Research On Vibration Response Separation Of Internal Excitation Source Of Rotating Machinery

Vibration signal is the carrier of mechanical failure information,and vibration signal analysis is the main method of condition monitoring and diagnosis.In the actual situation,the components of the vibration signal are complex,and it is the result of the joint action of multiple excitation sources.Meanwhile,the vibration signal is only a fuzzy reflection of the machine's fault state.Vibration source separation plays an important role in the accurate identification,tracking and quantitative expression of faults and reducing the uncertainty of the results.Usually,the mechanical equipment in operation will be affected by the combination of the environment and the internal excitation.In condition monitoring,the internal excitation source containing fault information is the useful vibration source that needs to be separated.The vibration response of the internal excitation source in rotary machine has three important properties: fragmentation,similarity and regularity of distribution.It is absolutely necessary to use the properties of the target vibration source to solve the problem of vibration source separation.The thesis carries out research from two different methods of common component reconstruction and low-rank restoration,and discusses the separation problem under variable speed conditions.The specific research content is as follows:(1)Research on the vibration response and properties of the internal excitation source.The mechanism of rotating machinery's vibration wear analyzed,and the excitation sources wear divided in space.Subsequently,the internal excitation source and its vibration response wear defined.From the perspective of small granularity,the vibration response properties of the internal excitation source wear analyzed and summarized.As said above,the internal excitation response has three properties including fragmentation,similarity and regularity of distribution.These properties can distinguish the vibration response of the internal excitation source and the environmental excitation source,as well as the vibration response of different internal excitation sources.They providing a theoretical basis for the separation of the vibration response of the internal excitation source.(2)Study on the separation method of common component reconstruction.This decomposes multiple small-granularity internal excitation response fragments,and uses the similar properties of the internal excitation response to select the reconstructed components to realize the separation of the internal excitation response.The expression ability and separation ability of different signal decomposition methods for target signals are explored.And the separation performance of the three decomposition methods was studied at the same time.The results show that the common components reconstruction separation method can effectively separate the internal excitation response in vibration signals.Specifically,only part of the target signals can been separated by a single decomposition method,and the performance of multiple decomposition methods is better than that of a single decomposition method.(3)Research on the separation method of low rank restoration.The response fragment was constructed as a matrix in columns.According to the similarity between the segments,the matrix was decomposed to obtain the low-rank matrix,and the internal excitation response signal was obtained by reconstruction.By analyzing the separation performance of the low-rank sparse model,it is found that the non-target signal satisfies the Gaussian distribution rather than the sparsity.Consequently,low-rank sparse model is not suitable for the separation of internal excitation response.Then a low-rank noise separation model is proposed,and it can constrain the target signal to be a low-rank term and the non-target signal to be noise term satisfied with Gaussian distribution.The simulation and experimental results show that the internal excitation response obtained by this model separation has a cleaner time domain waveform and a higher estimated SNR.Moreover,a study of a multi-low-rank noise separation model is carried out.Multiple internal excitation responses can be separated simultaneously by multi-low-rank noise separation model,and the residual components of other responses can be greatly reduced due to the increased use of mutual information among internal excitation responses.(4)Research on separation of internal excitation responses under variable speed conditions.This method maps the small-grained internal excitation response in the phase space to obtain the trajectory map,and then divides and weights the trajectory map to obtain the grid phase space.Under the framework of the separation of common components,the grid phase space is calculated for similarity.Determine the common component and reconstruct the component to realize the separation of internal excitation response under variable speed conditions.The influence of variable speed on internal excitation response separation is analyzed,and the geometric invariance of internal excitation response in phase space trajectory under the condition of variable rotation speed is studied.In addition,the optimal value of phase space embedding delay defined as ? was discussed.The results show that the separation method based on phase space can effectively separate the internal excitation response under the variable speed condition.