Variable structure model synthesis for switched systems

Model reduction methods exist for a large class of systems. However, to date there does not exist a well established model synthesis methodology for physically switched systems. A method that borrows ideas from model reduction and variable structure system (VSS) theory is presented for synthesizing models of switched systems. Metrics using relative error and activity analysis are given for measuring the relative fidelity of these models. These ideas are then extended to develop an approach for doing more computationally efficient simulations of vehicle models for virtual mission studies. The model fidelity metrics are used to derive reduced sub-models that predict the response of the system over specified time spans or dynamic modes. VSS theory is used to switch between these sub-models of varying fidelity. It is shown that by employing this variable fidelity approach simulation efficiency for moderately complex system models can be significantly improved.