Method and device for estimating angle of attack

Abstract

To provide an attack angle estimation method and the like capable of obtaining excellent estimation accuracy in a wide region.SOLUTION: An attack angle estimation method is configured to acquire lateral pressure Q of right and left wheels, longitudinal tangential force T, and a contact phase difference Δθ, which is an angle difference around an axle of contact positions between the right and left wheels and rails, as observation amounts based on output of strain gauges provided on right and left wheels 31 of a railway vehicle 1, acquire a rail curvature of the track on which the wheels travel, acquire the traveling speed of the railway vehicle, and estimate an attack angle and the contact positions of the right and left wheels by using a plant model and an observer of the right and left wheels.SELECTED DRAWING: Figure 2

Description

[Technical field]

[0001] The present invention relates to an angle-of-attack estimation method and an angle-of-attack estimation device for estimating the angle of attack of a railway vehicle wheel with respect to a rail. [Background technology]

[0002] Climbing derailments that occur on curved sections of railway tracks are closely related to wheelset motion variables, including the attack angle, which is the angle between the longitudinal direction of the rail and the direction of travel of the wheel. However, for many years, a practical method for measuring the attack angle has not been realized. For example, Non-Patent Document 1 describes a method for measuring the angle of attack using a laser displacement meter installed under the axle box. Furthermore, Non-Patent Document 2 describes a method for measuring the contact position between the wheel and the rail, etc. by processing images captured by a thermography camera. However, these methods are vulnerable to changes in illumination conditions, such as the effects of direct sunlight, and are not practical methods that can be used on actual railway tracks (operating lines, etc.). For this reason, in the past, the motion parameters of the wheelsets were not measured in normal driving tests, and were often not included as targets for evaluation in driving safety evaluations.

[0003] In response to this, Patent Document 1 describes a method of estimating the attack angle from the contact phase difference, which indicates the angular position shift around the axle (around the central axis of rotation of the wheels) of the contact positions of the left and right wheels with the rail, based on the output of a wheel load lateral pressure measuring device. Furthermore, Non-Patent Documents 3 and 4 propose a method for estimating the motion variables of a wheelset, including the angle of attack, using a Kalman filter to measure the three components of force, i.e., wheel load, lateral force, and front / rear tangential force, obtained by PQ measurement. Furthermore, Non-Patent Document 5 describes a method for estimating the contact phase difference Δθ, which expresses the amount of forward / rearward movement of the contact point as an angle, using a strain signal obtained by measuring the lateral wheel load using a PQ wheelset. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2019-81458 A [Non-patent literature]

[0005] [Non-Patent Document 1] Akinobu Kataori, Kenichi Doi, Hitoshi Iijima, Shuji Momozaki, Kenji Horioka, "Development of a continuous measuring device for the angle of attack and its measurement results", JR EAST Technical Review, No. 35, pp. 50-53, 2001. [Non-Patent Document 2] Takeshi Miyamoto, Hiroki Ikeda, Atsushi Furukawa, Hisayo Doi, Hiroaki Ishida, "Measurement of Wheelset Attack Angle During Curving", Proceedings of the Symposium on Railway Technology (J-RAIL), Vol. 9, pp. 97-100, 2002. [Non-Patent Document 3] Shoya Kuniyuki, Takatoshi Hondo, Mitsugu Suzuki, "Estimation of Wheelset Attack Angle by Kalman Filter Using Quasi-Static Single Wheelset Model of Railway Vehicles," Proceedings of the 16th Symposium on Motion and Vibration Control (MoViC2019), Vol. 16, Lecture No. C207, 2019. [Non-Patent Document 4] Shoya Kuniyuki, Takatoshi Hondo, Mitsugu Suzuki, "Effect of creep force saturation on the performance of railway wheelset attack angle estimation using Kalman filter with creep force constitutive law," Proceedings of Dynamics and Design Conference 2020, Lecture No. 504, 2020. [Non-Patent Document 5] T. Hondo and Y. Noguchi: “Measurement method for longitudinal displacement of wheel / rail contact point using strain gauges put on wheels,” Mechanical Engineering Journal, Vol. 7, No. 3, Doi:10.1299 / mej.19-00659, 2020. Summary of the Invention [Problem to be solved by the invention]

[0006] The technologies described in Patent Document 1 and Non-Patent Documents 3 and 4 make it possible to qualitatively capture the tendency of change in the angle of attack in curved sections within a relatively simple framework, without adding any additional measurement items to the existing wheel load lateral force measurement test using PQ wheelsets. However, the technology described in Patent Document 1 is premised on an assumption that the wheel is in a geometric contact state where it is in contact with the rail at the flange portion, and therefore has little versatility. In addition, in the technologies described in Non-Patent Documents 3 and 4, as described below, it was confirmed that the estimation performance of the attack angle deteriorated under some driving conditions. In addition, quantitative evaluation of the estimation results and consideration of the scope of application were insufficient.

