6th World Congresses of Structural and Multidisciplinary Optimization

In this paper, an attempt is made to develop a structural health monitoring system that can adapt to the structural systems and environments, by introducing the learning ability. By introducing the learning ability, it is not necessary to prepare any previous knowledge and examination for the underlying structures and environment. In other words, it is not necessary to use the precise modeling and analysis method before conducting the health monitoring. The proposed system can predict the next state of structural behavior such as displacement and velocity from the current state of structural behavior and external force by the fuzzy reasoning. The proposed system can make the tuning of the consequent part of rules of membership functions so as to predict the next step of structural behavior with accuracy, by using the steepest descent method. Because the system has the learning ability through the fuzzy reasoning, the error of the prediction will increase that when the dynamic characteristics of the structure change due to the deterioration and degradation in-service period. Thus, the system can identify the change of structural condition states. The system can have the robustness to accidental errors or noise and reduce the calculation time by avoiding the complex computation with the fuzzy reasoning. Several numerical examples are presented to demonstrate the efficiency and applicability of the proposed system. In the numerical examples, a multiple degree of freedom structure with the same number of sensors is used to examine the performance of detecting the degree and location of deterioration. Through the numerical examples, it is concluded that the proposed system can identify the change of structural characteristics and condition states.