Multi-channel fused wind power planetary gearbox fault diagnosis method

Abstract

The invention provides a multi-channel fused wind power planetary gear box fault diagnosis method, which comprises the following steps of: collecting multi-sensor vibration signals of key positions of a wind power planetary gear box, considering fault coupling association information between channels, designing global and local convolutional neural networks to automatically extract coupling characteristics between the channels and fault characteristics in the channels, and carrying out fault diagnosis on the wind power planetary gear box. Therefore, more complete fault information of the planetary gearbox is obtained, and the adaptive dynamic weighting module is utilized to screen fault high-correlation characteristics and suppress noise interference, so that the fault diagnosis precision is further improved. The method can automatically extract and fuse the spatio-temporal characteristics in the multi-channel vibration signals, inhibits interference, and improves the fault characteristic extraction capability and the fault diagnosis precision.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis of a wind power generator gearbox, in particular to a multi-channel fusion wind power planetary gearbox fault diagnosis method. Background technique [0002] Gear transmission system is an important part of large-scale doubly-fed wind turbines and the key to energy transmission and conversion. As the core component of the traditional gear system, the planetary gearbox is also one of the main sources of frequent failures of wind turbines. The internal structure of the planetary gearbox is complicated. Under the complex working conditions of low speed, heavy load, alternating load and strong gust impact for a long time, the key components such as internal bearings and gears are prone to wear, fracture and other failures, and even chain reactions evolve into Shutdown of the entire system. These failures and unplanned downtime have seriously affected the economic benefits of wind farms and t...

AI Technical Summary

Problems solved by technology

[0004] Existing research on fault diagnosis of wind power planetary gearboxes based on deep learning mostly only focuses on the fault feature extraction of components by a single sensor, and does not take into account that when the gearbox fails, it will produce multi-part coupling. The information can provide more comprehensive state information for planetary gearbox fault diagnosis and improve the diagnostic accuracy of the model

Images