Helical Gearbox Fault Diagnosis Using Adaptive Artificial Neural Network and Adaptive Coyote Optimization

Abstract

The Helical gearboxes (HG) are considered a significant part of providing power transmission of manufacturing administrations and are exposed to numerous failures because of their extended and intensive situation of acceleration. Therefore, to enhance the security and dependency of the HGs, monitoring the health condition and detecting different types of failures is essential. The estimation of HG failure detection majorly includes electric signals, the noise produced by airborne, lubricant examination, thermal images, and so on. Therefore, this research proposes an Adaptive Coyote Optimization-Adaptive Artificial Neural Network (A2CO-ANN) Gearbox fault diagnosis and missing data imputation for preventing the loss of significant data values. Moreover, the comparative analysis of the A2CO-ANN technique is examined using the available datasets DTS1 and DTS2 with the existing classifiers like Random Forest (RF), K-Nearest Neighbors (KNN), Decision tree (DT), Fuzzy, as well as Adaptive ANN is examined in terms of the performance metrics. Thus, the accuracy of the A2CO-ANN method on training percentage 80 for DTS1 and DTS2 is 91.54% and 90.05%, whereas the sensitivity rate is estimated as 98.26% and 98.35%, as well as the specificity rate, is valued as 84.08% and 81.09% respectively, which is increased than the traditional methods. © 2023 IEEE.