Article

Article title INFLUENCE OF MODEL PARAMETERS ON THE ADAPTATION PERIOD OF A SMOOTHING TIME SERIES IN PREDICTING CHANGES IN A SLOWLY VARYING PHYSICAL QUANTITY
Authors S. I. Klevtsov, A. B. Klevtsova
Section SECTION IV. MONITORING AND CONTROL IN TECHNICAL SYSTEMS
Month, Year 03, 2018 @en
Index UDC 681.3.062
DOI
Abstract When using a time series to predict the values of technical object parameters and physical variables, it is necessary to take into account the features of data extraction for forecasting associated with digital signal processing. The sampling step at data capture can be set to small. If the parameter change is insignificant for several consecutive steps, then we can use rows based on multiple exponential smoothing as the base model of the time series. To perform the prediction in the microcontroller in the background, a series with a simple implementation algorithm was chosen, but allows obtaining the result with an acceptable error. However, when predicting slowly varying physical quantities, the series has an adaptation period. The series within the adaptation period is characterized by a high error and is not suitable for forecasting. The task is to shorten the period of adaptation. This task can be solved with the help of setting the time series parameters. The paper presents the results of studies on the influence of tuning parameters on the duration and shape of the adaptation period. As parameters, the smoothing coefficient, the coefficients of the initial representation of the series, and the prediction step were considered. The effect of the series settings on the duration of the adaptation period is estimated. Based on the identified regularities, recommendations on the choice of parameters for setting the series have been determined, which will allow reducing the duration of the adaptation period. In particular, it is determined that the change in the smoothing coefficient does not significantly affect the prediction error at the initial segment. The analysis shows that in order to ensure good accuracy of forecasting with time series, it is necessary to choose a smoothing constant corresponding to the dynamics of the predicted process. To eliminate or reduce the time series adaptation and extend the prediction section, it is necessary to specify the initial values of the approximation coefficients of the original function.

