JP4825110B2 - Pantograph contact force estimation system and estimation method - Google Patents

Description

  The present invention relates to a system and method for estimating the contact force of a pantograph sliding plate to a trolley line.

  Since the train line equipment is a long equipment peculiar to railways, maintenance of overhead lines requires cost and time. At present, the trolley wire wear diameter is measured by the inspection vehicle, and the worker is checking the point of caution at night, and when the wear exceeds the allowable value, the maintenance is performed.

On the other hand, the contact force between the overhead line and the pantograph (hereinafter simply referred to as “contact force”), which has become measurable by recent research, is one of the important indexes for evaluating current collection performance. As a trend in recent years, there is a flow of measuring contact force with a pantograph such as an inspection car and using it for maintenance, and studies are also underway on methods of using contact force such as measurement of wave propagation velocity using contact force. (Non-Patent Document 1)
Further, in the current maintenance of train lines, suppression of trolley line wear is one of the important issues, but trolley line wear has a very complicated formation mechanism. The main causes of trolley wire wear formation include pantograph characteristics, pantograph current collection, sliding plate characteristics, travel speed, overhead wire type, overhead wire configuration, and trolley wire surface condition.

As described above, although the cause of friction formation of the trolley wire is extensive, it is considered useful to examine the causal relationship between the contact force and the wear of the trolley wire.
“Examination of train line diagnostic technique by measuring contact force between overhead line and pantograph” Second Symposium on Evaluation and Diagnosis, pages 105-110, December 2003

  However, the conventional contact force measurement is performed by driving the vehicle with an accelerometer or strain gauge on the pantograph hull of a specific vehicle such as an inspection vehicle and the pantograph hull in contact with the trolley wire. It is measured by letting. Therefore, the contact force between the overhead line and the pantograph can only be measured for a specific pantograph of a specific vehicle. Actually, since vehicles having a wide variety of pantographs travel on the overhead line, it is desirable to obtain these contact forces.

  An object of this invention is to provide the pantograph contact force estimation system and the estimation method which can acquire the contact force of various pantographs.

In the present invention, in order to obtain the contact force of various pantographs passing through a certain section, the contact force is estimated based on the information of sensors installed on the overhead line. Schematically, in the present invention, a contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire is a hanger force of a hanger connected to the trolley wire existing between measurement sections. The contact force f (t) is calculated from the inclination at both ends of the measurement section and the distribution of acceleration of the trolley line in the measurement section. More specifically, the object of the present invention is to provide a pantograph hull plate and a trolley. A contact force estimation system for estimating a contact force with a line,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval gradient _{(∂y / ∂x x = + ε} , ∂y / ∂x x = -ε), as well as data indicating continuous distribution of acceleration of the trolley wire in the measurement interval the (∂ ² y / ∂t ^2), respectively And a storage device storing
From the data showing the continuous distribution of the hanger force, tilt and acceleration, the following equation

(T: trolley wire tension, ρ: trolley wire linear density)
A contact force calculation means for calculating a contact force f (t) based on
This is achieved by a contact force estimation system characterized by comprising:

In a preferred embodiment, the storage device further includes data on force terms due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), and
The contact force calculation means is
From the data on the term of force due to the continuous distribution of the hanger force, inclination and acceleration, and the bending stiffness of the trolley wire at both ends of the measurement section,

(T: Trolley wire tension, ρ: Trolley wire linear density, EI: Trolley wire bending stiffness)
Based on the above, the contact force f (t) is calculated.

Further, an object of the present invention is a contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger on the trolley line in the measurement section by a predetermined distance A storage device storing data indicating (∂ ² y _j / ∂t ² : j = 1, 2,..., P),
From the data indicating the hanger force, tilt and acceleration, the following formula

(T: trolley wire tension, ρ: trolley wire linear density)
A contact force calculation means for calculating a contact force f (t) based on
This is achieved by a contact force estimation system characterized by comprising:

In a preferred embodiment, the storage device further includes data on force terms due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), and
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the measurement section, the contact force calculation means calculates the following formula:

(T: tension of trolley wire, ρ: linear density of trolley wire, w _j : inertia force correction coefficient, EI: bending rigidity of trolley wire)
Based on the above, the contact force f (t) is calculated.

