Failure diagnosis of linear arrays based on deep residual shrinkage network

A novel array antenna diagnosis method based on a deep residual shrinkage network (DRSN) is introduced in the case of linear arrangements. The failure of array elements often leads to significant changes in the far-field pattern of an array antenna. Therefore, the failure diagnosis of array antennas can be carried out according to the far-field pattern data. In this paper, the proposed method based on DRSN can automatically learn useful features from the sampled far-field patterns of the linear array and find the mapping relationship between different far-field patterns and failure scenarios, to find the locations and number of faulty array elements. Experimental results show that the proposed method based on DRSN has higher diagnostic accuracy in the noise environment at low signal-to-noise ratios by comparing with the traditional machine learning algorithm-support vector machine.