CN103838929A - Turning repair decision optimization method for rail transit vehicle wheel sets - Google Patents

Abstract

The invention relates to a turning repair decision optimization method for rail transit vehicle wheel sets. The turning repair decision optimization method includes: subjecting eight wheel of one carriage to wheel set abrasion data preprocessing; establishing wheel set abrasion data driving models such as a wheel diameter abrasion model, a rim thickness abrasion model and a wheel set turning repair proportion coefficient distribution model; performing Monte Carlo simulation on rail transit vehicle wheel set abrasion and turning repair to acquire expected wheel set service life under the combination of different rim thickness preventing turning repair value and turning repair recovery value. By means of a genetic algorithm in the simulation, the expected rim thickness value and wheel diameter value after one specific turning repair are optimized with the optimization objectives that the wheel diameter loss is as low as possible and is possibly close to the rim thickness recovery value after turning repair; optimal turning repair decisions, namely the optimized rim thickness preventing turning repair value and turning repair recovery value, can be finally analyzed according to the expected simulated service life of the wheel sets.

Description

The right Xuan of a kind of rail traffic locomotive vehicle wheel repaiies decision optimization method

Technical field

The Xuan that it is right that patent of the present invention relates to a kind of rail traffic locomotive vehicle wheel repaiies decision optimization method, and particularly a kind of rail traffic locomotive vehicle wheel based on taking turns abrasion data-driven model and Monte Carlo simulation is repaiied decision optimization method to Xuan.

Background technology

Enter 21 century, urban rail transit in China (being commonly called as subway) and Railway System are stepped into Rapid development stage.Along with the fast development of railway and urban track traffic, must higher requirement be proposed to the maintenance of the relevant device such as train and rail and life-span management.Wheel is to being called as one of railcar three large attrition components, the tread diameter that its tread causes with wheeling edge wearing transfinites, flange thickness transfinites fault and the improper faults such as the wheel flat, tread crackle, shelled tread that produce that contact of wheel track, and travel safety to railcar, riding comfort and rail have material impact serviceable life.Because wheel is in use constantly abrasion, flange thickness and wheel footpath (being tread diameter) can constantly diminish, and there is corresponding bound regulation in flange thickness and wheel footpath.In taking turns use: must carry out Xuan timely to it when flange thickness transfinites and repair, thereby recover flange thickness, but this Xuan process of repairing must be taken turns footpath as cost taking loss part; If transfinite in wheel footpath, must take turns changing.In addition, eight wheels in same joint compartment also must meet the poor requirement in wheel footpath.Therefore, at wheel, the Xuan in use procedure is repaiied to decision problem, be reduced to when to take turns at flange thickness Xuan is repaiied, and make wheel to which type of degree of caliper recovery, to extending wheel to serviceable life, reduce railcar maintenance cost, there is material impact.

Through retrieval, there is no to find about rail traffic locomotive vehicle wheel, Xuan to be repaiied the patent of decision optimization method both at home and abroad.Academia has delivered the relevant achievement in research of minority.Pascual and Marcos (Pascual F, Marcos J A.Wheel wear management on high-speed passenger rail:a common playground for design and maintenance engineering in the Talgo engineering cycle.Proceedings of the2004ASME/IEEE Joint Rail Conference, 193-199, 2004) for the rolling stock wheel of Talgo company of Spain to wear problem, by the wheel to for many years, wearing and tearing measured data is carried out to rough Statistics analysis, think and repair by Xuan again it is returned in 30.5mm situation in the time that wheel reduces to 27.5mm to flange thickness, take turns longer to serviceable life.The researchists such as domestic Wang Ling (Wang Ling, member China, Na Wenbo, Chen Xiai, Li Yuntang, the wheel based on abrasion data-driven model is repaiied policy optimization and residual life forecast, the system engineering theory and practice to Xuan, 31 (6), 1143-1152,2011.; Xu Hong, member China, Wang Ling, Na Wenbo, Xu Wenbin, Li Yuntang, railcar wheel based on Gaussian process is repaiied policy optimization to abrasion modeling and Xuan thereof, mechanical engineering journal, 46 (24), 88-95,2010) for Guangzhou Metro Cars wheel, abrasion and Xuan are repaiied to problem, provided corresponding abrasion model and Xuan and repaiied decision optimization method, but these achievements have just considered that to angle Xuan repaiies decision optimization problem from single wheel, the Xuan that can not directly apply to eight wheels in same joint compartment in solution reality repaiies decision optimization problem.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, eight wheel Xuan for same joint compartment repair decision problem, a kind of rail traffic locomotive vehicle wheel is provided, and right Xuan repaiies decision optimization method, make relevant maintenance support personnel wear away data and the method to history based on wheel, wheel is in the future repaiied to decision-making to Xuan and be optimized, thereby help to extend wheel to serviceable life.

