Application of principal component analysis for the formation of an elecric load mathematical model on railway transport

A.A. Komyakov

Abstract


The article is devoted to the approaches to the feature selection during the formation of an electric load mathematical model on railway transport. Among the known approaches to solving this problem the following methods adopted for the study: forward greedy algorithm, backward greedy algorithm, principal component analysis. The object for study – service locomotive depot of the South-Ural Railway, and the influencing factors – the volume of production activity and climatic factors. As a result of applying the method of principal components, two new feature is formed, which is justified by the Kaiser and scree criteria. The accuracy of the electric load regression model was estimated based on the average relative error, mean square error and the coefficient of variation for the test sample. Simulation results have shown that the accuracy of the model based on the principal component analysis was better than for the other methods. This allows us to recommend the principal component analysis for use in the simulation of electric load on railway transport.


Full Text:

PDF

References


Gapanovich V.A. Energy Strategy of JSC “Russian Railways”. Progress in the Implementation, Updating of the Basic Parameters of Investment and Innovation Priorities [Energeticheskaya strategiya OAO “RZD”. Khod realizatsii, aktualizatsiya osnovnykh parametrov, investitsionnykh i innovatsionnykh prioritetov], Zheleznodorozhnyy Transport [Railway Transport], 2007, no.7, pp. 1-20. (in Russ.)

Komyakov A.A. Analysis of Factors Affecting the Electrical Energy Consumption for Railway Transport [Analiz faktorov, vliyayushchikh na raskhod elektricheskoy energii na netyagovye nuzhdy strukturnykh podrazdeleniy zheleznodorozhnogo transporta], Nauchnye Problemy Transporta Sibiri i Dal'nego Vostoka [Scientific Problems of Transportation in Siberia and the Far East], 2009, no.1, pp. 151-153. (in Russ.)

Abdurahmanov A.M., Volodin M.V., Zybin E.Yu., Ryabchenko V.N. Forecasting methods in electricity distribution networks (review), Russian Internet Journal of Electrical Engineering, 2016, vol.3, no.1, pp. 3-23. (in Russ.). DOI: 10.24892/RIJEE/20160101

Yanovskiy L.P., Bukhovets A.G. Vvedenie v ekonometriku [Introduction to Econometrics], Moscow, KNORUS, 2007, 256 p. (in Russ.)

Ayvazyan S.A., Bukhshtaber V.M., Enyukov I.S., Meshalkin L.D. Prikladnaya statistika. Klassifikatsiya i snizhenie razmernosti [Classification and reduction of dimension], Moscow, Finansy i statistika, 1989, 607 p. (in Russ.)

Vorontsov K.V. Otsenivanie modeley i otbor priznakov [Evaluation models and feature selection]. Available at: http://www.MachineLearning.ru/wiki (accessed 7 September 2016). (in Russ.)

Borisova T.S. Metody vybora optimal'nogo nabora informativnykh priznakov dlya zadach klassifikatsii tekstov [Methods of selecting the optimal set of features for text classification tasks]. Available at: http://www.myshared.ru/slide/357026 (accessed 7 September 2016). (in Russ.)

Sklyar D., Trakhtenberg A. PHP. Retsepty programmirovaniya [PHP. Recipes programming], Saint-Petersburg, Piter, 2015, 784 p. (in Russ.)

Zhvalevskiy O.V. Selection of informative features: formulation of the problem and methods of its solution [Otbor informativnykh priznakov: postanovka zadachi i metodika ee resheniya], Trudy SPIIRAN [Scientific works of SPIIRAN], Saint-Petersburg, Nauka, 2007, Vol. 4, pp. 416-426. (in Russ.)

Gladkov L.A., Kureychik V.V., Kureychik V.M. Geneticheskie algoritmy [Genetic algorithms], Moscow, Fizmatlit, 2010, 317 p. (in Russ.)

Matveev M.G., Sviridov A.S., Aleynikova N.A. Modeli i metody iskusstvennogo intellekta. Primenenie v ekonomike [Models and methods of artificial intelligence. Application in the economy: the manual], Moscow, Finansy i statistika, 2011, 448 p. (in Russ.)

