【作者】 李美艳； 何兴； 徐贤兴； 金伟娅；

【Author】 Li Meiyan;He Xing;Xu Xianxing;Jin Weiya;Institute of Process Equipment & Control Engineering,Zhejiang University of Technology;

【机构】 浙江工业大学化工过程机械研究所；

【摘要】 针对数据缺乏导致预测精度不高问题,提出了一种关联支持向量回归(Relational Support Vector Regression,r-SVR)方法。该方法结合灰色关联分析计算因素之间关联度、优化数据权重的特点来改进支持向量回归预测小样本的精度。为进一步改进预测效果,引入交叉验证对r-SVR核函数的重要参数进行优化,避免人工选择的随机性,使模型更加稳定。本文利用r-SVR对蒸汽发生器传热管的爆破压力进行预测。结果表明,r-SVR的平均百分比偏差为1.14%,预测精度良好。更多还原

【Abstract】 In order to solve the problem of low prediction accuracy with rare data,a method of relational support vector regression(r-SVR)is proposed.In the method,gray correlation analysis is introduced to improve the prediction accuracy of the support vector regression with small samples,which is used to calculate correlation and optimize the data weight.To get better prediction accuracy,cross validation is introduced to optimize important parameters of kernel function in r-SVR,which can avoid the randomness of artificial selection and make the model to be more stable.The method of r-SVR is used to predict the bursting pressure of heat transfer tubes.The results show that the average percentage deviation of r-SVR is 1.14% and the prediction accuracy is fine.更多还原