Further, an object of the present invention is a contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger on the trolley line in the measurement section by a predetermined distance A storage device storing data indicating (∂ ² y _i / ∂t ² : i = 1, 2,..., M),
From the data indicating the hanger force, tilt and acceleration, the following formula

(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force)
A contact force calculation means for calculating a contact force f (t) based on
This is achieved by a contact force estimation system characterized by comprising:

In a preferred embodiment, the storage device further includes data on force terms due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), and
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the measurement section, the contact force calculation means calculates the following formula:

(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force, EI: trolley wire bending stiffness)
Based on the above, the contact force f (t) is calculated. Further, in a preferred embodiment, the storage device associates the acceleration data for learning, the inertial force data for learning, and the neural network model with a connection between neurons constituting the neural network. Stored neural network model data including the weight and threshold value of each neuron, and acceleration data for estimation,
The contact force calculating means is a learning means for making weights and thresholds included in the neural network model data appropriate values,
Read the initial neural network model data stored in the storage device, give the learning acceleration data to the input layer of the neural network model, and from the input layer to the output layer through the intermediate layer In turn, according to the neuron connection in the neural network model data, an output function based on a value obtained by subtracting a threshold from the sum of the multiplication results of the signal from the neuron closer to the input layer and the weight associated with the connection. Computing means for repeatedly calculating an output value and outputting the output value to a neuron coupled to the output layer;
Inertial force data for learning is applied to the output layer of the neural network model, an error signal is propagated from the output layer to the input layer through the intermediate layer, and at least the coupling in the neural network data. Learning means comprising: error backpropagation means for modifying associated weights and thresholds for the neurons, and storing the weights associated with the modified connections and thresholds for the neurons as modified neural network data; ,
Read out the neural network model data corrected by learning by the learning means stored in the storage device, give the estimation acceleration data to the input layer of the neural network model, and from the input layer through the intermediate layer A value obtained by subtracting a threshold value from the sum of multiplication results of signals from neurons closer to the input layer and weights associated with the connection in accordance with the connection of neurons in the neural network model data sequentially toward the output layer The output function is calculated based on the output function, the output value is repeatedly output to the neuron coupled to the output layer, and the value from the output layer is used as the estimated inertia force _fine . And estimation means for storing in the storage device.

  In one embodiment, for example, there are two measurement points between the connection points, and approximately ± 1/4 * from the connection point (distance between the connection points) (+ is the traveling direction of the train). Only a distance away.

  In another embodiment, there are two measurement points between the connection points, and approximately + 1/2 * (distance between connection points) and approximately −1 / 4 * (connection points) from the connection points. Are separated by a distance of

In addition, an object of the present invention is to provide a hanger force h _i (i = 1, 2,..., N) of a hanger connected to the trolley wire that exists at least during the measurement interval (−ε <x <ε). ), Slopes at both ends of the measurement interval (∂y / ∂x _{x = + ε} , −y / に お け_{る x x = -ε} ), and continuous distribution of trolley line acceleration in the measurement interval (∂ ² y / ∂). In an information processing apparatus including a storage device that stores data indicating each of t ² ), a method for estimating a contact force between a pantograph boat hull and a trolley wire,
From the data showing the continuous distribution of the hanger force, tilt and acceleration, the following equation

(T: trolley wire tension, ρ: trolley wire linear density)
This is achieved by a contact force estimation method characterized by calculating a contact force f (t) based on the above.

