Contemporary monitoring systems are a source of data streams. Processing of this data is an interesting issue from both a performance and data storage perspective. It is worth paying attention to the concept of stream database management systems, which are a hybrid that allows for efficient analysis of the data stream and provide a set of implemented statistical methods.This article presents the issues raised by embankment monitoring systems, a sensor network which generates a large stream of measured signals that should be analyzed in real-time. Warning or crash scenarios are also generated which are compared to the incoming data. A model of this data is presented and a construction of stream data analysis is proposed.