【作者】 赵艳霞；

【导师】 刘书贤；

【作者基本信息】 辽宁工程技术大学 ， 岩土工程， 2004， 硕士

【摘要】 本论文主要对基于神经网络的工程结构动力检测法的理论和应用进行了研究。通过理论及资料分析，得出工程结构在损伤前后动力特性参数即模态参数的改变包括了损伤位置和损伤程度两方面的信息。从而对模态参数进行合理的变换作为网络的输入值，并将混合优化策略引入到网络中，采用LMBP算法，提高了损伤识别网络的性能。论述了基于神经网络的多级损伤检测技术及实现策略，并对一框架结构进行了损伤模拟，应用多级神经网络进行损伤检测。结果表明，网络能对工程结构的损伤状况作出一定的判断。本文中并没有考虑到误差的影响，应用到实际中还需要做大量的工作。更多还原

【Abstract】 This paper focused on the theory and application of the method based on the combination of the dynamic diagnosis and neural network. Through the theory and reference material analysis, it showed that the change of the dynamic characteristic parameters can reflected the damage location and degree. In the paper, carried on rational variation as the input values of the network, took the mix optimization tactics as the networks algorithm. Those have improved the performance of the network. The paper also described the multi-level damage detection technology and tactics realized based on the neural network. Used computer to simulate different damage of a frame structure, detected the structure by the trained network. The result showed that the network could make good judgments to the damage state of the structure. Have not considered the influence of the error in the paper, so apply to reality need to do a large amount of work.更多还原