Impedance calculation of power ground grid by using artificial neural network

The power distribution network (PDN) is an essential component in high speed circuits, and the impedance calculation of PDN becomes a hot topic. In this paper, we proposed a method to calculate the impedance of power ground grid (PGG), which is one type of the PDN, by using artificial neural network (ANN). First, the PGG is equivalent to a simple power ground plane (PGP). The equivalent PGP has the same dimension and port location but different relative permittivity epsilon g and relative permeability mu g of the substrate compared to the original PGG. Then, ANN is used to calculate the equivalent epsilon g and mu g of the equivalent PGP. Thus, the impedance of equivalent PGP can be obtained by using Green's function. The proposed method has been validated by numerical and measurement examples. The agreement and difference of the impedance results in both examples are also analyzed.