Anomaly detection in IR images of PV modules using supervised contrastive learning (2021-12-06T00:00:00.000000Z)

TL;DR

This work serves the community with a more realistic view on PV module fault detection using unsupervised domain adaptation to develop more performant methods with favorable generalization capabilities.

Abstract

Increasing deployment of photovoltaic (PV) plants requires methods for automatic detection of faulty PV modules in modalities, such as infrared (IR) images. Recently, deep learning has become popular for this. However, related works typically sample train and test data from the same distribution ignoring the presence of domain shift between data of different PV plants. Instead, we frame fault detection as more realistic unsupervised domain adaptation problem where we train on labeled data of one source PV plant and make predictions on another target plant. We train a ResNet‐34 convolutional neural network with a supervised contrastive loss, on top of which we employ a k‐nearest neighbor classifier to detect anomalies. Our method achieves a satisfactory area under the receiver operating characteristic (AUROC) of 73.3% to 96.6% on nine combinations of four source and target datasets with 2.92 million IR images of which 8.5% are anomalous. It even outperforms a binary cross‐entropy classifier in some cases. With a fixed decision threshold, this results in 79.4% and 77.1% correctly classified normal and anomalous images, respectively. Most misclassified anomalies are of low severity, such as hot diodes and small hot spots. Our method is insensitive to hyperparameter settings, converges quickly, and reliably detects unknown types of anomalies making it well suited for practice. Possible uses are in automatic PV plant inspection systems or to streamline manual labeling of IR datasets by filtering out normal images. Furthermore, our work serves the community with a more realistic view on PV module fault detection using unsupervised domain adaptation to develop more performant methods with favorable generalization capabilities.

References80 items

Computer vision tool for detection, mapping, and fault classification of photovoltaics modules in aerial IR videos

Learning and Evaluating Representations for Deep One-class Classification

Self-Supervised Representation Learning for Evolutionary Neural Architecture Search

Contrastive Representation Learning: A Framework and Review

Driver Anomaly Detection: A Dataset and Contrastive Learning Approach

CSI: Novelty Detection via Contrastive Learning on Distributionally Shifted Instances

Contrastive Training for Improved Out-of-Distribution Detection

Deep Learning for Anomaly Detection

Joint Contrastive Learning for Unsupervised Domain Adaptation

Bootstrap Your Own Latent: A New Approach to Self-Supervised Learning

Anomaly Detection with Domain Adaptation

Deterioration Diagnosis of Solar Module Using Thermal and Visible Image Processing

Rethinking Assumptions in Deep Anomaly Detection

Unsupervised Anomaly Detection via Deep Metric Learning with End-to-End Optimization

Deep Subdomain Adaptation Network for Image Classification

Classification-Based Anomaly Detection for General Data

Supervised Contrastive Learning

Improved Baselines with Momentum Contrastive Learning

Multi-source Domain Adaptation in the Deep Learning Era: A Systematic Survey

Deep Nearest Neighbor Anomaly Detection

A Simple Framework for Contrastive Learning of Visual Representations

Multi-Source Domain Adaptation for Text Classification via DistanceNet-Bandits

Photovoltaic defect classification through thermal infrared imaging using a machine learning approach

Momentum Contrast for Unsupervised Visual Representation Learning

Unsupervised Domain Adaptation through Self-Supervision

Contrastively Smoothed Class Alignment for Unsupervised Domain Adaptation

Unsupervised Multi-source Domain Adaptation Driven by Deep Adversarial Ensemble Learning

Weakly Supervised Segmentation of Cracks on Solar Cells Using Normalized Lp Norm

Aligning Domain-Specific Distribution and Classifier for Cross-Domain Classification from Multiple Sources

Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty

Deep Semi-Supervised Anomaly Detection

Unsupervised Embedding Learning via Invariant and Spreading Instance Feature

An intelligent flying system for automatic detection of faults in photovoltaic plants

Deep Learning for Anomaly Detection: A Survey

Contrastive Adaptation Network for Unsupervised Domain Adaptation

AdaFlow: Domain-adaptive Density Estimator with Application to Anomaly Detection and Unpaired Cross-domain Translation

Adversarially Learned Anomaly Detection

Deep Anomaly Detection with Outlier Exposure

UMAP: Uniform Manifold Approximation and Projection

Unsupervised Fault Detection and Analysis for Large Photovoltaic Systems Using Drones and Machine Vision

UAV system for photovoltaic plant inspection

A semi-automated method for Defect Identification in large Photovoltaic power plants using Unmanned Aerial Vehicles

Representation Learning with Contrastive Predictive Coding

Deep One-Class Classification

DEEP CONVOLUTIONAL NEURAL NETWORK FOR AUTOMATIC DETECTION OF DAMAGED PHOTOVOLTAIC CELLS

Algorithms and Theory for Multiple-Source Adaptation

GANomaly: Semi-Supervised Anomaly Detection via Adversarial Training

Unsupervised Feature Learning via Non-parametric Instance Discrimination

Deep Anomaly Detection Using Geometric Transformations

Deep Visual Domain Adaptation: A Survey

Learning Deep Features for One-Class Classification

Decoupled Weight Decay Regularization

AUTOMATIC FAULT RECOGNITION OF PHOTOVOLTAIC MODULES BASED ON STATISTICAL ANALYSIS OF UAV THERMOGRAPHY

Real Time Fault Detection in Photovoltaic Cells by Cameras on Drones

PV plant digital mapping for modules’ defects detection by unmanned aerial vehicles

MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications

Improved Deep Metric Learning with Multi-class N-pair Loss Objective

Densely Connected Convolutional Networks

SGDR: Stochastic Gradient Descent with Warm Restarts

Beyond Sharing Weights for Deep Domain Adaptation

Automatic detection and analysis of photovoltaic modules in aerial infrared imagery

Deep Residual Learning for Image Recognition

Innovative Automated Control System for PV Fields Inspection and Remote Control

Supervised Representation Learning: Transfer Learning with Deep Autoencoders

The Precision-Recall Plot Is More Informative than the ROC Plot When Evaluating Binary Classifiers on Imbalanced Datasets

Learning Transferable Features with Deep Adaptation Networks

Very Deep Convolutional Networks for Large-Scale Image Recognition

Transfer learning with one-class data

On the importance of initialization and momentum in deep learning

Dimensionality Reduction by Learning an Invariant Mapping

Distance Metric Learning for Large Margin Nearest Neighbor Classification

Neighbourhood Components Analysis

Anomalous Example Detection in Deep Learning: A Survey

Effective End-to-end Unsupervised Outlier Detection via Inlier Priority of Discriminative Network

Transfer Anomaly Detection by Inferring Latent Domain Representations

Future of solar photovoltaic: Deployment, investment, technology, grid integration and socio-economic aspects

Outlier Detection with Autoencoder Ensembles

Variational Autoencoder based Anomaly Detection using Reconstruction Probability

Renewables 2016 Global status report

Anoma - lous example detection in deep learning : A survey Deep learning for anomaly detection : A survey , ” arXiv preprint arXiv : 1901