In a preferred embodiment, the storage device further includes data on force terms due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), and
From the data on the term of force due to the continuous distribution of the hanger force, inclination and acceleration, and the bending stiffness of the trolley wire at both ends of the measurement section,

(T: Trolley wire tension, ρ: Trolley wire linear density, EI: Trolley wire bending stiffness)
Based on this, the contact force f (t) is calculated.

Furthermore, an object of the present invention is to provide a hanger force h _i (i = 1, 2,..., N) of a hanger connected to a trolley wire that exists at least during the measurement interval (−ε <x <ε). ), Slopes at both ends of the measurement interval (∂y / ∂x _{x = ε} , ∂y / ∂x _{x = -ε} ), and a predetermined distance from the connection point with the hanger on the trolley line in the measurement interval Pantograph hull in an information processing apparatus including a storage device storing data indicating accelerations (∂ ² y _j / ∂t ² : j = 1, 2,..., P) at a plurality of measurement points. A method for estimating a contact force between a sliding plate and a trolley wire,
From the data indicating the hanger force, tilt and acceleration, the following formula

(T: trolley wire tension, ρ: trolley wire linear density)
This is achieved by a contact force estimation method characterized by calculating a contact force f (t) based on the above.

In a preferred embodiment, the storage device further includes data on force terms due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), and
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the hanger force, inclination, acceleration, and measurement section, the following formula

(T: tension of trolley wire, ρ: linear density of trolley wire, w _j : inertia force correction coefficient, EI: bending rigidity of trolley wire)
Based on the above, the contact force f (t) is calculated.

In addition, an object of the present invention is to provide a hanger force h _i (i = 1, 2,..., N) of a hanger connected to the trolley wire that exists at least during the measurement interval (−ε <x <ε). ), Slopes at both ends of the measurement interval (∂y / ∂x _{x = ε} , ∂y / ∂x _{x = -ε} ), and a predetermined distance from the connection point with the hanger on the trolley line in the measurement interval Pantograph hull in an information processing apparatus having a storage device storing data indicating accelerations (∂ ² y _i / ∂t ² : i = 1, 2,..., M) of a plurality of measurement points. A method for estimating a contact force between a sliding plate and a trolley wire,
From the data indicating the hanger force, tilt and acceleration, the following formula

(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force)
This is achieved by a contact force estimation method characterized by calculating a contact force f (t) based on the above.

In a preferred embodiment, the storage device further includes force term data (に よ_る ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x =-} ) at both ends of the measurement section. _ε )
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the measurement section, the contact force calculation means calculates the following formula:

(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force, EI: trolley wire bending stiffness)
Based on the above, the contact force f (t) is calculated.

In a preferred embodiment, the storage device includes, for the learning acceleration data, learning inertia force data, and a neural network model, weights associated with connections between neurons constituting the neural network. Storing neural network model data including threshold values of each neuron and acceleration data for estimation,
When estimating the contact force, the step of calculating the inertial force _fine includes:
Read the initial neural network model data stored in the storage device, give the learning acceleration data to the input layer of the neural network model, and from the input layer to the output layer through the intermediate layer In turn, according to the neuron connection in the neural network model data, an output function based on a value obtained by subtracting a threshold from the sum of the multiplication results of the signal from the neuron closer to the input layer and the weight associated with the connection. A calculation step of repeatedly calculating the output value and outputting the output value to a neuron coupled to the output layer side; and
Inertial force data for learning is applied to the output layer of the neural network model, an error signal is propagated from the output layer to the input layer through the intermediate layer, and at least the coupling in the neural network data. A learning step comprising correcting an associated weight and a threshold for the neuron, and storing a weight associated with the modified connection and the threshold for the neuron as modified neural network data;
Read out the neural network model data corrected by learning in the learning step stored in the storage device, give the estimation acceleration data to the input layer of the neural network model, and from the input layer through the intermediate layer A value obtained by subtracting a threshold value from the sum of multiplication results of signals from neurons closer to the input layer and weights associated with the connection in accordance with the connection of neurons in the neural network model data sequentially toward the output layer The output function is calculated based on the output function, the output value is repeatedly output to the neuron coupled to the output layer, and the value from the output layer is used as the estimated inertia force _fine . And an estimation step for storing in the storage device.