[0007] For example, there was a problem that the estimation accuracy decreased when the contact geometry used to estimate the attack angle differed from the actual situation or depending on the contact state between the wheel and the rail. In particular, when traveling on a sharp curve where the creep force becomes saturated with the friction force, the inverse function used in the estimation law becomes undefinable, resulting in a problem of deterioration in estimation accuracy. Furthermore, there was also a problem in that the estimation accuracy deteriorated when the contact geometry obtained from the lookup table (LUT) used in the estimation rule was traveling under driving conditions that differed from the assumed situation (for example, changes in slack, the angle of attack, the wheel and rail shapes, etc.). In view of the above-mentioned problems, an object of the present invention is to provide an attack angle estimation method and an attack angle estimation device that can obtain good estimation accuracy over a wide range. [Means for solving the problem]

[0008] In order to solve the above-mentioned problems, an attack angle estimation method according to one embodiment of the present invention is characterized in that, based on the output of strain gauges provided on the left and right wheels of a railway vehicle, the lateral pressure, front and rear tangential force, and contact phase difference, which is the angular difference around the axle of the contact positions of the left and right wheels with the rail, are obtained as observation quantities, the rail curvature of the track on which the wheels run is obtained, and the running speed of the railway vehicle is obtained, and the attack angle and the contact positions of the left and right wheels are estimated using a plant model and an observer of the left and right wheels. According to the present invention, by using an observer to estimate the wheel contact position and the attack angle simultaneously, it is not necessary to prepare in advance contact geometry information in the form of a lookup table, as in a method of determining the contact position using a lookup table, and the estimation accuracy of the attack angle can be improved in a wide range of applications. In addition, the contact phase difference can be calculated from the output signal of a strain gauge attached to the wheel using an existing wheel load lateral pressure measuring device (PQ wheel set), and since it is an observable quantity that can be obtained from general measuring equipment without adding any additional measurement items, it is possible to prevent the device configuration from becoming complicated. In the present invention, a configuration may be adopted in which a change in the geometric state of contact between the wheel and the rail is estimated simultaneously with the contact position by using a state space model.

[0009] In the present invention, a plant model integrating the left and right wheels can be used to estimate the attack angle and the contact positions of the left and right wheels using the observation quantities observed from the left and right wheels, respectively. In this case, by using the left and right speeds of the axle that connects the left and right wheels with an axle, a time differential equation for the contact position of the wheels can be approximately given, and the contact position of the wheels can be estimated using this time differential equation. According to each of these inventions, the contact position can be appropriately estimated using information obtained from a wheel load lateral force measuring device for a single axle.

[0010] In the present invention, the estimated value of the attack angle may be corrected using the contact phase difference in a region where the lateral creep force between the wheel and the rail is saturated. According to this, the contact angle α of the left and right wheels is r ,α l and for the contact phase difference Δθ,

number

[0011] As described above, according to the present invention, it is possible to provide an attack angle estimation method and an attack angle estimation device that can obtain good estimation accuracy over a wide range. [Brief description of the drawings]

[0012] [Figure 1] FIG. 11 is a diagram showing the concept of a contact phase difference used in an embodiment of the attack angle estimation method to which the present invention is applied. [Diagram 2] FIG. 1 is a diagram showing a flow from a measurement system to acquisition of an estimation result in an attack angle estimation method according to an embodiment. [Diagram 3] FIG. 1 is a diagram showing force elements taken into account in a vehicle model created for a simulation for verifying the validity of an embodiment of the present invention and a comparative example, as viewed from the front-rear direction of the vehicle. [Figure 4] FIG. 4 is a view taken along line IV-IV of FIG. [Diagram 5] FIG. 2 is a view taken along the arrow VV in FIG. 1. [Figure 6] FIG. 13 is a diagram showing the rail curvature of a test curve created for simulation. [Figure 7] FIG. 11 is a diagram showing an example of an attack angle estimation result by the attack angle estimation method of the embodiment, which is an estimation result at R100. [Figure 8]FIG. 11 is a diagram showing an example of an attack angle estimation result by the attack angle estimation method of the embodiment, which is an estimation result at R400. [Figure 9] FIG. 13 is a diagram showing the estimated average attack angle in a circular curve for each trial curve. [Figure 10] 11A to 11C are diagrams illustrating the results of lateral displacement estimation of the contact point on the R100 curve by the attack angle estimation method of the embodiment. [Figure 11] FIG. 13 is a diagram showing the result of estimating the lateral displacement of the contact point on the curve R400 using the attack angle estimation method of the embodiment. [Figure 12] FIG. 13 is a diagram showing the estimated results of the average lateral displacement of the contact point on the outer rail side in a circular curve for each of the test curves. [Figure 13] FIG. 13 is a diagram showing the result of attack angle estimation in the case of R100 when a contact phase difference is not used in the method of the embodiment. [Figure 14] FIG. 13 is a diagram showing the estimated results of outer rail lateral force at R100 depending on whether or not contact phase difference is used. [Figure 15] FIG. 13 is a diagram showing the estimation results of the contact phase difference in R100 with and without the use of the contact phase difference. [Figure 16] FIG. 13 is a diagram showing the estimation results of the outer rail side lateral creep force at R100 with and without the use of contact phase difference. [Figure 17] FIG. 13 is a diagram showing the estimation results of the outer rail side normal force at R100 depending on whether or not a contact phase difference is used. [Figure 18] 1 is a diagram showing an outline of the principle of an attack angle estimation method according to an embodiment; [Figure 19] FIG. 4 is a diagram illustrating a method for estimating an angle of attack in the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, an embodiment of an angle-of-attack estimation method and an angle-of-attack estimation device to which the present invention is applied will be described. In the present invention, a plant model is created using the relationships in Equations 1 and 2 that hold for each wheel, as shown in the existing attack angle estimation methods described in Non-Patent Documents 3 and 4.

