Premium
On‐line estimation of glucose and biomass concentration in batch fermentation process using particle filter with constraint
Author(s) -
Zhao Zhonggai,
Shao Xinguang,
Huang Biao,
Liu Fei
Publication year - 2012
Publication title -
asia‐pacific journal of chemical engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.348
H-Index - 35
eISSN - 1932-2143
pISSN - 1932-2135
DOI - 10.1002/apj.1626
Subject(s) - mathematical optimization , resampling , particle filter , biomass (ecology) , constraint (computer aided design) , iterated function , computer science , filter (signal processing) , mathematics , algorithm , mathematical analysis , oceanography , geometry , computer vision , geology
In most of fermentation processes, the substrate and biomass concentrations greatly influence the yield of the targeted product. In this paper, considering the nonlinear and non‐Gaussian nature of the fermentation processes, the particle filter method is introduced to estimate the substrate and biomass states, and an algorithm integrating optimization strategy with resampling is proposed to deal with constraints on states. The proposed approach investigates the constraint directly on the estimated value rather than on particles. If the estimation violates constraints, the optimal estimation is firstly obtained by the optimization approach, and then, the resampling and optimization strategy are iteratively used to randomly draw a particle each time among the violated posterior particles and project it onto the feasible region. The resampling and optimization steps are iterated until the state estimation, derived from posterior particles, reaches better performance than that through previous optimization formulation or all the violated posterior particles have been projected onto the feasible region. Compared with the other previous optimization approaches, the proposed method results in a better balance between reducing the on‐line computation load and obtaining more effective posterior particles. © 2012 Curtin University of Technology and John Wiley & Sons, Ltd.