For realizing object of the present invention, right Xuan repaiies decision optimization method to the invention provides a kind of rail traffic locomotive vehicle wheel, comprises following steps:

Step 1: for eight wheels in a joint compartment, take turns abrasion data pre-service, draw the sample data that wheel is repaiied scale-up factor to wheel rim and wheel footpath wear rate and Xuan;

Step 2: set up wheel to abrasion data-driven model, comprise that wheel footpath wears away model, flange thickness abrasion model and wheel Xuan is repaiied to scale-up factor distributed model etc.;

The wheel footpath abrasion model that the present invention adopts is a Weibull distribution of describing wheel footpath wear rate probability density characteristics.The flange thickness that the present invention adopts wears away model, has portrayed the correlated fitting relation of certain moment flange thickness wear rate and corresponding flange thickness value, is expressed as follows:

V _sd＝a×S _d ²+b×S _d+c+E _vsd (1)

Wherein V _sdfor flange thickness wear rate, S _dfor flange thickness value, E _vsdfor flange thickness wear rate random fit difference, E _vsdbe that an average is zero normal distribution random number, a, b and c are constant.It is that the gamma that a description Xuan repaiies scale-up factor k probability density characteristics distributes that the wheel that the present invention adopts is repaiied scale-up factor distributed model to Xuan.

Step 3: carry out rail traffic locomotive vehicle wheel to abrasion and the Monte Carlo simulation repaiied of Xuan, show that different flange thicknesses prevention Xuan values of repairing and Xuan repair the wheel of recovery value under combining to expectation serviceable life;

In this step, to repair wheel under recovery value combination as follows to the step in expectation serviceable life for simulation calculation specific one group of flange thickness prevention Xuan value of repairing and Xuan:

Step 3.1: all kinds of parameters of initialization;

Step 3.2 a: unit interval of simulation time stepping;

Step 3.3: according to Monte Carlo simulation Method And Principle, take turns abrasion data-driven model based on aforementioned, emulation produces wheel footpath abrasion value and the flange thickness abrasion value in the corresponding unit interval, and according to wheel footpath value and the flange thickness value in a upper moment, calculated wheel footpath value and the flange thickness value of current time;

Step 3.4: judge whether wheel footpath value is less than wheel footpath lower limit, if " being less than " enters step 3.5, otherwise carry out step 3.8;

Step 3.5: wheel adds 1 to emulation periodicity;

Step 3.6: judge that wheel that whether wheel be less than setting to emulation periodicity is to emulation total number of cycles, if " being less than " enters step 3.7, otherwise simulation optimization finishes, Output simulation result;

Step 3.7: current each wheel footpath value and each flange thickness value update all are right wheel footpath value and the flange thickness value of brand-new wheel, starts the emulation of next wheel to the emulation cycle, returns to step 3.2;

Step 3.8: judge whether each flange thickness value is less than or equal to the flange thickness prevention Xuan value of repairing, if " being less than or equal to " enters step 3.9, otherwise return to step 3.2;

Step 3.9: optimize a concrete Xuan and repair rear expectation flange thickness value and wheel footpath value, utilize genetic algorithm, considering under the prerequisite of the poor constraint in eight vehicle wheel footpaths, as far as possible little with the loss of wheel footpath, and Xuan repaiies rear rim one-tenth-value thickness 1/10, and to approach as far as possible flange thickness recovery value be optimization aim, flange thickness value and wheel footpath value that this Xuan repaiied to rear expectation are optimized;

