Stochastic model of thermal processes in the contact network at arc discharges occurring at high speeds of movement Viktoria V. Litvinovаa, Vladimir V. Moiseevb and Evgeniy V. Runevb a St. Petersburg Electrotechnical University «LETI», ul. Professora Popova 5, Saint Petersburg, 197376, Russia b Emperor Alexander I St. Petersburg State Transport University, 9 Moskovsky pr., Saint Petersburg, 190031, Russia Abstract The paper proposes a stochastic model, on the basis of which estimates are given of the parameters at which extreme situations occur due to the interruption of the electrical contact between the electro-rolling stock current collector (EPS) and a contact wire for the wear and tear of the contact network as a result of acts of arcing. The model takes into account the influence of random factors, which are temporary and sometimes repetitive. The probabilities of deviating from the coordinates of the breakdowns of the contact network from the values given in advance as a result of acts of arcs with defined repeatability periods are obtained. Keywords Weak contact, stochastic model, intensity function, repeatability period, probability asymptotics, characteristic function, multimodal distribution, unimodal distribution 1. Introduction when electric rolling stock moves along high-speed motorways, there are multiple disconnections of the A topical problem encountered in the operation of current conductors from the contact wire - a an electric rolling stock is the reduction of the wear violation of the mechanical and electrical contact, and tear of the contact network and the extension of which result in high-potential arc its useful life under the influence of electric arc discharges. Repeated acts of discharges result in discharges arising from the breakdown of severe wear on the surface of the overhead wire, mechanical contact1. leading to the breakdown of the electrical contact In the present operating conditions (soft soils, low and, in the worst case, the breakdown of the contact air temperature for most of the year, high humidity, network. icing and insufficient quality of contact suspension It should be noted that the above-mentioned surface treatment), it is not possible to improve the mechanical and electrical contact defects also lead to elasticity of the contact suspension. As a result, the deterioration of the traction equipment of the electric rolling stock [4]. According to the available static data on the Models and Methods for Researching Information overhaul of the contact suspension at the different Systems offices of the October Railway, the frequency of in Transport , December 11–12, 2020, Saint-Petersburg, major repairs for the replacement of the contact Russia suspension is on average from 1,5 to 3 months EMAIL: vlitvinova78@gmail.com (V.V. Litvinova); depending on the season of operation and the flow of moiseev_v_i@list.ru (V.I .Moiseev); jr_2010@mail.ru trains on the main line. (E.V. Runev); ORCID: 0000-0002-0628-0440(V.V. Litvinova); 0000- The paper proposes a stochastic model, which 0003-0558-6242 (V.I. Moiseev);0000-0001-6707- makes it possible to assess the important quantitative 888X(E.V. Runev) characteristics of said extreme situation, which is ©️ 2020 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0). temporary and sometimes repetitive. These CEUR Workshop Proceedings (CEUR- characteristics include the intensity function, the WS.org) repeatability period and the probability of deviation 84 of the disconnect coordinate from the specified value 2.2. Distribution intensity function [1], [3]. The above characteristics make it possible to assess the periods of inter-service service service and properties the probability of wear on the overhead wires, which includes breakages from electric arc discharges, to From (2) for the intensity function follows the p t  dF optimize the time and cost of drip repairs and to expression:  t    (here p t    adjust the repair plan for both the main lines and the 1  F t  dt sections of the road [9]. probability density function of the distribution  ), 2. Problem statement which specifies the following properties: 1. f the current probe position takes a value The key object in the study of extreme situations greater than a given value s , that is, the [5],[6],[7] arising from the breakage of the current distribution function  less than one: carrier from the contact wire is the random value position of the breakpoint on the overhead wire. The P  s  0  F s  1 for any end value   model parameter is the pair  , T , where  –  s , then   t dt , would be divergent, intensity function, аnd T – repeatability period of s the random distribution  . The distances between 1 provided that:  t   O  and the two disconnections of the current collector and t  the contact wire that exceed this are random. The  1  mean of these distances is the repeatability converging if:  t   O 1  ,   0 ; period. Another important parameter for the t  2. from equality   distribution of extremes is the intensity function [1]. Thus, the starting point of a mathematical model  t  1  F t   p t  the differential describing periodic extremes [15], [16] is the pair equation linking the intensity function and the    , T . density function follows: 1 d t  1 dp t   1   . 