number

number

number

[0014] In the existing attack angle estimation methods described in Non-Patent Documents 3 and 4, the wheel diameter r i and contact angle α i The left and right wheel axle displacement y iThe contact geometry is considered to be a function of the surface roughness, and is calculated in advance as a look-up table (hereafter referred to as LUT). The contact geometry is

number

[0015] One solution to this problem would be to consider appropriate contact geometry for each passing curve and provide an LUT, but this method would lack versatility. In this embodiment, in order to solve this problem, an attempt is made to improve the plant model so that estimation can be performed without assuming contact geometry. Furthermore, the existing attack angle estimation methods described in Non-Patent Documents 3 and 4 have a problem in that accuracy deteriorates when the creep force is saturated, but this embodiment also aims to improve this problem.

[0016] The reason why the contact geometry is given by the LUT is that the wheel diameter r at the contact position i and contact angle α i The left and right displacements of the contact point s needed to calculate i This is because it is difficult to analytically give Therefore, in this embodiment, the contact point lateral displacement s i Newly added, wheel diameter r i and contact angle α i The left and right displacement of the contact point s i We attempt to improve this situation by expressing it as a function of In this case, the wheel shape data is used in the LUT, and the wheel radius r is calculated without providing the contact geometry. i and contact angle α i can be calculated.

[0017] Contact point lateral displacement s i When adding to the state variables, it is necessary to give the corresponding differential equation. But s i Time derivative of

number

number

number

[0018] In this embodiment, the state variable x to be estimated is given by Equation 5.

number

number

number

number

number

number

number

number

number

number

number

[0019] In the existing attack angle estimation methods described in Non-Patent Documents 3 and 4, it has been confirmed that in addition to the contact geometry, saturation of creep forces has a negative effect on accuracy. This is because, in the existing attack angle estimation methods described in Non-Patent Documents 3 and 4, the observed quantity, the front and rear contact force T xi and lateral pressure Q i This was because when the creep force obtained by converting the above became saturated, the inverse function used to calculate the wheelset motion variables became undefinable. In this embodiment, this problem is also improved, and the attack angle ψ and the contact position s i In order to improve the estimation accuracy of , a new term is added to the observation equation. In recent years, a method has been proposed in which the contact phase, which represents the forward / backward movement amount at the contact point as an angle, is estimated by using the strain signal obtained by PQ measurement, as shown in FIG. 1, as in Non-Patent Document 5. FIG. 1 is a diagram showing the concept of a contact phase difference used in the attack angle estimation method of the embodiment. Δθ is approximately expressed by the following Equation 14.

number

[0020] Equation 14 is the relationship between the yaw angle ψ and the contact angle α i It links the relationships between these two, and adding it to the observables is expected to improve accuracy.

