Fault Diagnosis of Self-aligning Conveyor Idler in Coal Handling Belt Conveyor System by Statistical Features Using Random Forest Algorithm

Abstract

A coal handling system equipment is a bulk material handling system; it plays an important role in key mechanical industries. It holds the important aspects of a country economy in mining, smelter plants, thermal power plants, process industries, etc. Considering operation attributes of the coal handling belt conveyors, various parameters have to be taken into account while designing, as it has to convey materials from one location to another continuously for most part of the year. In broad spectrum, the flat belt coal conveyor has to be with maximum load handling ability, for conveying over long distance in a single stroke. Hence, it has to be steadfast in design, with easy operation and maintenance and high dependability in function. Self-aligning conveyor roller (SACR) is an important element in coal belt conveyor. It is placed between the carrying conveyor idlers to vary the sideways dislocation caused by imbalance loading which is difficult to avoid in harsh loading conditions. When the coal conveyor belt moves against the carrying rollers, there is a difference in frictional force between two sides, which will make the top strand of the coal belt conveyor to twist toward the center. Further, the crisscross movement, offset from the center line, and damage of coal conveyor belt were competently prevented by self-aligning conveyor roller. As SACR is found to be critical in coal belt conveyor systems, it becomes compulsory to supervise its smooth and continuous functioning. To make sure this certain, condition monitoring of self-aligning conveyor roller (SACR) should be done periodically which principally creates a classification or categorization problem. Self-aligning conveyor roller is made of vital elements like groove ball bearing, main central shaft, and the external shell. In this case, it is categorized with the below mentioned cases such as coal handling belt conveyor with SACR running in no-fault condition (NFC), with groove ball bearing fault condition (BBFC), with main shaft fault (MSF), with combined ball bearing fault condition and main shaft fault (BFC & MSF). A model investigational arrangement has been made as per the actual coal handling belt conveyor operating conditions and research requirement. Followed by the fabrication of SACR setup, the vibration signatures were obtained from the model for frequently occurring fault discussed earlier. These vibration signatures are fed to digital convertor and transformed to digital signals. From the digital vibration signals, statistical features were calculated. Then, effective statistical features were extracted and provided as input to random forest algorithm, followed by categorization, which was performed by random forest algorithm. In the current work, the random forest algorithm achieved 90.2% categorization accuracy, which summarizes the algorithm correctness in fault prediction self-aligning conveyor roller failure and life assessment.