13.7 Results

Consider first a situation where a buffer is added to the pH system randomly. The adaptive pH controller alters the membership functions it uses to enact its production rules (which do not change) although the process dynamics are altered when the buffer is added. This approach is similar to the subconscious actions of a human controller; a human changes his or her definition of the linguistic terms being used in conjunction with their informal rule-of-thumb approach. Figure 13.8 compares the performance of the adaptive GA-FC with a non-adaptive FC. The adaptive controller is able to achieve the objective much more efficiently than the non-adaptive FC because the adaptive controller is flexible enough to accommodate the changing process dynamics.

Figure 13.8  The adaptive controller is able to adjust to changes in the environment and thus out-performs the non-adaptive control system.

Next, consider a situation where the setpoint is changed. The adaptive pH controller must alter its membership functions in response to this change. Realize that declaring a new setpoint is actually changing the objective of the FC. Changing the objective of the controller often requires a modification of the FC rule set. However, the technique of using a GA to alter a set of membership functions is powerful enough to allow the FC to maintain a suitable level of control over the pH system by altering only the meaning of the fuzzy linguistic variables. Figure 13.9 compares the performance of an adaptive GA-FC with a non-adaptive FC. As in the previous example, the adaptive pH controller out-performs the non-adaptive FC.

Figure 13.9  The adaptive controller is able to overcome prescribed changes in the setpoint of the pH system.

Finally, consider a very disruptive change to the pH system, a case where the concentrations of the acid and base the FC is using to manipulate the pH are altered. This is perhaps the most severe change in process dynamics that could be implemented. The response of the system is now completely different: additions of acid or base induce changes in the pH of the system that are far different from the changes in pH that the very same additions of acid or base induced before their concentrations were changed. Figure 13.10 compares the performance of the adaptive pH controller with the performance of a non-adaptive FC. The adaptive GA-FC is able to attain a high degree of control over the pH system despite the dramatic changes in the environment.

Figure 13.10  The discrepancies between the adaptive and non-adaptive control system is emphasized when the pH system is buffered.

13.8 Summary and Conclusions

We have described an AI-based strategy for adaptive process control. This strategy uses GAs to fashion the three components necessary for a robust, comprehensive adaptive process control system: (1) a control element to manipulate the problem environment, (2) an analysis element to recognize changes in the problem environment, and (3) a learning element to adjust to changes in the problem environment. In this chapter, the strategy has been applied to the development of an adaptive controller for a laboratory acid-base pH system in which the process dynamics change in several different ways. Initially, the overall makeup of an adaptive control system was described. Next, the pH problem environment was introduced. Finally, the basic structure of each of the three individual components was developed, and results were provided demonstrating the merit of using GAs to drive the three components.

The results presented in this chapter demonstrate the power of adaptive control systems based on GAs and FCs. These adaptive control systems are able to recognize when the physical system has changed, to quantify the changes in the physical system, and to maintain a high degree of control over the physical system despite drastic changes in the system characteristics. Based on the results presented, it is concluded that adaptive GA-FCs allow industrial pH systems to be controlled via on-line changes to the membership functions used in the rule base associated with the control system.

Adaptive control systems are becoming vital to the efficient operation of today’s industrial plants. If the efficiency of such control systems is to increase, researchers must focus on the synergism of techniques from various fields of study. In this light, the field of AI contains a vast number of untapped resources. Specifically, GAs and FCs demonstrate characteristics that allow for the production of control systems that mimic the approach adopted by humans to the task of process control. And, in the final analysis, humans actually perform the task of adaptive control quite well, as does the adaptive control system presented in this chapter.

In the next two chapters we will apply the software architecture that was applied to a pH system to very different chemical processing systems. In Chapter 15 we will consider the adaptive control of hexamine production in which an exothermic chemical reaction must be effectively manipulated. In Chapter 16 we will apply the adaptive control strategy to a column flotation unit used to separate minerals. In both instances, the adaptive control system is effective.