微波加热制备部分稳定氧化锆及其稳定率预测研究  

Preparation of Partially Stabilized Zirconia by Microwave Heating and Study on Prediction of the Stability

作　　者：刘利军[1,2] 李东波[1,2] 彭金辉[1,2] 郭胜惠[1,2] 张利波[1,2] 陈菓[1,2] 

机构地区：[1]昆明理工大学冶金与能源工程学院,昆明650093 [2]昆明理工大学非常规冶金省部共建教育部重点实验室,昆明650093

出　　处：《材料导报》2011年第20期131-134,共4页Materials Reports

基　　金：国家重点基础研究发展计划(973计划)(2007CB613606)

摘　　要：针对微波加热制备部分稳定氧化锆过程的非线性、多变量、时变等复杂特点,运用机器学习方法中的LM-BP神经网络和支持向量机(SVM),以保温温度、升温速率、保温时间、降温速率和淬火温度作为输入量,稳定率为输出值,建立了微波加热制备部分稳定氧化锆的稳定率预测模型。分别利用两种预测模型进行稳定率的预测,通过与稳定率的测量值对比分析表明,两者均具有良好的预测能力,但SVM模型具有较高的预测精度。Due to complexities of the preparation of partially stabilized zirconia（PSZ） by microwave heating, i.e. non-linear, multi-variable, time-change and so on, two machine learning methods, i.e. Levenberg Marquardt- Back Propagation（LM-BP） neural network and support vector machine（SVM）, were used to build stability prediction models. The inputs were the holding temperature, rising rate of temperature, holding time, decreasing rate of temperature and hardening temperature. And the stability of PSZ was the only output. The stability value of PSZ was predicted by using two methods of LM-BP neural network and SVM. Comparing predicted value with measured value, the results indicate that both of the predicted results agreed with the measured value within reasonable experimental error and the predictive ability of SVM is better than that of LM-BP neural network.

关 键 词：部分稳定氧化锆 微波 稳定性 LM-BP神经网络 支持向量机(SVM) 预测 

分 类 号：TG146.4[一般工业技术—材料科学与工程] TP18[金属学及工艺—金属材料]