Abstract: With the rapid development of modern radar systems, anti-jamming technology has become critical for enhancing detection performance. Reconfiguration Intelligent Surface (RIS) offer significant advantages including low cost, minimal power consumption, and conformal deployment flexibility. However, when applied to anti-jamming beamforming applications, they still face fundamental challenges arising from the inherent trade-offs among discrete phase constraints and multi-dimensional optimization objectives. To address this challenge, this paper proposes an intelligent reflecting surface (IRS)-based radar anti-jamming optimization method using the linearly constrained minimum variance (LCMV) algorithm. Firstly, based on the discrete reconfigurable characteristics of RIS elements, the continuous phase distribution calculated by the LCMV algorithm is encoded into 1-bit and multi-bit discrete patterns. Furthermore, by leveraging the space-time coding capability of the IRS, the discrete coding are optimized using the minimum distance (MD) criterion, improving beam steering accuracy while maintaining low control complexity. Simulation results demonstrate that the RIS-based LCMV optimization scheme achieves high-gain beamforming in the desired direction while suppressing sidelobe levels below -25 dB and generating nulls with depths exceeding -45 dB in interference directions. The proposed method effectively enhances the anti-jamming performance of IRS-integrated radar systems.