[0021] When the contact phase difference Δθ is added to the observation, the observation is the front-rear tangential force T xi , Lateral pressure Q i and the contact phase difference Δθ, and the observation equation for the observation quantity y is given by Equation 15.

number

number

[0022] FIG. 2 is a diagram showing a flow from a measurement system to acquisition of an estimation result in the attack angle estimation method according to the embodiment. In the attack angle estimation method of the embodiment, first, a strain signal acting on the wheel is measured using a known wheel load lateral force measuring device using a PQ wheel set. The output of the bridge circuit made of strain gauges is then processed to obtain the necessary information shown below. The information to be obtained includes the wheel load P of the right wheel r , lateral force Q on right wheel r , the longitudinal tangential force T of the right wheel r , left wheel load P l , lateral force Q on the left wheel r , the longitudinal tangential force T of the left wheel l These contact forces include

[0023] In addition, from these strain signals, a contact phase difference Δθ, which is the angular difference around the axle between the rail contact positions of the left and right wheels, is obtained. Furthermore, the vehicle's running speed v can be obtained from, for example, a rotary encoder on the axle or the driver's seat of the vehicle. x Obtain information regarding the radius of rail curvature R from route registers, etc. Then, the attack angle ψ of the wheel set with respect to the rail and the contact positions s1, s2 of the left and right wheels with respect to the rail are simultaneously estimated using an observer.

[0024] The estimated performance of the present invention was investigated using simulated driving test data generated by numerical simulation. This is because it is possible to easily reproduce various curves and driving conditions, and it is easy to obtain information that is difficult to measure in experiments, such as contact position, making it easy to verify the validity. The simulation was performed using Simpack (ver. 2019.1) from Dassault Systems.

[0025] In the following discussion, we consider as an example a narrow-gauge (1067 mm gauge) conventional electric train that is common in Japan. 3 to 5 are diagrams showing force elements taken into consideration in a model created for simulation. FIG. 3 is a view of the vehicle as seen from the front-rear direction. FIG. 4 is a view taken along line IV-IV in FIG. 1 (a side view of the vehicle). 5 is a view taken along line VV in FIG. 1 (plan view of the vehicle). In the simulation, pseudo-running test data was generated using a one-vehicle model in which the car body, bogie frame, and wheelsets were rigid bodies. Table 1 shows the vehicle specifications related to the estimation algorithm. [Table 1]

[0026] As shown in Figures 3 to 5, the modeled railway vehicle 1 is composed of a car body 10, a bogie frame 20, a wheel set 30, an axle box suspension device 40, a secondary spring system 50, a yaw damper 60, a vertical movement stopper 70, a lateral movement stopper 80, a lateral movement damper 90, etc. The vehicle body 10 is a structure having a space, such as a vehicle cabin in which passengers are accommodated. The car body 10 is configured in a hexahedral shape with end structures and side structures rising upward from the front, rear and side edges of the underframe that forms the floor, and a roof structure provided on top of the end structures and side structures. The vehicle body 10 has a center pin 11 that protrudes downward from the underside of the floor. The central pin 11 functions as a base to which the bogie frame 20 is attached so that the bogie angle can be given. The center pin 11 is capable of transmitting longitudinal forces and the like between the center pin 11 and the bogie frame 20 via a known traction device.

[0027] The bogie frame 20, the wheel set 30 and the axle box suspension 40 cooperate to form a two-axle bogie bogie. The bogies are provided, for example, in a pair (two units), spaced apart from each other in the front-rear direction (travel direction) under the car body 10. The bogie frame 20 is a structural member that constitutes the main body of the bogie. The bogie frame 20 is, for example, a rectangular frame body in a plan view seen from above. The central part of the bogie frame 20 is connected to the center pin 11 via spring elements and damping elements (shown by symbols in Figures 3 to 5) possessed by the traction device so as to be capable of relative displacement in the up-down translation direction, left-right translation direction, pitching direction, rolling direction, and yawing (bogie angle application) direction.

[0028] As shown in FIG. 3 etc., the wheel set 30 is configured by fixing a pair of left and right wheels 31 to both ends of an axle 32 by press fitting or the like. Both ends of the wheel set 30 are supported by axle boxes (not shown) so as to be rotatable around the central axis of the axle 32. The axle boxes include bearings that hold journals formed at the ends of the axle 32, as well as a lubrication device for the bearings. In this embodiment, the wheel set 30 is a so-called PQ wheel set, which is a well-known type that can continuously measure the wheel load P, lateral force Q, and front / rear tangential force T acting between the wheel 31 and the rail independently for the left and right wheels 31 based on the output voltage of a bridge circuit consisting of multiple strain gauges attached to the plate portion of the wheel 31. Two sets of wheelsets 30 are provided for each bogie.

[0029] The axle box support device 40 connects the axle box to the bogie frame 20 via spring elements and damping elements so that the wheel set 30 can be displaced relative to the bogie frame in the up-down translation direction, the front-to-rear translation direction, the rolling direction, and the yawing (steer) direction. These spring elements and damping elements include the so-called primary spring (axle spring) system, which generates a reaction force corresponding to the vertical displacement of the axle box relative to the bogie frame and a damping force corresponding to the vertical displacement speed.

