Anomaly detection from slope surface strain using linear regression

In recent years, measurement devices and monitoring systems involving information and communications technology have been developed and used for slope monitoring in Japan. Although it is possible to measure slope monitoring data during slope failure, systems to determine slope anomalies are also required. In this study, anomalies were detected and verified from slope monitoring data using linear regression model based on machine learning. In this method, the data was predicted from time series of the slope surface strain by eight sensors installed in a slope failure experiment using a centrifuge, and verified with the measured data. For evaluation of the slope monitoring data, the machine learning method used slope monitoring data from normal state during slope stability as reference. When a pattern of data different from the normal state was detected, the slope was considered to be unstable and an alert was issued. Consequently, from the time series change in the number of anomalies detected by the eight installed surface strain sensors, it was confirmed that the anomalies on the slope can be detected before the collapse of the slope.