In above-mentioned steps of the present invention, corresponding optimization aim " wheel footpath loss is as far as possible little, and Xuan repaiies rear rim one-tenth-value thickness 1/10 and approaches as far as possible flange thickness recovery value " can be expressed as:

$\min [\underset{i=1}{\overset{8}{\mathrm{\Σ}}}{w}_{\mathrm{sd}}\·|S_{d,j,i}^{+}-S_{\mathrm{dR}}|+\underset{k=1}{\overset{8}{\mathrm{\Σ}}}{w}_D\·(D_{j,k}^{-}-D_{j,k}^{+})]---\left(\text{2}\right)$

Wherein

for t _jflange thickness value after moment i wheel Xuan repaiies, S _dRfor Xuan repaiies rear rim caliper recovery value,

for t _jmoment, k wheel Xuan repaiied front-wheel footpath value,

for t _jmoment, k wheel Xuan repaiied trailing wheel footpath value, w _sdand w _dfor being weight coefficient, meet w _sd+ w _d=1; Optimize the poor constraint in wheel footpath, Xuan that constraint comprises eight wheels in same joint compartment and repair the constraint of not transfiniting of rear rim one-tenth-value thickness 1/10 and wheel footpath value, and the flange thickness value of Xuan after repairing before being greater than Xuan and repairing flange thickness value and the wheel footpath value of Xuan after repairing be less than Xuan and repair the Xuan such as front wheel footpath value and repair front and back wheelset profile logical constraint.

Step 3.10: based on wheel, Xuan is repaiied to scale-up factor distributed model, sampling generates Xuan and repaiies scale-up factor, and the Xuan drawing according to the step 3.9 flange thickness value of repairing rear expectation calculates Xuan and repaiies rear rim one-tenth-value thickness 1/10 and wheel footpath value;

Step 3.11: accumulation Xuan repaiies number of times;

Step 3.12: judge whether each footpath value of taking turns is less than wheel footpath lower limit, if " being less than " returned to step 3.5, otherwise return to step 3.2.

Step 4: show that preferably Xuan repaiies decision-making, i.e. comparative analysis different flange thicknesses prevention Xuan value of repairing and Xuan repair the lower wheel of recovery value combination to expecting serviceable life, obtains preferably flange thickness prevention Xuan value of repairing and Xuan and repaiies recovery value and combine.

The Xuan that it is right that the present invention is a kind of rail traffic locomotive vehicle wheel repaiies decision optimization method, this method is the historical data to abrasion according to rail traffic locomotive vehicle wheel, take turns after wearing away data-driven model in data pre-service foundation, carry out rail traffic locomotive vehicle wheel to the Monte Carlo simulation wearing away and Xuan repaiies, thereby show that different flange thicknesses prevention Xuan values of repairing and Xuan repair wheel under the combination of recovery value to expectation serviceable life, the last wheel drawing according to emulation is to expecting result in serviceable life, analysis show that preferably Xuan repaiies decision-making, be that preferred flange thickness prevention Xuan value of repairing and Xuan repair recovery value combination, thereby help relevant maintenance personal to optimize wheel Xuan is repaiied to decision-making, extend wheel to serviceable life.

Brief description of the drawings

Fig. 1 is step schematic diagram of the present invention;

Fig. 2 is that simulation calculation of the present invention specific one group of flange thickness prevention Xuan value of repairing and Xuan repair wheel under recovery value combination to the expectation process flow diagram in serviceable life;

Fig. 3 is that different flange thicknesses prevention Xuan values of repairing in the specific embodiment of the invention and Xuan repair wheel under recovery value combination to expectation simulation result in serviceable life.

