摘要
支持向量机(SVM)作为一种机器学习算法,具有良好的非线性处理能力、理论全局最优及克服维数灾难等优点。文章采用高斯核函数(RBF)支持向量机回归模型(SVR),归一化和降维进行数据处理,交叉验证进行参数寻优,获得最佳预测模型,并对该预测模型进行测试。应用该预测模型分析工艺参数对高功率器件合格率的影响,结果表明其对合格率的预测精度较高,对功率器件质量的提升具有重要的指导意义。
As a machine learning algorithm, support vector machine（SVM） has the advantages of good nonlinear processing ability, theoretical global optimum and overcoming the curse of dimensionality. It presented a support vector machines regression model （SVR） with Gauss kernel function （RBF）. The best prediction model was obtained by normalization and dimensionality reduction for data and cross-validation for parameter optimization. The prediction model was tested and used to analyze the influence of process parameters on the qualified rate. Simulation results show that the SVR has high accuracy the quality of high power devices. in the prediction of qualified rate and great significance to improve
出处
《控制与信息技术》
2018年第1期62-64,86,共4页
CONTROL AND INFORMATION TECHNOLOGY
关键词
高功率器件
支持向量机
机器学习
数据降维
参数寻优
合格率
high power device
support vector machine （SVM）
machine learning
data dimensionality reduction
parameter optimization
qualified rate
