Knowledge computerbased systems

Computing technology offers a special opportunity to help with problem solving. In particular reason-based programs, commonly known as 'expert systems' , can be used in fault diagnosis and linked with corrective action. The Flour Milling and Baking Research Association at Chorleywood was the pioneer in applying such technology to the baking industry. Lately the work has continued in the Campden and Chorleywood Food Research Association.

Expert, or knowledge-based systems as they are now commonly referred to can combine facts and rules to solve problems. The ' rules' can take several forms including mathematical models, 'rules of thumb' and 'intuitive' rules. The latter may well take the form of ' if I increase the level of ingredient X then property Y in the product will change in a positive direction' . Such rules may not quantify the degree of change, only the direction.

Knowledge-based systems can contain many rules which should be capable of validation. They should not contain opinion but rather concentrate on facts. Such systems can perform a fault diagnosis within a few minutes and are capable of considering large information bases very quickly. They can consider many interactions and are often written to provide degrees of likelihood in the answers so that prioritising corrective actions is possible. Images and text can be integrated and displayed to provide pictorial display of product characteristics. In some cases it may be possible to diagnose faults with a knowledge-based system based solely on images run using touch-screen computing technology (Young, 1998a).

Unlike humans, knowledge-based systems never forget and always consider all the necessary information. However, they are not perfect because they rely on human programming and so are only as good as the information they contain. Nevertheless they can play an important role in aiding problem solving and offer a significant advantage over the classical written fault diagnosis text lists.

Knowledge-based systems have been applied for problem solving in the production of bread (Young, 1998a), cake (Petryszak et al., 1995; Young et al.,

1998) and biscuits. In addition to their application for problem solving they may be used in product development (Young, 1997), process optimisation, e.g. retarding (Young and Cauvain, 1994; Young, 1998b), and for training (Young, 1998a).

Continue reading here: The

Was this article helpful?

0 0