Volkova S.S., Sergienko R.B. Selection of features in nonparametric estimation of regression using genetic algorithms [Otbor informativnykh priznakov v neparametricheskoy otsenke regressii s ispol'zovaniem geneticheskikh algoritmov], Vestnik Tomskogo gosudarstvennogo universiteta [Bulletin of Tomsk State University], 2013, no.1(22), pp. 40-48. (in Russ.)

Gaydel A.V. Metody soglasovannogo otbora priznakov dlya klassifikatsii polutonovykh diagnosticheskikh izobrazheniy: diss. … kand. tekhn. nauk [Methods coherent feature selection for classification grayscale diagnostic imaging: dissertation of the candidate of technical sciences], Samara, 2015, 156 p. (in Russ.)

Kohavi R., John G. Wrappers for feature selection, Artificial Intelligence, 1997, 97(1-2), pp. 273-324. DOI: 10.1016/S0004-3702(97)00043-X

Manusov V.Z. Daily electrical load forecasting in power supply systems on the basis of fuzzy neural networks, Science and Technology, 2004, vol.1, pp. 197-201. DOI: 10.1109/KORUS.2004.1555318

Mestekemper Т., Kauermann G., Smith M. A comparison of periodic autoregressive and dynamic factor models in intraday energy demand forecasting, International Journal of Forecasting, 2013, vol.29, is.1, pp. 1-12. DOI:10.1016/j.ijforecast.2012.03.003.

Almeshaiei Eisa, Soltan Hassan, A methodology for Electric Power Load Forecasting, Alexandria Engineering Journal, 2011, vol.50, is.2, pp. 137-144. DOI:10.1016/j.aej.2011.01.015.

Vedernikova O.G. Development of a genetic algorithm for transportation problem [Razrabotka geneticheskogo algoritma dlya transportnykh zadach], Transport Aziatsko-Tikhookeanskogo regiona [Transport of Pacific Rim], 2015, no.3(5), pp. 11-14. (in Russ.)

Chernyshev Y.O., Vedernikova O.G. A method combining genetic search method and the method of simulated annealing [Sposob kombinirovaniya metoda geneticheskogo poiska i metoda imitatsii otzhiga], Vestnik Rostovskogo gosudarstvennogo universiteta putey soobshcheniya [Bulletin of Rostov State Transport University], 2002, no.1(7), pp. 116-120. (in Russ.)

Bureeva N.N. Mnogomernyy statisticheskiy analiz s ispol'zovaniem PPP “STATISTICA”. Uchebno-metodicheskiy material po programme povysheniya kvalifikatsii “Primenenie programmnykh sredstv v nauchnykh issledovaniyakh i prepodavanii matematiki i mekhaniki” [Multivariate statistical analysis using STATISTICA. Teaching material on the training program “Application of software tools in research and teaching of mathematics and mechanics”], Nizhniy Novgorod, 2007, 112 p. (in Russ.)

Komyakov A.A., Erbes V.V., Ivanchenko V.I. Application of artificial neural networks for electric load forecasting on railway transport, Environment and Electrical Engineering, 2015, pp. 43-46. DOI: 10.1109/EEEIC.2015.7165296. (in Russ.)

Cheremisin V.T., Erbes V.V., Komyakov A.A. Nonparametric statistical approach to evaluating the effectiveness of energy-saving devices, 14th International Conference on Environment and Electrical Engineering, 2014, pp. 58-60. DOI:10.1109/EEEIC.2014.6835836. (in Russ.)

Komyakov A.A., Nikiforov M.M., Erbes V.V., Cheremisin V.T., Ivanchenko V.I., Construction of electricity consumption mathematical models on railway transport used artificial neural network and fuzzy neural network, 16th International Conference on Environment and Electrical Engineering (EEEIC), 2016, Florence, Italy, pp. 1-4. DOI: 10.1109/EEEIC.2016.7555450




DOI: http://dx.doi.org/10.24892/RIJIE/20160305

Refbacks

  • There are currently no refbacks.


Copyright (c) 2016 A.A. Komyakov

© Russian Internet Journal of Industrial Engineering. ISSN 2310-0818

E-mail: indust.engineering.ru@gmail.com

Another version of the web site: http://indust-engineering.ru