地基结构及其相互作用系统参数识别理论结构工程专业论文.docxVIP

西安建筑科技大学博士论文测量信息的’隋况。用单元刚度矩阵基本值和模态应变能来选择基本模态，用修正的 西安建筑科技大学博士论文 测量信息的’隋况。用单元刚度矩阵基本值和模态应变能来选择基本模态，用修正的 Latin超立方采样技术和模态准入准则来产生网络的输入数据。框架子结构在动载作 用下的自振频率和模态作为网络的输入，子矩阵参与系数作为网络的输出，用 Levenberg—Marquardt算法训练网络。把动态子结构方法引入结构的参数识别中，从 而为自由度数目较多的大型结构的参数识别提供了一种可行的思路。 把以上建立的解决复杂结构参数识别的方法应用到贮仓一地基相互作用系统的 参数识别中，把上部结构和地基作为两个子结构，运用动态子结构参数识别方法分 别进行了参数识别。 关键词： 参数识别 神经网络滤波算法 动态子结构LM算法 动力相互作用 修正的Latin超立方采样技术 本文得到国家自然科学基金资助 】】 西安建筑科技大学博士论文Theory 西安建筑科技大学博士论文 Theory on Parameters Identification of Foundation， Structure，and Foundation-Structure Interaction system Specialty： Structural Engineering Graduate Student： Dong QingHua Instructor： Prof．Huang Yi ABSTRACT This paper makes series research on following aspects：parameters identification of foundation，structure，foundation—structure interaction system．1)The precision of the estimated parameters and convergence are enhanced by the introduction of modern optimizing algorithm． 2)Using new model from mathematics and mechnics，the reliability of parameters identification is improved than that of adopting chained—multi—degrees—of-freedom structures．3、On the basis of completed FEM methods，parameters of a frame are identified．With a curtain clue and solusion．this procedures also work well with a fairly large complex system．Hence，a new method to identify the parameters of foundation，structure，foundation—structure interaction system is being developed by summarizing the above research work． The relevant neural network of a new learning algorithm·---··backpropagation with Kalman filter is proposed in the context of system identification，The basic idea is that the adjusting weigh value of the neual network is taken as the identification of the parameters for the nonlinear system．So，the Kalman filtering technique was applied to modify the weight matrices of neural network．The calculuation programe is worked out on directly idenfifying the physical parameters for the silo，which combines neural network with new proposed learning algorithm within t