JP2019211263A - Wheel rotation angle estimation method - Google Patents

Abstract

To provide a wheel rotation angle estimation method capable of estimating the wheel rotation angle by utilizing the bridge output for wheel weight measurement without using a rotation angle sensor.SOLUTION: The wheel rotation angle estimation method is set to have a first wheel weight bridge circuit p1 and a second wheel weight bridge circuit p2 arranged at mutually different angles around the wheel axis 120 of a strain gauge while having a plurality of strain gauges 131A and 138B in which the wheel 110 of a railroad vehicle having sensitivity to the wheel weight P is approximated by a first wheel weight sensitivity function f(φ) and a second wheel weight sensitivity function f(φ) of wheel weight sensitivity, respectively, for the angle of rotation of the wheel, and the angle of rotation of the wheel is estimated based on the output pof the first wheel weight bridge circuit, the output pof the second wheel weight bridge circuit, the first wheel weight sensitivity function, and the second wheel weight sensitivity function.SELECTED DRAWING: Figure 14

Description

  The present invention relates to a wheel rotation angle estimation method for estimating a wheel rotation angle of a railway vehicle, and more particularly to a method capable of estimating a wheel rotation angle using a bridge output for wheel load measurement without using a rotation angle sensor.

The wheel load of a railway vehicle is a component force of an up-down direction component in a plane perpendicular to the rail direction among forces acting between the wheel and the rail.
The wheel load is expressed as a sum of a stationary wheel load as a force that supports the weight of the vehicle and a variation generated with the movement of the vehicle.
The wheel load fluctuates due to, for example, irregularity in the flatness of the track in the relaxation curve portion, torsion of the vehicle body, lack of cant, vehicle shake, vibration of unsprung mass, impact, and the like.
The wheel load is information necessary for grasping the running performance of the vehicle and the state of the track. Conventionally, wheel load measurement has been attempted by various methods.

  For example, Non-Patent Document 1 discloses a conventional bridge circuit configured by using a PQ wheel shaft with a plurality of strain gauges attached to wheels and sequentially connecting strain gauges. A PQ measurement technique called a so-called new continuous method is described in which the wheel load is continuously measured by correcting the output voltage with the rotation angle of the wheel.

Hiroyuki Kanemoto, “Measuring Driving Safety of Railway Vehicles” Railway Research Review Vol.66 No.6, June 2009

In the measurement of wheel load and lateral pressure using a PQ wheel shaft, the method is mainly divided into a method of measuring intermittent force (intermittent method) and a method of measuring continuous force (continuous method). The
In the measurement by the intermittent method, it is possible to measure with a simple measurement system configuration compared to the continuous method. On the other hand, the force cannot be measured continuously over the entire circumference of the wheel, so instantaneous fluctuations in the force are captured. Is difficult.
In addition, the intermittent method cannot evaluate the action time of the derailment coefficient, which is the ratio between wheel load and lateral pressure.
On the other hand, in the continuous method (including the new continuous method described above), it is essential to measure the wheel rotation angle in order to correct the influence that the sensitivity to the force of the bridge circuit for measurement changes according to the wheel rotation angle. Become.
In the conventional continuous wheel load and lateral pressure measurement method, a dedicated slip ring device equipped with a rotation angle sensor such as a rotary encoder is installed between the axle end of the axle and the axle box in order to measure the wheel rotation angle. It was necessary to measure it by attaching it to.
In such a configuration, there is a problem that the entire slip ring device is enlarged, and even if the strain gauge for PQ measurement is sound, measurement may be difficult due to a malfunction of the rotation angle sensor. It was.
In view of the above-described problems, an object of the present invention is to provide a wheel rotation angle estimation method capable of estimating a wheel rotation angle using a bridge output for wheel load measurement without using a rotation angle sensor.

In order to solve the above-described problem, the wheel rotation angle estimation method of the present invention is configured by sequentially connecting a plurality of strain gauges affixed to the wheels of a railway vehicle, and has sensitivity to wheel load. The wheel load sensitivity with respect to the rotation angle of the wheel in the first wheel load bridge circuit and the second wheel load bridge circuit arranged at different angular positions around the axle of the first wheel load sensitivity function , And the second wheel load sensitivity function, respectively, and the output of the first wheel load bridge circuit, the output of the second wheel load bridge circuit, the first wheel load sensitivity function, the second wheel load sensitivity function, The wheel rotation angle is estimated based on the wheel load sensitivity function.
According to this, by correlating the wheel load sensitivity of the wheel load bridge circuit that does not become a simple trigonometric function with respect to the wheel rotation angle, by approximating the wheel load sensitivity by a predetermined wheel load sensitivity function, the correlation between the outputs of the pair of wheel load bridge circuits is obtained. From this, the wheel rotation angle (contact phase) can be estimated.
This makes it possible to use the output of the wheel load bridge circuit even when the output of the lateral pressure bridge circuit with a relatively small lateral pressure and a simple trigonometric function of the sensitivity function shape cannot be used to estimate the wheel rotation angle. In addition, the wheel rotation angle can be estimated without using a rotation angle sensor such as a rotary encoder.
In particular, by applying this estimation of wheel rotation angle to the continuous measurement method for wheel load and lateral pressure, the configuration of the wheel load and lateral pressure measuring device is simplified and made compact, and it can be mounted and measured in vehicles. Work can also be facilitated.

The wheel rotation angle estimation method of the present invention is configured by sequentially connecting a plurality of strain gauges affixed to the wheels of a railway vehicle, has sensitivity to wheel load, and the angular position of the strain gauge around the axle. The wheel load sensitivity with respect to the rotation angle of the wheel in the first wheel load bridge circuit and the second wheel load bridge circuit arranged different from each other, the wheel load sensitivity to the wheel load sensitivity function, A front and rear tangential force sensitivity with respect to a rotation angle of the wheel in the first wheel load bridge circuit and the second wheel load bridge circuit is approximated by a wheel weight sensitivity function, respectively, and , Approximating each by a second longitudinal tangential force sensitivity function, estimating a longitudinal tangential force acting on the wheel, and outputting the first wheel load bridge circuit, the second wheel load bridge circuit, 1 A wheel rotation angle is estimated based on the wheel load sensitivity function, the second wheel load sensitivity function, the first front / rear tangential force sensitivity function, the second front / rear tangential force sensitivity function, and the estimated front / rear tangential force. It is characterized by doing.
According to this, in addition to the effect substantially similar to the above-described effect, by estimating the longitudinal tangential force acting on the wheel, by using the longitudinal tangential force sensitivity function, the output from the wheel load bridge circuit depends on the wheel load. Only the impact can be extracted.
For this reason, it is possible to ensure the estimation accuracy of the wheel rotation angle based on the wheel load bridge output even when the longitudinal tangential force is acting on the wheel during traveling of the railway vehicle.

