| Summary: | This work develops sparse polynomial models for investigating the response of electromagnetic filtering structures, when the design of the latter is affected by a number of uncertain variables. The proposed approach describes an improved implementation framework for contemporary Compressed Sensing techniques, which are known for their capacity to reconstruct sparse signals with a limited number of samples. Unlike typical implementations, the necessary set of basis functions is formulated after performing an initial estimation of partial variances that, despite being computationally cheap, provides sufficient information for the impact of each variable on the output. A number of numerical tests on different filter configurations verify the reliability of the presented methodology, display its efficiency, and unveil the performance of the considered structures, when operated under uncertainty.
|
|---|
