【摘要】 为了提高内燃机设计的可靠度 ,提出了应用神经网络 蒙特卡罗法求解发动机零部件可靠度的新方法 .该方法利用神经网络根据有限元分析样本逼近应力函数 ,结合应力 强度干涉模型 ,采用蒙特卡罗随机模拟求取可靠度 .将计算结果与常用的一次二阶矩法、设计验算点法相比较 ,表明该方法是可行的 ,而且适用面更广 ,可适用于多参数且服从于任何分布的问题更多还原

【Abstract】 In order to improve the reliability in turbocharger design, a new ANN(artificial neural networks) Monte Carlo method is presented to conduct the reliability analysis of blades in turbocharger. The BP neural network is used to conduct the stress function approximation. Combined with the Monte Carlo stochastic simulation, it can conveniently solve the reliability problem. Compared the result with several common methods, it is proved that the ANN Monte Carlo method is viable. And it can be used in more reliable problem such as multi dimension parameters submitted to any distribution form. The ANN Monte Carlo method is an important application of stochastic simulation to reliability analysis and engineering design更多还原