中文摘要：

摘要：首先构建了活塞的热机耦合模型，计算了活塞在机械负荷和热负荷下的变形情况，将此变形量代人到活塞动力学模型中．选取影响活塞敲击力的3个主要参数，每个参数取4个不同的数值，相互交叉组合共得到64种方案，运用动力学模型计算得到64种方案的活塞敲击力峰值，并综合比较了各因素对活塞敲击力峰值的影响趋势．而后利用BP神经网络理论建立了活塞敲击力的神经网络模型并进行了训练和验证，然后利用该模型对主要参数进行了优化．结果表明，运用热机耦合和神经网络相结合的方法进行活塞敲击力的优化设计，减少了工作量并能得到满足精度要求的结果，对活塞结构的优化设计有一定的指导意义．

英文摘要：

A piston coupling model of mechanical-heat load was firstly constructed and the piston deformation value under mechanical load and heat load was calculated. The deformation value was then taken into the piston dynamic model. Three major parameters affecting piston impinging force were selected, four different values for each parame- ter were taken, and 64 combinations were formed by cross interact. The piston peak impinging forces in the 64 cases were obtained by using the dynamic model. The influences of the factors on piston peak impinging force were com- pared comprehensively. The neural network prediction model of piston impinging force was built by BP neural net- work theory, which underwent training and certification. The main parameters affecting piston impinging force were optimized using this model. Results show that workload is reduced and the satisfactory results with required accuracy are obtained by combining mechanical-heat load coupling and neural network method to optimize the design of piston impinging force, which will provide some guidances for the optimization design of piston.