Seeking frequent episodes in baseline data of In-Situ Decommissioning (ISD) Sensor Network Test Bed with temporal data mining tools

Abstract

Savannah River National Laboratory (SRNL) has established an In-Situ Decommissioning (ISD) Sensor Network Test Bed—a unique, small scale, and configurable environment— for the assessment of prospective sensors on actual ISD system at minimal cost. The temporal data mining (TDM) technique can be employed to process the extensive data collected by the ISD sensors well because these data are time-specific, age-specific, and development stage-specific. This paper analyzed the baseline data collected by ISD Sensor Network Test Bed in recent years with the assistant of TDM algorithms to work out frequency episodes in the event stream. The results have confirmed that TDM techniques are effective tools to validate ISD performance, and the frequent episodes found in the data stream not only showed the daily cycle of the sensor responses, but also established the response sequences of different types of sensors, which was verified by the actual experimental setup. Some abnormal patterns may have the potential for prediction of system failures.