Research On Fault Diagnosis Of Rolling Bearings Based On Manifold Learning Algorithm And Developmental Network

In the modern industrialized production process,the fault diagnosis technology of mechanical equipment is deeply favored by people.Making full use of fault diagnosis technology can help enterprises avoid major safety accidents caused by mechanical failures,which has important research value.As one of the indispensable parts of rotating machinery,rolling bearing plays a very important role in the normal operation of machinery.It can not only support the rotating body of the machinery,but also ensure the rotation accuracy of the machine,and it can also reduce the friction in the operation of the machinery.coefficient.However,the operating environment of the rolling bearing is very harsh,and it is in the process of high load and high speed for a long time,so the defects and wear of the rolling bearing are inevitable.How to accurately distinguish the possible failures of rolling bearings has been arousing the enthusiasm of a large number of scholars for their research in recent years.With the development trend of complexity,large-scale,precision and intelligence of machinery and equipment,the types of machine faults are becoming more and more complicated,and the requirements of enterprises for the accuracy of fault judgment results are gradually becoming more demanding.Therefore,this paper discusses the fault diagnosis methods of rolling bearings,focusing on the signal filtering and noise reduction,fault feature extraction,dimensionality reduction and fault diagnosis recognition in the diagnosis process,and puts forward a complete and effective fault diagnosis method.On the basis of the research method,the pyqt5 development framework combined with sql server database technology was used to build a fault diagnosis system for rolling bearings.The specific research content is as follows:(1)After comparative analysis of vibration signal noise reduction methods,the best variational mode decomposition method is selected to realize the preprocessing of vibration signal noise reduction.In order to optimize the noise reduction effect of the variational modal decomposition algorithm,the particle swarm algorithm is introduced to optimize it,which greatly improves the noise reduction performance of the algorithm.(2)The relevant features of the signal are extracted using the feature extraction method combining the time domain and the frequency domain.On this basis,the manifold learning algorithm is used to reduce the dimensionality of the fault features,and the local linear embedded algorithm is improved.The kernel local linear embedded algorithm is proposed to reduce the dimensionality of high-dimensional and overlapping data samples.Processing,thereby greatly reducing the amount of calculation for subsequent model input.(3)Aiming at the fault diagnosis and identification method,the Developmental Network is applied to the field of rolling bearing fault diagnosis,constructing a network structure suitable for rolling bearing fault diagnosis,and establishing a diagnosis model combining manifold learning algorithms and developmental networks.Experimental results show that the model significantly improves the diagnostic accuracy of rolling bearing fault diagnosis.(4)Finally,based on the constructed model,the pyqt5 development framework combined with sql server database technology was used to build a rolling bearing fault diagnosis system,and the feasibility of the system was verified through examples.