Parameterized model order reduction for efficient time and frequency domain global sensitivity analysis of PEEC circuits

This paper presents a new parameterized model order reduction technique to efficiently perform global time- and frequency-domain sensitivity analysis of electromagnetic systems over the design space of interest. The partial element equivalent circuit (PEEC) method is adopted to build the electromagnetic system model at a set of initial samples in the design space. The block Laguerre-SVD algorithm is proposed to reduce the size of the original equations of the PEEC-based equivalent circuit along with those describing the port voltage and current sensitivities. Then, a multivariate cubic spline interpolation method is used to build a parameterized compact model of port voltages and currents along with their corresponding sensitivities over the entire design space of interest. Finally, two numerical examples are presented, which confirm the accuracy and efficiency of the proposed method.