Embodiment

As shown in Figure 1, the present invention includes four large steps: step 1: for eight wheels in a joint compartment, take turns abrasion data pre-service, draw the sample data that wheel is repaiied scale-up factor to wheel rim and wheel footpath wear rate and Xuan; Step 2: set up wheel to abrasion data-driven model, comprise that wheel footpath wears away model, flange thickness abrasion model and wheel Xuan is repaiied to scale-up factor distributed model etc.; Step 3: carry out rail traffic locomotive vehicle wheel to abrasion and the Monte Carlo simulation repaiied of Xuan, show that different flange thicknesses prevention Xuan values of repairing and Xuan repair the wheel of recovery value under combining to expectation serviceable life; Step 4: show that preferably Xuan repaiies decision-making, i.e. comparative analysis different flange thicknesses prevention Xuan value of repairing and Xuan repair the lower wheel of recovery value combination to expecting serviceable life, obtains preferably flange thickness prevention Xuan value of repairing and Xuan and repaiies recovery value and combine.

In step 1 of the present invention, for eight wheels in a joint compartment, take turns the pre-service of abrasion data, require first to remove the wheel that records imperfect and apparent error to abrasion data.In this example wheel to abrasion data recorded respectively eight wheels not same date wheel footpath value and flange thickness value and take turns Xuan repaiied, change working condition.In this specific embodiment, taking 30 days as one unit interval, calculate flange thickness and the wheel footpath wear rate sample data of certain wheel based on following formula:

$v_{\mathrm{Sd}}=\frac{S_{d,i+1}-S_{d,i}}{t_{i+1}-t_i}\·30---\left(1\right)$

$v_D=\frac{D_{i+1}-D_i}{t_{i+1}-t_i}\·30---\left(2\right)$

Wherein:

V _sdand v _dit is respectively an estimated value of flange thickness and tread diameter rate of depreciation;

T _iand t _i+1not have Xuan to repair in situation, the time value that twice of front and back flange thickness (or wheel footpath) is measured, unit is sky;

S _{d, i}and S _{d, i+1}respectively t _iand t _i+1the flange thickness value that moment measures;

D _iand D _i+1respectively t _iand t _i+1the tread diameter value that moment measures.

Calculate Xuan based on following formula and repair scale-up factor sample data:

$k=\frac{{D_j}^{-}-{D_j}^{+}}{{S_{d,j}}^{+}-{S_{d,j}}^{-}}---\left(3\right)$

Wherein:

K is the estimated sample value that Xuan repaiies scale-up factor;

D _j ^-and D _j ⁺it is respectively the tread diameter value measuring after repairing with Xuan before the j time wheel Xuan repaiies;

S _{d, j} ^-and S _{d, j} ⁺it is respectively the flange thickness value measuring after repairing with Xuan before the j time wheel Xuan repaiies.

In step 2 of the present invention, abrasion model in wheel footpath is a Weibull distribution of describing wheel footpath wear rate probability density characteristics, v _dsample meets v _d≤ 2, and according to sample histogram shape, select Weibull distribution to characterize v _ddistribution, suppose that d represents v _dstochastic variable, corresponding probability density function is:

$f\left(d\right)=\left\{\begin{array}{cc}\frac{{\mathrm{\&alpha;}}_1}{{\mathrm{\&beta;}}_1}{\left(\frac{-d+2}{{\mathrm{\&beta;}}_1}\right)}^{{\mathrm{\&alpha;}}_1-1}{e}^{-{\left(\frac{-d+2}{{\mathrm{\&beta;}}_1}\right)}^{{\mathrm{\&alpha;}}_1}}& -d+2\&GreaterEqual;0\\ 0& -d+2<0\end{array}\right.---\left(4\right)$

According to the sample data of wheel footpath wear rate, apply Maximum Likelihood Estimation Method, can obtain the scale parameter β of formula (4) Weibull distribution ₁=2.3617, form parameter α ₁=4.6743.Distribute according to this matching, wheel footpath rate of depreciation v _dbe about-0.16mm/30 days of mean value.

The flange thickness adopting in step 2 of the present invention wears away model, has portrayed the correlated fitting relation of certain moment flange thickness wear rate and corresponding flange thickness value.In this example, according to flange thickness wear rate sample data, flange thickness abrasion model can be expressed as follows:

V _sd＝-0.02606×S _d ²+1.538×S _d-22.75+E _vsd (5)

Wherein V _sdfor flange thickness wear rate, S _dfor flange thickness value, E _vsdfor flange thickness wear rate random fit difference, E _vsdbe that an average is zero normal distribution random number, its standard deviation is 0.3056.

