Automatic Faults Detection of Photovoltaic Farms: solAIr, a Deep Learning-Based System for Thermal Images

Abstract

Renewable energy sources will represent the only alternative to limit fossil fuel usage and pollution. For this reason, photovoltaic (PV) power plants represent one of the main systems adopted to produce clean energy. Monitoring the state of health of a system is fundamental. However, these techniques are time demanding, cause stops to the energy generation, and often require laboratory instrumentation, thus being not cost-effective for frequent inspections. Moreover, PV plants are often located in inaccessible places, making any intervention dangerous. In this paper, we propose solAIr, an artificial intelligence system based on deep learning for anomaly cells detection in photovoltaic images obtained from unmanned aerial vehicles equipped with a thermal infrared sensor. The proposed anomaly cells detection system is based on the mask region-based convolutional neural network (Mask R-CNN) architecture, adopted because it simultaneously performs object detection and instance segmentation, making it useful for the automated inspection task. The proposed system is trained and evaluated on the photovoltaic thermal images dataset, a publicly available dataset collected for this work. Furthermore, the performances of three state-of-art deep neural networks, (DNNs) including UNet, FPNet and LinkNet, are compared and evaluated. Results show the effectiveness and the suitability of the proposed approach in terms of intersection over union (IoU) and the Dice coefficient.

Machine Learning-Based Condition Monitoring for PV Systems: State of the Art and Future Prospects

“…In this case, many types of sensors can be used that rely on the information required by the ML developers. For example, image sensors, such as thermographic [46,47], X-ray [38,48], and electroluminescence cameras [49,50], are, in general, used to analyze the types of external defects related to degradation features by providing images of the surface of photovoltaic panels or cells. By comparison, traditional sensors such as I-V, P-V, temperature, and radiation sensors can be used to determine symptoms of both external and internal defects in the system.…”

“…In addition, rather than fine-tuning the CNN itself, a SVM classifier was fed with the feature maps to train the prediction model. Pierdicca et al [46] explored thermal images recorded through infrared sensors installed in a UAV to train a hybrid mask region-based CNN model for fault classification of a PV system under varying conditions. Accordingly, three fault modes (i.e., one anomaly, non-contiguous cells with anomalies, and contiguous cells with anomalies) of degradation were studied.…”

Using UAV to Detect Solar Module Fault Conditions of a Solar Power Farm with IR and Visual Image Analysis

“…Alsafasfeh, M., et al [2] yes yes Hwang, M.-H., et al [4] yes bypass diode Libra, M., et al [8] yes yes Niccolai, A., et al [9] yes digital map Vieira, R.G., et al [10] yes bypass diode Navid, Q., et al [11] yes Henry, C., et al [12] yes yes Pierdicca, R., et al [13] yes yes deep learning Jeong, H., et al [14] yes yes diagnosis Boulhidja, S., et al [15] yes Tsanakas, J.A., et al [16] yes yes Tsanakas, J.A., et al [17] yes Gallardo-Saavedra, S., et al [18] yes Ballestín-Fuertes, J., et al [19] EL 1 Herraiz, Á.H., et al [20] yes yes Fernández, A., et al [21] yes yes 1 Electroluminescence Technique.…”

Machine Learning for Photovoltaic Systems Condition Monitoring: A Review

“…but also hybrid types. R. Pierdicca et al [22] used thermal images obtained through infrared sensor installed in a drone to train a mask region-based CNN algorithm for patterns recognition problem under PV condition monitoring criteria.…”