2.1. Intensity function of the position of  t  dt 2 p t    t  dt the current probe xmo  3. if distribution fashion, then The intensity function is defined as [1],[3]. d Considers the probability that the position of the xmo   2 xmo  . current probe on the overhead wire will be equal to dt or greater than a certain value s : These properties make it easy to construct an P  s  1  F s , intensity function for models with modal (1) distributions. It should be noted that the function of where F  random value distribution function  . the density of the current probe is usually Enter the conditional probability that the unimodal. This is because the probability of position of the current collector’s disconnect on the withdrawal at certain points of the wire is lower contact wire will lie within the interval s, s    , because of mechanical tension and the position of the centre of mass of the portion of the wire considered. provided that its meaning is greater than or equals s , which is expressed by the following formula: 2.3. Period of repeatability of P  s, s       s separation positions P  s, s     s   : P  s  s  If you look at a series of observations in which :   t dt. the position of the current probe is greater than or s equal to , this deviation of the position to the right is (2). an event of interest to us. Let’s determine its The function  is called the intensity function probability P  s   1  F s  for p , and the or fault intensity function. It determines the probability of the opposite event – P  s   1  p . probability of a current probe being removed at a In the experiment, we’ll look at observations at point with a coordinate s more to the right s by regular intervals, and the experiment will stop as the amount величину  . 85 soon as we have an event of interest, namely the the right, with certain repeatability periods. In deviation to the right of the current probe s . addition, the formulae provide valid parameter To interpret the results, consider the random estimates by the maximum likelihood method value X   number of tests up to first to right (i.е. [11],[12],[14] Multimodal distributions should be   s ), it accepts values from set 0,1,2,... . X  is chosen as model distributions:  for partitions located on one block, it is subject to geometric distribution with parameter p sufficient to choose a unimodal distribution ( X  ~ G p ). Properties of this distribution are with density:  known: distribution series P X   k  p1  p  ;  k p t   f t  t0   10,L (8), where t 0  0, L , L – length of block - area; characteristic function  t   p ; expected 1  e 1  p  it  it is sufficient to use linear combinations of value EX   i   0  p , second starting point 1 p functions of the form (8) for breaks occurring on an extended section comprising several EX 2   0  1  p   2  p  allow to find the blocks. p2 2.4. Simulation example: case of one required model parameters [8],[10]. block - fixed length section Define the repeatability period as a mathematical expectation T : EX  [3]. The separation density function in this case It follows from the definition that for a given belongs to a two-parameter family h, t 0  model the repeatability period takes the form: distributions and has the form: T  F s  (3).   p t; h, t0   С t  t0   h  10,L , here t0  L 2 , h – 2 1  F s  the variation of the contact wire from the equilibrium It follows from formula (3) that for the position (can be determined by statistical repeatability period a bottom-up assessment is evaluation). Random density normalization constant fair: T  1.  , specified by the normalization condition:   Standard deviation from repeatability period is  p t dt  1 , p t; h, L  C t  L 2  h  10,L 2 given by 0, L  F s  expression:   : EX 2  E 2 X   , or 12 1  F s  and takes on the importance С  L L  12 h  2 . using formula (2) to accept: The distribution function has the form:   : Т 2  Т  (4). Then the probability of deviating right from the 1 3   F s; h, L   С     s  L 2 3 3 8    L   h  s  ,  set position of the current probe to the observation with the number k or in the room k : 0  s  L.   