method for modeling and / or control of a manufacturing process, a neural network controller for a manufacturing process using

摘要

A method for modelling and/or controlling a production process using a neural network comprising the steps of a. obtaining explicit knowledge about the process to be modelled or controlled from at least one expert operator; b. translating said explicit knowledge into a description based on fuzzy logic to obtain fuzzy logic knowledge, said description being structured in a modular manner and including a large number of small blocks of knowledge rules; c. providing a neural network structured in a manner corresponding to the modular structure of the fuzzy logic description of step b, said neural network including a large number of small subnetworks; d. injecting the fuzzy logic knowledge obtained in step b generally simultaneously in the subnetworks; e. obtaining measured data relating to the process; f. entering said measured data relating to the process into the fuzzy logic description of step b and into the neural network; g. training the neural network further on measured data. IMAGE