Article

Article title HYBRID EXPERT SYSTEM BASED ON PROBABILISTIC AND DETERMINISTIC MODELS
Authors N.A. Polkovnikova
Section SECTION IV. ARTIFICIAL INTELLIGENCE AND FUZZY SYSTEMS
Month, Year 06, 2015 @en
Index UDC 004.891
DOI
Abstract Development and implementation of intelligent hybrid expert systems is an important way in increasing reliability and operation efficiency of complex technical objects and allows applying monitoring results and automated solutions for operational tasks and information support of the operator. The concept of integrated use complex technical object with expert system improves the efficiency of monitoring and artificial competency of decision maker for solving operational problems and management. The development of knowledge base for hybrid expert system with calculation of reference and current stochastic models for the control parameters are considered as well as algorithms for automated intelligent information processing based on systematic approach to support the operator in making right decisions in operation of complex technical object on the example of ship"s main engine. To implement the diagnostic and predictive procedures in expert system developed: models, algorithms and software modules that enable at qualitatively new level use optimization methods and actual technical condition reserves to prevent failures and increase overhaul period. To formalize procedures of data evolution in the knowledge base of expert system developed method using stochastic models of different types and levels: standard and current (adaptive), local and integral, for which required database of measured parameters, software and models database. Developed model and algorithm of multiobjective optimization based on evolutionary algorithms in order to support the operator in choice of the main marine engine operation mode. The algorithm goal is to determine values of fuel supply parameters in order to minimize two objective functions: the content of nitrogen oxides in exhaust gases and specific fuel oil consumption.

Download PDF

Keywords Hybrid expert system; control object; decision maker; diagnostic and predictive procedures; stochastic models; multivariate polynomial regression; adaptation; evolutionary algorithm; multiobjective optimization; front Pareto.
References 1. Shishkin V.A. Analiz neispravnostey i predotvrashchenie povrezhdeniy sudovykh dizeley [Failure analysis and damage prevention of marine diesel engines]. Moscow: Transport, 1986, 192 p.
2. Polkovnikova N.A., Kureychik V.M., Polkovnikov A.K. Raschet stokhasticheskikh modeley dlya ekspertnoy sistemy. Programma dlya EVM. Svidetel'stvo o gos. registratsii No. 2014619228 ot 11.09.2014 [The calculation of stochastic models for expert systems. The PC program. The certificate of state registration No. 2014619228 from 11.09.2014.]. Publ. 20.10.2014. Official bulletin RU OBPBT, No. 10 (96), 2014.
3. Polkovnikova N.A., Kureychik V.M., Polkovnikov A.K. Ekspertnaya sistema dlya diagnostiki sudovykh dizeley «Dizel' ekspert». Programma dlya EVM. Svidetel'stvo o gos. registratsii № 2012615935 ot 28.06.2012 [Expert system for diagnosis of marine diesel engines "Diesel expert". The PC program. Certificate of state registration № 2012615935 dated 28.06.2012]. Publ. 20.09.2012. Official bulletin RU OBPBT, 2012, No. 3 (80), pp. 683-684.
4. Efremov L.V. Praktika veroyatnostnogo analiza nadezhnosti tekhniki s primeneniem komp'yuternykh tekhnologiy [Practice of probabilistic reliability analysis technique using computer technology]. St. Petersburg: Nauka, 2008, 216 p.
5. Ivanovskiy R.I. Teoriya veroyatnostey i matematicheskaya statistika. Osnovy, prikladnye aspekty s primerami i zadachami v srede Mathcad [Probability theory and mathematical statistics. Fundamentals, practical aspects, with examples and problems in Mathcad]. St. Petersburg: BKhV-Peterburg, 2008, 528 p.
6. Makarov I.M., Lokhin V.M., Man'ko S.V., Romanov M.P. Iskusstvennyy intellekt i intellektual'nye sistemy upravleniya [Artificial intelligence and intelligent control systems]. Moscow: Nauka, 2006, 333 p.
7. Ponomarev I.V., Slavskiy V.V. O geometricheskoy interpretatsii metoda naimen'shikh kvadratov [Geometric interpretation of least squares method], Izvestiya Altayskogo gosudarstvennogo universiteta [the News of Altai State University], 2012, No. 1 (73), pp. 119-121.
8. Nosko V.P. Ekonometrika. Vvedenie v regressionnyy analiz vremennykh ryadov [Econometrics. Introduction to regression analysis of time series]. Moscow: NFPK, 2002, 270 p.
9. Romanov V.N. Sistemnyy analiz dlya inzhenerov [System analysis for engineers]. St. Petersburg: SZGZTU, 2006, 186 p.
10. Shashkov V.B. Prikladnoy regressionnyy analiz. Mnogofaktornaya regressiya: Uchebnoe posobie [Applied regression analysis. Multivariate regression: a tutorial]. Orenburg: GOU VPO OGU, 2003, 363 p.
11. Vorob'ev S.N., Osipov L.A. Regressionnyy analiz: Ucheb.-metod. Posobie [Regression analysis: textbook]. St. Petersburg: SPbGUAP, 2000, 66 p.
12. Guzik V.F., Kidalov V.I., Samoylenko A.P. Statisticheskaya diagnostika neravnovesnykh ob"ektov [Diagnostics of non-equilibrium statistical objects]. St. Petersburg: Sudostroenie, 2009, 304 p.
13. Kureychik V.V., Kureychik V.M., Rodzin S.I. Teoriya evolyutsionnykh vychisleniy [The theory of evolutionary computation]. Moscow:Fizmatlit, 2012, 260 p.
14. Kenneth A. De Jong. Evolutionary computation. A unified approach. The MIT Press, 2006, 256 p.
15. Neumann F., Witt C. Bioinspired computation in combinatorial optimization. Springer, 2010, 216 p.
16. Ashlock D. Evolutionary computation for modeling and optimization. Springer, 2006, 571 p.
17. Brockhoff D., Zitzler E. Objective reduction in evolutionary multiobjective optimization: theory and applications, Evolutionary Computation, 2009, No. 17 (2), pp. 135-166.
18. Siegfried T., Bleuler S., Laumanns M., Zitzler E., Kinzelbach W. Multi-Objective Groundwater Management Using Evolutionary Algorithms, IEEE Transactions on Evolutionary Computation, 2009, No. 13 (2), pp. 229-242.
19. Xiao-Bing Hu, M. Wang, E. Di Paolo. Calculating complete and exact Pareto front for multiobjective optimization: a new deterministic approach for discrete problems, IEEE Transactions on cybernetics, 2013, Vol. 43, No. 3, pp. 1088-1101.
20. Hohm T., Zitzler E. A Multiobjective evolutionary algorithm for numerical parameter space characterization of reaction diffusion systems, International Conference on Pattern Recognition in Bioinformatics (PRIB 2009), Heidelberg, Germany, 2009. Springer, pp. 162-174.

Comments are closed.