P X   k  1  1  p   1  Fk s  (5) k Repeatability period in this model: 4s  L 2  L3  12 sh 3 3 If in the capacity of k choose T , for probability (5) there is an expression: T s; h, L   2 (9). L  4s  L 23  12hL  s  3 3 Т  1   P X   Т    1  1  2  (6)  1  Т  Probability of right deviation from a specified   value: Probability asymptotics (6) at large T :  P X   Т  s; h, L    T   has the form: Т 3 3   lim PX   T   1  1  0,63212 . (7)  L  4s  L 2 3  12 hs  (10), e  1  2    T  The formulas (5) and (6) make it possible to  3L L2  4h    estimate the probability of deviation from the   specified position of the current probe detachment to 86 From (11), (12), (13) it follows that when the 4s  L 2  3 3 L  12 sh 3 2 bends are small h : h  0  The repeatability period here T  . will accept the following values at the appropriate L  4s  L 2  12hL  s  3 3 3 2 points on the overhead wire: T1 h, L   1 2 , T2 h, L   1 , T3 h, L   2 . Formulas (9), (10) express the functional dependence The above dependency graphics are shown on of the repeatability period and the probability of the figure 1. right deviation of the current probe from a given Accepted here as L  1200 meters, then L 2  600 position to the observation with the number k or in meters. the room k from that of the s on a wire, here 0  s  L . These functional relationships are complex and cumbersome for numerical estimates, which is particularly important for applications, the type. Given the symmetry in the probabilistic model described by the unimodal distribution (8), limit values were found (9), (10). 2.4.1. Limit value of function T s; h, L when s  0 As a result of the cut-off s  0 repeatability period for the left end of the block: T1h, L  lim T s; h, L . Figure 1: Repeatability period dependent on s 0 contact position on overhead wire. L2 Here T1 h, L   1 Repeatability period limits are tabulated for ease    2 L  12h 2  24h L2 2 (11). of reference 1. Table 1 Repeatability period limit values 2.4.2. Limit value of T s; h, L when Limit value Extreme Limit value sL 2 T s; h, L contact wire when position h 0 For the middle of the block s  L 2 the s  0 T1 0, L   1 2 T1 h, L   1 repeatability period dependent on model parameters 2  24 h L2 will be: T2 h, L   lim T s; h, L   1 (12). T2 h, L   1 sL 2 T2 h, L   1 sL 2 s L0 T3 h, L   2 T3 h, L   2  24h 2.4.3. Limit value of T s; h, L when L2 The asymptotic formulas for the repeatability s L period indicate its limitation to a segment s  0; L , For the right end of block - section а s  L the corresponding to the length of the block - section ( L  1200 meters). This feature of the repeatability repeatability period of the model will be: T3 h, L   lim T s; h, L    2 L2  12h   period makes it possible to predict the frequency of major maintenance activities to replace worn-out s L  L2 (13). parts of the network. It should be noted that in the 24h operation of the network, the optimization of the cost  2 2 L of repairs is important, not only at the cost of the Due to the symmetry of the partitions, the limits work carried out, but also at the cost of the time of the repeatability period on the right and left ends spent. The latter means that it is more advantageous are linked by the ratio: T3 h, L  T1-1h, L . to repair several sections in parallel (in one period) than to replace the overhead wire consecutively (after some time to return the repair crew to the same 87 section). The above-mentioned mode of repair makes Thus, for this unimodal two-parameter model (8), it possible to substantially reduce the cost of idling of  probability limit values P X  Т j h, L when  electrified rolling stock. Thus, the replacement of the h  0  is not dependent on parameter L overhead wire on at least one block - the section distributions of type (8) and have a uniform form for leads to the dysfunction of a fairly long stretch of the the whole family of distributions. Localize the values of the probability function on a segment 0,42;0,56  railway network, resulting in economic losses for the enterprises using the company’s services «Russian Railways» as the main carrier. results in high accuracy forecast of wear periods and major maintenance of contact suspension. 2.5. Limit values for deviation from a specified value 2.6. Severance intensity function Intensity function of the probability of the current Based on the expression (10) for deviation probe being removed at a point with a coordinate s probabilities and repeatability time limits (11), (12), (13), the probability limits are determined from: to the right s by the amount  , introduced in T h , L  paragraph 2.1 for this model is:      P X   Т j h, L  1  1  1  j   1  T j h, L    s; h, L    C s  L 2  h  10, L  2  (12),   In the least case with low bending values h : 1  1  C  s  L 2  L 8  hs 3 3 3    h  0  refer: here 10, L – segment indicator function 0, L . Here PX   Т1h, L  (3  3) 3 ; PX  Т2 h, L 1 2 ; 12 С  L L2  12 h  – is the standard density constant of PX  Т h, L  5 9 . 