Research On Intelligent Diagnosis Method Of Rotating Machinery Based On Improved Wavelet Convolution And Transfer Learning

With the development of modern industrial manufacturing,mechanical equipment is developing in the direction of large-scale,automated,and complexity.Enterprises put forward higher requirements for high performance,safety and reliability,which makes rotating machinery fault diagnosis more important and more difficult at the same time.We will face complex working conditions and even special conditions that make it difficult to effectively install sensors.At present,the problems and challenges facing fault diagnosis are how to carry out equipment status collection and analysis,as well as fault diagnosis and health assessment.With the development of artificial intelligence technology,intelligent diagnosis technology based on deep transfer learning provides new ideas and methods for solving the above problems.In this context,we take the rotating equipment as the research object and based on deep learning and transfer learning theories.The researches are carried out from three aspects: the self-adaptive "feature enhancement" of the deep convolution model,the fault detection of the associated measuring points of rotating equipment,and the state monitoring and intelligent fault diagnosis of industrial robots.The main research contents of this paper are as follows:1)Aiming at the complexity and high task relevance of traditional artificial feature selection,a convolutional neural network based on wavelet convolution kernel is studied for adaptive feature extraction and enhancement of vibration signals of rotating equipment.We improved the convolutional layer and the pooling layer by analyzing the problems of the classic convolution model and combining the advantages of wavelet analysis in signal time-frequency characteristics and multi-scale analysis.It aims to mine the effective frequency band energy information in the vibration signal and enhance the adaptive feature extraction ability of the deep convolution model.We have carried out research on the fault diagnosis of rotating machinery and equipment under complex working conditions to provide a basis for fault diagnosis of the associated position of rotating machinery and equipment.2)Research on the fault diagnosis method of related location based on deep transfer learning.We analyze the data distribution adaptive algorithm in the field of transfer learning.Combining data distribution adaptation with an improved deep convolution model,a deep transfer model based on wavelet convolution energy is proposed.The wavelet convolution model is used to extract the characteristic information of the vibration signal of the associated position,and the data distribution adaptive method is used to reduce the distribution difference between different domains.To carry out experimental research on the associated position fault diagnosis of rotating machinery.It provides a new feasible solution for cross-measurement fault diagnosis of rotating machinery under complex working conditions.3)The robot condition monitoring system was developed.We studied the method of obtaining robot state data based on industrial protocol.The TCP/IP protocol network communication module is adopted.Based on the proposed wavelet convolution depth migration model,the robot state recognition function is developed to verify the effectiveness of the method proposed in this paper in robot state monitoring and evaluation.