Wind turbine generator bearing fault diagnosis method for multi-channel deep convolutional neural network

Abstract

The invention relates to a wind turbine generator bearing fault diagnosis method for a multi-channel deep convolutional neural network. The method comprises the steps of: simultaneously acquiring high-frequency vibration acceleration signals of a drive end and a non-drive end of a test bearing in various states by using a vibration acceleration sensor; analyzing the acquired vibration signals by using a time frequency analysis technology to obtain corresponding time frequency spectra; establishing a deep convolutional neural network diagnosis model, and training the diagnosis model by using the time frequency spectra and the states of the bearing as a training sample; evaluating the diagnosis model, and applying the diagnosis model to a bearing to be monitored. The method can realize automatic feature learning, avoids feature engineering, effectively utilizes multi-channel vibration signals, and has good universality and extensibility.

Description

Technical field [0001] The invention belongs to the field of wind turbine status monitoring and fault diagnosis, and particularly relates to a wind turbine bearing fault diagnosis method. Background technique [0002] Bearings are widely used in subsystems and other components of wind turbines, such as wind turbines, main shafts, gearboxes, generators, pitch systems and yaw systems. The most commonly used bearings in wind turbines are ball bearings. Ball bearing failure usually appears as wear or increased surface roughness of certain components and then develops into some major failure modes such as fatigue, cracks or fractures of the outer race, inner race, balls or cage. Bearing failure can lead to catastrophic failure of other components in the wind turbine subsystem. Serious failures of most mechanical components of wind turbines begin with bearing failure. Therefore, it is very necessary to diagnose early failures of wind turbine bearings. [0003] Bearing fault di...

AI Technical Summary

Problems solved by technology

However, unlike conventional units, the operating conditions of wind turbines are often much more complex and often operate in extremely harsh environments, and the vibration signals of bearings with initial faults usually contain a lot of noise, resulting in a very low signal-to-noise ratio ; Existing diagnostic methods usually only consider the vibration data of one channel, but lack the support for multi-channel data; wind turbine bearing fault diagnosis requires very strong prior knowledge and expert experience, by trying different feature designs and model parameter selection In order to obtain satisfactory results, there is currently no method that can intelligently learn the status characteristics of the unit from the low signal-to-noise ratio, complex and diverse monitoring data of wind turbines, and diagnose the status of the unit; the data set used for fault diagnosis is not Balanced data, the proportion of faulty samples is much smaller than that of normal samples, and the traditional pattern recognition algorithm with the overall diagnostic accuracy as the learning goal will pay too much attention to normal samples, which will reduce the diagnostic performance of faulty samples; The diagnosis model of the system is often not universal and generalizable, and it is difficult to apply to wind farms with a large number of units and different models, and it is also difficult to apply to the era of big data for operation and maintenance with increasing data volume.

Images