Integrated Assessment of Seismic Damage in Structures

A new approach is explored for assessing the safety condition of civil engineering structures following strong-motion earthquakes that utilizes artificial intelligence to collectively consider several factors in the assessment process. Following this approach, various types of information related to the damage condition of the structure are input into a neural network computer program that has been trained to generate safety assessments. The network acts on the input information according to weighting factors established from data on past seismic events to generate a condition assessment. This study focused on utilizing various quantitative damage indices in formulating a safety assessment. Development of a neural network for this purpose was accomplished using response data from structural systems tested in the laboratory. The neural network generated more reliable assessments than could be obtained using any single indicator or from a linear regression model that utilized all indicators. The network was used successfully to assess the condition of a real structure damaged in an earthquake. A network of this kind may be useful in automatic and immediate condition assessment following a seismic event and as an aid to an expert in determining such assessments.