Construction of mathematical models to predict M25 and M10 coke quality indices

Introduction. Mathematical models are developed to predict M25 and M10 coke quality. The calculations are carried out specifically for each of the coke batteries of the coke-chemical production (CCP) of Magnitogorsk Iron & Steel Works (MMK). The simulation is based on the charge factors: sum of inert components OK, %; vitrinite reflectance R0, %. These models are required to control the product quality and the optimization aimed at cost saving. The study objective is to construct adequate mathematical models for predicting М25 and М10 coke quality indices under the conditions of MMK CCP. Thus, it is assumed that MMK will obtain its own models that are highly competitive in forecast precision with the analogues used by other coke-chemical enterprises ofRussia.

Materials and Methods. Neural networks are used as a universal approximation for constructing mathematical models. When selecting their architecture, the authors emanated from the minimum number of neurons and the network layers. In addition, the minimization of the predictive error on a new sample was taken into account which was not used in training and testing.

Research Results. The development is based on the petrographic charge factors: sum of inert components OK (according to GOST 12112); vitrinite reflectance R0, (GOST 12113). With the help of artificial neural networks, one-dimensional mathematical models are constructed to predict impact coke strength indices of М25 and abrasion capacity of М10 (GOST 5953). The developed models are presented in a graphical form. Their predictive force is estimated.

Discussion and Conclusions. In the models developed within the framework of this study, only petrographic charge factors are used. The aggregate data on technical and plastometric analysis are not taken into account. This is the basic difference of the approach presented in this paper from the models implemented for other CCP, for example, at Nizhny Tagil Iron & Steel Works (NTMK), Novokuznetsk Iron & Steel Works (NKMK), and West-Siberian Iron & Steel Works (ZSMK). Even so, the adequacy of the obtained dependences is proved. It is proposed to use them to optimize the petrographic charge factors by various optimality criteria of the coke quality indices.

1. Kiselev, B.P., Lapshin, V.A. Syr'yevaya baza Rossii. 1. Retrospektiva. [Raw materials base of Russia. 1. Retrospective.] Coke and Chemistry, 1999, no. 11, pp. 2–9 (in Russian).

2. Kiselev, B.P. O nekotorykh aspektakh ugol'noy bazy koksovaniya Rossii v 2008–2009 godakh. [On some aspects of coal coking base of Russia in 2008-2009.] Coke and Chemistry, 2009, no. 12, pp. 9–15 (in Russian).

3. Stankevich, A.S., et al. Optimizatsiya sostava shikhty dlya koksovaniya i prognoz kachestva koksa po khimiko-petrograficheskim parametram. [Optimization of charge composition for coking and coke quality forecast on chemical-petrographic parameters.] Coke and Chemistry, 1998, no. 9, pp. 11–17 (in Russian).

4. Yevtushenko, V.F., et al. Postroenie i issledovanie algoritma prognozirovaniya pokazateley kachestva koksa s uchetom dinamiki nekontroliruemykh vozmushcheniy. [Construction and research of algorithm for predicting coke quality indicators taking into account of uncontrolled disturbances dynamics.] Coke and Chemistry, 1996, no. 6, pp. 21–26 (in Russian).

5. Smirnov, A.N., Petukhov, V.N., Alekseev, D.I. Analiz printsipov postroeniya matematicheskikh modeley dlya prognozirovaniya pokazateley kachestva koksa M25 i M10 s tsel'yu klassifikatsii i razrabotki kontseptsii gibridnoy modeli. [Classification of models for predicting coke quality (M25 and M10).] Coke and Chemistry, 2015, no. 5, pp. 13–18 (in Russian).

6. Stankevich, A.S., Myuller, I.P., Lelchuk, V.I. Raspredelenie ugley i sostavlenie ugol'nykh shikht dlya koksovaniya s prognozom kachestva koksa na osnove lineynogo programmirovaniya. [Distribution of coals and coal blending for coking with coke quality forecast based on linear programming.] Coke and Chemistry, 1981, no. 11, pp. 4– 8 (in Russian).

