Rotating Mechanical Non-stationary Signal Characteristics Analyzer Development

Rotating machine is widely exploited in applications.Especially the large size rotating machine usually serves as the key equipment in manufacturing enterprises.It plays an irreplaceable role in the fields of the national economy.It will result in serious aftermath in the event of an equipment fault.Therefore it is necessary to employ condition monitoring and fault diagnosis to the key equipment to prevent potential troubles.The condition monitoring and fault diagnosis is highly dependent on features analysis and extraction technology.The signal analysis and feature extraction function of traditional instrument on the stationary signal has been quite completive;however,it is insufficient for on non-stationary signals,which contain rich feature information.Hence,it is necessary to develop a virtual instrument with powerful analysis functions on non-stationary signals and traditional analysis functions.It will make up the deficiencies of annlysing functions in the instrument feld.This paper is a subproject of the Natural Science Foundation-funded projects of Yunnan Province,"Research on rotating machine non-stationary vibration signal analysis and characteristics extraction technology".Based on a brief introduction on the basic theory of rotating machinery fault diagnosis by illustrating the characteristics of different faults and the theory of linear time-frequency analysis by the thesis from engineering point of view, the core technique--time-variant filtering technology has been further studied.Following the basic signal processing steps,new approaches has been put forward and has solved some technical difficulties in time-varying filter schemes.Much research work has been spent on the preprocessing of non-stationary signal,which includes the arrangement of all preprocessing steps,the anti frequency aliasing filtering problem of non-stationary signal, the windowing processing of non-stationary signal in time-frequency analysis,the constraints to time-varying filtering with window function,and the selection criteria of analysis window function and synthesis window function for time-varying filtering.The thesis leads a further development on the obtaining of accurate and precious speed profile estimation with or without tacho.Main highlights in the thesis include that:it introduces a general expression of screening function for time-varying tracking filtering and its four common forms and it realizes the time-varying filtered signal synthesis by using the screened Gabor coefficients,which provids a new method to realize non-stationary signal tracking filtering--time-varying filtering based on Gabor Transform & Gabor Expansion (GT & GE).The thesis has also worked on the study of the post-processing of time-varying filtering. It has built a series fuctions such as Bode plot,Nyquist plot,orbit plot of single-component signal etc.And it has had further study on the even angle resampling of non-stationary signal and realized even angle resampling based on speed pulse and even angle resampling based on tacholess speed profile estimation.In addition for the problem of error accumulation from the resampling time mark calculation of the latter method,a new even angle resample method based on setting mark by zero points of a mono-component signal has been present in the thesis,which can effectively eliminate the error accumulation effect.Based on the achievement of former scholars and the new approaches studied in this thesis,a Virtual Instrument for rotating machinery vibration feature analysis has been developed with the intergration of present new approaches in the thesis.In the research work,many experiments have been done on a laboratory using test rig, which can simulate the running condition of routing machinery,and acquired the experimental data.It has also applied the developed instrument to do analysis and extract feature with the experimental data.And the powerful analysis functions of the virtual instrument have been revealed.