模式识别技术用于核电站松动件质量估计模型研究

Model Study of Mass Estimation of Loo se Parts in PWRs Using Pattern Recognition

介绍了一种利用模式识别理论和人工神经网络理论实现对核电站松动件质量进行估计的数学模型。利用模式识别理论的特征提取技术,实现对松动件碰撞信号的特征压缩,从而形成比较少的特征空间维数,经过神经网络的学习过程,实现对核电站松动件质量大小的估计。实验证明,该方法能够得到比较高的计算精度,其实现过程也比较简单方便,为核电站松动件质量估计提供了一种手段。

A new mathematics model was proposed based on the pattern recognition(PR)and neural network(NN)theories to estimate the masses of lo ose parts(LPs )in PWRs.The features existed in the i m- pact signals were extracted using th e PR technique.The compressed featu re space's dimensions can be much fewer than the original ones.Then th e learning procedure of NN can realiz e estimating the masses of LPs.The experiments carried out on t he simulator of reactor pressure vessel show that the model could be used to depict the complicated relationship between the mass and the impact information of LPs in NPPs.The measured precision can be h igh.And the realizing procedure is a lso simple.The mathematics model supplies a new solution for the mass estimation of LPs.