基于GM-SVM的三参数威布尔分布参数估计

3-PARAMETER WEIBULL DISTRIBUTION ESTIMATION BASED ON GM AND SVM

针对三参数威布尔分布模型复杂,在小样本情况下传统方法估计精度低的问题,提出了基于GM-SVM的参数估计方法。该方法将灰色模型(grey model,GM)和支持向量机(support vector machine,SVM)相结合。首先,利用灰色模型法在位置参数估计上的独特优势,其次,充分利用支持向量机在小样本情况下优良的估计效果,建立了高精度的GM-SVM威布尔分布参数估计模型,并将其应用于某型发动机叶片疲劳寿命分布的参数估计。仿真实验结果表明,该模型具有较高的参数估计精度,在小样本情况下优势较为明显,能够准确得出威布尔分布的三个参数。

Aiming at the fact that the 3-parameter Weibull distribution model is complex and traditional methods have a low accuracy rate in the case of small samples,a new parameter estimation method based on GM-SVM was proposed. Grey model（ GM） and support vector machine（ SVM） were combined in this method. Firstly,to take advantages of location parameter estimation based on GM method. Secondly,making full use of the accurate estimated results based on SVM in small samples.This method is applied to the establishment of a Weibull distribution parameter estimation model based on GM and SVM,and to the reliability life distribution parameter estimation of certain engine blade. The simulation test results show that this model has preferable parameter estimation accuracy,has obvious advantages in small samples. This model can accurately obtain the three parameters of Weibull distribution.