Fault Diagnosis Method for Rotating Machinery Gears Under Strong Coupling Impact Interference Decomposition
The working environment of rotating mechanical gears is quite noisy,and the collected multi-channel vibration signals are usually multi-component nonlinear time-varying data.Traditional modal decomposition methods decompose the vibration signals to obtain differentiated features in the form of stochastic resonance,and complete fault feature extraction.However,these methods ignore the coupling relation between impact interference signals,and when the coupling relation is complex it is difficult to effectively extract the impact components that reflect the faults of rotating machinery gears.In order to accurately determine the fault information of mechanical gears in complex environments,a fault diagnosis method for rotating mechanical gears based on interference coupling relation decomposition is proposed.A rotating mechanical gear vibration signal interference decomposition autoencoder is designed,and the Tianniu whisker search algorithm is introduced to decompose the coupling relation extracted by the autoencoder.The vibration signal interference of rotating machinery gears under strong coupling relation is decomposed into multiple coupling correlation results.A dual channel convolutional neural network fault diagnosis model for rotating machinery gears is constructed by combining convolutional neural networks and gated recursive units.The decomposed vibration signals are respectively fed into two channels of convolutional neural networks and gated recursive units to extract different vibration signal features and achieve fault diagnosis of rotating machinery gears.The experimental results show that this pre-filtering unit can eliminate random noise in the original signal and improve the decomposition.The proposed method has lower loss values,higher accuracy,and shorter time consumption.
NETL
NSTL
万方数据
