Towards the Automation of Infrared Thermography Inspections for Industrial Maintenance Applications

Pablo Venegas,Mario Ortega,Idurre Sáez de Ocáriz,Eugenio Ivorra

doi:10.3390/s22020613

Pablo Venegas, Mario Ortega + Show 2 more

Open Access

https://doi.org/10.3390/s22020613

Copy DOI

Abstract

The maintenance of industrial equipment extends its useful life, improves its efficiency, reduces the number of failures, and increases the safety of its use. This study proposes a methodology to develop a predictive maintenance tool based on infrared thermographic measures capable of anticipating failures in industrial equipment. The thermal response of selected equipment in normal operation and in controlled induced anomalous operation was analyzed. The characterization of these situations enabled the development of a machine learning system capable of predicting malfunctions. Different options within the available conventional machine learning techniques were analyzed, assessed, and finally selected for electronic equipment maintenance activities. This study provides advances towards the robust application of machine learning combined with infrared thermography and augmented reality for maintenance applications of industrial equipment. The predictive maintenance system finally selected enables automatic quick hand-held thermal inspections using 3D object detection and a pose estimation algorithm, making predictions with an accuracy of 94% at an inference time of 0.006 s.

Highlights

It is well known that the maintenance of industrial equipment, machines, or facilities tends to extend their useful life, ensure correct performance for a longer amount of time, improve efficiency, reduce the number of failures, and increase the safety of their use [1].Maintenance management has evolved significantly over time
The first stage in the development of the predictive model for electrical failures was to determine the components of the electronic board, whose behavior will enable the classification of the different operating conditions
Image classification technologies that process the entire image and that do not require one to determine areas from which features can be extracted are currently being developed. These technologies are part of the science of deep learning, and, in addition to finding limitations in dealing with infrared images, there is no public database of infrared images that can be employed for training in the industrial maintenance field, nor are there pretrained networks that employ transfer learning techniques. These factors, together with the necessity of powerful GPUs and long training times, makes the use of deep learning techniques difficult to recommend for a general infrared thermography (IRT) methodology in industrial maintenance

Summary

Introduction

It is well known that the maintenance of industrial equipment, machines, or facilities tends to extend their useful life, ensure correct performance for a longer amount of time, improve efficiency, reduce the number of failures, and increase the safety of their use [1].Maintenance management has evolved significantly over time. Corrective maintenance was the common rule at the early beginning of factory use, where equipment failures were solved as they occur This type of maintenance was accepted at early stages of the industry when downtime was not critical. After the second industrial revolution, maintenance evolved to apply preventive maintenance, where the equipment was periodically reviewed, and certain components were replaced based on statistical estimates, often provided by the manufacturer. The drawback of this concept of maintenance was the high associated costs due to strict replacement deadlines, which were often overestimated

Methods

Results

Discussion

Conclusion