In the present invention, the longitudinal tangential force is estimated by applying the least squares method in the time direction using the previous history of parameters having at least wheel load, longitudinal tangential force, wheel rotation angle, and wheel rotation speed. be able to.
According to this, it is possible to accurately estimate the longitudinal tangential force based on the time history of information obtained based on the output of a general wheel load / lateral pressure measurement axle, and improve the estimation accuracy of the wheel rotation angle. Can do.
In this case, initial values of wheel load, longitudinal tangential force, wheel rotation angle, wheel rotation speed are set, and the wheel rotation angle is based on values obtained by partial differentiation of the square error function with respect to the wheel rotation angle and the wheel rotation speed, respectively. And a first process for updating the wheel rotation speed and a second process for calculating the wheel load and the front-rear tangential force on the basis of the updated wheel rotation angle and wheel rotation speed. A configuration in which the front-rear tangential force is repeatedly estimated until the end condition is satisfied can be employed.

In the present invention, the first wheel load sensitivity function and the second wheel load sensitivity function may be configured to be periodic, continuous and differentiable functions.
According to this, the effect mentioned above can be acquired reliably.
In this case, the first wheel load sensitivity function approximates the wheel load sensitivity of the first wheel load bridge circuit by a finite Fourier series, and the second wheel load sensitivity function has a finite Fourier series. The wheel load sensitivity of the second wheel load bridge circuit can be approximated.

In the present invention, when the lateral pressure acting on the wheel is equal to or greater than a predetermined value, the wheel has a plurality of strain gauges affixed to the wheel and is sensitive to the lateral pressure, and the axle around the strain gauge The rotational angle of the wheel can be estimated based on the outputs of the first lateral pressure bridge circuit and the second lateral pressure bridge circuit arranged with different angular positions.
According to this, when the lateral pressure is large and the component caused by the lateral pressure is included as noise in the wheel load bridge output, the wheel rotation angle is estimated based on the output of the lateral pressure bridge circuit, The wheel rotation angle estimation based on the double bridge output and the wheel rotation angle estimation based on the lateral pressure bridge output can complement each other properly without using the rotation angle sensor under a wide range of driving conditions. .
For example, in the region where the lateral pressure is greater than or equal to a predetermined value, the estimated value of the wheel rotation angle is a value based on the output of the lateral pressure bridge circuit, or the estimated value based on the output of the wheel load bridge circuit and the lateral pressure bridge A configuration in which an estimated value based on the output of the circuit is synthesized with a predetermined weight to obtain a final estimated value, and the weight of the estimated value based on the output of the lateral pressure bridge circuit is increased in accordance with an increase in the lateral pressure It can be.

  As described above, according to the present invention, it is possible to provide a wheel rotation angle estimation method capable of estimating a wheel rotation angle using a bridge output for wheel weight measurement without using a rotation angle sensor.

It is a figure which shows typically the structure of the rail vehicle and track | orbit in embodiment of the wheel rotation angle estimation method to which this invention is applied. It is a figure which shows the structure of the wheel of the PQ axle in the wheel rotation angle estimation method of embodiment. It is a figure which shows the connection of the bridge circuit for wheel weight measurement in the wheel of FIG. It is a figure which shows the connection of the bridge circuit for lateral pressure measurement in the wheel of FIG. It is a figure which shows an example of the lateral pressure sensitivity calculated using the test result of a PQ wheel shaft. It is a figure which shows an example of the wheel load sensitivity calculated using the test result of a PQ wheel shaft. It is a figure which shows an example of the front-back tangential force sensitivity calculated using the test result of a PQ wheel shaft. It is a figure which shows the approximation result of the lateral pressure sensitivity by a finite Fourier series. It is a figure which shows the approximation result of the wheel load sensitivity by a finite Fourier series. It is a figure which shows the approximation result of the front-back tangential force sensitivity by a finite Fourier series. It is a figure which shows the result of having compared the Fourier series at the time of approximating lateral force sensitivity, wheel load sensitivity, and front-back tangential force sensitivity by a finite Fourier series. It is a graph which shows the example which calculated the phase calculation error by the influence of the front-back tangential force with respect to several front-back tangential forces and wheel load, Comprising: The correlation of the assumed phase and the calculated phase is shown. It is a graph which shows the example which calculated the phase calculation error by the influence of the front-and-back tangential force with respect to several front-and-back tangential forces and wheel load, Comprising: The correlation with the assumed phase and phase errors is shown. It is a flowchart which shows the procedure which estimates a parameter using the least square algorithm (PLSQ method) of embodiment. It is a figure which shows the update of the sample in the least square algorithm of embodiment. It is a figure which shows the wheel rotation angle (contact phase) estimated by each algorithm of Q method, P method, and PLSQ method, Comprising: It is a thing of the linear area of EN54E1 rail. It is a figure which shows the wheel rotation angle (contact phase) estimated by each algorithm of Q method, P method, and PLSQ method, Comprising: It is a thing of R120 curve area. It is a figure which shows the wheel rotation angle (contact phase) estimated by each algorithm of Q method, P method, and PLSQ method, Comprising: It is an R700 entrance relaxation curve. It is a figure which shows the wheel rotation angle (contact phase) estimated by each algorithm of Q method, P method, and PLSQ method, Comprising: It is a thing of the linear area of a JIS50kgN rail. It is a figure which shows the lateral pressure each calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by Q method and PLSQ method. It is a figure which shows the wheel load each calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by Q method and PLSQ method. It is a figure which shows the front-back tangent force each calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by Q method and PLSQ method.

Hereinafter, an embodiment of a wheel rotation angle estimation method to which the present invention is applied will be described.
In the wheel rotation angle estimation method of the embodiment, the wheel load of the PQ wheel shaft in which a strain gauge constituting a wheel load measurement bridge circuit (wheel load bridge) is attached to the wheel without using a rotation angle sensor such as a rotary encoder. The wheel rotation angle (contact phase with the rail) is measured using a bridge output or the like.

FIG. 1 is a diagram schematically illustrating the configuration of a railway vehicle and a track in a wheel rotation angle estimation method according to an embodiment.
In the embodiment, the railway vehicle 1 is a passenger vehicle such as a train having a two-axis bogie on the front and rear of the vehicle body as an example.
As shown in FIG. 1, the railway vehicle 1 includes a vehicle body 10, a carriage frame 20, a pillow spring 30, a vertical motion damper 40, a shaft box 50, a shaft box support device 60, a PQ wheel shaft 100, and the like. .

The vehicle body 10 is a structure having a space in which passengers and the like are accommodated.
The vehicle body 10 has a frame that is provided at the bottom and to which a floor board is attached, a side edge of the frame, a side structure raised upward from the front and rear ends, a wife structure, a roof structure provided at the top, and the like. It is substantially formed in a hexahedral shape.

The cart frame 20 is a member that constitutes a main body of a two-axis bogie that is attached to the lower portion of the vehicle body 10.
The cart frame 20 is formed as a frame whose planar shape when viewed from above is substantially rectangular.
The bogie frame 20 is attached so as to be capable of relative rotation about a vertical axis and to be able to be displaced relative to the vehicle body 10 in the vertical direction so that a bogie angle can be given to the vehicle body 10.
The relative displacement in the front-rear direction between the vehicle body 10 and the carriage frame 20 is restricted by a traction device (not shown).
The carriage frame 20 is mounted with a brake device (not shown), and in the case of an electric vehicle, a main motor and the like.

