The stability evaluation method of the SIMU group is single, which seriously affects the use efficiency of the use unit. Aiming at the weak linear correlation between the two error coefficients of the SIMU, this paper applies the principle of kernel in statistical learning theory to extend the linear feature extraction algorithm to nonlinear feature extraction algorithm and improve it for evaluating to the stability of the SIMU. The numerical experiments show that the proposed Kernel Factor Analysis has a good evaluation effect.