Statistical tools for optimal dynamic model building

Abstract A general, systematic procedure is presented to support the development and statistical verification of dynamic process models. Within this procedure, methods are presented to address several key aspects, such as structural identifiability and distinguishability testing, as well as optimal design of dynamic experiments for both model discrimination and improving parameter precision. A novel optimisation-based approach is introduced for testing of model structural identifiability and distinguishability, involving semi-infinite programming and max-min problems. The design experiments is cast an optimal control problem within a framework that enables the calculation of optimal sampling points, experiment duration, fixed and variable external control profiles, and initial conditions of a dynamic experiment subject to general constraints on inputs and outputs. Within this framework, methods are presented to provide experiment design robustness, accounting for parameter uncertainty. The procedure is demonstrated through a practical biotechnology example.