The pillow spring 30 is an air spring provided between the lower part of the vehicle body 10 and the upper part of the carriage frame 20.
The pillow spring 30 supports the vehicle body 10 and transmits a load to the carriage frame 20, and generates a spring reaction force corresponding to the vertical displacement between the vehicle body 10 and the carriage frame 20.
A pair of pillow springs 30 are provided, for example, spaced apart in the sleeper direction at the center in the front-rear direction of the carriage frame 20.

The vertical motion damper 40 is a damping element such as a hydraulic damper that generates a damping force corresponding to the vertical relative speed between the vehicle body 10 and the carriage frame 20.
The vertical motion dampers 40 are provided on the left and right sides of the carriage frame 20, respectively.

The axle box 50 rotatably supports a PQ wheel shaft 100 described later.
The axle box 50 includes a bearing that rotatably supports journal portions formed at both ends of the axle 120, a lubricating device thereof, a vehicle speed generator (not shown) that detects the rotational speed of the axle 120, and the like. Has been.
Further, the axle box 50 is provided with a slip ring device (not shown) for applying an input voltage to each bridge circuit provided on the PQ wheel shaft 100 and extracting an output voltage.

The axle box support device 60 is attached to the carriage frame 20 and supports the axle box 50 and the PQ wheel shaft 100 so as to be capable of relative displacement with respect to the carriage frame 20.
The axle box support device 60 allows the vertical displacement of the axle box 50 with respect to the carriage frame 20 and also allows the steering operation in which the PQ wheel shaft 100 rotates (yaws) relative to the carriage frame 20 around the vertical axis.
The shaft box support device 60 includes a shaft beam 61, a shaft spring 62, a shaft damper 63, and the like.

The shaft beam 61 is an arm-like member whose one end is connected to the carriage frame 20 so as to be swingable.
The axle box 50 is attached to the end of the axle beam 61 opposite to the carriage frame 20 side.
An elastic body bush 61a is provided at a connection portion between the shaft beam 61 and the carriage frame 20, and the steering of the wheel shaft 100 is permitted by elastic deformation thereof.

The shaft spring 62 is a spring element that generates a spring reaction force corresponding to the vertical displacement of the axle box 50 relative to the carriage frame 20.
The shaft spring 62 is provided in the upper part of the shaft box 50, and an upper end part and a lower end part are connected to the carriage frame 20 and the shaft box 50, respectively.

The shaft damper 63 is a damping element such as a hydraulic damper that generates a damping force according to the vertical relative speed of the axle box 50 with respect to the carriage frame 20.
The shaft damper 63 is disposed in parallel to the side of the shaft spring 62 and has an upper end portion and a lower end portion connected to the carriage frame 20 and the shaft box 50, respectively.

The wheel rotation angle estimation method according to the embodiment includes a wheel weight / wheel shaft measuring device having a PQ wheel shaft (wheel weight / lateral pressure measuring wheel shaft) 100 to which wheels 110 having strain gauges attached to left and right ends of an axle 120 are attached. Use.
Drawing 2 is a figure showing the composition of the wheel of the PQ wheel axle in the wheel rotation angle estimating method of an embodiment.
FIG. 2A is a view seen from the rotation center axis direction of the wheel and the vehicle width direction outside, and FIG. 2B is a cross-sectional view taken along the line bb in FIG.
In FIG. 2 (a), 32 characters a, b, c,..., As symbols indicating positions on the circumference, are equally spaced around the center of the wheel 110 (center angle is 10 ° intervals). It is distributed and attached.

The PQ axle 100 is configured by press-fitting a pair of left and right wheels 110 in the vicinity of both ends of the axle 120 and fixing them.
The journal portions provided at both ends of the axle 120 are supported by the axle frame support device 60 so as to be capable of relative displacement in the vertical direction and the yaw direction (steer direction) with respect to the carriage frame 20.

The track includes a rail R, a sleeper S, a ballast B, and the like.
The rail R is a member on which the wheels 110 are placed, and a pair of left and right are separated from each other by a predetermined distance and are arranged substantially in parallel.
The sleeper S is a member that is disposed below the rail R and is responsible for holding the rail R between the gauges and transmitting the load from the rail R to the roadbed.
The sleepers S are, for example, prestressed concrete sleepers (PC sleepers), and a plurality of sleepers S are arranged at predetermined intervals in the traveling direction of the vehicle.
The rail R is attached to the sleeper S via an elastic fastening device (not shown).
Further, a track pad (not shown) is disposed between the lower surface portion of the rail R and the upper surface portion of the sleeper S.
The ballast B is, for example, crushed stone or the like, and constitutes a road bed on which the sleeper S is placed.

  As shown in FIG. 2, the wheel 110 is an integrated wheel in which a hub portion 111, a rim portion 112, a tread surface 113, a flange 114, a plate portion 115, and the like are integrally formed.

The hub portion 111 is a portion that is provided at the center portion of the wheel 110, and into which the axle 120 is inserted and fixed by press-fitting.
The hub portion 111 is formed with an increased thickness in the rotation axis direction (longitudinal direction of the axle 120) with respect to the plate portion 115, and a boss hole into which the axle 120 is press-fitted is formed at the center.

The rim portion 112 is a portion provided on the outer peripheral edge of the wheel 110 and formed in an annular shape concentric with the axle 120.
The rim portion 112 constitutes a so-called tire portion that rolls in contact with the rail.
The rim portion 112 is formed with an increased thickness in the rotation axis direction with respect to the plate portion 115.

The tread surface 113 is a surface portion that is formed on the outer peripheral surface portion of the rim portion 112 and contacts the upper surface of the rail R (see FIG. 1).
The tread 113 is provided with a predetermined conical tread slope in order to smoothly pass the curve with a difference in turning radius between the left and right wheels when passing the curve.

The flange 114 is a portion that projects from the vicinity of the inner end of the tread 113 to the outer diameter side in a brim shape.
The flange 114 has a function of restricting the amount of left-right movement with respect to the track of the PQ wheel shaft 100 and preventing wheel removal.
The flange 114 abuts against the side surface of the rail on the outer track side when passing through the curve, and has a function of restricting displacement in the left-right direction (sleeper direction) of the PQ wheel shaft 100.

The plate portion 115 is a substantially flat portion that connects the outer peripheral surface of the hub portion 111 and the inner peripheral surface of the rim portion 112.
An opening 116 is formed in the plate portion 115.
The opening 116 has a circular shape formed so as to penetrate the plate portion 115 in the axle direction.
For example, eight openings 116 are provided at substantially equal intervals in the circumferential direction of the plate portion 115.

  The wheel 110 includes strain gauges 131A, 131B, 132A, 132B, 133A, 133B, 134A, 134B, 135A, 135B, 136A, 136B, 137A, 137B, 138A, 138B, and a lateral pressure measurement for wheel load measurement. The strain gauges 131a, 131a ′, 133a, 133a ′, 135a, 135a ′, 137a, 137a ′ are attached.

The strain gauges 131A, 131B, 132A, 132B, 133A, 133B, 134A, 134B, 135A, 135B, 136A, 136B, 137A, 137B, 138A, and 138B for measuring the wheel load are arranged on the inner peripheral surface of the opening 116 on the wheel 110. Are sequentially affixed in the circumferential direction.
The strain gauges 131A and 131B are arranged on the inner peripheral surface of the opening 116 provided at a position corresponding to the symbol (a) (phase / angular position around the axle) in the circumferential direction of the wheel 110.