Download PDF

Keywords Time series; model; forecasting; technical parameter; adaptation period; microprocessor; real time.
References 1. Klevtsova A.B. Parametricheskaya zonnaya otsenka sostoyaniya tekhnicheskogo ob"ekta s ispol'zovaniem rezhimnoy karty [Parametric conditioning assessment of the condition of a technical object with the use of a modal map], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2010, No. 5 (106), pp. 107-111.
2. Matuszewski J. Application of clustering methods for recognition of technical objects// Modern Problems of Radio Engineering, Telecommunications and Computer Science (TCSET), 2010 International Conference, 2010, pp. 39-40.
3. Klevtsova A.B., Klevtsov G.S. Modeli parametricheskoy ekspress-otsenki sostoyaniya tekhnicheskogo ob"ekta [Models of parametric rapid assessment of the state of a technical object], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2008, No. 11 (88), pp. 15-19.
4. Detlev W. Gross “Partial Discharge Measurement and Monitoring on Rotating Machines”, IEEE Int. Sym. On Elect. Insul, Boston MAUSA, April 7-10, 2002, pp. 33-41.
5. Yaroshenko I.V. Matematicheskaya model' i metod klassifikatsii tekhnicheskogo sostoyaniya vysokovol'tnykh mekhatronnykh moduley [Mathematical model and method of classification of technical condition of high-voltage mechatronic modules], Inzhenernyy vestnik Dona [Engineering journal of Don], 2014, No. 2. Available at: ivdon.ru/ru/magazine/archive/n2y2014/2330.
6. Evtikhiev N.N., Karp V.P., Pudova N.V. Intellektual'nye sistemy podderzhki prinyatiya resheniy i optimizatsii upravleniya v slozhno organizovannykh dinamicheskikh ob"ektakh [Intelligent decision support and management optimization systems in complex dynamic objects], Pribory i sistemy upravleniya [Devices and control systems], 1996, No. 3, pp. 35-40.
7. Lihua Sun, Yingjun Guo, Haichao Ran. A New Method of Early Real-Time Fault Diagnosis for Technical Process, Electrical and Control Engineering (ICECE), 2010 International Conference, 2010. Wuhan, China, pp. 4912-4915.
8. Vovk S.P., Ginis L.А. Modelling and forecasting of transitions between levels of hierarchies in Difficult formalized systems, European Researcher, 2012, Vol. (20), No. 5-1, pp. 541-545.
9. Klevtsov S.I., Klevtsova A.B., Burinov S.V. Model' parametricheskoy kachestvennoy ierarkhicheskoy otsenki sostoyaniya tekhnicheskoy sistemy [A parametric hierarchical model for quality assessment technical systems], Inzhenernyy vestnik Dona [Engineering journal of Don], 2015, No. 3. Available at: ivdon.ru/ru/magazine/archive/n3y2015/3088.
10. Klevtsov S.I. Modelirovanie algoritma kratkosrochnogo prognozirovaniya izmeneniya bystromenyayushcheysya fizicheskoy velichiny v real'nom vremeni [Modeling of short-term forecasting algorithm of changes in the rapidly changing physical quantity in real time], Inzhenernyy vestnik Dona [Engineering journal of Don], 2012, No. 3 (21), pp. 199-205.
11. Klevtsov S.I. Otslezhivanie izmeneniya sostoyaniya dinamicheskogo ob"ekta v real'nom vremeni s ispol'zovaniem mikroprotsessornogo modulya [Monitoring of changes in the state of a dynamic object in real time using a microprocessor module], Vserossiyskaya nauchno-tekhnicheskaya konferentsiya «Problemy razrabotki perspektivnykh mikro- i nanoelektronnykh sistem (MES)»: Sb. trudov [all-Russian scientific and technical conference "Problems of development of advanced micro- and nanoelectronic systems (MES)": proceedings], 2012, No. 1, pp. 684-687.
12. Klevtsov S.I. Predvaritel'naya otsenka sostoyaniya sovokupnosti parametrov tekhnicheskogo ob"ekta s ispol'zovaniem intellektual'nogo mikroprotsessornogo modulya [Preliminary assessment of the state of a set of parameters of a technical object using an intelligent microprocessor module], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2010, No. 5 (106), pp. 43-48.
13. Klevtsov S.I. Prognozirovanie izmeneniy fizicheskoy velichiny v real'nom vremeni s ispol'zovaniem lineynogo adaptivnogo fil'tra [The prediction of changes of physical quantities in real time using a linear adaptive filter], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2013, No. 5 (142), pp. 180-185.
14. Peter J. Brockwell, Richard A. Davis ITSM: An Interactive Time Series Modelling Package for the PC. Springer New York, 1991, 105 p.
15. Sidorov S.G., Nikologorskaya A.V. Analiz vremennykh ryadov kak metod postroeniya prognoza potrebleniya elektroenergii [Analysis of time series as a method of forecasting electricity consumption], Vestnik IGEU [Vestnik IGEU], 2010, Issue 3, pp. 1-3.
16. Klevtsov S.I. Identification of the State of Technical Objects Based on Analyzing a Limited Set of Parameters, 2016 International Siberian Conference on Control and Communications, SIBCON 2016 – Proceedings, 2016, pp. 749-752.
17. Darkhovsky B., Piratinska A. Novel Methodology for Segmentation of Time Series Generated by Different Mechanisms, Proceedings of International work-conference on Time Series (ITISE-2014). Iss. 1. Granada: Copicentro Granada S.L., 2014, pp. 273-285.
18. Darkhovsky B., Brodsky B. Asymptotically Optimal Methods of Early Change-point Detection, Sequential Analysis, 2013, No. 32, pp. 158-181.
19. Sergienko A.B. Tsifrovaya obrabotka signalov [Digital signal processing]. Saint Petersburg: Piter, 2002, 608 p.
20. Boyle C. Mastering Statistics with your Microcomputer. Macmillan Master Series. Macmillan Education UK .1986, 155 p.
21. Klevtsov S.I. Osobennosti vybora parametrov nastroyki modeli sglazhivayushchego vremennogo ryada dlya osushchestvleniya kratkosrochnogo prognozirovaniya izmeneniya fizicheskoy velichiny [Features of choice of parameters of adjustment of model of the smoothing time series for implementation of short-term forecasting of change of physical quantity], Izvestiya YuFU. Tekhnicheskie nauki [Izvestiya SFedU. Engineering Sciences], 2011, No. 5 (118), pp. 133-138.
22. Lukashin Yu.P. Adaptivnye metody kratkosrochnogo prognozirovaniya vremennykh ryadov [Adaptive methods of short-term time series forecasting]. Moscow: Finansy i statistika, 2003, 416 p.
23. Brillindzher D.R. Vremennye ryady. Obrabotka dannykh i teoriya: monografiya [Time series. Data processing and theory: monograph], ed. by A.N. Kolmogorova: translation from English. Moscow, 1980, 536.
24. George E.P. Box, Gwilym M. Jenkins, Gregory C. Reinsel. Time series analysis: forecasting and control. 4th ed. A JOHN WILEY & SONS, INC., PUBLICATION, 2015, 712 p.

Comments are closed.