Optimization and Control of Dynamic Bioprocesses

For many applications, such as commodity chemical production, ore leaching, waste treatment, and environmental remediation, bioprocesses can be less expensive, more energy-efficient, and more environmentally friendly than conventional processes. The key to effective implementation of bioprocesses is rational design, optimization, and process control. We are applying the tools of intelligent systems to develop supervisory systems for the optimization and control of continuous, dynamic, and uncharacterized bioprocesses. We have designed, built, and evaluated hierarchical hardware and software systems for the control of microbial oxidation of soluble iron in a continuous stirred tank reactor. The supervisory control module uses stochastic learning to determine what system parameters (i.e., pH, dilution rate, and temperature) should be, based on the state of the system. An expert-based flow-rate controller optimizes reactor performance for each set of system parameters. Theoretically, high reactor conversion can be obtained much faster by varying these multiple parameters simultaneously as opposed to the traditional method of varying a single parameter at any given instance.