The strain gauges 132 </ b> A and 132 </ b> B are disposed on the inner peripheral surface of the opening 116 provided at a location corresponding to the symbol “e” so as to face the circumferential direction of the wheel 110.
The strain gauges 133A and 133B are disposed on the inner peripheral surface of the opening 116 provided at a position corresponding to the reference symbol so as to face the circumferential direction of the wheel 110.
The strain gauges 134A and 134B are disposed on the inner peripheral surface of the opening 116 provided at a location corresponding to the symbol W so as to face the circumferential direction of the wheel 110.
Strain gauges 135 </ b> A and 135 </ b> B are disposed on the inner peripheral surface of the opening 116 provided at a location corresponding to the reference mark so as to face the circumferential direction of the wheel 110.
The strain gauges 136 </ b> A and 136 </ b> B are arranged on the inner peripheral surface of the opening 116 provided at a location corresponding to the sign N so as to face the circumferential direction of the wheel 110.
The strain gauges 137A and 137B are disposed on the inner peripheral surface of the opening 116 provided at a position corresponding to the reference numeral so as to face the circumferential direction of the wheel 110.
The strain gauges 138A and 138B are arranged on the inner peripheral surface of the opening 116 provided at a position corresponding to the symbol m so as to face the circumferential direction of the wheel 110.

The strain gauges 131a, 131a ′, 133a, 133a ′, 135a, 135a ′, 137a, and 137a ′ for measuring the lateral pressure are the surface of the plate portion 115 (side surface of the wheel 110) and the region on the inner diameter side of the opening 116. Is arranged.
The strain gauges 131a and 131a ′ are arranged on the inner diameter side of the opening 116 provided with the strain gauges 131A and 131B.
The strain gauge 131a is affixed to the outer surface of the wheel 110 in the vehicle width direction.
The strain gauge 131a ′ is affixed to the inner surface of the wheel 110 in the vehicle width direction.
The strain gauge 131a and the strain gauge 131a ′ are disposed so as to face the rotation axis direction of the wheel 110 with the plate portion 115 interposed therebetween.

The strain gauges 133a and 133a ′ are arranged on the inner diameter side of the opening 116 provided with the strain gauges 133A and 133B.
The strain gauge 133a is affixed to the outer surface of the wheel 110 in the vehicle width direction.
The strain gauge 133a ′ is affixed to the inner surface of the wheel 110 in the vehicle width direction.
The strain gauge 133a and the strain gauge 133a ′ are disposed so as to face the rotation axis direction of the wheel 110 with the plate portion 115 interposed therebetween.

The strain gauges 135a and 135a ′ are disposed on the inner diameter side of the opening 116 provided with the strain gauges 135A and 135B.
The strain gauge 135a is affixed to the outer surface of the wheel 110 in the vehicle width direction.
The strain gauge 135a ′ is affixed to the inner surface of the wheel 110 in the vehicle width direction.
The strain gauge 135a and the strain gauge 135a ′ are disposed so as to face the rotation axis direction of the wheel 110 with the plate portion 115 interposed therebetween.

The strain gauges 137a and 137a ′ are disposed on the inner diameter side of the opening 116 provided with the strain gauges 137A and 137B.
The strain gauge 137a is affixed to the outer surface of the wheel 110 in the vehicle width direction.
The strain gauge 137a ′ is affixed to the inner surface of the wheel 110 in the vehicle width direction.
The strain gauge 137 a and the strain gauge 137 a ′ are disposed so as to face the rotation axis direction of the wheel 110 with the plate portion 115 interposed therebetween.

FIG. 3 is a diagram showing the connection of the wheel load measurement bridge circuit in the wheel of FIG.
As shown in FIGS. 2 and 3, the wheel load measuring bridge circuit (wheel load bridge) is provided with two systems, a p1 system and a p2 system.
The p1 system and p2 system wheel load measurement bridge circuits are configured such that the angular positions of the strain gauges included in the circuits are shifted by 90 ° from each other.
The p1 system wheel load measurement bridge circuit is configured by sequentially connecting strain gauges 131A, 132A, 136B, 135B, 132B, 131B, 135A, and 136A in an annular shape.
In wheel load measuring bridge circuit p1 strains, strain gauges 132A, and between 136B, strain gauge 131B, thereby applying the input voltage _{V in} and between the 135A, strain gauges 131a, and between the 131a ', strain A voltage between the gauges 135a and 135a 'is defined as an output voltage _Vout .

The p2 system wheel load measurement bridge circuit is configured by sequentially connecting strain gauges 133A, 134A, 138B, 137B, 134B, 133B, 137A, and 138A in an annular shape.
In wheel load measuring bridge circuit p2 strains, strain gauges 134A, and between 138B, strain gauge 133B, thereby applying the input voltage _{V in} and between the 137A, strain gauges 133a, and between the 133a ', strain A voltage between the gauges 137a and 137a ′ is defined as an output voltage _Vout .
The output voltages V _out (p ₁ , p ₂ ) of the p1 system and the p2 system are transmitted to a recording device installed on the vehicle via a slip ring (not shown).

FIG. 4 is a diagram showing the connection of the lateral pressure measurement bridge circuit in the wheel of FIG.
As shown in FIGS. 2 and 4, the lateral pressure measurement bridge circuit (lateral pressure bridge) is provided with two systems, q1 system and q2 system.
The bridge circuits for measuring the lateral pressure of the q1 system and the q2 system are configured such that the angular positions of the strain gauges included in the circuits are shifted by 90 ° from each other.
The bridge circuit for measuring the lateral pressure of the q1 system is configured by sequentially connecting strain gauges 131a, 131a ′, 135a ′, and 135a in an annular shape.
In q1 strains, strain gauges 131a, and between the 135a, the strain gauges 131a ', 135a' while applying the input voltage _{V in} and between the strain gauges 131a, 131a 'and between the strain gauges 135a, 135a' _{Is the} output voltage _Vout .

The q2 system lateral pressure measurement bridge circuit is configured by sequentially connecting strain gauges 133a, 133a ′, 137a ′, and 137a in an annular shape.
In q2 strains, strain gauges 133a, and between the 137a, the strain gauges 133a ', 137a' while applying the input voltage _{V in} and between the strain gauges 133a, 133a 'and between the strain gauges 137a, 137a' _{Is the} output voltage _Vout .

In this embodiment, prior to the proposal of the contact phase difference estimation method using the wheel bridge output, the bridge sensitivity function is approximated by a finite Fourier series as a premise thereof.
FIG. 5 is a diagram illustrating an example of the lateral pressure sensitivity calculated using the PQ wheel shaft test result.
FIG. 6 is a diagram illustrating an example of wheel load sensitivity calculated using the PQ wheel shaft test result.
FIG. 7 is a diagram showing an example of the longitudinal tangential force sensitivity calculated using the PQ wheel shaft test result.

As shown in FIG. 5, the relationship between the lateral pressure sensitivity and the loading phase is approximately a trigonometric function.
On the other hand, as shown in FIG. 6, the wheel load sensitivity has a shape having a plurality of peaks with respect to the loading phase, and the contact phase cannot be calculated correctly by the method using the arctangent function.
Therefore, in this embodiment, calculation equivalent to the arctangent function in a single trigonometric function is performed by approximating the wheel load sensitivity function with a finite Fourier series.

