Font Size: a A A

Deep Learning And Its Application In Mechanical Equipment Health Monitoring And Fault Diagnosis

Posted on:2022-09-13Degree:MasterType:Thesis
Country:ChinaCandidate:Z H ZhangFull Text:PDF
GTID:2492306602458104Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
Rotating machinery has been widely used in industrial fields.The bearing is the key part of rotating equipment and its working condition directly affects the stability and efficiency of the equipment.Rotating equipment generally works in a harsh environment,which makes the performance of bearing very easy to be destroyed.In generally,it is difficult to directly observe the subtle fault.If the existing fault cannot be solved in time,it will cause an accident including the economic loss and causality.Due to the complex structure of mechanical equipment,the operating parameters are dynamic and time-varying.Therefore,the fault usually has complex and uncertain characteristics.In addition,the vibration signal acquired by the sensor is often corrupted by the background noise,which makes it difficult to analyze the fault directly.The intelligent detection technology for fault and quickly detect various damages of the equipment,thereby improving the reliability and reducing maintenance cost.In the paper,the mechanical equipment is analyzed in order to improve the real-time monitoring level,fault classification accuracy and predicting the remaining useful life accurately.Based on the analysis for the fault mechanism and industrial big data,the signal processing algorithm and deep learning technique are utilized to extract the fault characteristics in the paper in order to achieve the operation monitor and the remaining useful life prediction.In addition,it is also necessary to perform noise suppression on vibration signal in order to improve the accuracy of fault diagnosis.The research content of paper is mainly divided into the following three parts:(1)Research on equipment condition monitoring and fault diagnosis based on deep learning.A combined convolutional neural network model is constructed according to the fault pattern and damage level,then the parallel network is used to extract the time and frequency information simultaneously,and the fault pattern and damage level are identified by adjusting the hyperparameter and training the model parameter to obtain real-time operating status and achieve the fault diagnosis.Experimental data under different working conditions are used to verify.(2)Research on denoising for vibration signal based on deep encoder-decoder neural network.The paper presents an encoder-decoder convolutional neural network with skip connection to learn sparse representation of signal.Inspired by the autoencoder,the established network is comprised of encoder and decoder in order to make sure the input and the output have the same length.Meanwhile,the skip connection can avoid the gradient disappearance and improve the noise removal.Finally,the reconstructed signal has higher accuracy and lower error.(3)Research on remaining useful life prediction for engine based on deep learning.In the paper,the life cycle data of aero engine is used to predict the remaining useful life.On the other hand,the convolutional neural network combined with BIGRU’s aero-engine remaining useful life estimation network architecture is constructed to analyze the simulated data of aero engine,and compare with the existing remaining useful life prediction model.The proposed method can map the collected data to the value of remaining useful life.Besides,the presented method does not rely on the professional skills,and it can simply be applied in practice.The paper takes the CWRU bearing data and C-MAPSS dataset as experimental objects.The models for fault diagnosis,noise reduction and remaining useful life prediction are proposed,and the effectiveness is verified,which can provide technical support for the stable operation of the equipment.
Keywords/Search Tags:deep learning, fault diagnosis, neural network, remaining useful life prediction, mechanical equipment
PDF Full Text Request
Related items