A systematic approach to mechanistic model building: Developing and testing complex reaction mechanisms.

摘要

Many engineering activities including control, design, experimental planning, fault diagnosis, optimization, parameter estimation, and prediction all depend on the existence of physical system models. Current research in computer aided design involves the development of faster procedures for solving models, but speeding the development of models can also reduce the elapsed time between the identification of a problem and its solution. This thesis presents a systematic approach to developing and testing models for isothermal homogeneous reaction mechanisms. As a discovery tool, this approach suggests species and reactions that may be missing from an existing mechanistic hypothesis, and it solves an integer program to create alternative hypotheses. As an experimental planner, this approach uses a hierarchy of mechanism representations to identify qualitative experiments that can distinguish among the alternatives.;The systematic approach to building mechanisms is applied to ten example problems drawn from combustion and atmospheric chemistry. Known mechanisms for HBr formation and