偏最小二乘法在成分提取中的应用研究

Study on the Algorithm of PLS in the Component Concentration

介绍了偏最小二乘的理论基础,指出了该方法在小样本、多自变量、变量间存在多重相关性时的独特优势,并以航空发动机大修费用为例,进行数据的仿真。结果表明:该方法能有效地进行数据成分的提取,及应变量和自变量之间函数关系的判别。

The algorithm of PLS（Partial Least-square Regression） is introduced,and the advantage of the algorithm that is accomplished in deal with the small stylebook,too much independent variable and severe correlation between the variable is pointed out,also a simulation of the algorithm which is about the maintain expenses of the aeroengine is accomplished which shows that the algorithm is effective in component concentration and distinguishing the function correlation.