MESA: an interactive modeling and simulation environment for intelligent systems automation

Describes MESA, a domain-independent interactive tool for the development of reusable causal models and model-based-reasoning applications. Our current efforts are focused on developing automated sensor monitoring applications for NASA flight projects. MESA supports model development from a component-centered approach and provides a graphical editor for rapidly prototyping components and connections from a small set of modeling primitives that describe structure and behavior (e.g. quantity, mechanism). Causal models, as used in artificial intelligence, contain interdependent structural and behavioral descriptions of the system being modeled. MESA also provides a model-based discrete-event simulator that can predict future behavior of the system from knowledge of the current behavior of the model. MESA thereby allows models to be tested, debugged and deployed within one generic environment.<>