基于部分线性变系数模型的飞行训练成绩综合评定

Comprehensive Evaluation of Flight Training Performance Based on Partially Linear Varying Coefficient Model

摘要: 针对当前飞行训练成绩综合评定工作中耗费大、效率低、过度依赖人工等问题，以飞行学员飞行训练成绩数据为研究对象，构建了部分线性变系数模型，通过数据拟合，对参数和非参数分量进行了估计和假设检验，并完成了变量的有效选择，最后使用模型评价指标决定系数$R^2$ 和残差标准差$\widehat{\sigma }$ ，对部分线性变系数模型和多元线性模型进行了比较。结果表明，部分线性变系数模型在对飞行训练成绩综合评定中要明显优于多元线性模型。

Abstract: Aiming at the problems of high cost, low efficiency, and excessive reliance on manual work in the comprehensive evaluation of flight training performance, this study takes flight training performance data of cadets as the research object and constructs a partially linear varying coefficient model. Through data fitting, the parameter and non-parameter components were estimated and subjected to hypothesis testing. Effective variable selection was also performed. Finally, the model evaluation metrics—coefficient of determination (R²) and residual standard deviation—were used to compare the partially linear varying coefficient model with the multiple linear model. The results show that the partially linear varying coefficient model significantly outperforms the multiple linear model in the comprehensive evaluation of flight training performance.