A kind of gear box composite fault component separation diagnosis method and device

Abstract

The invention discloses a method and a device for separating and diagnosing components of a composite fault of a gearbox. Starting from the typical failure modes of the main components of the gearbox and gears, combined with the convolution sparse theory, a multi-scale composite regularization convolution sparse separation diagnosis algorithm is proposed. Separation of linear aliasing and coupled composite faults. According to the fault dynamics difference between bearings and gears, the idea of ​​multi-scale convolution kernel is proposed to better match the fault components. A composite regularization sparse model is proposed for the statistical differences of each component in the composite fault signal. The optimization problem of the model is transformed into the frequency domain to realize the fusion solution of multi-scale convolution kernels, and the frequency domain optimization problem is decomposed into sub-problems alternately iteratively solved by using the alternate iterative solution framework in the alternating direction multiplier method. The experiment verifies the effectiveness of the method. Feasibility and effectiveness. Effective separation and diagnosis of compound faults are realized, and Gaussian noise and slowly varying harmonic interference are avoided.

Description

technical field [0001] The invention belongs to the technical field of gearbox composite fault diagnosis, and in particular relates to a method and device for separating and diagnosing components of a gearbox composite fault. Background technique [0002] As the most widely used rotary component system in many mechanical equipment, gearboxes play an irreplaceable role in mechanical transmission. With the increasingly severe operating environment of mechanical equipment, the rotating body parts such as bearings and gears in the gearbox are easily damaged. It is very important to effectively monitor and diagnose the operating status of the above-mentioned rotating parts. At present, most of the research focuses on the fault detection of single components such as bearings or gears. Most of the traditional signal processing methods cannot analyze the composite faults of multiple components at the same time. In addition, because the vibration sensor is generally arranged on the ...

AI Technical Summary

Problems solved by technology

However, the general sparse representation based on predefined dictionaries needs to set a redundant dictionary satisfying the unique sparsity of the class according to the prior expert information, and perform sparse reconstruction of the latent components according to the sparse representation of dictionaries of different classes, resulting in There are problems such as low accuracy caused by insufficient matching and noise interference, and the need to select a suitable dictionary based on experience.

Images