Model-Based System Verification: A Formal Framework for Relating Analyses, Requirements, and Tests

Abstract

As modern systems become increasingly complex, there is a growing need to support the systems engineering process with a variety of formal models, such that the team of experts involved in the process can express and share knowledge precisely, succinctly and unambiguously. However, creating such formal models can be expensive and time- consuming, making a broad exploration of different system architectures cost-prohibitive. In this paper, we investigate an approach for reducing such costs and hence enabling broader architecture space exploration- through the use of model transformations. Specifically, a method is presented for verifying design alternatives with respect to design requirements through automated generation of analyses from formal models of the systems engineering problem. Formal models are used to express the structure of design alternatives, the system requirements, and experiments to verify the requirements as well as the relationships between the models. These formal models are all represented in a common modeling language, the Object Management Group’s Systems Modeling Language (OMG SysML^TM). To then translate descriptive models of system alternatives into a set of corresponding analysis models, a model transformation approach is used to combine knowledge from the experiment models with knowledge from reusable model libraries. This set of analysis models is subsequently transformed into executable simulations, which are used to guide the search for suitable system alternatives. To facilitate performing this search using commercially available optimization tools, the analyses are represented using the General Algebraic Modeling System (GAMS). The approach is demonstrated on the design of a hydraulic subsystem for a log splitter.