[0030] The secondary spring system 50 includes, for example, bolster springs provided on the lower part of the car body 10 and the upper part of the bogie frame 20 . The bolster springs are, for example, air springs that have the function of supporting the weight of the car body 10, and are provided as a pair spaced apart in the sleeper direction. The secondary spring system 50 has spring elements, damping elements, etc. that generate a spring reaction force corresponding to the vertical displacement of the carbody 10 and the bogie frame 20 and a damping force corresponding to the vertical displacement speed.

[0031] The yaw damper 60 is provided between the side of the bogie frame 20 and the car body. The yaw damper 60 has a spring element and a damping element that generate a spring reaction force and a damping force corresponding to the rotation of the bogie frame 20 relative to the carbody 10 in the yaw direction. The vertical movement stopper 70 regulates the relative ascent of the bogie frame 20 with respect to the car body 10 (the relative descent of the car body 10 with respect to the bogie frame 20). The vertical movement stopper 70 has a spring element that acts when the bogie frame 20 rises above a predetermined amount relative to the car body 10 and generates a spring reaction force according to the amount of further rise of the bogie frame 20.

[0032] The lateral movement stopper 80 regulates the relative displacement between the car body 10 and the bogie frame 20 in the sleeper direction. The lateral movement stopper 80 has a spring element that acts when the relative displacement between the car body 10 and the bogie frame 20 exceeds a predetermined amount from the neutral position, and generates a spring reaction force according to the further amount of relative displacement. The lateral movement damper 90 is a damping element that generates a damping force according to the relative displacement speed between the car body 10 and the bogie frame 20 in the sleeper direction.

[0033] In each simulation described below, the test curve was a curve pointing to the right with respect to the traveling direction. FIG. 6 is a diagram showing the rail curvature of a test curve created for the simulation. In FIG. 6, the horizontal axis indicates the travel distance of the vehicle, and the vertical axis indicates the curvature. As shown in Figure 6, the section from the start of driving to 100 m is a straight line, followed by a 100 m entrance transition curve, a 100 m circular curve, a 100 m exit transition curve, and finally a 100 m straight line.

[0034] Using the simulation described above, an investigation was carried out into the extent to which the estimation performance of the present invention depends on the curve radius. Therefore, a simulation was performed by changing the radius of the circular curve between R100 and R1000 in increments of 100 m. The running speed and cant were fixed at 10 km / h and 0 mm (no cant), and the slack was set to 0 mm. The virtual driving data obtained by the simulation was used as the measurement data after adding noise. In this embodiment, the sampling time of the measurement data is set to, for example, 1 ms.

[0035] Using virtual measurement data created by simulation, an attempt was made to estimate the state of the wheelset motion variables using the method of the comparative example. The observer was designed using a known technique called Unscented Kalman Filter (UKF), which is a type of nonlinear Kalman filter.

[0036] The validity of the method according to the embodiment will be verified below. The verification is performed using driving data obtained by driving the pseudo test curves having R100 to R1000 generated by the above-mentioned Simpack. The measurements are supplemented with normal white noise with a mean of zero. In addition, the initial values ​​of the state quantities to be estimated are as shown in Equation 16.

number

[0037] FIG. 7 is a diagram showing an example of an attack angle estimation result by the attack angle estimation method according to the embodiment, which is an estimation result at R100. FIG. 8 is a diagram showing an example of an attack angle estimation result by the attack angle estimation method according to the embodiment, which is an estimation result at R400. In Fig. 7, it can be seen that the attack angle estimation value is qualitatively well captured regardless of whether it is a transition curve or a circular curve section. Furthermore, from Fig. 8, it can be confirmed that the attack angle estimation value is qualitatively well captured even in R400, regardless of whether it is a transition curve or a circular curve section. Furthermore, it was confirmed that the attack angle estimation results were qualitatively valid for curves other than the two curves mentioned above. Therefore, it can be said that this embodiment provides a qualitatively valid attack angle estimation result regardless of changes in contact geometry.

[0038] Next, we consider the quantitative performance of the angle-of-attack estimation. FIG. 9 shows the estimated average attack angle in the circular curve for each of the trial curves. The horizontal axis indicates the curve radius, and the vertical axis indicates the attack angle and its root mean square error (root mean square: RMS). As shown in FIG. 9, it was confirmed that the estimated angle of attack had an RMS of 0.04 deg or less from the true value within the range of curve radii R100 to R1000, indicating that the estimation was performed with almost no error. That is, in this embodiment, it can be said that a quantitatively valid attack angle estimation result is obtained regardless of changes in contact geometry.

