Research On High Resolution Time Domain Fault Diagnosis Technology And Its Application

Rotating machinery fault diagnosis is of critical importance in ensuring the safe and efficient operation of industrial equipment.Rotating machinery,such as turbines,pumps,and motors,are key components in various industries such as power generation,oil and gas,manufacturing,and transportation.Due to wear and tear,misalignment,imbalance,and other factors,these machines can experience various types of mechanical failures.Early detection and diagnosis of these failures can prevent catastrophic breakdowns,reduce downtime,and increase the lifespan of equipment.Vibration signal analysis is an important tool in rotating machinery fault diagnosis.By analyzing vibration signals,various types of faults that may occur in rotating machinery can be identified,such as imbalance,misalignment,bearing defects,gear defects,and more.Various techniques used for analyzing vibration signals include time-domain analysis,frequency-domain analysis,and time-frequency analysis,with time-frequency analysis providing more detailed signal information including frequency and time resolution.It can capture frequency changes in the signal over different time periods,thus offering advantages in analyzing non-stationary or transient signals.In addition,time-frequency analysis can improve the accuracy and reliability of signal measurements by providing higher frequency and time resolution.Research on the application of time-frequency analysis in mechanical fault diagnosis includes:1.Considering the advantages of the spectral kurtosis method,such as filtering the signal in a short time,quickly detecting peaks and valleys in the signal,and processing non-linear signals,this paper introduces the concept of spectral kurtosis and two algorithms based on it,namely,the spectral kurtosis algorithm based on short-time Fourier transform and the spectral kurtosis algorithm based on fast binary-tree FIR filters.The advantages and disadvantages of each algorithm in terms of computational efficiency and accuracy are discussed.The spectral kurtosis method is effective in locating fault bands and designing filters based on them,but a single strong impulse disturbance can affect the accuracy of fault band location by the spectral kurtosis algorithm.To solve this problem,an iterative spectral kurtosis method is proposed in this paper,which combines iterative algorithms with statistical thresholds to resist the influence of a single strong impulse disturbance on the spectral kurtosis algorithm.Simulation analysis verifies the effectiveness of the iterative spectral kurtosis method.2.In mechanical systems,fault characteristics are usually manifested as transient pulses,which require a more concentrated time-frequency representation than traditional methods.The recently proposed time-frequency reassigned multi-synchronous squeezing transform(TMSST)effectively extracts and represents these transient pulses,but TMSST is based on short-time Fourier transform with fixed time-frequency resolution,which may result in unclear highfrequency imaging.To overcome this limitation,TMSST is combined with the S transform and the local maximum method to produce better resolution results.In addition,this paper proposes an extraction-reconstruction algorithm that combines the maximum value of the spectral envelope with signal reconstruction.To evaluate the effectiveness of this method,three signals are simulated,and Rényi entropy is used as the performance target to confirm the effectiveness of the TFA method and reconstruction technique,addressing the limitations of traditional fault diagnosis methods,and its effectiveness is verified by simulation results.3.Based on this research,vibration signal analysis software has been developed,which integrates commonly used time-frequency analysis methods,including short-time Fourier transform,continuous wavelet transform,S transform,Wigner-Ville distribution,and others.The software provides a user-friendly interface for signal analysis,visualization,and processing,and can be used in practical mechanical fault diagnosis applications.