3 a random distribution  , specified by the These limits for probabilities indicate that the normalization condition:  p t dt  1 , probability of a deviation increases with the 0, L  coordinate of the detachment. On figure chart of deviation probability from  p t; h, L  C t  L 2  h  10,L . 2  coordinates s . Following, on figure 3 is the graph of the intensity function  for the following values of the distributions: h  0,05 meters, L  1200 meters. Figure 3: Intensity function graph Figure 2: Probability of deviation chart along the overhead wire. This relationship reflects the fact that at the end of repeatability, as is the case for most contact networks, the probability of deviation varies on a subset of the segment 0,1 , but it does not accept Analyzing the intensity function and its graph, you can see that it has at least in the middle of the values close to 0 or 1. Probability as a function s – segment 0, L , which is fully consistent with the slowly changing in segment 0, L , i.е. by the length type of probability distribution and the presence of of the wiring function. At the point L 2 probability the latter mode also at a point t 0  L 2 . function is bent, which means increasing the rate of The expression for the intensity function (12) has growth of the function when approaching the right a rather cumbersome and difficult form for analysis. end. 88 Give an asymptotic intensity function in the vicinity conditional distribution P s, s     s of a fashion point t 0  L 2 : probabilities of lead.  t; h, L   24h Note also that in this model the intensity function  L L  12h 2   1 has no property:  t   O  , t   . This is due 576 h 2 t    t  L 2   L2 L2  12h  2 (1 to the fact that it is set in a non-trivial way on a compact segment 0, L , outside which it lasts 0: the  3   24 1  1728 h   t  L 22  breakdowns of the current collector occur in a fixed- 2  L L  12h  L L  12h  3 2 3    length block, so the asymptotics at the large coordinates of the separation points are not   o t  L 2 2  meaningful. 3). The main part of the formula (13) asymptotics should be designated ~ : 2.7. Conclusion ~ t ; h, L   24 h   L L  12 h 2  The task of estimating the probability characteristics of an extreme situation occurring 576 h 2  t  L 2  during the operation of the contact network as a  (14).  L2 L2  12 h  2 result of the detachments of the current collector from the contact wire during the movements of high-  3  speed electric rolling stock was defined and solved,  24 1  1728 h   t  L 22  L L  12 h  2 3 2  L L  12 h  3   as a consequence of electric arc discharges between the overhead wire and the current collector. Graph of the main part of the asymptotics of the Precise formulas and their asymptotic intensity function in the vicinity of the point t 0  L 2 expressions have been found in important extreme in the figure below 4. cases for the probabilities of deviations from a given value, repeatability periods and intensity function for Figure 4: Graph of the main part of the a special type of unimodal distributions, in asymptotics of the intensity function in the accordance with the load distribution over the length vicinity of the point t 0  L 2 of the wire. This makes it possible to assess the most important characteristics and to predict the occurrence of said extreme situation - breakage of the wire as a result of heavy heating with an electric arc. It is equally important to draw up an optimal plan of work for periodic major maintenance, thus minimizing the cost. Note that this type of distribution of possible straps from the overhead wire, although approximate, is for evenly stretched contact suspensions without severe altitude variations and the absence of soft soils and underground floating lakes, It describes the processes of cutting off the current probes with sufficient precision. The work, according to the idea of the authors, ~ t  shows that the Dependency graph analysis  has a natural continuation, where the type of  distribution will be specified according to the main part of the intensity function in the geometrical characteristics of the contact wire, such neighborhood of the mode of the distribution has a as the coordinates of the attachment points, the quadratic view. Minimum value ~ t  is at a point  amount of bending, the curvature, the natural profile. t 0  L 2 . This is fully consistent with the fact that In addition, it is intended to assess the parameters of the working distributions on the basis of statistical the intensity function acts as the density of the data from different offices and company roads «Russian Railways» [17]. 