Software-Sensor Based on Moving Horizon State Estimation with Application in Biotechnology

摘要

In biotechnology information concerning the status of the bioprocess is of vital importance to optimize and control bioprocesses. Software-sensors can be used to provide this necessary information. A design approach for software-sensors based on moving horizon state estimation is presented. This approach is an optimization-based strategy for state estimation that can be used for nonlinear systems. A finite number of past measurements is used to estimate the current state of the system. Important features which make this approach well suited for biotechnological processes are the option to include additional constraints on system states and to deal with multiple sample rates. To show its performance this software-sensor is applied to a production process of the bacterium Photorhabdus luminescens, where the production rate of carbon dioxide and glucose concentration are used as measurements to estimate the biomass a crystalline inclusion protein and two other important substrates within the cultivation medium.