基于PLSR的针叶材纸浆得率影响因素相关性分析  

Correlative analysis on the influential factors to pulp yield of softwood based on PLSR method

作　　者：莫卫林 杨浩[1] 熊智新[1] MO Weilin;YANG Hao;XIONG Zhixin(Jiangsu Provincial Key Lab of Pulp and Paper Science and Technology,Nanjing Forestry University,Nanjing 210037,China)

机构地区：[1]南京林业大学江苏省制浆造纸科学与技术重点实验室,南京210037

出　　处：《中华纸业》2018年第10期11-14,共4页China Pulp & Paper Industry

基　　金：国家林业局948项目"农林剩余物制机械浆节能和减量技术引进"(2014-4-3)

摘　　要：纸浆材的纸浆得率大小与其化学组成有较密切的相关性,通常可以用多元线性回归(MLR)分析方法研究这种相关性的密切程度。但当各性能指标间存在严重的多重共线性时,MLR方法难以克服这一缺陷。研究利用18种针叶材制浆性能典型案例数据及偏最小二乘回归(PLSR)方法探讨了化学指标和纸浆得率之间的关系,并和传统的MLR方法进行比较。研究结果表明:PLSR能较好地克服纸浆材性能指标之间的多重共线性影响,预测结果更为准确,模型解释也比MLR更为合理,为纸浆材制浆性能综合评价提供了一条有效途径。Pulp yield is closely related to the chemical composition in the wood, which can be usually analyzed based on multiple linear regression （MLR）. But MLR method cannot be effective enough when there is serious multicollinearity among various performance indices. In the paper, the relationship between chemical indices and pulp yield was discussed by means of PLSR based on the research data in the pulping process of 18 kinds of softwoods. And comparison is made between PLSR and traditional MLR analyzing method. It’s proved that compared with MLR, PLSR can provide a more accurate prediction and reasonable explanation because it’s able to overcome the multicollinearity among various performance indices. Thus, an effective way is put forward for comprehensive evaluation of pulping performances.

关 键 词：纸浆得率 化学指标 多重共线性 偏最小二乘回归 

分 类 号：TS7[轻工技术与工程—制浆造纸工程] TS743