Knowledge-Based Expert System Control

Evolving from efforts [22] to use the best features of trial-and-error, process model, expert system, and expert model approaches, QPA [23-25] combines KBES traits with online dielectric, pressure, and temperature data to implement autoclave curing control. QPA combines extensive sensor data with KBES rules to determine control actions. These rules determine curing progress based upon process feedback, and implement control action. QPA adjusts production parameters on-line as such—within the limits of its heuristics—QPA can accommodate batch-to-batch prepreg variations. 

Because it considers no analytical model, QPA does not make explicit use of heat transfer dynamics. Nonetheless, QPA does reduce the autoclave curing cycle durations in several experimental autoclave curing runs. 

The objective of the Springer KBES is twofold To ensure a high-quality part in the shortest autoclave curing cycle duration. This KBES is similar to QPA in that sensor outputs are combined with heuristics not with an analytical curing model. The rules for compaction dictate that dielectrically measured resin viscosity be held Constant during the First temperature hold in the autoclave curing run. The autoclave temperature is made to oscillate about the target hold temperature in an attempt to attain constant viscosity. Full pressure is applied from the cure cycle start. 

Perry and Lee [28,29] offer an enhancement of QPA, based upon use of dual heat flux sensors and additional thermocouples in autoclave curing. This enhancement entails determining heat transfer properties during the cure, then using these properties in conjunction with PID regulatory control of autoclave temperature. Using the additional sensors, Perry and Lee employ an on-line Damkohler number in lieu of the second time-derivative of temperature to avoid exothermic thermal runaway within the prepreg stack thermoset resin. The Damkohler number is defined as  

Da = L2(Miryl,)rRFV = rate of heat generation k(Tt — T0) rate of heat conduction 

---

Contents indude mathematical modeling, process control, statistics, hierarchical control systems, artifidal Intelligence techniques, knowledge-based expert systems, and modeling of large scale systems using the general purpose simulator. 

Rules seemingly have the same format as IF.. THEN.. statements in any other conventional computer language. The major difference is that the latter statements are constructed to be executed sequentially and always in the same order, whereas expert system rules are meant as little independent pieces of knowledge. It is the task of the inference engine to recognize the applicable rules. This may be different in different situations. There is no preset order in which the rules must be executed. Clarity of the rule base is an essential characteristic because it must be possible to control and follow the system on reasoning errors. The structuring of rules into rule sets favours comprehensibility and allows a more efficient consultation of the system. Because of the natural resemblance to real expertise, rule-based expert systems are the most popular. Many of the earlier developed systems are pure rule-based systems. 

Fuzzy logic systems grew out of a desire to quantify rule-based expert systems. Fuzzy set theory had provided us with an effective framework for dealing with fuzzy information and for translating control strategies based on an expert knowledge into an automatic control strategy. 

J. Klaessens, J. Sanders, B. Vandeginste and G. Kateman, LABGEN, expert system for knowledge based modelling of analytical laboratories. Part 2, Application to a laboratory for quality control. Anal. Chim. Acta, 222 (1989) 19-34. 

The facts in a knowledge base include descriptions of objects, their attributes and corresponding data values, in the area to which the expert system is to be applied. In a process control application, for example, the factual knowledge might include a description of a physical plant or a portion thereof, characteristics of individual components, values from sensor data, composition of feedstocks and so forth. 

Such a system would include a program similar to that of Storer and Comish-Bowden to do equilibrium calculations. A communication-control subprogram would be Hiked to an expert model by using the EXPERT knowledge-base shall (or system-builder) which is advantageous here because it can interact with procedures such as those written in FORTRAN for numerical computation. Additional programs and a small data base, which EXPERT can handle, would keep track of which chemical was what array element, and other requirements mentioned above. 

Other researchers have improved on both knowledge base and sensors, often using different programming tools, but with much the same results. Most of these systems have not been transitioned to commercial application because they were developed by researchers who had no interest in the development of user interfaces and the continued support that is necessary for a commercial software package. As a consequence of the success of these experimental systems, however, the developers of commercial supervisory control packages are moving to incorporate support for expert system logic into their software [24,25],... 

A second approach to the problem of difficult to obtain measurements is knowledge-based or model-based control. Knowledge-based systems attempt to use various types of knowledge of the biological process (rules etc.) to supplement traditional mathematical control approaches.16 Expert systems are one type of knowledge-based control. Model-based control systems use a model of the process as part of the control algorithm their reliability depends on the accuracy of the model. 

---