Research On The Residual Stress Detection And Evaluation Based On Variational Mode Decomposition

Mechanical components with high precision and reliability are prerequisites for ensuring the normal operation of mechanical equipment.For mechanically processed components such as welding,casting,rolling,cutting,bending,etc.,residual stresses often occur,and most residual stresses are detrimental,which can seriously affect the structural strength and performance of mechanical components.Therefore,scholars at home and abroad have been working on a more accurate,convenient,fast and non-destructive method for detecting residual stress of components.In this paper,the relationship between the natural frequency and the residual stress of the mechanical system is analyzed.The influence of the difference of the vibration characteristics on the three-dimensional joint time spectrum is explored.The energy-time-frequency distribution characteristic value based on the three-dimensional joint time spectrum of the component vibration signal is proposed.It is used as a residual stress discriminant index to detect the residual stress of the component.The above method was verified by taking welded steel plate as an example.The results show that the method is effective.The main contents are as follows:Aiming at the shortcomings of Hilbert Huang transform(HHT)method for non-stationary signal analysis,an adaptive frequency band decomposition model of signal-variational mode decomposition(VMD)is introduced.Compared with HHT mode decomposition model,this method is compared with HHT mode decomposition model.The decomposition result is more satisfactory in dealing with the modal aliasing problem.The VMD method was then verified using the simulated signal.For the parameter setting problem in VMD program,the influence of key decomposition parameters-modal number and penalty factor on the accuracy of decomposition results is analyzed.By studying the spectrum map,center frequency curve and decomposition time,the selection method of sum is proposed.The feasibility of this method is confirmed by simulation signals.The state information of the residual stress of the component is reflected by detecting the state of the vibration characteristic parameter of the component,and the residual stress state of the detecting component is converted into the vibration characteristic parameter identification of the component.Through the exploration of the simulation signal,the three-dimensional joint time spectrum based on the vibration signal of the component can truly reflect the difference of the vibration characteristics of the component.To moreaccurately and effectively distinguish the difference of the three-dimensional joint spectrum,a three-dimensional joint based on the vibration signal of the component is proposed.The energy-time-frequency distribution eigenvalue method of the spectrum.The process of the method is as follows: firstly,the vibration signal of the mechanical component is collected by the pulse excitation method,and then the signal is decomposed into several IMF components which can effectively characterize the vibration characteristics of the component by the VMD method,and then the Hilbert transform is performed on each IMF component and solved.The three-dimensional joint time spectrum of the reconstructed signal is calculated.Finally,the amplitude center of gravity,the frequency center of gravity and the time center of gravity of the three-dimensional joint time spectrum are calculated,and the energy-time-frequency distribution characteristic value of the three-dimensional joint time spectrum is obtained by using the Euclidean distance,and is used as the detecting machine.Discrimination index of residual stress of components.Welded steel plates of three different states were examined by blind hole method:welding residual stress of untreated steel plate,first stress-relieving steel plate and second stress-relieving steel plate,and residual stress distribution map was made.The energy-time-frequency distribution eigenvalue method based on the three-dimensional joint time spectrum of component vibration signals is used to analyze the vibration signals of welded steel plates under different conditions.By comparing the changes of the energy-time-frequency distribution eigenvalues of the three-dimensional joint time spectrum,the stress state of the welded steel plate is accurately determined.The method is consistent with the detection results of the blind hole method,which proves the effectiveness of the proposed method.It provides a feasible method for achieving lossless residual stress detection.