A high-voltage circuit breaker mechanical fault diagnosis method based on multidimensional zoom statistical analysis includes the following steps: S1, through fault simulation test, using an IEPE acceleration sensor to obtain vibration signal waveforms of a high-voltage circuit breaker of a certain model under different machine conditions, and forming a vibration signal database of the high-voltage circuit breaker; S2, carrying out band-pass filtering on collected vibration signals according to the frequency characteristics of the IEPE acceleration sensor; S3, doing statistics of the amplitudedistribution of the vibration signals; S4, reducing a distance matrix A from a dimension p to a dimension p' through a multidimensional zoom method; and S5, using a support vector machine to comparethe p'-dimension coordinates of a real-time measured vibration signal of the high-voltage circuit breaker with the coordinates in the basic waveform database, in order to diagnose the mechanical faultstate of the high-voltage circuit breaker. The calculation is simple. The influence of local dispersion of signals can be avoided. The error rate of fault diagnosis can be reduced.