The Construction And Implementation Of Intelligent Fault Diagnosis Framework For Gearbox Based On CNN And GAN

Gearboxes are key transmission components and widely used in various industrial applications.Due to the possible operational conditions,such as varying rotational speeds,long period of heavy loads,etc.,gearboxes may easily be prone to failure.Condition Monitoring(CM)has been proved to be an effective methodology to improve the safety and reliability of gearboxes.Deep learning approaches,nowadays,further enable the CM with more powerful capability to exploit faulty information from massive data and make intelligently diagnostic decisions.However,for most of conventional deep learning models,such as Convolutional Neural Network(CNN),a large amount of labelled training data is a prerequisite,while to obtain the labelled data is usually a laborious and time-consuming job and sometimes even unattainable.To tackle these issues,based on machine learning and deep learning algorithms in the field of artificial intelligence research,this paper proposes three gearbox intelligent fault diagnosis frameworks to cope with gearbox fault diagnosis under the conditions of small samples or even without fault samples.The specific research content of this article can be summarized as follows:(1)A modified convolutional neural network(MCNN)is proposed by integrating global average pooling(GAP)to reduce the number of trainable parameters and simplify the architecture of deep learning model.The proposed MCNN improves the traditional CNN's ability in fault diagnosis with limited labelled data.Two experimental gearbox datasets are utilized to demonstrate the effectiveness of the proposed MCNN method.Compared with traditional deep learning approaches,namely LSTM,CNN and its variant methods,the experimental results show that the proposed MCNN with higher discrimination and generalization ability in fault classification and diagnostics under the scenario of limited labelled training samples.(2)An automatic fault detection framework for key components of planetary gearboxes is proposed in this work,in which the finite impulse response(FIR)band-pass filter is adopted to extract the frequency components of the sun gear,planet gear,planet carrier,and ring gear,respectively,and then train the deep convolutional generative adversarial networks(DCGAN)model with the frequency components of each part separately.Afterwards,when completing the models training,the four FIR filter models and DCGAN models are combined to be a fault detection framework for key components of planetary gearboxes.The input of the framework is the vibration signals measured from a planetary gearbox,and the output are key component health conditions of the monitored planetary gearbox.The key advantage of this framework is to simultaneously and automatically identify faults,namely sun gear,planets,planet carriers and ring gear faults,with one detection.(3)A novel transfer diagnosis framework based on deep adversarial convolutional neural networks is proposed in this paper,which aims at adapting the feature distributions between training and testing data,to utilize the adequate labeled fault data from training set to help diagnosing unlabeled testing data.By adding the domain discriminant classifier into the convolutional neural network,the domain adversarial loss can be introduced into objective function,and the parameters of convolutional layers can be updated to learning domain-invariant feature representation with the adversarial training.