Surrogate-Assisted Tolerance Analysis of Microstrip Linear Arrays with Corporate Feeds

The paper addresses the problem of fast statistical analysis of linear array antennas with corporate feeds using surrogate-assisted methods. Realistic design has to take into account manufacturing tolerances that normally lead to a degradation of array performance with respect to both electrical and field parameters. Tolerances may be pertinent to antenna elements, element spacing, geometry parameters of the corporate feed, as well as material parameters of the substrate. The main performance figure affected by these parameter deviations is the sidelobe level (SLL). Accurate evaluation of SLL requires full-wave EM analysis of the array which is very expensive in computational terms. Consequently, performing statistical analysis directly at the level EM array model is impractical. Here, we discuss simple techniques that permit speeding up the process. These include various surrogate modeling methods as well as correction techniques that are used to align the fast surrogates with the respective EM models.