Computational Intelligence in Design and Manufacturing

Citation

Andrew, A.M. (2001), "Computational Intelligence in Design and Manufacturing", Kybernetes, Vol. 30 No. 9/10, pp. 1333-1341. https://doi.org/10.1108/k.2001.30.9_10.1333.1

The term “computational intelligence” is used to denote what is more usually termed AI, or a subset of it. Presumably, and justifiably, the author has a semantic objection to the more usual term. Most of the chapters of the book deal with aspects of design and manufacturing that are amenable to systematic treatment, and another three deal with topics of an essentially AI nature.

In his Preface and first chapter the author refers to the need for modern manufacturing industry, in its competitive environment, to be rapidly adaptable and, as he puts it, agile. He argues that this requires overall computer control, and he particularly stresses integration of all aspects. He sees product design integrated with the planning of production facilities, and with other aspects such as:

• •

  liaisons with suppliers;

• •

  distribution;

• •

  marketing; and

• •

  customer relations.

An enterprise in not sufficiently agile if, for example, it finalises the design of a product before starting to plan how to make it. Many discussions by Stafford Beer, especially referring to his work in Chile, make the same point using other terminology.

In the integrated computerised management system that is visualised, the necessary data processing and decision‐making are likely to be intractable by algorithmic means and this is where the heuristics and other features of an AI approach are needed.

There is some correspondence to the representation by Beer (2000) of an enterprise as a viable system with (Beer, 1962) the human nervous system as a paradigm. This is without special reference to computerisation, but there is a similar emphasis on integration, referred to as holistic management. It is also interesting that heuristics were applied to an aspect of production planning in a very early paper (Tonge, 1963) in the AI literature.

However, despite the enthusiastic advocacy of integrated operation based on AI techniques, the subsequent treatment in the book is disappointing in this respect. Of the three chapters devoted to essentially AI topics, one is on “Knowledge‐based systems, one on neural networks” and one on “Data Mining”. In each of these the respective topic is developed in the management context, fairly comprehensively on the whole though at least one weak spot is the treatment of “resolution” as a strategy for an inference engine (p. 38). However, the treatments do not support the claim of an integrated approach based on AI. The chapter on “Neural Nets”, for example, ends with description of an application to fault diagnosis, which is undoubtedly valid and useful but unrelated to overall system integration.

The book is written in textbook style, with exercises for the reader at the end of each chapter (without provision of answers). It is based on material taught by the author and on his experience in consultancy. The criticism levelled here, that he has not fully achieved his vision of full integration through machine intelligence, is tempered by the reflection that his students are encouraged and equipped to work towards it.

Apart from the issue of full integration, the book has a wealth of material on systematic approaches to production planning, including:

• •

  choice of equipment;

• •

  factory floor and warehouse layout;

• •

  various aspects of scheduling including the assembly‐line balancing problem; and

• •

  supplier evaluation.

One chapter has the enigmatic title of “Kanban systems”, where “Kanban” is a Japanese word meaning “visual record” and is a way of supporting the just‐in‐time production concept.

Much attention is given to “agility”, or the ability of a company to supply a variety of products quickly and at low cost, an aspect that is readily illustrated by reference to car production and that requires choices between early and late differentiation in production. Here, aspects of the emphasis on integration can be seen in a practical context since the required flexibility can be allowed for in design of the product as well as in the arrangements for its manufacture and considerations of marketing and customer relations also enter.

Allied to the matter of integration of design and manufacture, there is an interesting attempt in the third chapter to treat aspects of product design by a formal representation similar to that used in the methods of production planning. This is at an early stage and seems to be an area pioneered by the author of the book. It seems rather likely that there could useful parallels in the descriptions of organisms in proposals for simulated evolution such as those in Varela and Bourgine (1992) and other publications on artificial life.

There is certainly a great deal of useful material here, both on account of its immediate usefulness and for its pointers to future developments.