论文部分内容阅读
针对不确定性参数对结构力学性能的随机影响,该文利用混合神经网络良好的小样本学习和泛化能力构建结构响应复杂的函数关系,采用改进的混沌粒子群算法优化网络寻址结构。结合蒙特卡洛法对结构进行随机性分析,并根据该文提出的新的灵敏度度量参数计算随机变量的全局灵敏度系数。通过数学算例和工程算例验证了所提方法的可行性,且结构响应的概率分布曲线也可以真实的反应实际情况。同时,利用该文所提出的随机灵敏度计算方法可以更好的反应各随机变量对结构响应的相关性和敏感性。
Aiming at the stochastic influence of uncertainty parameters on the mechanical properties of structures, this paper uses the small sample learning and generalization ability of hybrid neural networks to construct the complex function relationship of structure response, and uses the improved chaotic particle swarm optimization algorithm to optimize the network addressing structure. Combined with Monte-Carlo method, the structure is analyzed stochastically, and the global sensitivity coefficients of random variables are calculated according to the new sensitivity measure parameters proposed in this paper. The feasibility of the proposed method is verified by mathematical examples and engineering examples, and the probability distribution curve of structural response can also truly reflect the actual situation. At the same time, using the stochastic sensitivity calculation method proposed in this paper can better reflect the correlation and sensitivity of each random variable to the structural response.