Budget Amount *help |
¥5,500,000 (Direct Cost: ¥5,500,000)
Fiscal Year 1991: ¥5,500,000 (Direct Cost: ¥5,500,000)
|
Research Abstract |
This paper is intended to establish the control method of performing optimum water injection in ceramics mulling process. In this paper, the synthetic mullite is used as ceramics materials. In our laboratory, we have studied the characteristic analysis of ceramics mulling. The optimum mulling time and the optimum moisture content have been clarified. And the settling moisture content is possible to control by a neural network (NN) model. However, the reasonable estimated value of the settling moisture content, have not been obtained from the NN model built in the previous study. In this paper, a modified the NN model is proposed to determine the quantity of water injection. Furthermore, in order to improve the computation time in the model building, the accuracy of the estimated value and generality of the control model, the NN-driven fuzzy reasoning system is applied to the ceramics mulling process. The results are obtained as follows ; 1) NN-driven Fuzzy Reasoning system has more follow-up efficiency than the NN model. Further, it is shown that this system has high reasoning performance if the objective process is complex such as the single NN model can not express the behavior sufficiently. 2) From the control experiment, it is shown that NN-driven Fuzzy Reasoning system is effective for the control of water injection process. The present study gives useful informations for mulling control in the practical plant.
|