It is that the gamma that a description Xuan repaiies scale-up factor k probability density characteristics distributes that the wheel that the present invention adopts is repaiied scale-up factor distributed model to Xuan.Be 3.782 because Xuan repaiies scale-up factor sample k minimum value, distribute to characterize distribution character so adopt to the γ after right translation 3.7, suppose k ₁be to represent that Xuan repaiies the stochastic variable of scale-up factor k, probability density function is expressed as follows:

$g\left(k_1\right)={\mathrm{gampdf}}_{a_1,b_1}\left(k_1\right)=\left\{\begin{array}{cc}\frac{{k_1}^{a_1-1}{\&CenterDot;e}^{-\frac{k_1-3.7}{b_1}}}{\mathrm{\&Gamma;}\left(a_1\right)\&CenterDot;{b_1}^{a_1}},& k_1-3.7\&GreaterEqual;0\\ 0,& k_1-3.7<0\end{array}\right.=\left\{\begin{array}{cc}\frac{{k_1}^{a_1-1}\&CenterDot;e^{-\frac{k_1-3.7}{b_1}}}{\mathrm{\&Gamma;}\left(a_1\right)\&CenterDot;{b_1}^{a_1}},& k_1\&GreaterEqual;3.7\\ 0,& k_1<3.7\end{array}\right.---\left(6\right)$

Wherein a ₁> 0 is form parameter, b ₁> 0 is scale parameter, Γ (a ₁) be gamma function:

$\mathrm{\&Gamma;}\left(a_1\right)={\&Integral;}_0^{+\&infin;}{x}^{a_1-1}\&CenterDot;e^{-u}\mathrm{du},a_1>0---\left(7\right)$

Therefore, k ₁mean value is a ₁b ₁+ 3.7, variance is a ₁b ₁ ².

Repairing scale-up factor sample application Maximum Likelihood Estimation Method according to Xuan can obtain, and the form parameter distributing about the γ of k is 1.1779, and the mean value that scale parameter is 2.6615, k is 1.1779 × 2.6615+3.7=6.8350.

In step 3 of the present invention, the simulation flow providing according to Fig. 2, the simulation calculation certain rim thickness prevention Xuan value of repairing S _dPrepair recovery value S with Xuan _dRwheel under combination is to expecting serviceable life.

Suppose at t _jmoment is measured flange thickness and the wheel footpath of each wheel, is designated as S _{d, j, 1}, S _{d, j, 2}..., S _{d, j, 8}and D _{j, l}... D _{j, 8}, subscript 1,2 ..., 8 and the first, the second ..., the 8th wheel correspondence.First and second wheels (or the third and fourth, or the 5th and the 6th, or the 7th and the 8th) on same axle.The the first, the second, three, the 4th wheel (or the five, six, seven, eight) is on same bogie.In this example, under wheel footpath allows, be limited to 770mm, be above limited to 840mm; Flange thickness is limited to 26mm under allowing, and is above limited to 32mm.Require coaxial wheels to be no more than 2mm to wheel footpath difference simultaneously, be no more than 4mm with steering framing wheel footpath difference, be no more than 7mm with joint wheel footpath, compartment difference.

In step 3.1 " all kinds of parameters of initialization ", input t _j=0, S _{d, j, i}=32mm, D _{j, k}=840mm, i, k=1,2 ..., 8, make Xuan repair times N _rep=0, order wheel is to emulation periodicity N _lC=0, and the wheel of setting is N to emulation total number of cycles _lCtotal=50, it is enough large for the optimal value of estimation wheel life expectancy length.Input S simultaneously _dPand S _dRvalue.In addition,, in view of the importance of flange thickness, set w _sd=1, w _d=0.The time step of emulation is a unit interval, namely 30 days.