  ADVANTAGE OF THE INVENTION According to this invention, it becomes possible to provide the pantograph contact force estimation system and the estimation method which can acquire the contact force of various pantographs.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. FIG. 1 is a diagram schematically showing an example of the structure of an overhead wire. As shown in FIG. 1, the boat body 3 of the pantograph 2 contacts the trolley wire 10 of the overhead line 1. In the example of FIG. 1, the vehicle (not shown) provided with the pantograph 2 travels in the direction of the arrow.

  As shown in FIG. 1, a suspension line 13 is stretched between support points 11 and 12 of a column (not shown) that supports the overhead line. Further, hangers (see, for example, reference numerals 14 and 15) hang from the suspension line 13 at equal intervals. The lower end of the hanger is connected to the trolley wire 10.

In the present embodiment, the distance between the support points 11 and 12 is 50 m, and ten hangers are provided at intervals of 5 m. In the present embodiment, the measurement interval is between the support points, and the sensor is arranged at a predetermined position on the trolley wire 10 in the measurement interval. For example, both ends of the measurement section, that is, positions aligned with the support points 11 and 12 on the trolley wire 10 are the first measurement points 101 and 102, respectively. At each of the first measurement points, two accelerometers are arranged at a predetermined interval, for example, 0.1 m apart. Further, in the example of FIG. 1, predetermined positions on both sides of the connection point between the trolley wire 10 and the hanger are the second measurement points (reference numerals 111 and 112). An accelerometer is also disposed at each of the second measurement points.
In the present embodiment, the arrangement of two sensors (also referred to as “first sensors”) at each of the first measurement points at both ends of the measurement section is the same in all examples. There are various modes of arrangement of other sensors at the second measurement point (sensors arranged on the trolley line, between the hangers, or at the connection points with the hangers: also referred to as “second sensors”).

  FIGS. 2A to 2C are diagrams showing the arrangement of the second sensor. The arrangement position of these second sensors is a measurement point of acceleration. As shown in FIG. 2A, in the first mode, five accelerometers 211 to 215 are arranged between the hangers 201 and 202 at equal intervals. In this aspect, the distance from the hanger 201 to the second sensor 211 is +0.5 m, the distance between adjacent second sensors is +1 m, and the distance from the hanger 202 to the second sensor 215 is −0.5 m. .

  As shown in FIG.2 (b), in the 2nd aspect, the 2nd sensor is provided in the position of +/- 1.3m both sides of the connection point with a hanger (refer code | symbol 221 and 222). Further, as shown in FIG. 2C, in the third mode, the second sensor is provided at the positions of +2.5 m and −1.3 m of the connection point with the hanger (see reference numerals 231 and 232). ). The position of the second sensor in FIG. 1 corresponds to that of the second mode shown in FIG.

Although not shown in FIG. 1, a strain gauge is arranged on each hanger so that the hanger force (h ₁ , h ₂ ,..., H _{10 in} FIG. 1) can be measured.

Hereinafter, the principle of contact force estimation according to the present invention will be described. In FIG. 1, the contact force f of the pantograph 2 acts on one place, and is an infinite length string supported by n hanger forces h _i (i = 1, 2,..., N). Formulate the equation of motion. When ρ is the line density of the trolley line, T is the tension of the trolley line, xi is the x coordinate of the i-th hanger point, and v is the train speed, the behavior y ( x, t) can be approximately expressed as in equation (1).

Assume that the measurement interval is in the range of “−ε <x <ε” and that there is a pantograph in the measurement interval. Under this assumption, if the equation (1) is integrated, the following equation (2) can be obtained.

In the formula (2), the first term on the right side is the hanger force, which can be obtained from the expansion and contraction of the hanger measured by the strain gauge provided on each hanger. Alternatively, the hanger force can be directly measured by attaching a load cell to the hanger.

