This study proposes an irregularity detection system of sabo facility foundation using a monitoring data of groundwater level utilizing the multipattern neural network. The proposed system predicts the daily groundwater level of the measurement point at the lowest sea level by a neural network function associated with the other measurement points. If the difference between the prediction and the observation of a groundwater level exceeds a certain threshold level, the system recognizes that an irregularity might occur. The proposed system shows better fitting than the conventional method, because the observed data includes multipatterns relationship in the observation.