フーリエ係数Ａ_ｎｉ及びＢ_ｎｉは、ＰＱ軸検定結果を用いて、例えば最小二乗法により決定する。 Sensitivity functions corresponding to the wheel bridge circuits p1 and p2 with the contact phase φ as a factor are f ₁ (φ) and f ₂ (φ), respectively, and the actual bridge outputs are p ₁ and p ₂ .
Assuming that the wheel load actually acting is P, the relationship between the bridge output, the sensitivity function, and the wheel load can be expressed as Equations 1 and 2 in the state where the front-rear tangential force is not acting.

p ₁ = Pf ₁ (φ) (Formula 1)
p ₂ = Pf ₂ (φ) (Formula 2)

In this way, the sensitivity function f _i (φ) (i = 1 or 2) when expressing the wheel load bridge output is expressed by a finite Fourier series of the maximum order D as shown in Equation 3.

The Fourier coefficients A _ni and B _ni are determined by, for example, the least square method using the PQ axis test result.

FIG. 8 is a diagram showing an approximation result of lateral pressure sensitivity by a finite Fourier series.
FIG. 9 is a diagram showing an approximation result of wheel load sensitivity by a finite Fourier series.
FIG. 10 is a diagram showing an approximation result of tangential force sensitivity by a finite Fourier series.
The plot in FIG. 8 is the lateral pressure bridge output per 1 kN load calculated from the test result, and the curve is the result of approximating it with a finite Fourier series of order D = 11.
9 and 10 show the results of approximating wheel load sensitivity and front-rear tangential force sensitivity by the same method.
As shown in FIG. 9, it can be seen that the characteristic shape of the waveform of the wheel load sensitivity can be expressed as a continuous and differentiable function with guaranteed periodicity by the finite Fourier series.

FIG. 11 is a diagram showing a result of comparing Fourier series when lateral force sensitivity, wheel load sensitivity, and front-rear tangential force sensitivity are approximated by a finite Fourier series.
As for the lateral pressure bridge, the other coefficients are significantly smaller than the Fourier coefficient of order 1.
In the bridge circuit q1, whereas first order coefficient _{B 11} of cos function is dominant in the bridge circuit q2, a primary coefficient _{A 12} of the sin function is dominant.
This confirms that the PQ wheel shaft of this embodiment does not cause a large error even in the conventional method of handling the two systems of lateral pressure bridge outputs as a simple trigonometric function having a phase difference of 90 °. Yes.
On the other hand, in the wheel load bridge circuits p1 and p2, coefficients other than the first order are large, and in particular, odd-order high-order Fourier coefficients are identified as large values compared to the lateral pressure bridge.

The wheel load is P, the lateral pressure is Q, the longitudinal tangential force is T, and the sensitivity functions for each force are f _i (φ), h _i (φ), and g _i (φ), respectively. When the lateral sensitivity coefficient is e _i (φ), the wheel load bridge output p _i and the lateral pressure bridge output qi can be expressed by the following equations 4 to 7.

p ₁ = Pf ₁ (φ) + Tg ₁ (φ) + Qe ₁ (φ) (Formula 4)
p ₂ = Pf ₂ (φ) + Tg ₂ (φ) + Qe ₂ (φ) (Formula 5)
q ₁ = Qh ₁ (φ) (Formula 6)
q ₂ = Qh ₂ (φ) (Expression 7)

As described above, there are four equations related to the bridge output, and the number of unknowns is also four, P, Q, T, and φ. Therefore, if p1, p2, q1, and q2 are given, this combination is basically the same. Equations can be solved for unknowns.
As a simple method, first, Equation 6 and Equation 7 are simultaneously solved for Q and φ, and Equation 4 and Equation 5 are solved using the obtained φ.
If the sensitivity function of the lateral pressure measurement bridge can be regarded as sinusoidal, it can be easily solved using the arc tangent function, but the above equation based on the finite Fourier series expression can be solved by a numerical solution such as Newton's method. Q and φ can also be obtained by solving.

However, the direction of action of the lateral pressure Q may vary during travel.
For example, the sign of the bridge output is reversed between when a positive lateral pressure is applied to the wheel and when a negative lateral pressure is applied.
This corresponds to a 180 ° phase shift when viewed in phase space.
That is, the contact phase φ obtained by the above method cannot be distinguished from a position shifted by 180 ° from the actual contact phase.

とし、これを式６及び式７に代入して輪重Ｐ及び前後接線力Ｔを計算する。
横圧の作用方向に起因して、推定された位相が実際の接触位相から１８０°ずれている場合には、推定された位相を用いて計算した輪重Ｐは負の値となり、横圧の符号反転を判定することができる。
この場合には、推定された位相を１８０°シフトさせた位相

が正しい接触位相である。 In order to solve this, a bridge circuit for wheel load measurement is utilized.
Note that, unlike lateral pressure, the wheel load always acts in the positive direction.
Phase estimated from the bridge output for lateral pressure measurement

By substituting this into Equation 6 and Equation 7, the wheel load P and the longitudinal tangential force T are calculated.
When the estimated phase is shifted by 180 ° from the actual contact phase due to the acting direction of the lateral pressure, the wheel load P calculated using the estimated phase becomes a negative value, Sign inversion can be determined.
In this case, a phase obtained by shifting the estimated phase by 180 °

Is the correct contact phase.

In the above-described method, it is difficult to estimate the phase using the lateral pressure measurement bridge output in a state where the lateral pressure is hardly applied.
In that case, it is necessary to obtain the contact phase φ using the output of the wheel bridge.
By using the finite Fourier series expression described above, if the bridge outputs p ₁ and p ₂ are given, the simultaneous equations consisting of Equations 1 and 2 can be obtained by Newton's method or the like if the longitudinal tangential force is not acting. By solving, it is possible to calculate the contact phase φ and the wheel load P.

を作る。 In the following, the effect of the longitudinal tangential force on the phase calculation results will be examined.
In order to investigate the influence of unknown longitudinal tangential force on contact phase calculation using only the wheel load measurement bridge output, the following calculation was tried.
First, the simulated bridge output is based on the data when the PQ wheel axis is verified for the longitudinal tangential force.

make.

このとき、Ｐ及びＴは、適当に設定する。 That is, assuming that the lateral pressure is Q = 0 and the assumed contact phase is 0 ≦ φ ≦ 2π, the following equations 8 and 9 are calculated.

At this time, P and T are set appropriately.

及び、輪重推定値

を計算する。
すなわち、以下の式１０、式１１を、これらの各推定値について解く。
これにより、ブリッジ出力には前後接線力の影響が含まれているものの、接触位相を計算する方程式には前後接線力が考慮されていないという状況を模擬する。 Next, using the above-described equations 1, 2, 8, and 9, the estimated value of the contact phase

And estimated wheel load

Calculate
That is, the following equations 10 and 11 are solved for each of these estimated values.
This simulates a situation in which the bridge output includes the influence of the longitudinal tangential force, but the equation for calculating the contact phase does not consider the longitudinal tangential force.

