744 controller performance is better as compared with PI and conventional fuzzy controllers. This fuzzy control system was applied to the optimal production of glutathione (GSH) in yeast fed-batch culture.
160 Expert System Aid Control in a Penicillin Fermentation Process Wang Shuqing, Zou Wei, pp 439-440 In recent years, microcomputers have been widely used in antibiotic fermentation processes. Environmental variables have been successfully controlled by microcomputers, but some main variables are still difficult to control, because they cannot be measured online. The process is too complex, and a suitable dynamic model is difficult to fred. So, in an industrial penicillin fermentation process, the biomass concentration is controlled by operators adjusting the glucose feedrate, on the basis of past experience. An expert system including hierarchical intelligent control has been developed to identify the penicillin fermentation situations and make decisions to control the glucose fee&ate.
161 Hybrid Modelling of Bioprocesses within a Network Transformation Framework: A Software Engineering Concept W. Wiechert, T. H6ner, C. Hausmann, M. M611ney, M. Kinder-Theiflen, pp 441-444 Various modelling paradigms have been applied to bioprocess control. To achieve robust control strategies for practical application, different methods for simulation, analysis, validation and control have to be combined. The arising "hybrid process models" usually have a high complexity and are difficult to maintain. A software engineering approach is presented that allows the representation of hybrid process knowledge in a wellstructured way, based on the notion of abstract frames, networks and information transformations. First results from work on an object-oriented computer implementation of the described concept ("BioProcess ToolKit") are summarized. Some applications (i.e. online bioprocess optimization, NMR spectral deconvolution) are discussed.
162 Improvement of Medium Composition using an Expert System M. Kishimoto, pp 445-448 An expert system which included a database of rule sets and frame sets was constructed for the improvement of cell growth in culture experiments. The inference procedures were stored as production rules in the database, and the data of media compositions, the content of chemical elements of each composition, and the other culture conditions were stored as frames. The culture experiments were carried out with the addition of the medium components which contained the amount of chemical elements inferred to be insufficient, and the growth of E. coli was about 5 times higher than that of the primary culture experiment without the addition.
163 The Debates M.N. Pons, M.N. Karim, pp 449-452 Validation was certainly the keyword of the round-table discussion aimed to define the necessary balance between the new methods (Artificial Intelligence, Expert Systems, Artificial Neural Networks, Principal Component Analysis) and the need to develop and improve direct measurement probes.
Abstracts
164 Studies on On-Line State and Parameter Estimation Through a Real-Time Process Simulator S. Feyo de Azevedo, P. Pimenta, F. Olivelra, E. Ferreira, pp 453-458 A PC-based simulator of non-linear multiple-input multiple-output dynamic systems was employed for studies concerning the on-line estimation of both 'nonmeasured' state variables and multiple specific growth rates of a virtual baker's yeast fed-batch fermentation. The identification algorithms were implemented in an 'Observer Computer' which received the 'measured' state variable generated by the process simulator in the 'Process Computer'. The state observer showed a robust behaviour under different situations. The simulation package was seen to be an efficient and cheap tool for the simulation of non-linear processes with time-varying parameters, providing a realistic environment for tests of identification, estimation and control strategies.
165 MACROBAL: A Program for Robust Data Reconciliation and Gross Error Detection C. Heilinga, B. Romein, pp 459-460 A set of linear constraint equations provides the mathematical basis for relating conversion rates in (bio)chemical processes. In the computer program MACROBAL, running on any type IBM-PC, such equations can be defined by summing up all relevant compounds and the conserved quantities in an 'elemental' composition matrix. MACROBAL uses the concept of the redundancy matrix for analyzing the system of equations. If possible, measured rates will be balanced and non-measured rates will be calculated. A Chi-square test is performed for detecting errors in the system description and/or measured rates. The serial elimination method can be used to trace the error location.
166 Bioboss - Information and Data Management Software for Fermentation Process Development M. Pokkinen, R. Oinas, E. Hamaker, A. Saari, M. Salkola, M. Denicolai, A. Holmberg, pp 461-464 Bioboss is a new information and data management software designed to assist researchers in the development of fermentation processes. Bioboss offers tools for the description and design of experiments, for the analysis, and storing of experimemal data, and for the storage and management of information. Bioboss's facilities are especially useful in large and timeconsuming process and product development projects.
167 A Program for Error Diagnosis using a Sequential Vector Comparison Test B. Romein, R.T.J.M. van der Heijden, pp 465-466 Verification of data measured in biomchnological processes is essential in research and for industrial monitoring and control. The newly developed "vector comparison test" is more powerful in detecting and locating errors than the serial elimination method, especially when it is applied sequentially. Both methods use linear constraint equations based on the conservation principles. A program (running on IBM-PCs) is under development as an implementation of the vector comparison test. Process-specific information must still be changed in the source code. The program was used successfully for the detection and location of errors in