To address the challenges of labor-intensive guidewire manipulation and limited automation in current robotic systems for endovascular surgeries,this paper proposes an autonomous guidewire control algorithm based on diffusion policy.This approach aims to enhance the success rate and stability of surgical robots in performing such underactuated control tasks.First,we model the guidewire control task as a Markov Decision Process (MDP),and employ imitation learning to train the robotic system using expert demonstration data,enabling efficient operation across various vascular environments.The expert control strategy is subsequently modeled as a conditional probability distribution using a denoising diffusion probabilistic model,which generates control actions based on observed states.Simulation tests for interventions in both left subclavian artery and brachiocephalic trunk were conducted in an aortic arch environment,with a comparative analysis against existing imitation learning methods.Experimental results demonstrate that the proposed method achieves significant performance improvements in both tasks,exhibits low sensitivity to hyperparameter settings,and maintains superior training stability.