[0039] Next, we consider the estimation performance of the lateral displacement of the contact point, which has been newly added to the state quantities. 10 and 11 are diagrams showing the results of lateral displacement estimation of the contact point on the curves R100 and R400 using the attack angle estimation method of the embodiment. The horizontal axis indicates the travel distance, and the vertical axis indicates the estimated value, the true value, and the curvature of the curve of the lateral displacement of the contact point. 10 and 11, the following facts can be confirmed regarding the estimation of the position of the contact point. -In straight sections, the left and right wheels were not properly estimated. - Regardless of the curve radius, the inner gauge side is not properly estimated in curved sections. - In circular curve sections, the outer rail side is estimated appropriately regardless of the curve radius.

[0040] The reason why the estimated results for both wheels in the straight section and the inner rail side in the curved section deviate from the true values ​​is thought to be that in the tread area, even if the position of the contact point changes, there is no significant change in the wheel diameter and contact angle. In other words, as long as it is possible to identify the fact that the contact point exists in the tread area, there will be no significant error in the calculation of the observation quantity using the prior estimate shown in Equation 15. This is thought to have caused the variance to not converge, resulting in inappropriate estimation results.

[0041] In contrast, in the flange region, the wheel diameter and contact angle are highly dependent on the location of the contact point. Therefore, the contact position of the outer rail when traveling on a curve must be clearly identified, and this can be seen in Figures 10 and 11(b). For these reasons, it is believed that the attack angle was estimated appropriately, as shown in Figures 7 and 8.

[0042] FIG. 12 is a diagram showing the estimated results of the average lateral displacement of the contact point on the outer rail side in the circular curve for each of the test curves. FIG. 12 plots the average of the estimated lateral displacement of the contact point on the outer rail side in a circular curve, the average of the true lateral displacement, and the RMS of both on the vertical axis, and the curve radius on the horizontal axis. From FIG. 12, it can be seen that as the curve radius increases, the RMS also tends to increase. The RMS from the true value is 1 mm or less in the range of curve radii R100 to R800, but increases to approximately 2.5 mm at R900, and then increases significantly to approximately 14.5 mm at R100.

[0043] Furthermore, when looking at the true value of the lateral displacement of the contact point at R1000, the value has decreased significantly compared to the results for previous curve radii. In other words, it was found that no flange contact occurred in the R1000 case. This situation means that contact occurs in the tread area on the outer rail as well as at the contact points on the straight and inner rail, which means that the sensitivity of the wheel diameter and contact angle in the tread area decreases, resulting in a deterioration in estimation accuracy.

[0044] From the above results, it is confirmed that the improved proposed method can obtain an attack angle estimate with qualitative and quantitative validity without using contact geometry information. It was also confirmed that quantitatively valid estimations of the lateral displacement of the outer rail contact point were possible when flange contact occurred.

[0045] In this embodiment, taking into consideration the increase in state variables to be estimated and in anticipation of improved accuracy when the creep force is saturated, the contact phase difference Δθ is newly added to the observation quantities. Below, we will examine whether this actually changes the estimation accuracy by comparing it with data for R100, which is considered to be in a highly saturated state. FIG. 13 is a diagram showing the attack angle estimation result in the case where the contact phase difference is not used in the method of the embodiment, in the case of R100. The horizontal axis represents the traveled distance, and the vertical axis represents the estimated angle of attack, the true value, and the curvature. As shown in FIG. 13, it can be seen that the estimated angle of attack values ​​tend to be qualitatively similar to the true values, but are quantitatively different. In other words, it can be confirmed that if the contact phase difference is not added to the observation quantity, a quantitatively valid attack angle cannot be estimated.

[0046] The cause of this will be considered from the perspective of observables. FIG. 14 is a diagram showing the estimation results of the outer rail lateral force at R100 depending on whether or not the contact phase difference is used. FIG. 15 is a diagram showing the estimation result of the contact phase difference in R100 with and without the use of the contact phase difference. Here, in an actual estimation algorithm, if the contact phase difference Δθ is not included in the observation amount, there is no problem in not calculating the estimated value of the contact phase difference Δθ, but the calculation is performed here for comparison purposes.

[0047] Figure 14 shows the results of lateral force estimation using the prior estimate. Looking at Figure 14, we can see that the estimated results for both methods up until after passing through the curve show the same trend, and that the values ​​are close to the true values ​​if viewed casually. When the contact phase difference Δθ was not added to the observations, the estimated angle of attack lacked quantitative validity. However, from Figure 14 it can be confirmed that the lateral force Q is estimated validly regardless of whether the estimated angle of attack is quantitatively valid or not.

[0048] In addition, regarding the estimated results of the contact phase difference shown in Figure 15, in the case of Figure 15(b) where it is used as an observation quantity, the true value can be reproduced on average, whereas in the case of Figure 15(a) where it is not used as an observation quantity, it can be confirmed that the true value during passing through a curve cannot be reproduced on average. The cause of this can be explained from the estimated results of the lateral creep force and normal force shown in Figures 16 and 17. FIG. 16 is a diagram showing the estimation results of the outer rail side lateral creep force at R100 with and without the use of the contact phase difference. FIG. 17 is a diagram showing the estimation results of the outer rail side normal force at R100 depending on whether or not the contact phase difference is used.

