This paper proposed a robust beamforming method that jointly estimated the steering vector and covariance matrix to address the impact of expected signal steering vector mismatch and covariance matrix error on algorithm performance in adaptive beamforming. Firstly, the algorithm decomposed the Capon power spectrum features and reconstructd the covariance matrix of interference and noise; Then, considering the impact of interference guidance vector estimation errors on algorithm performance, an uncertain set constraint model for interference guidance vector estimation was established to obtain an optimized estimate of the interference vector; Finally, an uncertain set constraint model for the expected steering vector was established according to the criterion of maximizing the output signal power of the array. The optimal estimation value was obtained by using convex quadratic constraint quadratic optimization iteration and extreme value search method. The simulation results show that the algorithm proposed in this paper improves the beamforming performance under conditions of guidance vector mismatch and array error.