Existing structures suffer from an actual operating environment, excessive use, overloading, exposure to climatic variations and lack of sufficient maintenance and so on. In addition, structures can deteriorate due to the natural aging process or when subjected to unexpected loads, i.e. strong earthquake or high winds. The problem has added significance since the resources to build new structures are dwindling and the existing structures are aging; in some cases their design lives may have already been exhausted. Furthermore, there are many important structures in operation that were designed using provisions that are less rigorous than current practices, older structures whose properties may have changed over the years. Therefore, it is important to develop simple, inexpensive, and nondestructive evaluate procedures, which can be used routinely for the in-service condition assessment of existing and retrofitted structures.

This study is to develop the method to locate the damaged spots and to evaluate the damage. The proposed method is based on the fact that the values of structural parameters, in terms of mass, stiffness and damping characteristics change according to the shifts in the physical status of the structure. Details of these changes are reflected in changes of the structure properties. Thus, identification of structure parameters can be used to locate damaged spots and quantify the behavior of a structure.

A regular system identification procedure requires the measurement of input excitation forces and output responses. Unfortunately, the input excitation information is often difficult to obtain and hard to measure accurately. The system identification method not to require the input excitation information is needed, and the unknown input excitation could be of any type, including harmonic or random force of white noise.

Most system identification methods assume that the information of mass is known, Compared to the system parameters and the input load, mass is calculated relatively easily from the design references. However, there are live loads by the use of the structure as well as dead loads which is calculated from the design references in existing structures. If the use of the buildings or structures is changed or they are repaired, it is difficult to calculate the exact masses. Also sometimes, only part of a structure may have the mass information. It is needed to identify the system parameters (i.e. stiffness and damping) of the partial structure or the whole structure using only limited mass information of pre-selected locations.

The behavior of structure during an earthquake is greatly influenced by the property of the ground. The natural vibration characteristic of the ground may change the property of the earthquake. It increases or decreases the seismic load which excites the structure. If the deformation of the ground is occurred by the earthquake, rocking and sway may happen. And it may influence the behavior of the upper structure. The soil-structure interaction (i.e. rocking and sway) should be considered to apply to the existing structures affected by the earthquake.

When a structure suffers significant damage visible to the naked eye, there is an obvious problem. However and perhaps more seriously, the presence of defects after a natural disaster may not be obvious. Damages, especially by earthquake excitation, are occurred during an earthquake. The behavior of structure under strong earthquake excitation is easy to be nonlinear. Accordingly nonlinear system identification method is needed. Most nonlinear system identification methods in time domain should previously determine the hysteretic nature of the restoring force to identify the nonlinear behavior of structure (Yoshida, I. and Sata, T., 2004 & Lee, H. H. and Kim, S. S., 2002). In other words, they identify the coefficients for the decided hysteretic nature of the restoring force instead of identifying directly the unknown structural parameters. A system identification method of structural systems exhibiting inelastic restoring forces is needed to identify directly time-varying structural parameters and to estimate the restoring force-displacement relation successfully.

Including the proposed Advanced-ILS method, some system identification methods need data of displacement, velocity, and acceleration at each degree of freedom. Measuring the acceleration is relatively convenient and it does not need a set of fixed references. After measuring the acceleration response on the necessary point of structure, a digital method to double integrate accelerometer data in order to measure displacements is tried (Ribeiro et al, 1997). However, the acceleration data measured from the accelerometer during earthquake include the various noises. The velocity and displacement data calculated from the acceleration data including these noises may result in the distorted output different from the behavior of the real structure. Therefore, the procedure of eliminating noises using the efficient and proper process is important.

