Extreme Learning Machine Optimization Based on Eel and Grouper Optimizer for Gear Fault Diagnosis
M. K. Harith, M. Firdaus Isham, R. Amirulaminnur, Mohd Syahril Ramadhan Mohd Saufi, W. A. A. Saad, M. D. A. Hasan, Mat Hussin Ab Talib · Journal of Physics Conference Series · 2025
Abstract The reliability, dependability, and sustainability of machinery are essential for enhancing industrial productivity and efficiency. Any instances of machine components failing or malfunctioning can lead to unforeseen downtime and financial repercussions. In response, this research introduces a maintenance method for mechanical components that focuses on diagnosing gear failures through the utilization of an extreme learning machine (ELM) optimization technique known as Eel and Grouper Optimizer (EGO). A series of gear vibration signals sourced from an online repository, comprised of both operational and faulty data, were utilized to assess the proposed methodology. The EGO methodology was implemented to ascertain an optimal configuration for the ELM approach, specifically focusing on determining the appropriate number of neurons, input weight, and bias range values. The results suggest that the proposed strategy improves the classification accuracy of ELM by 14% compared to conventional method. This approach is also transferable to other industries seeking to enhance the dependability and sustainability of their facilities.