基于BP神经网络的汽水分离器应力在线计算模型

On-line Stress Calculation Model of Steam-Water Separator Based on BP Neural Network

汽水分离器的寿命监测是关系到直流锅炉运行安全性的重要课题,实现寿命监测的关键是寻找实时应力的计算模型。针对模型的需要,建立了分离器的三维有限元模型,根据运行工况进行了应力计算。利用BP神经网络优越的非线性逼近能力,建立了介质压力和筒体金属壁温序列与分离器应力的在线计算模型。经过仿真验证,模型的计算精度能够满足于工程应用。

Life monitoring of the steam-water separator（SWS） is a very important content of Ultra Supercritical Boiler（UCB） safety management.Its service life is related to the monitoring of the security issue of the UCB.Based on the structure of the steam-water separator,a 3-D finite element model of thermal-pressure coupling stress field is proposed.In order to realize the stress online calculation,the neural network and finite element method are combined together.The calculated results are the training sample.The non-linearity of BP neural network is used in constructing the mapping relation between stress and temperature series of barrel wall and pressure of the SWS.By simulation,the results accuracy can meet the industrial demand and this model can be used in practice.