Research on Fault Diagnosis of Motor Rolling Bearing Based on Improved Multi-Kernel Extreme Learning Machine Model
Guojun Zhang, Tong Zhou · Artificial Intelligence Advances · 2023
Motors play a crucial role in energy conversion and are essential components of mechatronic systems. However, diagnosing faults in rolling bearings during motor operation presents significant challenges, making it difficult to achieve high accuracy in fault identification. To address these challenges, this paper introduces a novel intelligent diagnostic method based on an enhanced multi-kernel extreme learning machine (ELM) model. While the ELM model is widely used for diagnosing motor rolling bearing faults, it often struggles to classify complex vibration data. To improve its performance, this study proposes a multi-kernel ELM (MKELM) model that integrates three traditional kernel functions: Gaussian, polynomial, and perceptron kernels. Additionally, to overcome the challenges posed by the numerous parameters and the risk of local optima in the MKELM model, the kernel parameters were optimized using the Grey Wolf Optimization (GWO) algorithm, resulting in the GWO-MKELM algorithm. Finally, the GWO-MKELM algorithm was applied to diagnose motor rolling bearing faults. Experimental results show that this method achieves a 99.6% accuracy rate and effectively identifies various types of bearing faults.