Step 3.2 of the present invention: unit interval t of simulation time stepping _j=t _j+ 1;

Step 3.3: according to Monte Carlo simulation Method And Principle, the wheel based on aforementioned formula (4)-(7) provide is to abrasion data-driven model, and emulation produces t _jwheel footpath abrasion value V in the unit interval before moment _{d, j, k}with flange thickness abrasion value V _{sd, j, i}, and according to the wheel footpath value D in a upper moment _{j-l, k}with flange thickness value S _{d, j-l, i}, the wheel footpath of calculating current time is worth D _{j, k}with flange thickness value S _{d, j, i}, be expressed as follows:

D _j，k＝D _j-l，k+V _D，j，k×1 (8)

S _d，j，i＝S _d，j-l，i+V _Sd，j，i×1 (9)

Wherein i, k=1,2 ..., 8

Step 3.4: judge D _{j, k}< 770mm?, wherein k=1,2 ..., 8, if " being less than " enters step 3.5, otherwise carry out step 3.8;

Step 3.5:N _lC=N _lC+ 1;

Step 3.6: judge N _lC< N _lCtotal, " if being less than ", enter step 3.7, otherwise simulation optimization finishes, the note simulation optimization time finish time is t _jtotal, the life expectancy cycle EL of wheel can be expressed as:

$\mathrm{EL}=\frac{t_{\mathrm{jtotal}}}{N_{\mathrm{LC}}}---\left(10\right)$

Step 3.7: current each wheel footpath value and each flange thickness value update all are right wheel footpath value and the flange thickness value of brand-new wheel, even D _{j, k}=840mm, S _{d, j, i}=32mm, wherein i, k=1,2 ..., 8, start the emulation of next wheel to the emulation cycle, return to step 3.2;

Step 3.8: judge S _{d, j, i}≤ S _dPwherein i=1,2 ..., 8, if " being less than or equal to " enters step 3.9, otherwise returns to step 3.2;

Step 3.9: optimize a concrete Xuan and repair rear expectation flange thickness value and wheel footpath value, utilize genetic algorithm, considering under the prerequisite of the poor constraint in eight vehicle wheel footpaths, as far as possible little with the loss of wheel footpath, and Xuan repaiies rear rim one-tenth-value thickness 1/10, and to approach as far as possible flange thickness recovery value be optimization aim, and this optimization aim can be expressed as:

$\min [\underset{i=1}{\overset{8}{\mathrm{\&Sigma;}}}{w}_{\mathrm{sd}}\&CenterDot;|S_{d,j,i}^{+}-S_{\mathrm{dR}}|+\underset{k=1}{\overset{8}{\mathrm{\&Sigma;}}}{w}_D\&CenterDot;(D_{j,k}^{-}-D_{j,k}^{+})]---\left(\text{11}\right)$

Wherein

for t _jflange thickness value after moment i wheel Xuan repaiies, S _dRfor Xuan repaiies rear rim caliper recovery value,

for t _jmoment, k wheel Xuan repaiied front-wheel footpath value, utilized for t _jmoment, k wheel Xuan repaiied trailing wheel footpath value, w _sdand w _dfor being weight coefficient, meet w _sd+ w _d=1, this embodiment, in view of the importance of flange thickness, is set w _sd=1, w _d=0.

The constraint condition of optimizing is as follows:

(a) the poor constraint in wheel footpath: $0\&le;|D_{j,u}^{+}-D_{j,w}^{+}|\&le;2\mathrm{mm},0\&le;|D_{j,d}^{+}-D_{j,f}^{+}|\&le;4\mathrm{mm},$

wherein u, w ∈ { 1,2,3,4,5,6,7,8}, d, f ∈ { 1,2,3,4} or d, f ∈ { 5,6,7,8}, d ≠ f, g, h ∈ { 1,2,3,4,5,6,7,8}, g ≠ h.

(b) Xuan repaiies the constraint of not transfiniting of rear rim one-tenth-value thickness 1/10 and wheel footpath value: $770\mathrm{mm}\&le;D_{j,k}^{+}\&le;840\mathrm{mm},$ k＝1，2，…，8

(c) Xuan repaiies front and back wheelset profile logical restriction:

with

i, k=1,2 ..., 8.

