Abstract:
A data collection system pertaining to one embodiment of the present invention collects time-series data which is output from sensors installed on equipment which is a monitored object and carries out detecting an abnormality of the equipment. The data collection system stores plural fault models as data for comparison with time-series data and, in a learning process, determines a range to examine within time-series data by comparing the time-series data with each one of the fault models. An abnormality detection process includes extracting a partial frequency spectrum to examine from the frequency spectrum of time-series data, using information on the range to examine within the time-series data determined through the learning process, and carrying out detecting an abnormality of the equipment using the extracted frequency spectrum.
Abstract:
In a data collection system in which a server and a gateway are connected to each other and which includes a plurality of sensors installed on a subject of monitoring and outputting sensor data, the server selects a collection rule to change the granularity of the sensor data when it is detected on the basis of the sensor data that an abnormality or a sign of abnormality has occurred in the subject of monitoring, and the server transmits the collection rule to the gateway. The gateway acquires and manages states of the sensors, selects a first condition according to the collection rule received from the server, changes a setting for acquiring the sensor data on the basis of the first condition and the states of the sensors, acquires sensor data from the sensors on the basis of the setting, and transmits the sensor data to the server.