The vibration signal of gearbox can directly reflect its operation status, so the gearbox fault diagnosis based on vibration signal is more effective than other gearbox fault analysis methods. Before using vibration signals to diagnose gearbox faults, it is necessary to extract valuable characteristic signals from the collected vibration signals. Therefore, the time domain and frequency domain analysis of gearbox vibration signal is particularly important.
Time domain signal reflects the situation that waveform changes with time. The key to analyze time domain signal is to study the relationship between signal amplitude and time. By analyzing the time domain signal, we can usually get the mean value, maximum value, skewness index, kurtosis index, margin index and other eigenvalues
Where Xi is the amplitude of the signal at the ith sampling point; N is the total number of sampling points.
According to past experience, kurtosis index and margin index are more sensitive in the early stage of fault occurrence, but the stability decreases with the time of fault occurrence. The peak index and skewness index are stable although they are not sensitive at the initial stage. Therefore, using these eigenvalues together in fault diagnosis can improve the recognition rate of the model. To analyze the vibration signal in frequency domain, it is necessary to decompose the time domain signal into frequency domain signal with frequency as abscissa through certain transformation, so as to obtain the amplitude and phase information of frequency components. The commonly used frequency domain characteristic indexes include correlation factor and harmonic factor, and their calculation formulas are as follows:
Where FI is the frequency corresponding to the power spectrum at time I; PI is the power spectrum amplitude at time I.