A Research On Roating Machinery Fault Diagnosis Based On Variable Speed Condition

Bearings and gearboxes are two most critical elements in rotating machinery which health condition can affect the overall performance and reliability of the machine.An early detection of incipient defects can prevent a component defect from deteriorating further to cause a complete machine breakdown.Vibration based condition monitoring techniques are often employed to monitor the health condition of a machine during operation due to it often contains a lot of state information.Vibration signals acquired from practical industrial sources often demonstrate non-stationary characteristics either due to the change of load or speed during a machine operation,or due to the impulses generated by mechanical defects.Time-frequency analysis(TFA)techniques are effective tools to extract the defect related time-frequency features from non-stationary signals for machine fault diagnostic.This paper analyzes the limitations of existing TFA and further improves the TFA method based on the specific fault signal of rotating machinery.The main contributions are as follows:(1)Aiming at the problem that second-order multi-synchrosqueezing transform(SMSST)is sensitive to noise,this paper constructs a new time-frequency operator(TFO)to compare and screen each time-frequency point on the SMSST time-frequency plane.And finally the energy most relevant to fault characteristics is retained and the irrelevant energy is removed.The improved second-order multi-synchrosqueezing transform(ISMSST)was verified by a set of multi-component simulation signals and variable speed fault signals of rolling bearing and planetary gearboxes in two sets of rotating machinery.The results show that the proposed method can extract fault features accurately and reduce the interference of noise to the algorithm effectively.(2)Aiming at the problem that synchrosqueezing transform(SST)has limited capacity in the handling of strong time-varying signals or signals with low signal-to-noise ratio(SNR).A parameterized iterative SST is proposed in this study by utilizing a parameterized chirplet transform to capture the fast-varying instantaneous frequencies(IF)of a strong time varying signal in conjunction with an iterative SST to produce an energy concentrated time frequency result for an accurate machine fault diagnosis.The effectiveness of the current technique is examined using two simulated non-stationary signals and two sets of machine defect data acquired under varying speed condition.The results show that the current technique can produce a highly energy concentrated time frequency result for an accurate mechanical fault diagnosis under either strong speed variation or/and noise contamination conditions.(3)In view of the limited ability of existing TFA to analyze impulse signals,this paper uses wavelet threshold denoising method to initially process the original signal to highlight its fault characteristics,and then time-reassigned local maxima synchrosqueezing transform was proposed to analysis in the direction of time axis by short-time Fourier Transform(STFT).The transient time-frequency feature extraction operator(TFM)is used to extract the most relevant features from TLMSST results.The effectiveness of the proposed technique is verified by a set of simulated impulse signals and two sets of fault data of rotating machinery with obvious impact characteristics.The results show that the proposed method can effectively highlight the impact characteristics and realize effective diagnosis.