Considering the challenges posed by high-speed, highly maneuverable targets in interception scenarios, this paper presents a switching guidance law based on the segmentation of situational assessment. The guidance strategy divides the interception process into a long-range stage and a close-range stage. In the long-range stage, a nonlinear kinematic model is used to assess the target’s reachable set (RS) and select virtual points to maintain dynamic coverage. In the close-range stage, a linear kinematic model is used to assess the zero effort miss distance (ZEM) and the maneuverable RS of the interceptors to cater to the necessary interception conditions. Furthermore, a two-stage guidance strategy is proposed and a convolutional neural network (CNN) is utilized to optimize the stage switching timing. Simulations show that the proposed guidance strategy achieves an interception success rate of 78.1% and an average miss distance of 5.338 m, which outperforms proportional navigation (PN), coverage-based cooperative guidance (CBCG) and other methods.