with no emissions. However, PV systems are susceptible to operational and environmental stresses. Moreover, due to their no supervisory control, PV panels monitoring is necessary to keep its performance and efficiency intact. Therefore, this study monitors PV panels based on health into three sub-classes: healthy, hotspot, and faulty through infrared thermography. Thermographs key points are selected using an 8x8 uniform pixel grid and speed-up robust features (SURF) are extracted from grid intersection points. Afterwards, k-mean clustering algorithm due to its simplicity, creating single level clusters based on actual observations similarities and similar observations closeness within cluster and dissimilarity to other clusters observations, is used to transform features into visual words. Finally, shallow classifiers because of low training time and high prediction speed are utilized. After extensive testing and compressive analysis, the proposed approach was found economical, fast, and showed high testing accuracy of 97% through a multi-class shallow classifier (support vector machine) with low computational complexity and less storage size. Thus, this approach can be employed to monitor mega-watt PV systems with high accuracy and to keep performance and emissions mitigation potential high while lowering payback time.

Advanced impedance mismatch technique for detecting faults in photovoltaic systems

“…These techniques range from visual inspections to identify issues, such as shading, dust accumulation, and soiling, 1 to infrared and electroluminescence imaging combined with current-voltage (I-V) characterization for assessing module degradation. [2][3][4][5][6][7] Despite their proven effectiveness, these methods face limitations, including the requirement for costly equipment, the labor-intensive nature of inspections, and the challenge of managing large data sets in learning-based approaches. [8][9][10] Thermal imaging and I-V curve analysis play pivotal roles in fault detection, [11][12][13] albeit with limitations in the precision of fault localization.…”

“…The development of PV fault detection methodologies has made significant strides, concentrating on a variety of techniques for identifying and diagnosing defects within PV systems. These techniques range from visual inspections to identify issues, such as shading, dust accumulation, and soiling, 1 to infrared and electroluminescence imaging combined with current–voltage ( I – V ) characterization for assessing module degradation 2–7 . Despite their proven effectiveness, these methods face limitations, including the requirement for costly equipment, the labor‐intensive nature of inspections, and the challenge of managing large data sets in learning‐based approaches 8–10 …”

Thermal image analytics for industrial anomaly detection

“…Ahmed et al extracted thermal image features using the SURF method, and utilized the K-Means clustering algorithm for creating a visual vocabulary. Shallow classifiers like SVM, Naive Bayes, k-Nearest Neighbours are then utilized with this vocabulary for classifying images of photovoltaic panels as healthy, hotspot, or faulty [52]. Following this work, Ahmed et al merged the benefits of pre-trained deep learning models like ResNet18, SqueezeNet, and GoogleNet through transfer learning and of shallow classifiers like SVM, k-Nearest Neighbours, naive Bayes to classify thermal im-ages of photovoltaic panels as healthy or as one of the five different types of defects [53].…”