Human labour remains widely used in many production systems, although this is often not ideal due to such factors as inconsistencies, high costs, and physical dangers. However, replacing a specific individual is often expensive, even when the individual's role appears to be simple and relatively mundane. Tasks that appear to be very simple, such as simple visual inspection, can often be the most difficult to automate. Conversely tasks that to the layman appear very complex, such as the interpretation of mass spectography data, can be performed by a computer 'Expert System' relatively easily. In between these two domains lies a third area; the fuzzy grey terrain of the professional craftsman. It is in these first and third areas, which to date have proved intractable to conventional computer approaches to automation, that neural methods have shown particular promise. This chapter will describe two industrial problems and illustrate how neural computing is being used to help in their automation.
Neural Networks in Automation Procedures, Page 1 of 2
< Previous page Next page > /docserver/preview/fulltext/books/ce/pbce051e/PBCE051E_ch21-1.gif /docserver/preview/fulltext/books/ce/pbce051e/PBCE051E_ch21-2.gif