A parallel processing scheme in power systems computation

摘要

This paper focuses theoretically to develop a new parallel computation algorithm by exploiting the natural structure of power systems. The proposed approach not only improves the efficiency of computation but also ensures the accurate integration. In other words, we consider a power system as one big group of DAEs (Differential Algebraic Equations) and many small groups of DAEs. The big group of DAEs represents the commonly connected network, while each small group of DAEs may express a generator, a control device, a small relatively independent power system, or a dynamic load, e.g. motor, which is connected to the network. Then, whole system is naturally divided into a main system corresponding to the big group of DAEs and may subsystems corresponding to each small group of DAEs respectively. Each subsystem is processed parallel while the coordination among groups is implemented only by the variables of the main system based on the implicit function theory. Since the decomposition exploits the natural structure d power systems, our parallel scheme can fully utilize the existing software resources, which in turn significantly reduces coding, program maintenance and future expansion works.