[0049] Theoretically, the lateral creep force is a force that is highly dependent on the angle of attack, but as can be seen from Figure 16, the estimated results for the lateral creep force are similar, even though the angle of attack is a reasonable estimate. This is because, regardless of the angle of attack, the creep force is saturated as the vehicle passes through a sharp curve. In an embodiment, the lateral pressure is calculated from the lateral creep force and the normal force. Looking at the normal force estimation results in Figure 17, we can see that the two estimates are almost identical on average. In other words, the lateral pressure calculated from the lateral creep force and the normal force will be almost the same value if the attack angle is above a certain value and the creep force is saturated.

[0050] The fact that the lateral force ends up being the same value regardless of whether the estimated angle of attack is valid means that the results of comparing the estimated and measured lateral force values ​​are not fed back into the estimated angle of attack. This is the reason why the lateral force is a valid estimate regardless of whether the estimated angle of attack is quantitatively valid, and is the cause of the previous issue of a decrease in accuracy when the creep force saturates.

[0051] This error is corrected by information on the contact phase difference Δθ. In the estimation result of the contact phase difference shown in FIG. 15, it can be confirmed that when the contact phase difference is added to the observation amount, the estimated value becomes the true value. This means that when an error occurs with respect to the true value as shown in FIG. 15(a), the attack angle, which is the factor that produces the error, can be corrected by adjusting the observer gain. In other words, it was shown that the decrease in accuracy due to creep force saturation can be avoided by adding the contact phase difference Δθ to the observation quantity.

[0052] Hereinafter, it will be described how the decrease in accuracy can be improved by incorporating the measured value of the contact phase difference Δθ into the estimation system. FIG. 18 is a diagram showing an outline of the principle of the attack angle estimation method according to the embodiment. In FIG. 18, the horizontal axis represents the attack angle ψ, and the vertical axis represents the lateral creep force. In the existing method, the relationship between the lateral creep force (effectively the lateral pressure) shown by the dashed line and the angle of attack was utilized to obtain an estimate of the angle of attack from the lateral creep force. This method had the problem that the attack angle could not be uniquely determined in the region where the lateral creep force was saturated at the friction force μN. In this case, the attack angle estimate remains near the limit where the lateral creep force becomes saturated with the friction force.

[0053] In this embodiment, the contact phase difference information is used to avoid this problem. The contact phase difference has a linear relationship with the attack angle, as shown by the solid line in Fig. 18. The proportionality coefficient is determined by the contact position s between the wheel and rail, and the larger the contact position s, the larger the slope in Fig. 18 becomes. In this embodiment, information on both the lateral creep force and the contact phase difference is used to estimate the angle of attack ψ.

[0054] FIG. 19 is a diagram showing a method for estimating the angle of attack in this embodiment. FIG. 19(a) shows the state where the creep force is not saturated. In this case, the angle of attack can be determined inversely from each of the two diagrams. Using these two attack angles and the plant model, the attack angle is estimated by a known observer.

number

[0055] FIG. 19(b) shows the state where the creep force is saturated at the friction force μN. In this case, the attack angle estimate is determined in the same way as in the non-saturated condition, but the lateral creep force F y The attack angle ψ obtained from Fy Since the reliability of is low, the observer estimates the angle of attack.

number

number

[0056] To accurately use the contact phase difference information, it is essential to know the contact position s. In this embodiment, the time differential equation of the contact position s, which is difficult to express analytically, is approximated using the lateral speed of the wheel set, thereby incorporating the contact position s into the estimated quantity. In the method of the comparative example, when determining the contact position s, the contact geometry between the wheel and the rail is calculated in advance and the result is obtained as a lookup table. In the present embodiment, however, the estimated value of the contact position s is determined in a manner similar to that of the attack angle, based on the time differential equation of the contact position s, using the measured values ​​of the front and rear tangential force T, the lateral force Q, and the contact phase difference Δθ.

[0057] The attack angle estimation method of the above-described embodiment can be executed by an attack angle estimation device of an embodiment described below. The attack angle estimation device of the embodiment includes an observation quantity acquisition unit that acquires the wheel load, lateral force, front and rear tangential force of the left and right wheels, and the contact phase difference, which is the angular difference around the axle of the contact positions of the left and right wheels with the rail, based on the output of strain gauges provided on the left and right wheels of a railway vehicle, a rail curvature acquisition unit that acquires the rail curvature of the track on which the wheels run, a running speed acquisition unit that acquires the running speed of the railway vehicle, and an estimation calculation unit that estimates the attack angle and the contact positions of the left and right wheels using a plant model and an observer of the left and right wheels.

