A novel measurement-based procedure for load dynamic equivalent identification

The distribution network modeling is of great importance on power system analysis, considering both power quality and network stability. Computational effort reduction is one of the main targets for the network equivalent representation, even if results reliability remains a key point to be verified. In this paper an improved model is proposed, based on the most commonly used composite load models. Its main property is the generalization ability: the equivalent model can be applied to networks in various configurations, e.g. with different power absorption or different share of load types. A measurement-based procedure for load modeling is presented and discussed with simulations. Thanks to the combination of two optimization algorithms, namely the genetic algorithm and the grey-box approach, the procedure reaches high accuracy levels with a huge variety of load representations.