Determining frequency dependent equivalents through evolutionary algorithms

The aim of this paper is to present an alternative methodology for obtaining frequency dependent equivalents from power networks. The methodology focus on determining rational functions that suitably approximate the frequency response of admittance present in networks equivalents. Through this methodology, part of the system under analysis can be replaced by a Frequency Dependent Network Equivalent (FDNE), which simplifies system representation and makes possible the accurate evaluation of large power networks. The methodology is based on Evolutionary Algorithms (EAs). Basically, different set of values for the rational function parameters are simultaneously assessed. The EA suitably changes the values for the function parameters, decreasing the RMS error between the system´s original frequency response and the response provided by the rational function. The results obtained through the proposed methodology are compared with the ones obtained through the Vector Fitting methodology, which is the common approach used for solving similar problems.