Abstract
To introduce the performance design into the decision making problem of the repair or renewal of the damaged armor block on seawall, a performance evaluation system of seawall with the accumulative damage of armor is necessitated. In this study, numerical experiments and model tests are conducted to investigate the reflection coefficient of the seawall under condition of the damage progression of armor blocks on crown and slope of seawall. The regressive prediction properties between the reflection coefficient and the damage parameter could not be recognized from those results. The performance evaluation system is developed using a neural network system, which evaluates the causal relation of the reflection coefficient and the accumulative damage parameter. The neural network with a three-layer calibrated by the numerical experiments can predict the reflection coefficient of seawall under the damage progression of armor layer within about 10% errors.
