A CRANE GEARBOX FAULT DIAGNOSIS METHOD INCORPORATING SVM AND MULTI-CASE MIGRATION LEARNING

Xin Zheng

Keywords

Grey wolf algorithm, variational modal decomposition, migration learning, speed reducer, failure diagnosis

Abstract

Reducer is the core component of a crane. The health is directly related to the safe operation of the whole crane. Due to the high load work content of cranes, the reducer becomes the main faulty component. The crane reducer vibration signal is difficult to extract and the hidden information is difficult to mine. Therefore, in this study, a method for extracting gearbox fault features based on the Grey Wolf algorithm combined with variational modal decomposition is proposed. Aiming at the less vibration signals of the reducer fault, a single-case migration learning fault diagnosis model is proposed. For the unstable performance of the diagnostic model using single-case source domain data samples migration training, a multi-case migration learning fault diagnosis model is proposed. The experimental results indicate that the diagnostic accuracy of the migration learning fault diagnostic model in single working condition is more than 90%. The diagnostic accuracy of the migration learning fault diagnostic model in multiple working conditions is more than 90%. The computation time is less than that of other algorithmic models. The results indicate that the designed fault diagnosis model has certain reference significance for the fault diagnosis of speed reducer in the actual production environment.

Important Links:

Go Back