Neural Networks in Manufacturing

Abstract

Attention focuses on minimizing the time to train a neural network so that it recognizes a specified set of a system's input parameters. In training the neural network, the error function must be minimized. This is important in expert assessment of solutions generated by a smart system for the design of manufacturing processes. In such a system, solutions are generated by the combined operation of numerous modules on the basis of logical rules. The system to be designed will generally be complex and may contain subsystems of different types that function according rules described by fuzzy logic and systems of precedents [1].