| Gear 、 rolling bearing and other parts as the core elements of the mechanical equipment,its state is directly related to the performance of the whole mechanical equipment.At present,scholars have carried out researches on this phenomenon and made some achievements.However,when mechanical equipment fails in industrial production,it will be accompanied by strong background noise,and even multiple fault signals will be coupled to each other,resulting in complex faults,which makes it more difficult to monitor the status of equipment and diagnose mechanical faults.Corresponding fault diagnosis methods are needed for different fault types.Therefore,based on the symplectic geometry mode decomposition(SGMD)method,this paper proposes the improved symplectic geometry mode decomposition(ISGMD),the weighted sliding symplectic geometry mode decomposition(WSSGMD)and the weighted symplectic geometry matrix machine(WSSMM)fault diagnosis method,aiming at the problems in the fault diagnosis field such as composite fault and strong background noise.This paper is supported by the national key research and development program "Research on Service Quality Inspection and Maintenance Quality Control Technology of Major Complex Electromechanical System"(No.2016YFF0203400)and "Development of Automobile High Reliability On-board Power Electronics Integrated System"(No.2018YFB0104600).Two improved symplectic geometric mode decomposition methods and weighted symplectic support matrix machine are proposed in the paper,then we probe into the effectiveness of these methods.The main work and innovation of this paper are as follows:(1)SGMD method has strong noise robustness and is suitable for processing nonstationary signals.However,this method also has shortcomings.In the process of symplectic geometry reconstruction,the method uses the similar period and frequency to artificially set the termination conditions to reconstruct the final symplectic geometry components,which will lead to the uncertainty of the analysis results.Therefore,the ISGMD method is proposed.Hierarchical Cluster method is used to restructure the initial Symplectic Geometry components,and then the final Cluster Symplectic Geometry Components(CSGCs)are obtained.The experimental results show that the ISGMD method can effectively extract the features of complex fault signals of rotating machinery and improve the accuracy of fault diagnosis.(2)The effect of SGMD method in the processing of early gear fault signals is not ideal,which is because of the following two shortcomings of SGMD method:(1)The method directly reconstructs the trajectory matrix through the original time series,and its weak fault features are submerged in the global time series.(2)This method is to reduce noise by eliminating the components with small energy,but the components with small energy may contain weak fault information,which makes this method not suitable for early fault feature extraction.Therefore,the WSSGMD method is proposed in this paper.Firstly,a sliding window add to the original one dimension time series to reconstruct time sequence,and put forward the variable entropy to weight Symplectic Geometry components.Then we will get the weighted symplectic geometry components(WSGCs)with most of the fault information.The results of emulational and experimental signal analysis show that the WSSGMD can effectively extract the characteristics of early gear fault signals.(3)Support Matrix Machine(SMM)is a new type of classifier,and takes matrix as input.However,SMM has some limitations when dealing with complex classification problems.This is because the input matrix contains a lot of background noise,which seriously affects the accuracy of fault diagnosis.Therefore,the WSGMM is proposed in this paper.Firstly,the eigenvector matrix is obtained by symplectic geometry decomposition.Then the weighted symplectic geometry coefficient matrix is obtained through the variable entropy weighting process.Finally,the weighted symplectic geometry feature matrix is input into the SMM for classification.The signal analysis of planetary gear experiment shows that WSGMM can effectively diagnose gear fault types. |
