基于神经网络与果蝇优化算法的涡轮叶片低循环疲劳寿命健壮性设计

在对涡轮叶片低循环疲劳寿命概率分析的基础上,将广义回归型神经网络(generalized regression neural network,GRNN)与果蝇优化算法(fruit fly optimization algorithm,FFOA)结合,利用果蝇优化算法的多点全局的快速搜索能力来优化影响疲劳寿命的随机变量,进行涡轮叶片低循环疲劳寿命健壮性优化设计.优化结果表明:疲劳寿命的概率区间减小17.9%,对随机变量的敏感度降低,从而可以更精确地对疲劳寿命进行估计.计算结果验证了该方法在工程应用中的可行性. 

Robust design of turbine-blade low cycle fatigue life based on neural networks and fruit fly optimization algorithm

By combining generalized regression neural network (GRNN) with fruit fly optimization algorithm (FFOA) and using FFOA multi-point global search ability to optimize the random variable which affects the fatigue life, a robust optimization design for low cycle fatigue life of turbine-blade can be made on the base of probability analysis for turbine-blade low cycle fatigue life.Optimization results show that the probability interval of fatigue life decreases 17.9%,and the sensitivity of the low cycle fatigue life of the random variable can be reduced,so the fatigue life can be estimated more accurately.Optimization results indicate that the proposed method is available and feasible for the engineering application.