GBRT技术在MLS模拟训练成绩评估中应用

Application of GBRT Technology in Performance Evaluation of Military Logistics Support Simulation Training

综合成绩评估是军事后勤模拟训练系统中的一个重要功能模块,为了实现模拟训练自动化成绩评估,在模拟训练系统成绩评估中提出使用梯度渐进回归树(GBRT)智能技术。在对某科目的后勤模拟训练成绩数据分析基础上,找出线性无关评估指标特征集合"人员配置成绩、完成任务时间和装备操作正确率"作为输入数据,建立基于GBRT成绩评估预测模型,最后通过优化参数组合的GBRT评估模型预测模拟训练综合成绩。经过某科目的模拟训练成绩数据样本测试,基于GBRT模型评估综合成绩与实际训练成绩基本一致,从而验证GBRT技术在军事后勤保障模拟训练成绩评估中应用的可行性和有效性,为军事后勤保障模拟训练系统自动化成绩评估增添新的途径和方法。

The comprehensive performance evaluation is an important function module in the military logistics simulation training system.In order to realize the evaluation of simulation training automation,gradient progressive regression tree(GBRT)intelligent technology is proposed in the performance evaluation of simulation training system.Based on the analysis of the performance data of logistics simulation training for a certain subject,this paper found out the feature set of linear independent evaluation index"personnel allocation performance,time to complete tasks and equipment operation accuracy"and used it as the input data,established a GBRT-based performance evaluation and prediction model,and finally predicted the comprehensive performance of simulation training through the evaluation model.The results of the GBRT model are basically consistent with the actual training results,which verifies the feasibility and effectiveness of the method,and provides an automated performance evaluation method for the military logistics support simulation training system.