ObjectivesThe aim of the research is to develop the principle of storing data templates to take their temporal natureinto account, making it possible to reduce decision-making times.In order to describe and identify temporal patterns in fuzzy time series behaviour in real time, the task was set to develop a hybrid data structure that allows for a consideration of sequences of fuzzy values formed from clear observable data as well as a determination of the length of these sequences and possible uneven time intervals between the observations.MethodsThe article discussesan approach to formalising the description of temporal cause-effect relationships between events occurring at the object location as well as that of its environment, based on a set of singly-connected lists of triplets. Each triplet contains a fuzzy linguistic variable, the duration of its observation and the permitted interval of observation of insignificant data.ResultsAn algorithm for detecting knowledge base patterns in real time was developed, taking into account the possibility of a time shift in observing long sequences of identical values of the observed value. The possibility of partial data overlapping corresponding to triplets of different patterns is taken into account. The proposed hybrid pattern makes it possible to accelerate the detection of temporal regularities in the data.ConclusionScientific results are presented by the developed structure for storing information on temporal regularities in data, based on a singly linked linear list, as well as an algorithm for finding regularities in observational data using a set of OLS-patterns. The advantage of this structure and algorithm in comparison with the known ways of storing and analysing temporal data is a reduction in the amount of memory necessary for storing templates in the knowledge base, as well as the possibility of applying OLS patterns for decisionmaking purposes.
Real-Time False Data Detection in Smart Grid Based on Fuzzy Time Series
