Test conditions as excitation intensity, excitation duration and engagement force between horn and test object are important factors affecting defect thermal signal in sonic infrared imaging. Experimental statistics and simulated results show that the increase of the excitation intensity and the excitation duration will enhance crack heating. However, the increase of the engagement force will enhance firstly and then weaken crack heating. Moreover, the effects of the test conditions on crack heating are interactive, and inappropriate test conditions will induce a test-stopping alarm. Based on the effect analysis of testing conditions on crack thermal signals and alarm incidents, a multivariate nonlinear regression model and a logistics regression model were adopted to estimate probability of detection and alarm during test, which will finally determine the value range of test conditions. The optimization methods proposed in this paper will provide theoretical direction for test scheme in sonic infrared imaging technology.