Discussion of positive fraction vector fitting for frequency-dependent network equivalents
The positive fraction vector fitting (PFVF) is a special method to guarantee the passivity of rational models such as frequency-dependent network equivalents. It involves constraints that enforce each fraction of the rational model to be passive, which are much stricter than the original passive requirements. PFVF lacks theoretical foundation but works well in practise. This study explains the rationality of PFVF by revealing important features of rational models that the complex-pole fractions corresponding to dominant resonance peaks can be adjusted passive through a minor change. The numerical case corroborates the theoretical analysis.