Automated Identification of Overheated Belt Conveyor Idlers in Thermal Images with Complex Backgrounds Using Binary Classification with CNN

Abstract

Mechanical industrial infrastructures in mining sites must be monitored regularly. Conveyor systems are mechanical systems that are commonly used for safe and efficient transportation of bulk goods in mines. Regular inspection of conveyor systems is a challenging task for mining enterprises, as conveyor systems’ lengths can reach tens of kilometers, where several thousand idlers need to be monitored. Considering the harsh environmental conditions that can affect human health, manual inspection of conveyor systems can be extremely difficult. Hence, the authors proposed an automatic robotics-based inspection for condition monitoring of belt conveyor idlers using infrared images, instead of vibrations and acoustic signals that are commonly used for condition monitoring applications. The first step in the whole process is to segment the overheated idlers from the complex background. However, classical image segmentation techniques do not always deliver accurate results in the detection of target in infrared images with complex backgrounds. For improving the quality of captured infrared images, preprocessing stages are introduced. Afterward, an anomaly detection method based on an outlier detection technique is applied to the preprocessed image for the segmentation of hotspots. Due to the presence of different thermal sources in mining sites that can be captured and wrongly identified as overheated idlers, in this research, we address the overheated idler detection process as an image binary classification task. For this reason, a Convolutional Neural Network (CNN) was used for the binary classification of the segmented thermal images. The accuracy of the proposed condition monitoring technique was compared with our previous research. The metrics for the previous methodology reach a precision of 0.4590 and an F1 score of 0.6292. The metrics for the proposed method reach a precision of 0.9740 and an F1 score of 0.9782. The proposed classification method considerably improved our previous results in terms of the true identification of overheated idlers in the presence of complex backgrounds.

Advanced Image Analytics for Mobile Robot-Based Condition Monitoring in Hazardous Environments: A Comprehensive Thermal Defect Processing Framework

“…The whole series of experiments and data collection experiments are carefully described in our previous work [ 17 , 35 , 41 ]. The IR videos were captured by a FLIR T640 camera (FLIR Systems, Wilsonville, OR, USA), which has a resolution of 640 × 480 pixels, a maximum frame rate of 30 frames per second, and a thermal sensitivity of 0.035 °C.…”

“…Sequences of frames are captured from IR videos that let us accurately localize the idlers. To reduce computation consumption and to raise the precision of detection, a square-shaped ROI with a size of pixels is defined on the original frames [ 35 , 41 ].…”

“…In region-based methods, a certain region in the processed image is assigned to a possible pre-defined feature. In our previous work [ 41 ], we proposed a binary classification method based on a CNN architecture to classify frames with signs of thermal defects on BC idlers. However, due to the limitations of the region-based classification method, we could not cluster the defected idlers based on their degree of severity.…”

CECvT: Initial Diagnosis of Anomalies in Thermal Images

“…Moreover, the efficacy of the CNN was validated for diagnosing anomalies in rotating machinery based on infrared thermographic imaging [25]. Infrared imagery has been employed in conveyor systems for binary classification of thermographic images to monitor the status of belt conveyor idlers using CNN [26]. Many solar power plants face challenges due to numerous defects that cause nonnegligible power losses; to address these, drone-based thermographic imaging was utilized along with CNN-based anomaly detection [27].…”

A mixed sample data augmentation method based on sparse adversarial attack

“…Deep neural networks have achieved state-of-the-art performance in numerous fields, particularly in computer vision tasks such as image classification 1 , semantic segmentation 2 , and object detection 3 . However, it is well-known that these models tend to memorize training data and make overconfident predictions, often leading to overfitting 4 .…”