一种基于改进后的孤立森林机器学习方法在飞行器控制系统试验电源数据包络分析中的应用

An Application of an Improved Isolated Forest Machine Learning Method to Envelopment Analysis of Aircraft Control System Power Test Data

为了弥补当前试验数据包络分析方法存在的缺陷,提出并实现了一套新的数据包络分析方法。针对控制系统设计方式相同的不同飞行器多条数据一致性判断问题,采用改良后的孤立森林方法进行数据包络分析,快速找出异常电源电压数据。这种方法还可以推广到类似其他采样数据的数据包络分析场景。在此基础上开发了数据包络分析软件,并进行了多次验证试验。结果表明,提出的数据包络分析方法能有效判断出飞行器控制系统电源所涉及链路中的隐性或显性问题。

In order to make up for the defects of current test data envelopment analysis methods,a new data envelopment analysis method is proposed and implemented.In order to judge the consistency of multiple data of different aircraft in the same control system,the after improvement isolated forest method is used for data envelopment analysis.Aiming at the consistency judgment of multiple data of different aircraft with the same control system design,the improved isolated forest method is used for data envelopment analysis to find out unqualified power supply voltage data.This method can also be extended to sample data such as otherdata enveloping analysis scenarios.On this basis,a data envelopment analysis software was developed,and several verification tests were carried out.The results show that the data envelopment analysis method proposed in this paper can effectively judge the link problems involved in the aircraft.