A prototype neural network supervised control system for Bacillus thuringiensis fermentations

Abstract

This article discusses the development of a prototype neural network-based supervisory control system for Bacillus thuringiensisfermentations. The input pattern to the neural network included the type of inoculum, operation temperature, pH value, accumulated process time, optical density in fermentation medium, and change in optical density. The output from the neural network was the predicted optical density for the next sampling time. The control system has been implemented in both a computer simulation and a laboratory fermentation experiment with promising results. 0 1994 John Wiley & Sons, Inc.

Bioprocess monitoring and computer control: Key roots of the current PAT initiative

“…Facts were I/O connections and relationships of cascaded loops (e.g., lower airflow causes low DO); rules were production rules (e.g., if-then, fault, etc.). Concise way to store and access troubleshooting manual Miles 1988Schlager (1988 Expert system supervision along with aeration-agitation control for 20% reduced energy costs Streptomyces amylofaciens production of spiramycin Zhejiang University/Hangzhou 1989 Qi et al (1989) Rule based system with computer system processing linguistically formulated simple rules that simulates user observation of state of the process E. coli production of enzyme Universität Hannover 1989 Lübbert et al (1989) Combined supervisory knowledge base (i.e., G2 system), scheduling knowledge base, and online relational database 1992Jørgensen et al (1992 Fuzzy control to regulate sugar feed rate; feed control strategy altered for each of five cultivation phases Hosobuchi et al (1993b On-line estimates for biomass and protein using dynamically modified, feed-forward ANN; expert supervisory system developed for penicillin G transferred to clavulanic acid production, a similar but less well understood process Recombinant E. coli production of protein ZENECA/University of Newcastle Glassey et al (1994 Neural network that predicted optical density at subsequent sampling time based on inputted measured optical density at prior sampling times Zhang et al (1994 Hybrid model combined dynamic differential equations, ANN, and a fuzzy expert system…”

An efficient model development strategy for bioprocesses based on neural networks in macroscopic balances: Part II

“…2c, BBM 1) the initial PenG concentration ([PenG] 0 ), the temperature (T k ), and the amount of added base (B k ) were taken as inputs. For (fed-)-batch processes often the time is given as input to black box models (Albiol et al, 1995;Montague and Morris, 1994;Warnes et al, 1996;Zhang et al, 1994). Therefore, for the second BBM (Fig.…”

Neural simulation of an unsteady state continuous recombinant fermentation with imperfect mixing

“…Earlier studies (Link0 and Zhu, 1992; Montague et al, 1992;Zhang et al, 1994) suggest that a three-layer feed-forward neural network trained by the backpropagation algorithm is suitable. The first layer receives and transmits the inputs, which are the initial conditions for yj(j = 1,2, .…”