Variable to be optimized is the flange thickness after Xuan repaiies

with tread diameter

i, k=1,2 ..., 8.

Because genetic algorithm is widely known by the people, repeat no more concrete optimizing process here.

Step 3.10: based on wheel, Xuan is repaiied to scale-up factor distributed mode pattern (6) and (7), sampling generates a Xuan and repaiies scale-up factor k, and show that according to step 3.9 result Xuan repaiies rear rim one-tenth-value thickness 1/10 S _{d, j, i} ⁺with wheel footpath value D _{j, k} ⁺,

S _d，j，i ⁺＝S _d，j，i ⁺ (12)

D _j，k ⁺＝D _j，k ^--k×(S _d，j，i ⁺-S _d，j，i ^-) (13)

Step 3.11: accumulation Xuan repaiies times N _rep=N _rep+ 1;

Step 3.12: judge D _{j, k}< 770mm? wherein k=1,2 ..., 8, if " being less than " returned to step 3.5, otherwise returns to step 3.2.

Repeatedly repeating step 3.1 is to step 3.12 as required, and simulation calculation draws the different flange thickness prevention Xuan value of repairing S _dPrepair recovery value S with Xuan _dRthe lower wheel of combination is to expectation serviceable life, thereby obtains different S _dPand S _dRthe lower wheel of combination is to expectation EL in serviceable life, as shown in Figure 3, and wherein 26mm≤S _dP< S _dR≤ 32mm and S _dP, S _dR∈ { 26,26.5,27,27.5,28,28.5,29,29.5,30,30.5,31,31.5,32mm}.As shown in Figure 3, when selecting { S _dP=26mm, S _dR=30mm}, { S _dP=26mm, S _dR=30.5mm}, { S _dP=26mm, S _dR=31mm}, { S _dP=26mm, S _dRwhen the combination such as=31.5mm}, wheel is larger to expecting serviceable life.

Utilize VC++ software to programme based on above-mentioned steps, the relative computer software that exploitation wheel is repaiied decision optimization to Xuan, can automatically realize rail traffic locomotive vehicle wheel Xuan is repaiied to decision optimization.

Claims (3)

1. right Xuan repaiies decision optimization method to the present invention relates to a kind of rail traffic locomotive vehicle wheel, comprises following steps:

Step 1: for eight wheels in a joint compartment, take turns abrasion data pre-service, draw the sample data that wheel is repaiied scale-up factor to wheel rim and wheel footpath wear rate and Xuan;

Step 2: set up wheel to abrasion data-driven model, comprise that wheel footpath wears away model, flange thickness abrasion model and wheel Xuan is repaiied to scale-up factor distributed model etc.

Step 3: carry out rail traffic locomotive vehicle wheel to abrasion and the Monte Carlo simulation repaiied of Xuan, show that different flange thicknesses prevention Xuan values of repairing and Xuan repair the wheel of recovery value under combining to expectation serviceable life;

In this step, to repair wheel under recovery value combination as follows to the step in expectation serviceable life for simulation calculation specific one group of flange thickness prevention Xuan value of repairing and Xuan:

Step 3.1: all kinds of parameters of initialization;

Step 3.2 a: unit interval of simulation time stepping;

Step 3.3: according to Monte Carlo simulation Method And Principle, take turns abrasion data-driven model based on aforementioned, emulation produces wheel footpath abrasion value and the flange thickness abrasion value in the corresponding unit interval, and according to wheel footpath value and the flange thickness value in a upper moment, calculated wheel footpath value and the flange thickness value of current time;

Step 3.4: judge whether wheel footpath value is less than wheel footpath lower limit, if " being less than " enters step 3.5, otherwise carry out step 3.8;

Step 3.5: wheel adds 1 to emulation periodicity;

Step 3.6: judge that wheel that whether wheel be less than setting to emulation periodicity is to emulation total number of cycles, if " being less than " enters step 3.7, otherwise simulation optimization finishes, Output simulation result;

Step 3.7: current each wheel footpath value and each flange thickness value update all are right wheel footpath value and the flange thickness value of brand-new wheel, starts the emulation of next wheel to the emulation cycle, returns to step 3.2;

