Gear Fault Feature Variable Selection Model Based on IGWO-VMD Optimization

With the increasing scale of the wind power industry, the maintenance of wind turbine equipment is a major issue currently facing us. Gears are the core components for energy conversion and transmission in wind turbines, and are also prone to faults. Therefore, fault diagnosis of gears is necessary. This article takes the gears of wind turbines as the research object, uses vibration analysis method to achieve gear fault diagnosis, and conducts in-depth research on feature selection and classification of vibration signals. The data obtained from the decomposition of the gear fault diagnosis model based on IGWO-VMD was used to extract multiple fault features of the gear vibration signal, including time-frequency domain, energy entropy, sample entropy, and multi-scale arrangement entropy. In order to remove redundant and overlapping information from the extracted fault features, this paper uses CARS, IRIV, and BOSS algorithms to perform feature selection on the extracted feature variables. Finally, through an example, based on the comprehensive analysis of the number of filtered feature variables and the accuracy of the classification model, the IRIV algorithm has better feature selection performance, effectively reducing the complexity of the model.