The manufacturing industry is considered one of the biggest industries in the world. Efficient and continuous operation is crucial in manufacturing industries in order to meet output demands. Hence, any machine or mechanical equipment failure could lead to unplanned stops and losses. This study proposes a maintenance strategy for mechanical equipment, specifically a gear fault diagnosis approach based on the integration of an extreme learning machine (ELM) and a geometric mean optimizer (GMO). The proposed method was tested using sets of gear vibration signals from an online database, which consists of healthy and faulty data. The GMO method was used to select an optimized parameter for the ELM method. The result shows that the proposed method is able to improve the classification performance of ELM by 10% as compared with conventional-ELM. The proposed method is also capable of being used in any manufacturing industry for maintaining their mechanical equipment.