Step 3.8: judge whether each flange thickness value is less than or equal to the flange thickness prevention Xuan value of repairing, if " being less than or equal to " enters step 3.9, otherwise return to step 3.2;

Step 3.9: optimize a concrete Xuan and repair rear expectation flange thickness value and wheel footpath value, utilize genetic algorithm, considering under the prerequisite of the poor constraint in eight vehicle wheel footpaths, as far as possible little with the loss of wheel footpath, and Xuan repaiies rear rim one-tenth-value thickness 1/10, and to approach as far as possible flange thickness recovery value be optimization aim, flange thickness value and wheel footpath value that this Xuan repaiied to rear expectation are optimized;

Step 3.10: based on wheel, Xuan is repaiied to scale-up factor distributed model, sampling generates Xuan and repaiies scale-up factor, and the Xuan drawing according to the step 3.9 flange thickness value of repairing rear expectation calculates Xuan and repaiies rear rim one-tenth-value thickness 1/10 and wheel footpath value;

Step 3.11: accumulation Xuan repaiies number of times;

Step 3.12: judge whether each footpath value of taking turns is less than wheel footpath lower limit, if " being less than " returned to step 3.5, otherwise return to step 3.2.

Step 4: show that preferably Xuan repaiies decision-making, i.e. comparative analysis different flange thicknesses prevention Xuan value of repairing and Xuan repair the lower wheel of recovery value combination to expecting serviceable life, obtains preferably flange thickness prevention Xuan value of repairing and Xuan and repaiies recovery value and combine.

2. the right Xuan of a kind of rail traffic locomotive vehicle wheel according to claim 1 repaiies decision optimization method, it is characterized in that, described wheel footpath abrasion model is a Weibull distribution of describing wheel footpath wear rate probability density characteristics; Described flange thickness wears away model, has portrayed the correlated fitting relation of certain moment flange thickness wear rate and corresponding flange thickness value, is expressed as follows:

V _sd＝a×S _d ²+b×S _d+c+E _vsd (1)

Wherein V _sdfor flange thickness wear rate, S _dfor flange thickness value, E _vsdfor flange thickness wear rate random fit difference, E _vsdbe that an average is zero normal distribution random number, a, b and c are constant; It is that the gamma that a description Xuan repaiies scale-up factor k probability density characteristics distributes that described wheel is repaiied scale-up factor distributed model to Xuan.

3. the right Xuan of a kind of rail traffic locomotive vehicle wheel according to claim 1 repaiies decision optimization method, it is characterized in that, the concrete Xuan of described optimization repaiies rear expectation flange thickness value and wheel footpath value, corresponding optimization aim, for wheel footpath loss is as far as possible little and Xuan repaiies rear rim one-tenth-value thickness 1/10 and approaches as far as possible flange thickness recovery value, can be expressed as:

$\min [\underset{i=1}{\overset{8}{\mathrm{\&Sigma;}}}{w}_{\mathrm{sd}}\&CenterDot;|S_{d,j,i}^{+}-S_{\mathrm{dR}}|+\underset{k=1}{\overset{8}{\mathrm{\&Sigma;}}}{w}_D\&CenterDot;(D_{j,k}^{-}-D_{j,k}^{+})]---\left(\text{2}\right)$

Wherein

for t _jflange thickness value after moment i wheel Xuan repaiies, S _dRfor Xuan repaiies rear rim caliper recovery value,

for t _jmoment, k wheel Xuan repaiied front-wheel footpath value,

for t _jmoment, k wheel Xuan repaiied trailing wheel footpath value, w _sdand w _dfor being weight coefficient, meet w _sd+ w _d=1; Optimize the poor constraint in wheel footpath, Xuan that constraint comprises eight wheels in same joint compartment and repair the constraint of not transfiniting of rear rim one-tenth-value thickness 1/10 and wheel footpath value, and the flange thickness value of Xuan after repairing before being greater than Xuan and repairing flange thickness value and the wheel footpath value of Xuan after repairing be less than Xuan and repair the Xuan such as front wheel footpath value and repair front and back wheelset profile logical constraint.

Images