の差を評価する。
図１２、図１３は、複数の前後接線力Ｔと輪重Ｐに対して、前後接線力の影響による位相計算誤差を計算した例を示すグラフである。
図１２において、横軸は想定した位相を示し、縦軸は計算された位相を示している。
図１３において、横軸は想定した位相を示し、縦軸は位相誤差を示している。 Next, the assumed contact phase φ and the estimated contact phase

Evaluate the difference.
FIGS. 12 and 13 are graphs showing an example in which the phase calculation error due to the influence of the longitudinal tangential force is calculated for a plurality of longitudinal tangential forces T and wheel loads P. FIG.
In FIG. 12, the horizontal axis represents the assumed phase, and the vertical axis represents the calculated phase.
In FIG. 13, the horizontal axis indicates the assumed phase, and the vertical axis indicates the phase error.

In a state where the longitudinal tangential force does not act (T = 0), the phase error is negligible.
On the other hand, when the longitudinal tangential force is acting, it turns out that the phase calculation result is distorted.
The smaller the wheel load is relative to the longitudinal tangential force, the greater this distortion.
It can also be seen that there are regions that are susceptible to the influence of the front-rear tangential force and regions that are less likely to be affected by the assumed phase.
Compared to FIG. 9, it can be seen that it is particularly susceptible to the influence of the longitudinal tangential force near the phase where the bridge sensitivity of either p1 or p2 is maximum.

To summarize the above-described contents, the simultaneous equations (formulas 4 to 7) that normally melt sufficiently including the contact phase φ become singular when the lateral pressure Q approaches zero, and P, Q, T, φ It is a problem that it becomes impossible to solve about.
In other words, when the lateral pressure Q is close to zero, information is insufficient only with the bridge outputs p ₁ (t), P _{2 (} t), q ₁ (t), and q ₂ (t) at a specific time t. Means.

This is because, although there are three physical quantities to be obtained, P, T, and φ, there are only two equations 4 and 5 that can be practically used.
In the PLSQ method according to the present embodiment, a bridge output sequence in the time direction (time series) is used to compensate for the shortage of information.
Hereinafter, a method will be described in which P, T, φ, and wheel rotation speed ω are regarded as parameters, and these parameters are sequentially estimated by applying the least square method continuously in the time direction.

とする。
ただし、ｓは、最小二乗法を適用するデータサンプル数である。
また、インデックスｋを負値としているのは、基準時刻ｔ_０を起点として、時間を遡る方向のデータを使用することを明記するためである。
取り出したｓ個のサンプルの間は、輪重Ｐ、前後接線力Ｔ、回転角速度ωは変化しないと仮定する。
言い換えれば、ｓ個のサンプルにおける平均的な輪重、前後接線力、回転角速度の推定値

を求める。 The wheel load bridge output at time t = t _k (k = 0, −1, −2,..., −s + 1)

And
Here, s is the number of data samples to which the least square method is applied.
Also, what the index k and negative values, starting from the reference time t _0, in order to specify the use of the direction of the data back in time.
It is assumed that the wheel load P, the longitudinal tangential force T, and the rotational angular velocity ω do not change during the s samples taken out.
In other words, an estimate of the average wheel load, longitudinal tangential force and rotational angular velocity in s samples

Ask for.

とすると、理想的にはｋ番目のサンプルに対応するブリッジ出力は、以下の式１２、式１３のように表せる。

ただし、簡略化のため、輪重ブリッジの横圧に関する横感度項は無視している。 The sampling period is Δt, and the phase estimation value at time t ₀ is

Then, ideally, the bridge output corresponding to the k-th sample can be expressed by the following equations 12 and 13.

However, for the sake of simplicity, the lateral sensitivity term related to the lateral pressure of the wheel bridge is ignored.

が既知の場合に、

を最小二乗法により計算する方法について述べる。
これは、パラメータを

基底関数ｆ_１，ｇ_１，ｆ_２，ｇ_２とする線形最小二乗問題であり、最終的に線形連立方程式に帰着できる。 First,

Is known,

A method of calculating by the least square method will be described.
This is a parameter

This is a linear least square problem with basis functions f ₁ , g ₁ , f ₂ , and g _2, and can ultimately be reduced to a linear simultaneous equation.

ただし、

である。
二乗誤差関数を、

に関して偏微分すると、以下の式１５、式１６を得る。

The square error function E is defined as the following Expression 14.

However,

It is.
The square error function is

The following equations 15 and 16 are obtained by partial differentiation with respect to.

を解けばよい。
式１５、式１６の左辺を０とおいて、

について整理すると、以下の式１７を得る。

式１７を

について解くことで、輪重及び前後接線力の推定値を計算できる。 The parameter to be obtained is a parameter that stops the square error.

Can be solved.
With the left side of Equation 15 and Equation 16 set to 0,

The following formula 17 is obtained.

Equation 17

By solving for, it is possible to calculate the estimated values of wheel load and longitudinal tangential force.

が既知の場合に、

を求める問題を考える。
二乗誤差関数は、

を求める場合と同じであるが、この場合パラメータ

が基底関数に関して線形分離されていないため、勾配法による反復解法を適用する。
二乗誤差関数を

について偏微分すると、以下の式１８、式１９を得る。

ただし、関数ｆ，ｇに対して、ｆ’，ｇ’は、位相に関する常微分を表す。 continue,

Is known,

Think about the problem you want.
The square error function is

Is the same as

Is not linearly separated with respect to the basis function, so the iterative solution by the gradient method is applied.
The square error function

Is partially differentiated, the following equations 18 and 19 are obtained.

However, for the functions f and g, f ′ and g ′ represent ordinary differentials with respect to the phase.

を用いて、以下の式２０、式２１により、接触位相及び回転角度推定値の更新を行う。

ただし、α_φ，α_ω＞０は、計算の安定化のための係数である。 Calculated

Is used to update the contact phase and the rotation angle estimated value according to the following Expression 20 and Expression 21.

However, α _φ and α _ω > 0 are coefficients for stabilizing the calculation.

FIG. 14 is a flowchart illustrating a procedure for estimating a parameter using a least square algorithm.
Hereinafter, the steps will be described step by step.

の初期パラメータを、適当に設定する。
その後、ステップＳ０２に進む。 <Step S01: Initial parameter setting>

Set the initial parameters appropriately.
Thereafter, the process proceeds to step S02.

を更新する。
その後、ステップＳ０３に進む。 <Step S02: Update Wheel Rotation Angle and Wheel Rotation Speed Estimated Value>
Estimated values of wheel rotation angle and wheel rotation speed using the above recurrence formulas (Formula 20 and Formula 21)

Update.
Thereafter, the process proceeds to step S03.

を用いて、輪重、前後接線力の推定値

を計算する。
その後、ステップＳ０４に進む。 <Step S03: Calculate wheel load and longitudinal tangential force estimate>
Estimated values of wheel rotation angle and wheel rotation speed updated in step S02

Is used to estimate wheel load and longitudinal tangential force

Calculate
Thereafter, the process proceeds to step S04.

