8. Advanced Controllers

The most effective control system can be visualized as the one with an advanced digital control system using a model predictive controller. One of the most beneficial results of computer technology revolution has been the ease in process control. With all the modern hardware and software in place, it is extremely simple to solve complex mathematical models on - line.²² State estimation techniques make the heart of these advanced controllers which are discussed in next subsection.

8.1 State estimation techniques
 
Briefly, state estimation techniques are the techniques of obtaining complete and accurate characterization of a process's current condition from an incomplete, noisy set of process measurements. These techniques are absolutely essential to modern control of fermentation systems, due to the unavailability of adequate on – line sensors for many biochemical process states. Reference should be made to references 4, and 23 - 26 for thorough knowledge of the subject.

8.1.1 Component balancing techniques

Component balancing technique is used to provide on - line estimates of the  fermentor state variables which then are used by the control algorithm. The reader is suggested to look into references 23 - 26 to pursue the subject further.

8.1.2 Filtering theory

After using component balancing techniques to generate growth rate estimates, integration of the rates can be performed to estimate other states such as biomass concentration. This approach does not work well in the practice since all real measurements are associated with error. This issue is addressed by classical state estimation theory, and particularly in this case by filtering techniques.⁴

8.2 Control and dynamic optimization

Optimization and control of a bioreactor can best be accomplished with a model – predictive control technique. In this technique, a dynamic model of the process calculates the response as a function of initial conditions, input, and/or setpoint changes. Future compensation is computed to maximize a profit function or to optimize a control function such as the sum of squares of the off - set residuals.²²