89 3. Acknowledgements The authors express their gratitude to their [8] Kudarov R.S., Kudarov R.S., Kukharenko colleagues at the University of Petersburg for the L.A. About properties of some distributions communication ways of Emperor Alexander I for and possibilities of their applications. System many years of fruitful cooperation, which led to analysis and analytics. 2019. 1 (9). pp. 76-83. the emergence of interesting ideas in approaches [9] Sidak A.A., Kornienko A.A., Glukhov A., to solving various applied problems, particularly Diasamidze S.V. Categorization and relevant in today’s environment, such as Evaluation of Critical Railway Information simulating the reliability and sustainability of Infrastructure Dual Technologies. 2019. 1 traffic support systems and intelligent transport (86). pp. 88-93. systems. [10] Kukharenko L.A. Mathematical We also express our gratitude to the simulation of extreme situations / L.A. departments «Higher Mathematics» and Kukharenko, V.A. Ksenofontov, E.V. Runev «Informatics and Information Security» for the // Problems of mathematical and natural friendly warm atmosphere, constant discussions science training in engineering education. and creative search in solving the emerging Collection of works of the IV International applications problems. Scientific and Methodological Conference. Note the high level of contribution to the Vol. 2. / under Dr. Techn. Sciences, Prof. V. organization and conduct of the seminar A. Khodakovsky. - St. Petersburg, 2017. - C. «Models and Methods for Researching 96-103. - eLIBRARY ID: 32558836. Information Systems in Transport» employees of [11] Xenofontova V.A. About moments and the department «Information and computing momentary characteristics of one extreme systems». limit law. / B.A. Xenophotova, L.A. Kukharenko, E.V. Runev // Systems analysis References and analysts. - 2018. - 1 (6). - pp. 9-14. - eLIBRARY ID: 32762842. [12] Kuharenko L.A. Some properties of [1] E. Gumbel. Extreme value statistics. M.: exponential law and gamma distribution. / World, 1965. L.A. Kuharenko, E.V. Runev // Systems [2] B.Gnedenko. Sur la distribution limite du analysis and analysts. - 2018. - 3(8). - pp. 109- terme maximum d'une serie aleatoiire//Ann. 118. - eLIBRARY ID: 36415642. Math. Statist. 44, 1943. [13] Gubareva T.S. Distribution types when [3] David G. Order-based statistics / Translation estimating extreme floods. / T.C. Gubarev // from English. V.A. Egorova, V.B. Nevorova; Geoecology. Engineering Geology, Podred. V.V. Petrova. - Moscow: Science, Hydrogeology, Geology. - 2010. - 5. - pp. 1979. - 335 p. 446-457. - eLIBRARY ID: 15188509. [4] L.A. Kukharenko. Study of electrical strength [14] Kukharenko L.A. Mathematical of insulation Monolith // Inform electro 30, simulation of extreme situations / L.A. ser. Electrical materials, P. 9, 1975. Kukharenko, V.A. Ksenofontova, E.V. Runev [5] V. G. Degtyave, L. A. Kukharenko, V. A. // Problems of mathematical and natural Khodakovsky. The Stochastic Transport science training in engineering education. System Challenges//SGS Newsletter. Issue 1- Collection of works of the IV International Spb.: 2003. Scientific and Methodological Conference. [6] L.A. Kuharenko, T.S. Moseeenko. About one Vol. 2. / under Dr. Techn. Sciences, Prof. V. model of extreme situations. Materials of the A. Khodakovsky. - St. Petersburg, 2017. - pp. Scientific and Practical Conference Current 96-103. - eLIBRARY ID: 32558836. Problems of the Economy. -Spb.: MBI. 2002. [15] Rusanova E.V., Runev E.V. on the [7] L.A. Kukharenko. About the composition of assessment of the parameters of the model of the distribution of extremums. The works of extreme events caused by floods in the the international scientific methodical Siberian rivers. / In the compendium: Arctic: conference Mathematics in the university. modern approaches Materials of the Kostroma, June 1996. 90 International Scientific and Practical Conference. Tyumen, 2020. - pp. 312 – 316. [16] Kuharenko L.A., Ksenofontova V.A., Runev E.V. Mathematical simulation of extreme situations. - Problems of Mathematical and Natural Science Training in Engineering Education. Collection of works of the IV International Scientific and Methodological Conference. Volume 2. - Spb, 2017, pp. 96-103. [17] V. G. Degtyarev, L. A. Kucharenko, Kudarov R.S., Kudarov R.S. Mathematical justification of the hazard test information system. Systems analysis and analyst. 2020. 1 (12), pp. 117-132 91