EXPERT SYSTEMS AND ON-LINE CONTROL

Marginal improvements in the control of composites manufacturing processes, although useful in the short term, will not provide the needed levels of quality, reliability, or economy of production. Lately there has been an increasing emphasis on building quality control into the manufacturing operation (National Materials Advisory Board 1989; Karbhari et al. 1992). The aim of intelligent manufacturing is to use immediate feedback to achieve direct control of product characteristics. The feedback is both to previous steps in the process to adjust control variables, and to subsequent process stages for anticipated changes. Applicable control theories and their relative rankings, as well as their brief descriptions, are given in Table 6.10.

The most advanced work in the area of expert systems and on-line control has been in the area of autoclave cure optimization (some examples of investigators are Abrams, Saliba, Beris). Cure simulation analysis has been coupled to incremental plate theories in order to study the relationships between complex gradients in temperature and degree of cure, so as to enable the use of an expert system linking process modeling and heuristics for on-line control of the cure cycle. Further linking of micromechanics models has enabled correlation of residual stresses, deformation, and fiber undulation to cure state and performance.

In its final state, the optimal system as hypothesized by a team of researchers at the University of Delaware Center for Composite Materials is described schematically in Figure 6.17, wherein process models are used to construct heuristics and rule-based schemas that can be used in the real-time expert system to control the process based on input from sensors.

This is done keeping in mind that the use of process models is computer-intensive in time requirements, and hence it may not be possible to use them on-line to control the process in real time. Future applications of such procedures include the use of expert systems in autoclave processing (Trivisano et al. 1992, 1104), injection molding of composites, pultrusion, and even RTM (Kranbuehl et al. 1992, 907). The closest analog of this system within the Japanese environment is represented by the schematic in Figure 6.16. Although individual aspects of the modules in Figure 6.17 may be at a more advanced stage in Japan, it was not obvious that an integrated structure for intelligent manufacturing as represented above actually existed or was even in use at a primitive level.


Figure 6.16. A Scheme for Overall Quality and Reliability Assessment

Table 6.10
Ranking of Control Mechanisms


Figure 6.17. Use of AI and Expert Systems for Intelligent Manufacturing


Published: April 1994; WTEC Hyper-Librarian