Fault Diagnosis Of Scroll Compressor Based On Time-Frequency Diagram And Convolutional Neural Network

With the wide application of scroll compressors in various fields,as an important part of the working system,the quality of its operation directly affects the efficiency and quality of the entire operation.However,because the scroll compressor belongs to the emerging fluid machinery,its special The non-linear and non-stationary fault signals generated by the body structure have not established a complete database,so it is very important to study the faults of the scroll compressor.Combining the operation principle of scroll compressor and the characteristics of vibration signal,The neural network is introduced into the fault diagnosis research of scroll compressor to realize the fault diagnosis of scroll compressor.this paper proposes a fault diagnosis method based on time-frequency diagram and convolutional neural network.Classification and identification of fault types of rotary compressors.The primary research results can be shown as follows:（1）The working principle of the scroll compressor and the vibration mechanism generated by the vibration signal are studied.Firstly,the operating principle and internal structure are described.The components of the vibration signal are analyzed by studying the structure mechanism and movement mode.Finally,the difficulties in the fault diagnosis of scroll compressors are explained.（2）The Time-frequency analysis method and convolutional neural network are studied.The time-frequency analysis methods such as short-time Fourier transform,wavelet transform and Hilbert-Huang transform are introduced,two sets of simulation signals are constructed,and the characteristic expression effects of different Time-Frequency analysis methods are compared.The results show that the complex Morlet wavelet basis function has the best Secondly,it expatiates the fundamentals and framework of convolutional neural network,and exounds the training process of convolutional neural network.In view of the excellent recognition ability of convolutional neural network for two-dimensional images,it can be used as vortex compression Neural Network for Machine Fault Diagnosis.（3）Time-frequency diagram combined with convolutional neural network to solve the problem of scroll compressor fault diagnosis.A signal acquisition system composed of a software pIatform,sensors and other related hardware is built to collect fault vibration signals of scroll compressors,and apply wavelet transform to transform the vibration signals into time-frequency graphs,which are preprocessed as input to the convolutional neural network,Training convolutional neural network,In order to realize the identification and classification of fault types.Through the results,we can know that the fault diagnosis method proposed in this article has achieved better accuracy rate.