<Step S04: Determination of Completion of Iterative Calculation>
It is determined whether or not a preset iteration end condition is satisfied. If the iteration end condition is satisfied, a series of processes is terminated.
If the iteration end condition is not satisfied, the process returns to step S02 and the subsequent processing is repeated.
For example, that the repetition of certain number n _i is completed, it is possible to repeat end condition.

FIG. 15 is a diagram illustrating sample update in the least square algorithm of the embodiment.
After the iteration end condition is satisfied and the calculation at time t ends, as shown in FIG. 12, a sample at time t + Δt is added, and the oldest sample among the data used for parameter estimation at time t is deleted.
That is, s samples in the direction of going back from the parameter estimation target time t are used for parameter estimation at each time.
Table 1 shows each parameter of the least square algorithm in the present embodiment.

  Hereinafter, a verification result of the wheel rotation angle estimation method of the present embodiment using wheel load / lateral pressure waveform data at the time of a running test performed by running an actual railway vehicle will be described.

Comparison of the phase estimation results by each algorithm is shown in FIGS.
16 to 19 are diagrams showing wheel rotation angles (contact phases) estimated by the respective algorithms of the Q method, the P method, and the PLSQ method.
The Q method is a method for estimating the wheel rotation angle based on the lateral pressure bridge output.
The P method is a method of estimating the wheel rotation angle using Equations 1 and 2 based on the wheel load bridge output.
The PLSQ method is a method of estimating by the least square method based on the wheel load bridge output.
All phase estimations were performed offline after applying a second-order linear low-pass filter with a cutoff frequency of 200 Hz to the original waveforms of the wheel load bridge output and the lateral pressure bridge output.

FIG. 16 shows a case where powering is performed in the EN54E1 rail straight section.
The tie rate gradient of the EN54E1 rail straight section is 20: 1, and in the case of the combination with the modified arc tread, the wheel-rail contact point moves to the outside of the track as compared to the combination with the JIS 50 kgN rail.
For this reason, the equivalent tread gradient is large, and even in a straight section, the lateral pressure is relatively large.
Therefore, although it is a straight line, the phase estimation result by the Q method is good.
On the other hand, when the P method and the PLSQ method, which are two estimation methods using the wheel bridge output, are compared, the P method shows the same distortion as the above-described trial calculation, but the estimation result of the PLSQ method is the Q method described above. The PLSQ method is almost the same as the estimation result by the PLS method, and a more appropriate phase estimation result is obtained for the P method.

FIG. 17 shows a case where the vehicle travels at an approximately constant speed in the R120 circular curve section.
In this section, a sufficiently large lateral pressure is acting, and this is one of the sections that can be relied upon in the estimation result by the Q method.
In this section, the phase estimation result by the PLSQ method is slightly distorted, which is presumed to be because the lateral pressure sensitivity of the wheel bridge output is not considered in the PLSQ method.
However, as described above, a reliable phase estimation result can be obtained by the Q method in the section where the lateral pressure is applied, and it is considered that more appropriate phase estimation can be performed by using both in combination.

FIG. 18 is for the R700 inlet relaxation curve.
FIG. 19 shows a straight section using a JIS 50 kgN rail.
All the data are intervals in which the lateral pressure is close to zero and a reliable phase estimation result cannot be obtained by the Q method.
In particular, although the phase estimation by the Q method is not stable near the time when the lateral pressure crosses zero, it can be confirmed that stable phase estimation can be performed by the PLSQ method even in such a situation.
From the above results, it is clear that even when the lateral pressure Q is small and a reliable phase estimation result cannot be obtained by the Q method, a reasonable phase estimation can be performed by the PLSQ method using the least square method. It was.

FIG. 20 to FIG. 22 show the results of wheel load, lateral pressure, and longitudinal tangential force calculation by the same calculation method as the so-called new continuous method, using the estimated wheel rotation angle (contact phase).
FIG. 20 is a diagram illustrating the lateral pressure calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by the Q method and the PLSQ method, respectively. .
FIG. 21 is a diagram showing wheel weights calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by the Q method and the PLSQ method, respectively. .
FIG. 22 is a diagram showing front and rear tangential forces calculated using the wheel rotation angle (contact phase) measured by the rotation angle sensor, the wheel rotation angle (contact phase) estimated by the Q method and the PLSQ method, respectively. is there.

20 to 22, the upper graph uses the contact phase measured by the rotation angle sensor (rotary encoder).
The middle and lower graphs use contact phases estimated by the Q method and the PLSQ method, respectively.
Note that the P method was excluded from the scope of this study because distortion due to the longitudinal tangential force occurred in the section where the longitudinal tangential force occurred.

For the lateral pressure shown in FIG. 20, almost the same calculation result is obtained regardless of the phase estimation (measurement) technique.
This is thought to be because the lateral pressure bridge output is approximately a single trigonometric function, and the dependence of the lateral pressure calculation result on the wheel rotation angle is extremely small.
On the other hand, in the wheel load shown in FIG. 21 and the longitudinal tangential force shown in FIG. 22, the difference between the estimation methods is large. Generally, the Q method is a curved section (large lateral pressure Q), and the PLSQ method is a straight section ( The lateral pressure is small (Q), which is compatible with the new continuous method using a rotary encoder.
This corresponds to the stability of the phase estimation for each section described above.

Therefore, in the present embodiment, a predetermined threshold is provided for the lateral pressure Q detected based on the wheel load bridge output, and the wheel rotation angle (contact phase) is determined by the PLSQ method in the region where the lateral pressure Q is less than the threshold. It is also possible to make a configuration in which the wheel rotation angle is estimated by the Q method in a region where the lateral pressure Q is greater than or equal to the threshold.
Further, the wheel rotation angles estimated by the PLSQ method and the Q method are respectively weighted to obtain the final estimated wheel rotation angle (contact phase), and the estimated value by the Q method is increased according to the increase in the lateral pressure Q. It is also possible to increase the weight.

