SVR在航空发动机基线挖掘中的应用研究

The Application Research of Support Vector Regression in Aero-engine's Baseline Mining

针对航空发动机基线难以获取的问题,利用支持向量回归机(support vector regression,SVR)算法,采用厂家监控系统数据和飞机快速存储记录器(quick access recorder,QAR)数据两种方式对基线进行挖掘分析,提供了获取基线的多种途径和方法,取得了比较可靠的结果。支持向量回归机在处理非线性回归分析时具有快速、准确的优点,能够进行单参数及多参数的基线回归分析,通过计算结果比较分析,多参数基线回归与单参数基线回归、一元线性基线拟合相比具有偏差小、精度高的优势,能够有效提高发动机基线监控的准确性。

As the aero-engine＇ s baseline is difficult to get,Support Vector Regression（SVR） is used with data from monitoring systems of manufacturers and Quick Access Recorder（QAR） to do baseline mining analysis,providing a variety of ways to get baseline and the results turned out to have a good reliability. SVR is fast and precise in dealing with nonlinear regression analysis. Single parameter regression and multi-parameter regression can be processed by using SVR. The conclusion can be drawn from analysis of calculation results that multi-parameter regression of SVR has a lot of advantages such as small deviation and high accuracy compared with linear regression and single parameter regression of SVR. As a result,the accuracy of the aero-engine＇s monitoring can be improved effectively.