Detection of nonconformities in monitoring system´s measurements

This paper focuses on the use of intelligent computational methods and data mining technology applied to interpretation of measurements obtained from instruments of a hydroelectric dam that will signals the occurrence of an unusual value. This analysis is done considering the latest measurement collected and comparing this individual value with previous measurements already saved at historical database. This analysis is periodically done by an intelligent algorithm that produces a warning report if the latest value collected from the instrument is not compatible with the historical behavior learned from the past measurements. A modified HOT SAX algorithm is proposed for this analysis since this method was originally conceived for a fixed size time series and not for continuous data acquisition.