求解超高压筒形容器爆破压力的神经网络方法  

ANN-based prediction of bursting pressure under ultra-high pressure for cylindrical vessel

作　　者：袁格侠[1,2] 刘宏昭[1] 钱学梅[3] 范彩霞[1,4] 钱学军 秦丽柏[3] 

机构地区：[1]西安理工大学机械与精密仪器工程学院,陕西西安710048 [2]宝鸡文理学院机电研究所,陕西宝鸡721007 [3]中国兵器科学研究院宁波分院,浙江宁波315103 [4]焦作大学,河南焦作454150 [5]内蒙古第一机械制造(集团)有限公司,内蒙古包头0314034

出　　处：《兵器材料科学与工程》2010年第2期31-34,共4页Ordnance Material Science and Engineering

基　　金：宝鸡文理学院重点科研项目(ZK0727)资助

摘　　要：将BP和RBF神经网络的理论和算法应用于预测超高压容器爆破压力的研究中。选用MATLAB神经网络工具箱建立预测爆破压力的神经网络模型,研究模型中影响爆破压力的主要参数,内外径比值和材料的强度极限,屈服极限,屈服强度与强度极限的比值;选用Faupel、Crossland和Bones等文献中的爆破实验数据对神经网络模型进行训练,用训练好的神经网络模型对爆破压力进行预测。预测结果表明,用BP和RBF神经网络方法建立的模型能够对超高压筒形容器的爆破压力进行较为准确的预测。The theory and the algorithm of BP and RBF neural network are applied in the research for predicting the bursting pressure of ultra-high pressure vessel.First,the neural network model has been established for predicting bursting pressure by using MATLAB Neural Network Tools in consideration of the main factors of influencing bursting pressure.The factors include ultimate strength,yield stress,ratio of outer radius to inner radius of the pressure vessel cylinders and yield ratio.Then the established neural network model is trained by choosing a large amount of bursting experimental data from Faupel,Crossland and Bones,and some references.Finally,the trained neural network model is used to predict the bursting pressure.The prediction results show that the bursting pressure model using the BP and RBF neural network method can predict bursting pressure exactly.

关 键 词：超高压容器 爆破压力 BP神经网络 RBF神经网络 预测 

分 类 号：TQ051[化学工程] TP391.75[自动化与计算机技术—计算机应用技术]