As we all know, conformal arrays can provide smaller radar cross-section (RCS) and greater angle coverage than conventional linear arrays and planar arrays. In this paper, we develop a robust adaptive beamforming (RAB) approach using the cylindrical uniform conformal array. Firstly, we reconstruct the interference-plus-noise covariance matrix (INCM) by estimating the steering vectors (SVs) and powers of all interferences. Specifically, we estimate the interference SVs by exploiting the MUSIC spectrum and corrected by solving a minimization problem based on the orthogonality between the interferences SVs and noise subspace, and the interferences powers are further estimated by utilizing the non-correlation between different signal SVs. With the reconstructed INCM, the SV of the signal-of-interest is estimated by solving a quadratically constrained quadratic programming problem. Numerical results demonstrate that the proposed approach is obviously superior to the existing RAB techniques.