Damage estimation using AI was examined to facilitate the restart of nuclear power plants after a large earthquake. Cumulative absolute velocity (CAV), in addition to PGA, PGV, and response spectrum, was used as training data to estimate the ductility factor. A multivariate analysis was also performed to confirm the consistency with the AI and to compare the results. The multivariate analysis confirmed that the response spectrum at a structural period of 2 times and a damping coefficient of 0.20, which are equivalent period and equivalent damping of response, are influential parameters to the results. The most accurate results were obtained in the case where the relationship between the data other than CAV and the ductility factor was learned．