Considering the facts mentioned above, the primary objective of this study is to develop the system identification method to locate the damaged spots and to evaluate the damage in existing structures (buildings and similar structures that can be represented by a finite element algorithm) without disrupting their normal operations. To meet this primary objective, the proposed study aims to do the following:

1)to develop the system identification method to identify the structural parameters (i.e. stiffness and damping), where the input loading is not required to be measured and output response measurements are available at all degree of freedoms;

2)to develop the system identification method to identify the structural parameters of the partial structure or the whole structure using only limited mass information of pre-selected locations;

3)to develop the system identification method to consider the soil-structure interaction (i.e. rocking and sway) to apply to the existing structures affected by the earthquake;

4)to develop the system identification method to identify the time-varying structural parameters and to estimate the restoring force-displacement relation of the nonlinear structure;

5)to develop the procedure of eliminating noises from the measured acceleration data which are integrated to obtain the velocity and displacement data;

6)to verify the proposed system identification method by investigating numerical examples and shaking table tests.

This study includes eight chapters and an appendix. Chapter 1 is the introduction. It includes the statement of problem, the objection of research and the summary of chapter.

In Chapter 2, the literature review on system identification methods is given. The literature review includes the concept of system identification as well as currently available time domain system identification methods for the linear system and the nonlinear system.

Chapter 3 is about the proposed Advanced-ILS method which can identify element-level structural parameters (i.e. stiffness and damping) of a structure and a soil-structure interaction system using the structural responses at all degree of freedoms. It includes the structural model and the procedure of the proposed method. Then, the proposed method is applied to shear-type buildings. Five simulated numerical examples with different unknown input excitations including sinusoidal, random signal, and earthquake, are considered in this chapter. Both noise-free and noise-included responses are used in the identification processes of all numerical examples.

Chapter 4 is about the proposed Advanced-ILS method which can identify the system parameters of the partial structure or the whole structure using only limited mass information of pre-selected locations. It includes the partial structural model, substructure model, and the procedure of proposed method. Then, the proposed method is applied to shear-type buildings. Two numerical examples about the identification of the partial structure and two numerical examples about the identification of the whole structure are considered in this chapter. Both noise-free and noise-included responses are used in the identification processes of all numerical examples.

Chapter 5 is about the proposed Advanced-ILS method which can identify the time-varying system parameters and the restoring force-displacement relation of the nonlinear structure. It includes the nonlinear structural model of the proposed method. Then, the proposed method is applied to shear-type buildings. Three simulated numerical examples with different restoring force characteristics are considered in this chapter.

Chapter 6 is about the integral method of the measured acceleration data. It includes the error of measured acceleration data and the method (i.e. frequency domain filtering) to eliminate noises from the measured acceleration data which are integrated to obtain the velocity and displacement data. Two examples which use the measured acceleration data from the shaking table test are considered.

Chapter 7 is about the shaking table tests and the system identification using the measured acceleration data from the tests. 1×1span-three story shear-type structure is tested on the shaking table. Tests are two types. Test #1 is about the shear-type structure which is assumed that the base of structure is fixed. Four accelerometers are placed at each floor and shaking table with horizontal direction and three laser displacement sensors measure the horizontal displacement of each floor to compare with the integrated velocity and the double integrated displacement from the measured acceleration. Test #2 is about the shear-type structure which considers the soil-structure interaction. Two accelerometers and four laser displacement sensors are added at the base of structure with vertical direction. The measured acceleration data from the shaking table tests are integrated to obtain the velocity and displacement data after data processing which is explained in Chapter 6. These data are applied to the proposed Advanced-ILS method.

Chapter 8 is about the summary and conclusion of this study.

本論文は、外乱による振動特性の測定に基づいた建築構造物の構造同定手法に関するものであり、次の8章から構成されている。

第1章「INTRODUCTION」では、本研究の背景として、既存建物が供用中に環境や過荷重および劣化によって生じる構造性能の低下を非破壊的に検知する手法として、構造物の同定が脚光を浴びており、特に地震による建物を構成する部材の損傷を同定するヘルスモニタリングは重要であるとしている。本研究では、外乱による振動特性の計測により、損傷の生じた構造物の同定や、損傷部分の同定および安全性の評価のための手法の開発を行うものとしている。