According to the embodiment described above, the following effects can be obtained.
(1) The wheel load sensitivity of the wheel load bridge circuits p1 and p2 is approximated by wheel load sensitivity functions f ₁ (φ) and f ₂ (φ), and the wheel load bridge outputs p ₁ and p ₂ and the wheel load sensitivity function f By estimating the wheel rotation angle (contact phase) φ based on ₁ (φ) and f ₂ (φ), the lateral pressure Q is relatively small, and the outputs of the lateral pressure bridge circuits q1 and q2 are estimated as the wheel rotation angle φ. Even if it cannot be used, the wheel rotation angle φ can be estimated without using a rotation angle sensor such as a rotary encoder by using the outputs of the wheel bridge circuits p1 and p2.
In particular, by applying such estimation of the wheel rotation angle φ to a known continuous measurement method of wheel load / lateral pressure, the rotation angle sensor is unnecessary and the slip ring is miniaturized. The configuration of the measuring device can be simplified and made compact, and mounting on a vehicle and measurement work can be facilitated.
(2) The longitudinal tangential force sensitivity of the wheel bridge circuits p1, p2 is approximated by the longitudinal tangential force sensitivity functions g ₁ (φ), g ₂ (φ), and the estimated longitudinal tangential force T and the longitudinal tangential force sensitivity function By extracting the wheel rotation angle φ using g ₁ (φ) and g ₂ (φ), only the influence of the wheel load P is extracted from the outputs p ₁ and p ₂ of the wheel load bridge circuits p ₁ and p _2. Can do.
For this reason, even when the longitudinal tangential force T is acting on the wheel 110 during traveling of the railway vehicle, it is possible to ensure the estimation accuracy of the wheel rotation angle φ based on the wheel load bridge outputs p ₁ and p _2. it can.
(3) The longitudinal tangential force T is estimated by applying the least squares method in the time direction using a conventional history of parameters including the wheel load P, the longitudinal tangential force T, the wheel rotational angle φ, and the wheel rotational speed ω. More specifically, the initial values of the wheel load P, the longitudinal tangential force T, the wheel rotation angle φ, and the wheel rotation speed ω are set, and the square error function E is deviated by the wheel rotation angle φ and the wheel rotation speed ω, respectively. A first process for updating the wheel rotation angle φ and the wheel rotation speed ω based on the differentiated value, and calculating the wheel load P and the longitudinal tangential force T based on the updated wheel rotation angle φ and the wheel rotation speed ω. Based on the time history of the information obtained based on the output of the general PQ wheel shaft 100 by repeating the second process to repeat until a predetermined iteration end condition is satisfied, T is estimated accurately and the wheel rotation angle φ is estimated accurately. The degree can be improved.
(4) The wheel load sensitivity functions f ₁ (φ) and f ₂ (φ) are cyclic and continuous and differentiable with respect to the wheel load sensitivity of the wheel bridge circuits p1 and p2 by a finite Fourier series. By approximating as a function, the above-described effects can be obtained with certainty.
(5) In the region where the lateral pressure P is large, the wheel rotation angle φ is estimated by using the Q method using the outputs q ₁ and q ₂ of the lateral pressure bridge circuits q1 and q2, so that the wheel load bridge output is obtained. The wheel rotation angle φ can be appropriately estimated under a wide range of driving conditions without using a rotation angle sensor by mutually complementing the wheel rotation angle estimation based on the wheel pressure angle estimation based on the lateral pressure bridge output.

(Other embodiments)
In addition, this invention is not limited only to embodiment mentioned above, Various application and deformation | transformation can be considered.
(1) The vehicle type and configuration of the railway vehicle to which the present invention is applied, the wheel shape of the PQ wheel shaft, the application pattern of the strain gauge, the connection of the bridge circuit, and the like are not limited to the embodiment and can be changed as appropriate.
(2) Numerical calculation methods and specific mathematical formulas are not limited to the embodiment, and can be changed as appropriate.
For example, the method for generating the wheel load sensitivity function is not limited to the one using the finite Fourier series as in the embodiment, and may be another method. For example, spline interpolation may be used.
Further, the method for minimizing the error by the least square method is not particularly limited.
(3) The wheel rotation angle estimation method of the present invention can be applied not only to wheel load / lateral pressure measurement by the new continuous method as described in the embodiment but also to other uses.
For example, it is also possible to calculate the wheel rotation angles (contact phases) of the left and right wheels of the same wheel axis, and to estimate the attack angle, which is the yaw angle with respect to the wheel track, based on the deviation.
(4) In the embodiment, the wheel load bridge output and the lateral pressure bridge output are recorded while the vehicle is running, and then the calculation for estimating the wheel rotation angle is performed offline. It is good also as a structure performed by.

DESCRIPTION OF SYMBOLS 1 Railway vehicle 10 Car body 20 Bogie frame 30 Pillow spring 40 Vertical motion damper 50 Shaft box 51 Acceleration sensor 60 Shaft box support device 61 Shaft beam 61a Elastic body bush 62 Shaft spring 63 Shaft damper 100 PQ wheel shaft 110 Wheel 111 Hub part 112 Rim part 113 Tread surface 114 Flange 115 Plate portion 116 Opening 120 Axle 131A to 138A, 131B to 138B Strain gauge for measuring wheel load 131a to 137a, 131a 'to 137a' Strain gauge for measuring lateral pressure R Rail S Sleeper B Ballast

Claims (7)

A plurality of strain gauges affixed to the wheels of the railway vehicle are sequentially connected to each other, have sensitivity to wheel load, and are arranged with the angular positions of the strain gauges around the axle different from each other. Approximating the wheel load sensitivity to the rotation angle of the wheel in the wheel load bridge circuit and the second wheel load bridge circuit by the first wheel load sensitivity function and the second wheel load sensitivity function,
Estimating the wheel rotation angle based on the output of the first wheel load bridge circuit, the output of the second wheel load bridge circuit, the first wheel load sensitivity function, and the second wheel load sensitivity function. A method for estimating a wheel rotation angle. A plurality of strain gauges affixed to the wheels of the railway vehicle are sequentially connected to each other, have sensitivity to wheel load, and are arranged with the angular positions of the strain gauges around the axle different from each other. Approximating the wheel load sensitivity to the rotation angle of the wheel in the wheel load bridge circuit and the second wheel load bridge circuit by the first wheel load sensitivity function and the second wheel load sensitivity function,
The first anteroposterior tangential force sensitivity function and the second anteroposterior tangential force sensitivity function are defined as the longitudinal tangential force sensitivity with respect to the rotation angle of the wheel in the first wheel load bridge circuit and the second wheel load bridge circuit. Are approximated by
Estimating the longitudinal tangential force acting on the wheel,
An output of the first wheel load bridge circuit, an output of the second wheel load bridge circuit, the first wheel load sensitivity function, the second wheel load sensitivity function, the first longitudinal tangential force sensitivity function, A wheel rotation angle estimation method, wherein the wheel rotation angle is estimated based on the second longitudinal tangential force sensitivity function and the estimated longitudinal tangential force. The longitudinal tangential force is estimated by applying the least square method in the time direction using at least the previous history of parameters having wheel load, longitudinal tangential force, wheel rotation angle, and wheel rotation speed. 2. The wheel rotation angle estimation method according to 2. Set initial values for wheel load, front / rear tangential force, wheel rotation angle, wheel rotation speed,
A first process for updating the wheel rotation angle and the wheel rotation speed based on a value obtained by partially differentiating a square error function by the wheel rotation angle and the wheel rotation speed, respectively, and the updated wheel rotation angle and 4. The longitudinal tangential force is estimated by repeating the second process of calculating the wheel load and the longitudinal tangential force based on a wheel rotational speed until a predetermined iteration end condition is satisfied. Wheel rotation angle estimation method. The first wheel load sensitivity function and the second wheel load sensitivity function are periodic, continuous, and differentiable functions, respectively. The wheel rotation angle estimation method according to item. The first wheel load sensitivity function approximates the wheel load sensitivity of the first wheel load bridge circuit by a finite Fourier series,
The wheel rotation angle estimation method according to claim 5, wherein the second wheel load sensitivity function approximates the wheel load sensitivity of the second wheel load bridge circuit by a finite Fourier series. When the lateral pressure acting on the wheel is equal to or greater than a predetermined value, it has a plurality of strain gauges affixed to the wheel and is sensitive to lateral pressure, and the angular position of the strain gauge around the axle is determined. The rotation angle of the wheel is estimated based on outputs of the first lateral pressure bridge circuit and the second lateral pressure bridge circuit that are arranged differently. 7. The wheel rotation angle estimation method according to item 1.

Images