Abstract
Offset lithographic printing is the most widely used commercial printing process. The goal of the process is to attain the desired level of print quality and to maintain this level consistently for all subsequent prints. The role of the press operator is crucial in maintaining the print consistency, as this requires an understanding of the effects of the on-line adjustments made to the process on the outcoming prints. This research suggests the use of connectionist networks to dynamically capture the relationships between the process variables, and therefore, provide the knowledge base of an expert system with updated knowledge that is difficult to articulate and may vary throughout the lifetime of the machine. The properties of fault tolerance and generalization from noisy and/or incomplete data inherent to connectionist networks makes them practical for direct learning from actual process data.
Original language | English (US) |
---|---|
Pages (from-to) | 295-306 |
Number of pages | 12 |
Journal | Computers in Industry |
Volume | 21 |
Issue number | 3 |
DOIs | |
State | Published - Apr 1993 |
Keywords
- Expert systems
- Lithographic printing
- Neural networks
ASJC Scopus subject areas
- General Computer Science
- General Engineering