支持向量机在航空发动机起动模型辨识中的应用研究

The Study of Support Vector Machines Used in Identification for Engine Start Model

支持向量机(SVM)可以优化网络,有效降低模型复杂性,不存在维数灾难和局部极小问题。本文以某型发动机起动调整试验的试车数据为样本,使用SVM对某一大气条件下的发动机起动模型进行了辨识,并使用另外一组试车数据,通过辨识模型对起动过程进行了仿真;最后,比较了SVM和RBF神经网络起动模型的辨识精度。结果表明,用SVM辨识发动机起动过程模型,方法简单,学习速度快,辨识精度较高。

SVM can optimally create network structure automatically, efficiently reduce the modeling complexity, and without local minima, dimension disaster. So SVM are used to identify the start model at an atmospheric condition. With such identification, it has simulated a start process using other test data.At last, comparison has been made between SVM and RBF neural networks. The results show that SVM has a better identification of start model with features of simplicity, fast to learn and precision.