ANN based Process Control in Manufacturing

Artificial Neural Network (ANN) based expert system in manufacturing control is presented as a new approach used for discrete system. Process planning control and process scheduling control are discussed in detail by multilayer ANN. The weights series of Wii "i" the layers number, "j" the neurons number, are got initially by learning samples through mapping function and then revised further by competitive learning. The mechanism of competitive learning includes two steps: response and competition, which is only worked among these excited neurons by means of learning rules to revise the weights. The learning rules involve the weights calculation by iterating and the convergence by balance criterion. The expected values in ANN learning system in process planning are obtained by sample learning with mark-giving experiment. But in scheduling control the quantitated items in long term scheduling strategy are used for expected values.