動態濃縮結構矩陣、基因演算法與交叉模型交叉模態法應用於結構系統參數修正

Abstract

一個結構系統，可以利用結構系統參數以及結構模態參數進行的描素 ，而這些參數正是我們了解結構系統的依據。結構系統參數修正方法已經 發展多年，而當中多數的研究皆是以模態資訊已知為前提進行分析，然而 大部分的模態取得方法存在著許多缺點。因此本論文不進行模態資訊的量 測，將提供一種新的模態資訊取得方法，並搭配合適的結構系統參數修正 方法，並加以修正改良使結構系統參數修正結果更加準確。研究方法將藉 由兩個策略:1.使用逆疊代動態濃縮方法將勁度矩陣和質量矩進行濃縮2. 藉由頻率資訊搭配基因演算法尋找模態參數，改善模態資訊量測不精確的 問題。前者有效的降低了參數的數量，有利於高度密集的搜尋；後者則搭 配前者的動態濃縮特性以及的模態資訊，配合交叉模型交叉模態方法進行 結構參數修正，達到結構系統參數修正的目的。最後完成一個僅須獲得結 構完工後的頻率，即可執行結構系統參數修正的有效數值方法，解決傳統 模態資訊量測不準確的缺點，並能達到結構系統參數修正的目的。

A structural system can be functionally experience by structural and its corresponding modal parameters. Likewise, the parameters are the basis of describing a system. The way of model updating structural system parameters has been developed for years; most of studies are analyzed under a presupposition that mode shape is already known. Therefore, the errors and disadvantages are embedded in those studies. Our thesis is designed without measuring mode shape and provides an alternative way to acquire mode shape. Furthermore, the study is accompanied by a suitable way of revising structural system parameter in order to enhance the accuracy of the result. The method is based on two strategies: 1. Concentrating stuffiness matrix and mass matrix by Iterative method for dynamic condensation of structural matrices. 2. Improving the accuracy when measuring mode shape by setting modal parameters in a way of combining frequency information and Genetic Algorithms. Respectively, one reduces the amount of parameters and advances the search in high intensity; and the other, accompanied with the property of Iterative method for dynamic condensation of structural matrices and mode shape from the previous strategy, revises structural parameters by Cross-model cross-mode method for model updating. In conclusion, the effective value of revising structural system parameter can be executed by only a decisive frequency of the whole structure. This solves the disadvantages in traditional studies of measuring mode shape , and fulfills the purpose of revising structural system parameters.