Abstract: Due to complexity and high cost, traditional phased array antennas are no longer suitable for current research applications. Unconventional array antennas based on subarray technology have achieved a good compromise between complexity and radiation performance, and have received widespread attention. This paper proposes an array synthesis method based on total variation compressed sensing to solve the problem of continuous subarray partitioning in a uniform linear phased array. By transforming the excitation of different subarray antenna units into the gradient domain to make them sparse, the optimization problem of subarray layout and excitation in subarray partitioning is converted into the problem of maximizing the sparsity of the excitation gradient domain. The unconstrained cost function minimization problem is constructed by augmenting the Lagrangian formula, and the deterministic alternating algorithm is used for optimization. Finally, the partitioning of a continuous subarray in a uniform linear array is achieved. Through several numerical examples of typical array subarray partitioning, the effectiveness of the proposed method is verified through numerical examples of several typical array subarray partitioning, and by comparing it with some traditional subarray partitioning methods (such as genetic algorithms and K-means clustering methods) in terms of direction diagram matching error, array performance parameters, and computational efficiency.