AUTOMATED REASONING ABOUT MACHINES

Numerical simulation is often used in predicting machine behavior, a b asic capability for many tasks such as design and fault diagnosis. How ever, using simulators requires considerable human effort both to crea te behavioral models and to analyze and understand simulation results. I describe algorithms which automate the kinematic and dynamical anal ysis needed to create behavioral models and which automate the intelli gent control of computational simulations needed to understand a machi ne's behavior over both short and long time scales. The input is a des cription of a machine's geometry and material properties, and the outp ut is a behavioral model for the machine and a concise qualitative/qua ntitative prediction of the machine's long-term behavior. My algorithm s have been implemented in a working program which can predict a machi ne's behavior over both short and long time periods. At present this w ork is limited to mechanical devices, particularly clockwork mechanism s.