HYBRID APPROACHES TO THE FORMALISATION OF EXPERT KNOWLEDGE CONCERNING TEMPORAL REGULARITIES IN THE TIME SERIES GROUP OF A SYSTEM MONITORING DATABASE

Objectives. The presented research problem concerns data regularities for an unspecified time series based on an approach to the expert formalisation of knowledge integrated into a decision-making mechanism. Method. A context-free grammar, consisting of a modification of universal temporal grammar, is used to describe regularities. Using the rules of the developed grammar, an expert can describe patterns in the group of time series. A multi-dimensional matrix pattern of the behaviour of a group of time series is used in a real-time decision-making regime in the expert system to implements a universal approach to the description of the dynamics of these changes in the expert system. The multidimensional matrix pattern is specifically intended for decision-making in an expert system; the modified temporal grammar is used to identify patterns in the data. Results. It is proposed to use the temporal relations of the series and fix observation values in the time interval as ―From-To‖, ―Before‖, ―After‖, ―Simultaneously‖ and ―Duration‖. A syntactically oriented converter of descriptions is developed. A schema for the creation and application of matrix patterns in expert systems is drawn up. Conclusion. The advantage of the implementation of the proposed hybrid approaches consists in a reduction of the time taken for identifying temporal patterns and an automation of the matrix pattern of the decision-making system based on expert descriptions verified using live data derived from relationships in the monitoring data. 

1. Parsaye K.A. Characterization of Data Mining Technologies and Processes. The Journal of Data Warehousing. 1998;1.

2. Suchkova L.I., Starikov E.S. K voprosu o metode vyyavleniya zakonomernostey v dannykh monitoringa. Materialy 14 Mezhdunarodnoy konferentsii "Izmerenie, kontrol, informatizatsiya". Barnaul: AltGTU; 2013. 155-157. [Suchkova L.I., Starikov E.S. On the question of the method of identifying patterns in monitoring data.Proceedings of 14th International Conference ―Measuring, Control and Automation‖. Barnaul: AltSTU; 2013. 155-157. (In Russ.)]

3. Starikov E.S. Formalizatsiya opisaniya zakonomernostey v gruppakh vremennykh ryadov. Sbornik po rezultatam XXI zaochnoy nauchnoy konferentsii "Mezhdunarodnyy nauchnoissledovatelskiy zhurnal". Chast 1. Ekaterinburg; 2013;11(18):131-132. [Starikov E.S. The formalization of regularities description for time series groups. Collection of the results of the 21st Scientific Conference ―Research Journal of International Studies‖. Pt. 1. Ekaterinburg; 2013;11(18):131132.(In Russ.)]

4. Pavlov S.I. Sistemy iskusstvennogo intellekta: uchebnoye posobie. Tomsk: Tomskiy gosudarstvennyy universitet sistem upravleniya i radioelektroniki; 2011;1:175. ISBN 978-5-4332-0013-5. [Pavlov S.I. Systems of artificial intelligence: tutorial. Tomsk: Tomsk State University of Control System and Radio Electronics; 2011;1:175 (In Russ.)] Available from: http://biblioclub.ru/index.php?page=book&id=208933

5. Yarushkina N.G., Afanasyeva T.V., Perfilyeva I.G. Intellektualnyy analiz vremennykh ryadov: uchebnoye posobie. Ulyanovsk: UlGTU; 2010. 320 s. [Yarushkina N.G., Afanasyeva T.V., Perfilyeva I.G. Intelligent analysis of time series: tutorial. Ulyanovsk: UlSTU; 2010. 320 p. (In Russ.)]

6. M.Yorchen, F., Ultsch, A. Mining Hierarchical Temporal Patterns in Multivariate Time Series. In: Biundo, S., Fruhwirth, T., Palm, G. (Eds.), KI 2004. Advances in Artificial Intelligence, Proceedings 27th Annual German Conferencein. AI, Ulm. Germany: Springer, Heidelberg; 2004, 127-140. Available from: http://www.mybytes.de/papers/moerchen04mining.pdf

7. Boks Dzh., Dzhenkins G. Analiz vremennykh ryadov. Prognoz i upravlenie. M.: Mir; 1974. 406 s. [Box G., Jeniks G. Time series analysis. Prediction and control. Moscow: Mir; 1974. 406 p. (In Russ.)]

8. Starikov E.S. Suchkova L.I., Koreshkov Yu.N., Koreshkova A.A. Universalnyy podkhod k opisaniyu i ispolzovaniyu ekspertnoy informatsii v sistemakh monitoringa na baze patternov povedeniya. Polzunovskiy almanakh. 2015;1:31-35. [Starikov E.S. Suchkova L.I., Koreshkov Yu.N., Koreshkova A.A. Universal approach to the description and use of expert information in monitoring systems based on the patterns of behavior. Polzunov‘s Anthology. 2015;1:31-35. (In Russ.)]

9. Suchkova L.I. Algoritmicheskoye obespechenie monitoringa neshtatnykh sostoyaniy obekta kontrolya na osnove mnogomernykh patternov. Izvestiya AGU. 2013;1/2:118-122. [Suchkova L.I. Algorithmic support of the improper control object states on the basis of multidimensional patterns. News of ASU. 2013;1/2:118-122. (In Russ.)]

10. Suchkova L.I. Chumakov I.A., Yakunin A.G. Identifikatsiya vozdeystviy v priborakh okhrany uprezhdayushchego tipa: monografiya. Doitschland, Saarbrücken, Palmarium Academic Publishing; 2013. 181 s. [Suchkova L.I. Chumakov I.A., Yakunin A.G. Identification of influences in the devices of protection anticipating type: monograph. Doitschland, Saarbrücken, Palmarium Academic Publishing; 2013. 181 p. (In Russ.)]