基于多元统计分析绝缘纸工频击穿强度影响因素的研究

Study on the Factors Affecting AC Breakdown Strength of Insulating Paper based on Multivariate Statistical Analysis

中文关键词:  油纸绝缘  工频击穿强度  灰色关联分析  主成分分析  最佳子集选择

Key Words:oil-paper insulation  AC breakdown strength  grey relational analysis  principal component analysis  best subset selection

中文摘要:

      为提高特高压变压器用绝缘纸的制造技术和质量水平、研究“材料-结构-性能”之间的相关性，提出一种基于灰色关联分析的绝缘纸工频击穿强度影响因素量化分析模型，并采用主成分分析和最佳子集选择方法进行多参数优化，构建绝缘纸工频击穿强度的多元线性回归模型。结果表明，绝缘纸工频击穿强度影响因素的灰色关联顺序为：纤维长度>细小纤维含量>纤维宽度>紧度>透气度>孔隙率>厚度。主成分分析表明，提取的3个主成分能够解释原来参数95.76%的信息，最佳子集选择将参数优化为3个参数；绝缘纸工频击穿强度的多元线性回归模型拟合度较高，模拟样本和验证样本预测结果的相对偏差基本在10%以内，表明回归模型具有良好的预测能力。

Abstract:

      In order to improve the manufacturing technology and quality level of insulating paper for ultra-high voltage （UHV） transformers and study the correlation of "material-structure-performance"， a quantified analysis model of the factors affecting the AC breakdown strength of insulating paper based on gray relational analysis （GRA） was proposed， and the principal component analysis （PCA） and the best subset selection method were used for multi-parameter optimization to construct a multiple linear regression model for the AC breakdown strength. The results showed that the gray correlation order affecting AC breakdown strength of insulating paper was： fiber length>fines content>fiber width>density>air permeability>porosity>thickness. Principal component analysis showed that 95.76% of the messages in the original parameters could be explained by three extracted principal components， and the best subset selection method optimized the parameters into three. The multiple linear regression model had a high degree of fitting and the relative deviation of the prediction results between the simulated sample and the verified sample was basically within 10%， indicating that the regression model was of good prediction capability.