部附件送修费用的偏最小二乘回归模型及敏感性分析

为了克服部附件送修费用的一般模型不能方便进行敏感性分析的缺点,考虑设计阶段下输入数据采集的可行性,应用偏最小二乘回归理论,通过变量投影重要度分析,筛选出重量、价格、平均非计划拆卸时间、平均车间修理时间、SRU的个数5个参数。采用对数线性关系式,构建了部附件送修费用的参数模型,并通过实际值与回归值的误差分析验证了该模型的可靠度。根据得到的参数模型和工程经验,确定了各参数的单一函数表达式及其取值范围;结合切线斜率即敏感度的事实,仿真计算了各个参数变化的显著敏感区间。结果表明,该模型精度较高,能方便进行敏感性分析,可作为飞机设计阶段工程人员选择合适部附件的实用工具。

Partial Least-Squares Regression Model of Component Repair Cost and Sensitivity Analysis

In order to overcome the shortcoming of the general model for component repair cost that the parameter cannot be used to analyze sensitivity expediently five parameters namely weight price mean time between unscheduled removals average shop time per overhauls and number of SRUs were selected by using Partial Least-Squares Regression (PLSR) and taking the feasibility of inputting the gathered data into consideration. The logarithm linear model was established for parameters of the component repair cost. The reliability of the model was validated through the analysis of the error between the true result and regression result. Through the parameter model and engineering experience the function expression and value range of all parameters were confirmed. Based on the truth that the tangent slope is the sensitivity salient sensitivity interval was calculated out under the change of parameter. The result shows that this model has a high accuracy and can be used to analyze sensitivity expediently which could be used as a practical tool for engineers who want to choose right component in design.