Abstract: The lack of apriori information on the restricted isometry property (RIP) of the observation matrix in nearfield scenario can not guarantee an accurate diagnosis with a high probability when using the l1 norm minimization. In order to overcome this deficiency, a fast diagnosis method with nonconvex compressed sensing and planar nearfield measurements for array diagnosis utilizing iteratively reweighted least squares algorithm is explored in this paper. Taking into account that the number of failed elements is far less than that of the total array elements, the nearfield data of a healthy array and a failed array are acquired by the probe using the random undersampling strategy. Then the differential array is constructed. Finally, the sparse incentive is recovered through the proposed method and the goal of array diagnosis is achieved. Numerical simulation results indicate that the proposed approach not only avoids the adverse influence on the performance of diagnosis due to the lack of RIP information, but also overcomes the problem of local minima of the nonconvex norm, therefore reduces the diagnosis time and improves the probability of the success rate of diagnosis effectively.

Key words: array diagnosis, nonconvex compressed sensing, sparse recovery, nearfield measurements, lp(0