A knowledge-based pattern recognition approach for real-time diagnosis and control of fermentation processes as variable structure plants

A fundamental property of microbial systems, which are the central part of fermentation plants, is their structural variability. From the viewpoint of process control, this imposes serious difficulties because the problem of controlling variable structure plants is analytically unsolved. However, the amalgamation of some nonstandard methodologies offers possibilities for handling the variable structure phenomenon. Of special interest is the knowledge-based pattern recognition approach, which provides a natural basis for classification of the structural state of the microbial systems, and is well-suited to their uncertain characteristics. The development of such knowledge-based techniques and their application to the control of an amino acid production process are discussed.>