Integrated time sampling design and measurement set selection for biochemical systems

Abstract

The optimal experimental design (OED) for observation strategy is investigated in this paper to collect the most informative experimental data for parameter estimation. The aim is to determine the best sampling time points and also select the most valuable measurement state variables through OED. The two design objectives are integrated together as a single-objective optimisation problem in which the variables and their sampling times are weighted in an expanded time sampling framework. Three optimisation methods, i.e., the Powell's method, the sequential selection method, and the sequential quadratic programming method, are employed to solve the optimisation problem. Their computation efficiencies are compared using a biodiesel case study system simulation. Simulation results demonstrate the effectiveness of the proposed method in reducing parameter estimation uncertainties as well as reducing parameter correlations. It can also be observed that the integrated OED doesn't cost extra computation efforts when variable selection is considered together with the time sampling task.