卫生陶瓷凝胶注模成型干燥收缩的人工神经网络预测

Dry Shrinkage Prediction of Sanitary Ceramics in the Gelcasting Process by an Artificial Neural Network

卫生陶瓷凝胶注模成型工艺的复杂性使其很难控制材料成型后的干燥收缩率,尝试用人工神经网络技术对卫生陶瓷凝胶注模成型工艺过程进行识别和仿真,采用Levenberg-Marquardt算法建立了卫生陶瓷凝胶注模成型工艺中单体、引发剂、交联剂、催化剂含量和坯体干燥收缩的映射网络模型,从而可利用该模型来预测在一定有机成型添加剂含量下卫生陶瓷的干燥收缩。结果表明,其预测平均误差小于0.65%,而且该模型可以分析任意2种工艺因素对陶瓷干燥收缩的偶合作用。

Complexity of the gelcasting process for sanitary ceramics makes it difficult to control the dry shrinkage of the green body. We try to simulate the gelcasting process of sanitary ceramics based on improved BP artificial neural net using Levenberg-Marquardt training algorithm. The nonlinear relationship between monomer content, initiator content, catalyzer content, crosslinker content and dry shrinkage is established. Dry shrinkage performance of sanitary ceramics can be predicted by means of the trained neural net. The results show that the average prediction error of dry shrinkage is low than 0. 65% ;and the as-establlshed model is suitable for the analysis of the co-effect by any two factors on the dry shrinkage of sanitary ceramics.