In order to achieve the degree detection of the crack damage degree of a spur gear system, a combination analysis method of principal component analysis (PCA), and the gray theory analysis method are adopted in the paper. First, a dynamic model of a spur gear system is established, so that it is possible to obtain the simulation signals under different degrees of crack. The next step is to capture the crack sensitive characteristic parameters of these signals by modern signal analysis methods including time domain, frequency domain, etc. Then the characteristic parameters is optimized by the combined analysis method of the PCA and the gray theory proposed in this paper to reduce its dimensionality. Finally, the degree detection is achieved depending on the parameter of correlation by the method. Through the analysis of the simulation signals from the model, it is proved that the accuracy increased by 16% than the general method of fixed threshold. The method is effective.