7. Stankevich, A.S., et al. Model' optimizatsii pokazateley prochnosti koksa na osnove khimikopetrograficheskikh parametrov ugley i nelineynogo programmirovaniya. [Model of optimization of coke strength indicators on the basis of coal chemical-petrographic parameters and non-linear programming.] Coke and Chemistry, 2000, no. 5, pp. 21–29 (in Russian).

8. Stankevich, A.S., et al. Sostavlenie shikht dlya koksovaniya na osnove optimizatsii i prognoza prochnosti koksa po khimiko-petrograficheskim pokazatelyam ugley. [Composition of coal charge materials and prediction of coke strength.] Coke and Chemistry, 2002, no. 3, pp. 9–15 (in Russian).

9. Bazegsky, A.E., et al. Model' prognoza kachestvennykh kharakteristik metallurgicheskogo koksa na osnove khimiko-petrograficheskikh pokazateley ugol'noy shikhty. [Model for predicting the characteristics of metallurgical coke.] Coke and Chemistry, 2002, no. 9, pp. 15–22 (in Russian).

10. Stankevich, A.S., et al. Ratsional'noe raspredelenie ugley i optimizatsiya sostava shikht dlya koksovaniya. [Rational distribution of coals and optimization of charge materials for coking.] Coke and Chemistry, 2003, no. 9, pp. 8–16 (in Russian).

11. Stankevich, A.S., et al. Prognoz prochnosti koksa na osnove khimiko-petrograficheskikh parametrov ugol'nykh shikht s uchetom ikh pnevmomekhanicheskoy separatsii. [Forecast of coke strength on the basis of chemicalpetrographic parameters of coal blends taking into account their pneumomechanical separation.] Coke and Chemistry, 2005, no. 12, pp. 14–21 (in Russian).

12. Stankevich, A.S., et al. Model' prognoza pokazateley CSR i CRI na osnove khimiko-petrograficheskikh parametrov ugol'nykh shikht i usloviy ikh koksovaniya. [Forecast model of CSR and CRI indices on the basis of chemical-petrographic parameters of coal charges and coking conditions.] Coke and Chemistry, 2008, no. 9, pp. 37–44 (in Russian).

13. Stankevich, A.S., Bazegsky, A.E. Optimizatsiya kachestva koksa OAO «EVRAZ ZSMK» s uchetom osobennostey ugol'noy syr'yevoy bazy. [Optimizing coal production at OAO EVRAZ ZSMK on the basis of the available coal.] Coke and Chemistry, 2013, no. 10, pp. 14–21 (in Russian).

14. Vukolov, E.A. Osnovy statisticheskogo analiza. Praktikum po statisticheskim metodam i issledovaniyu operatsiy s ispol'zovaniem paketov STATISTICA i EXCEL. [Fundamentals of statistical analysis. Workshop on statistical methods and investigation of operations using STATISTICA and EXCEL packages.] 2nd revised and enlarged ed. Moscow: Forum. 2008, 464 p. (in Russian).

15. Osovskiy, S. Neyronnye seti dlya obrabotki informatsii. [Neural networks for information processing.] Moscow: Finansy i statistika, 2002, 344 p. (in Russian).

16. Ventsel, E.S. Teoriya veroyatnostey. [Theory of probability.] 10th reimpression. Moscow: Akademiya, 2005, 576 p. (in Russian).

17. Smirnov, A.N., Alekseev, D.I. Sopostavlenie i analiz adekvatnosti matematicheskikh modeley dlya prognozirovaniya pokazateley kachestva koksa M25 i М10. [Comparison and adequacy analysis of the mathematical models for predicting the M25 and М10 coke quality indices.] Vestnik of Nosov Magnitogorsk State Technical University, 2017, vol. 15, no. 3, pp. 62–67 (in Russian).