振動特性の計測による構造物の同定には多くの手法が実用化されている。例えば、最小自乗法による振動特性による同定法では、構造物の振動に関する加速度や変位に関する時刻履歴の他、振動を励起する入力や質量分布に関する情報を使って剛性を評価することが可能である。

しかし建物の同定では、利用できる外乱が限られている。その上、風に対する建物の振動では励起する入力の計測が困難であり、一方で、地震に対する構造物の同定のように、建物地盤の相互作用によるスウェイやロッキング振動の場合のように、基礎下の入力や基礎質量や回転質量が推定が困難である。このような外乱の下での振動による同定は、従来の方法では必要な情報が不足して適用できないことがある。そのため、本研究では、改良繰返し最小自乗法(Advanced Iterative Least Square Method: Advanced ILS)を提案し、情報が不足した場合にも同定を可能とする方法に関する検討を行うとしている。

第2章「LITERATURE REVIEW」においては、構造物の同定に関する既往の研究のうち特に時間領域での同定法についてを調査分類している。ここでは、振動を励起する入力が未知の場合、構造物が非線形の場合の同定手法について詳しく述べておりそれらの比較分析を行って、入力と一部の質量が未知な場合、あるいは地震動に対する振動における地盤建物の相互作用に関する同定に関する研究は過去には行われていないとしている。

第3章「ADVANCED-ILS METHOD FOR EXITING STRUCTURE」においては、構造物の剛性や減衰および構造物地盤の相互作用を同定するための改良繰返し最小自乗法の理論について述べ、建物を表す仮想の5自由度せん断型振動モデルを例として、入力が未知の場合の同定を行った例を示している。

第4章「ADVANCED-ILS METHOD USING LIMITED INFOMRATION」においては、振動を励起する入力に加えて構造物を表すせん断型振動モデルの一部の質量に関する情報は未知とする条件における同定手法の理論について述べ、仮想の構造物に対する数値解析例を示している。

第5章「ADVANCED-ILS METHOD FOR NONINEAR STRUCTURRE」においては、非線形履歴特性を有する構造物において、構造パラメータが時間によって刻々変動する場合の同定方法の理論について述べ、鉄骨構造や鉄筋コンクリート構造の部材の履歴特性を有する履歴モデルからなる仮想の架構モデルに、本同定手法を適用した場合の数値解析例を示している。

第6章「INTEGRAL METHOD OF ACCELERATION DATA」においては、実現象では、得られた測定データが加速度記録であることが多いため、それらを積分して速度記録、変位記録に変換して同定に用いる場合を想定し、加速度記録に含まれる雑音やエラーが同定結果に及ぼす影響を検討し、雑音やエラーを除去する手法について検討している。

第7章「SYSTEM IDENTIFICATION USING DATA OF SHAKING TABLE TEST」では、ここに提案された同定手法を、振動台による多層せん断型振動モデルおよびロッキングスウェイ型振動モデルの模型試験体の振動実験に適用して、構造物の同定を行い、良い対応が得られたものとしている。せん断型振動モデルは、3層1スパンで基礎固定とし、加速度計とレーザー変位計により水平方向の計測を行っている。ロッキングスウェイ型振動モデルは、基礎下にロッキングバネを挿入したもので、水平方向の計測の他に基礎の鉛直方向の加速度・変位を計測している。第6章で提案された 雑音やエラーを除去する手法を本振動台実験で観測された記録に適用して、構造パラメータの同定を行い、良い対応が得られたとしている。

第8章「CONCLUSION」では、本研究で提案した 改良繰返し最小自乗法(Advanced Iterative Least Square Method: Advanced ILS) の推定精度について総括するとともに、今後の課題に関して取り纏めている。

このように、本研究は、建築構造物ならびに風圧力・地震動という特殊性に着目し、さらに損傷の進展の時刻歴を非破壊手法で計測するための構造物の同定手法を提案しており、構造物の同定手法を、建築構造物の健全性をモニターするための実用的な手法へと高めるための道を開いたものとして、極めて有用な研究である。

よって、本論文は博士(工学)の学位請求論文として合格と認められる。