指数外推法和支持向量机相结合的压气机特性扩展方法

针对航空发动机起动建模中压气机低转速特性计算精度较低的问题,提出了指数外推法和支持向量机(SVM)相结合的特性扩展计算方法.首先分析了指数外推法的计算方法及其局限性,然后将压气机已知的高转速特性作为SVM的训练集,以指数外推法获取的低转速特性作为测试集,同时将压气机特性转换为按出口气流参数表示以降低SVM原始数据的非线性,利用交叉验证算法选择SVM参数并进行模型训练,预测并获得压气机低转速特性.通过与单纯使用指数外推法获取的特性对比分析表明:指数外推法和SVM相结合的压气机特性扩展计算方法,最大相对误差减小了约2.8%,有效提高了特性扩展计算精度. 

Method of compressor characteristic extension combining exponent extrapolation method with support vector machine

Considering the poor accuracy computation of the compressor lower speed characteristic in aero-engine startup modeling, a method of compressor characteristic extension combining exponent extrapolation method with support vector machine (SVM) was proposed. Firstly, the mechanism and limitation of exponent extrapolation method were analyzed. Then the known data of compressor higher speed characteristics were used as SVM training set, and the data of lower speed characteristic obtained by exponent extrapolation method were used as the test set. Meanwhile, in order to reduce the nonlinearity of the original data in SVM, the compressor characteristic was converted into the characteristic represented by outlet flow parameters. The lower speed characteristics were predicted and obtained by using a kind of cross validation algorithm to select the parameters of SVM and train the SVM model. The comparative analysis with the characteristics obtained by using only exponent extrapolation method shows that using the method of compressor characteristic extension combining with exponent extrapolation method with SVM, the maximum relative error is reduced by about 2.8%, and the characteristic extension precision is effectively improved.