A Real-Time Knowledge-Based System for Fermentation Control

The ability to control a highly non-linear and time variant fermentation process is of considerable importance to the biotechnological industries which are continually striving to obtain higher yields and improved uniformity of production. Many industrial fermentations are operated in a fed-batch mode using a pre-defined (open loop) substrate feeding regime based on years of empirical development. Experienced operators continually monitor the process and hold microbial growth to an "ideal" trajectory using heuristics and process knowledge to make adjustments to the feeding strategy in order to compensate for variations between batches. These modifications can be quite significant since even a small increase in product yield can considerably affect process profitability by making more efficient use of carbon substrate. This paper describes the application of a real-time knowledge based system (RTKBS) to provide supervisory control of fed-batch fermentation processes. The system also performs sensor validation, fault detection/ fault diagnosis and incorporates relevant expertise and experience drawn from both the process engineering domain and the control engineering domain. The system has been developed using the G2 Real-Time Expert System (Gensym Corp.) and runs on a SUN 3/60 workstation which is linked to pilot scale fermenters via an Ethernet connection to a IBM PS/2. The PS/2 carries out local monitoring and control of the fermentation process through an RS232 serial connection to a number of industrial signal processors which are in turn directly linked to the process.