GADF and Improved CNN Gearbox for Fault Diagnosis in Complex Environments
Aiming at the problem of fault diagnosis of gear box under variable load and complex environment containing noise,a new method based on Gram Angle Difference Field was proposed(for short,GADF),and an improved method of gearbox fault diagnosis with Multiple Attention Mechanism Convolutional Neural Network(for short,MACNN).Firstly,the fault data set of one-dimensional vibration signal collected from the gearbox is preprocessed.Then the one-dimensional data set is converted into two-dimensional image data by Gram Angle difference field.Secondly,the number of two-dimensional data sets is input into the improved convolutional neural network model with multiple attention mechanism for training.Finally,Softmax was used to clas-sify the gearbox fault data set.Through experimental verification,the proposed method can reach more than 99.80%in two load data sets,and the improved model training efficiency is higher and the time is shorter.Meanwhile,the proposed method is better than some published gear box fault diagnosis methods.In addition,the proposed method can solve the problem of gear box in the complex environment of variable load and noise at the same time.Still has a strong fault diagnosis ability.
万方数据
维普
