基于支持向量机的航空发动机叶片超声检测

支持向量机(SVM)是一种具有出色学习性能的新型机器学习方法,它能够较好地克服神经网络容易出现的过学习、网络结构难以确定以及局部极小等缺点。研究了小波包变换提取发动机叶片缺陷特征向量的问题,提出一种基于支持向量机的航空发动机叶片超声检测方法。实验表明,基于小波包分解提取特征向量结合支持向量机的识别方法,能够有效地区分发动机叶片部件的几种典型缺陷。

ULTRASONIC TEST OF AERO-ENGINE BLADES BASED ON SUPPORT VECTOR MACHINE

Support Vector Machine(SVM)is a new type of machine study method which has excellent study propensity.It can overcome the disadvantages of Neural Network(NN)such as over-studying,uncertainty of the structure and local minimization problems.Feature vectors of flaws in aero-engine blades were extracted by wavelet packet transform(WPT).A new type of ultrasonic test based on SVM was presented in the detection of aero-engine blades.Through the real tests of aircraft blades,the method could classify the typical flaws in the engine blades.