[0058] As the observation quantity acquisition unit, a wheel load lateral pressure measuring device using a known PQ wheel set can typically be used. For example, the above-mentioned Patent Document 1 describes an example of such a wheel load lateral force measuring device, and the entire text of Patent Document 1 is incorporated herein by reference.

[0059] As described above, in this embodiment, the method of estimating the contact position using a lookup table, which is used in existing methods such as the existing attack angle estimation methods described in Non-Patent Documents 3 and 4, is eliminated, and instead, changes in the contact geometric conditions are incorporated into the state space model, making it possible to estimate the contact position and the changes in the contact geometric conditions simultaneously. In the comparative example, the plant models constructed separately for the left and right wheels are integrated, and the attack angle and the like are estimated using the measured values ​​of the left and right wheels. In this embodiment, the contact phase difference Δθ is added to the observation amount. As a result, the effectiveness of the embodiment was verified using simulation data obtained by running 10 curves from R100 to R1000, and the following results were obtained. (1) The estimated angle of attack was qualitatively and quantitatively valid, regardless of the radius of the curve on which the vehicle was traveling. (2) When the wheel and rail flanges were in contact while traveling on a circular curve, the RMS deviation of the estimated lateral displacement of the outer rail contact point was within 3 mm of the true value. In particular, for curves with R800 or less, the RMS was less than 1 mm, and quantitatively valid estimation results were obtained, similar to the angle of attack. (3) By adding the contact phase difference Δθ as an observable quantity, the accuracy of estimating the attack angle at the time of creep force saturation could be improved.

[0060] (Other embodiments) The present invention is not limited to the above-described embodiments, but various applications and modifications are possible. For example, the configuration of the railway vehicle is not limited to the model configurations shown in FIGS. 3 to 5, and can be modified as appropriate. For example, in the embodiments, a narrow-gauge electric train with a track gauge of 1067 mm (equivalent to so-called conventional railway lines in Japan) is considered, but the present invention can also be applied to railways with other track gauges (such as the Shinkansen with a standard track gauge of 1435 mm). In addition, the form of each equation and the values ​​of various parameters used in the estimation can be changed as appropriate. [Explanation of symbols]

[0061] 1 Railway vehicles 10 Car bodies 11 Center pin 20 Bogie frame 30 Axle 31 Wheel 32 Axle 40 Axle box support device 50 Secondary spring system 60 Yaw damper 70 Vertical movement stopper 80 Horizontal movement stopper 90 Left and right movement stopper

Claims

1. obtaining, as observation quantities based on outputs of strain gauges provided on the left and right wheels of the railway vehicle, a lateral force of the left and right wheels, a front-rear tangential force, and a contact phase difference which is an angular difference around an axle between contact positions of the left and right wheels and a rail; Obtaining a rail curvature of a track on which the wheels run; Acquire a traveling speed of the rail vehicle; Using a plant model and an observer for the left and right wheels, an angle of attack and a contact position of the left and right wheels with the rail are estimated. The method for estimating an angle of attack is characterized by:

2. Using an observer, the attack angle ψ of the wheel set with respect to the rail and the contact positions s1, s2 of the left and right wheels with respect to the rail are simultaneously estimated, 2. The method of claim 1, further comprising:

3. Using a plant model integrating the left and right wheels, the attack angle and the contact positions of the left and right wheels with the rail are estimated using the observation quantities observed from the left and right wheels, respectively.

3. The method for estimating an angle of attack according to claim 1 or 2, characterized in that:

4. By using the left and right speeds of a wheel set that connects the left and right wheels with an axle, a time differential equation for the contact position of the wheel with the rail is approximately given, and the contact position of the wheel with the rail is estimated using the time differential equation.

4. The method of claim 3, further comprising the steps of:

5. In a region where the lateral creep force between the wheel and the rail is saturated, the contact phase difference is used to correct the estimated value of the attack angle. The method for estimating an angle of attack according to any one of claims 1 to 4, characterized in that

6. an observation quantity acquiring unit that acquires the wheel loads, lateral forces, front and rear tangential forces of the left and right wheels, and a contact phase difference that is the angular difference around the axle of the contact positions of the left and right wheels with the rail, based on outputs of strain gauges provided on the left and right wheels of the railway vehicle; a rail curvature acquisition unit for acquiring a rail curvature of a track on which the wheels run; a travel speed acquisition unit that acquires a travel speed of the railroad vehicle; an estimation calculation unit that estimates an attack angle and a contact position of the left and right wheels with a rail using a plant model of the left and right wheels and an observer; An attack angle estimation device comprising:

Images