| As an important power transmission equipment,rotating machines are widely used in various fields such as wind turbine,helicopters,marine and other large engineering machines.Rolling element bearings and gears are the most critical parts of a rotating machines which health condition can affect the overall performance and reliability of a rotating machine,and their failure will not only affect the running state of the mechanical equipment,but also cause casualties in serious cases.Therefore,realizing the condition monitoring and fault diagnosis of the key components of rotating machines are extremely important to ensure the safety of the entire operation.Condition monitoring signals acquired from rotating machinery in practical applications often exhibit strong nonstationary characteristics due to load or speed changes,or during a machine startup or shutdown,as a result,the traditional signal processing and fault diagnosis techniques dealing with stationary signals such as Fourier transform,envelope analysis are no longer applicable.Time-frequency analysis(TFA)is an effective tool to disclose the changing dynamic in a non-stationary signal.Therefore,this study discusses the application of time-frequency analysis in fault diagnosis of rotating machinery at variable speeds,and proposes improved techniques for problems associated with poor noise robustness and time-frequency energy divergence in time-frequency analysis.(1)This study proposes a fault diagnosis technique for varying speed rolling bearings based on a combined ensemble empirical mode decomposition(EEMD)and synchosqueezing transform(SST).Firstly,EEMD technique is employed to decompose the envelope signal of a bearing condition monitoring signal into various intrinsic mode functions(IMFs)and to improve the signal-to-noise ratio of the signal through the noise reduction in the process.Secondly,SST is utilized to obtain the energy concentrated timefrequency distribution of each IMF,the time frequency distributions of the IMFs are then added to obtain a noise-reduced overall time frequency distribution of the bearing defect signal for an accurate fault diagnosis.(2)This study presents an analysis method for non-stationary signals,which combined multi-synchrosqueezing transform and ridge extraction(MSST-RE).MSST is based on SST,and using multi-synchrosqueezing iteration process to gradually concentrate the time-frequency energy of strong variation signal.However,the noise contained in a signal will be compressed at the same time as the number of iterations increases in the transform,which will interfere with the analysis signal.In order to solve this problem,the RE technique is used to extract each mono component contained in the TFA results of the MSST in turn.All the extracted mono-components time frequency distributions by superimposing to obtain a noise-free time frequency representation.(3)This study proposes a fault diagnosis technique for varying speed rotating machine based on a combined polynomial chirplet transform and multi-synchrosqueezing transform(PCT-MSST).PCT is a common TFA method for processing non-stationary signals,which based on the short time Fourier transform(STFT),it improves instantaneous frequency estimation by introduce frequency rotating operator and frequency shifting operator.However,the time-frequency spectrum obtained from the PCT exist blurry time-frequency energy.To disclosing the changing dynamic in fault features for mechanical detection,based on the PCT,this paper uses the MSST to concentrate the blurry time-frequency energy in a stepwise manner to further enhance the time frequency energy distribution in the signal.(4)Aiming at the fault diagnosis of rolling bearing and planetary gearbox,a fault diagnosis system is developed based on matlab App Designer,and the practicability of the system is verified by analyzing the fault experimental signals. |
