A new method based on multi-task learning for the design of sparse conformal array is presented.Considering the different orientations of antennas and variable curvature radius of the platform

array manifold model for 3D conformal array is established.Combined with the framework of multi-task learning (MTL)

the pattern of uniform conformal array antenna scatting on a cone aperture is studied.In accordance with the identified sparse vector supporting area

the under-determined steering vector matrix equation is transformed into the over-determined conversion feature matrix equation.The synthesis performance comparisons with matrix pencil method and genetic algorithm are performed to demonstrate the superiority of the proposed algorithm over traditional algorithm in the literature.And a sparse conformal array antennas design with favorable pattern matching is obtained by MTL method.