Aiming at the problem that the current eggshell crack detection method has poor precision and completeness, a large-scale eggshell crack detection method based on impulse response signal is proposed. The eggshell shock response signal collecting device is composed of exciter, egg, base and acceleration sensing equipment's signal acquisition terminal. The obtained unsteady signal is decomposed into several inherent mode functions by EMO, and the main components of the signal are transformed by Hilbert Huang to obtain the energy distribution atlas of time-frequency. It is added to principal component analysis method to calculate the characteristic contribution rate of signal. The singularity index of each layer of eggshell signal is obtained, and the standard value of singularity index of corresponding layer is given. The singularity index of each layer is compared with the standard value, if the singularity index is greater than the standard value, it is a cracked eggshell; otherwise, it is a good eggshell. By layer comparison, a large-scale eggshell crack detection is realised. The experimental results show that the precision rate and recall rate of this method are both high and reasonable. [ABSTRACT FROM AUTHOR]