ISSUE 180, article 3

Posted on

by admin

DOI:https://doi.org/10.15407/kvt180.02.025

Kibern. vyčisl. teh., 2015, Issue 179, pp 25-33.

Savchenko Evgenia A., PhD (Engineering), Senior Researcher of Department of Information Technologies of inductive modeling, International Research and Training Center for Information Technologies and Systems of National Academy of Sciences of Ukraine and Ministry of Education and Science of Ukraine, av. Acad. Glushkova, 40, Kiev, 03187, Ukraine, e-mail: savchenko_e@meta.ua

TECHNOLOGY FOR SOLVING THE PROBLEM OF MODELING AND FORECASTING BASED ON INDUCTIVE APPROACH

Introduction. The advantage of inductive algorithms is in their ability to automatically find dependencies hidden in a sample of experimental data. Combinatorial algorithms of GMDH (group method of data handling) are the main inductive modeling algorithms. These algorithms applied to real problems showed that it’s not always possible to unambiguously determine a model by one criterion. Method of a model after-determination based on the Combinatorial GMDH algorithm is developed for such case. A technology based on the combinatorial GMDH algorithm and the after-determination method was developed for the modeling and forecasting.

The purpose of this article is to develop the methodology and technology for modeling and forecasting on the experimental data sample based on the combinatorial algorithm GMDH method and the after-determination method. They will help to find the optimal model in real applications.

Results. A technology for solving the problem of modeling and forecasting on the basis of the inductive approach was developed and described. This approach is based on the combinatorial algorithm GMDH method and completions. This technology, based on a sample of experimental data, automatically finds the object model or process using two external selection criteria: accuracy and bias. The developed computer technology was tested in solving applied problems of modeling and prediction: in problems of diabetes in a home-based monitoring and problems of modeling the interaction of ions with the surface of the jet gas materials.

Conclusion. A computer technology that provides an effective solution for the problems of modeling and prediction of the experimental data was developed and described. Numerical examples demonstrate its efficiency. This technology provides increased noise immunity models due to the consistent application of external criteria GMDH: the criterion of regularity and bias. This technology was used in real applications for modeling and forecasting and its effectiveness has been confirmed.

Keywords: inductive approach, combinatorial algorithm of group method of data handling, modeling, forecasting, technology.

Download full text (ru)!

References

  1. Ivakhnenko A.G. Noise–immunity of modeling. Kiev: Naukova Dumka, 1985, 215 p.(in Russian).
  2. Madala H.R. Inductive Learning Algorithms for Complex Systems Modeling. Boca Raton: CRC Press Inc., 1994, 384 p.
  3. Ivakhnenko A.G. Group method of data handling as competitor to the method of stochastic approximation. Automatics, 1968, no. 3, pp. 64–78 (In Russian).
  4. Stepashko V.S. Combinatorial GMDH algorithm with optimal scheme of models sorting. Automatics, 1981, no. 3, pp. 31–36 (In Russian).
  5. Ivakhnenko A.G. Investigation of Efficiency of Additional Determination Method of the Model Selection in the Modeling Problems by Application of the GMDH Algorithm. Journal of Automation and Information sciences. Begell House: Inc. Publishers, 2008, vol. 40, no. 3, pp. 47–58.
  6. Savchenko E.A. Analytical and Numerical Study of the Selective Properties of the Errors Unbiasedness Criterion in the Problems of Inductive Modeling. Journal of Automation and Information sciences. Begell House: Inc. Publishers, 2012, vol. 44, no. 4, pp. 1–12.
  7. Ivahnenko A.G. Conception of the successive algorithmic approaching (lowering) to the exact decision of interpolation tasks of artificial intelligence. Cybernetics and computing engineering, 1999, vol. 124, pp. 40–60 (In Russian).
  8. Savchenko E.A. Preprocessing of data sample in inductive modeling problem. Control Systems and Computers, 2015, N2, pp. 82–87.
  9. Ivakhnenko A.G. Application of Algorithms of the Method of Batch Assessment of Arguments for Recovering Missed Data and Prediction of the Glucose Level in Blood on at Home Diabetes Monitoring. Journal of Automation and Information sciences. Begell House: Inc. Publishers, 2002, vol. 34, no. 6, pp.123–133.
  10. Savchenko E.А. Analysis of the selective properties of the GMDH criteria when applying them consistently. Modeling and control as ecological and economic systems of the region. Kiev: IRTC ITS, 2008, no. 4, pp. 199–210 (In Russian).

Received 25.03.2015