Further, in the expression (2), the second term on the right side is a vertical component (component in the y direction) due to the trolley line tension. Further, ∂y / ∂x _{x = + ε} and ∂y / ∂x _{x = −ε} are inclinations at the end of the measurement section, respectively. The second term on the right side can be obtained by measuring inclinometers (first sensors) arranged at both ends of the measurement section on the trolley line. The third term on the right side of the equation (2) is the inertial force of the trolley wire. Actually, it is obtained from a continuous distribution of trolley line vertical acceleration.

  The third term on the right side of the equation (2) needs to acquire a continuous acceleration distribution, and its measurement is not easy. Therefore, an approximate expression in which the third term on the right side is represented by acceleration at a finite location (p location) is shown in Equation (3). The p limited positions correspond to the arrangement positions of the second sensors shown in FIGS. 1 and 2, and the acceleration measured by the second sensor is used.

Here, w _j is an inertial force correction coefficient for the acceleration at the j-th point. The inertial force correction coefficient w _j also includes a plurality of modes. For example, in the first aspect, the inertia correction coefficient w _j is set to the following value.

w _j = ρL / p (4)
(P: number of measurement points, L: length of measurement section)
The case where the equation (4) is used is referred to as “uniform uniform mass condition”.

It was also considered to identify w _j so that the estimation result and the simulation result match. The identification of w _j will be described in detail later. The case where the identified w _j is used is referred to as “equivalent mass identification condition”.

Furthermore, the above equation (3) is transformed into the equation (5), and it is considered that the nonlinear function _fine for estimating the inertial force is identified by the neural network (NN). As will be described later, in the NN, the input of the NN is acceleration at each point, an inertial force is applied to the teacher signal, and learning is performed at each time using the error back propagation method. However, the output function of the NN intermediate layer was a sigmoid function, and the others were linear functions.

A specific system configuration for estimating the contact force f described above will be described in more detail. FIG. 3 is a block diagram showing the configuration of the contact force estimation system according to the first to third embodiments of the present invention. As shown in FIG. 3, the contact force estimation system 20 according to the present embodiment includes a storage device 22 that stores various data, an interface 24, a correction coefficient identification processing unit 26 for identifying correction coefficients, and a storage. A contact force calculation unit 28 that calculates the contact force using the data stored in the device 22 and the correction coefficient is provided.

  Further, the contact force estimation system 20 includes a storage medium read / write device 34 for reading / writing a portable storage medium 35 such as an input device 30, a display device 32, a CD-ROM, a CD-R / W, and a DVD-RAM. Data is exchanged with the interface 24.

  The storage device 22 stores a hanger force data file 36, a data file of vertical component data (hereinafter referred to as “trolley line tension data file”) 38, and an acceleration data file 40. Further, the calculation result file 42 which is a calculation result by the contact force calculation unit 28 is also stored in the storage device 22 after the processing. Further, the storage device 22 stores various data generated in the calculation process (not shown).

  In the first embodiment, data indicating an almost continuous acceleration distribution of the overhead line is used instead of the acceleration data of the measurement point.

In the second embodiment, since the inertia force correction coefficient under the uniform equivalent mass condition is used, the correction coefficient identification processing unit 26 only calculates the correction coefficient w _i according to the equation (4). On the other hand, in the third embodiment, since the inertia correction coefficient of the equivalent mass identification condition is used, the correction coefficient identification processing unit 26 executes a calculation described in detail later.

The hanger force data file 36 stores the hanger force h _i (i = 1, 2,..., N) in the measurement section. Further, the trolley line tension data file 38 includes, for example, vertical component data ((2) based on the trolley line tension based on the inclination calculated from the accelerations of the accelerometers arranged at two measurement points at both ends of the measurement section. The second term on the right side of the equation is stored. Alternatively, in the trolley tension data file 38, the inclinations (∂y / ∂x _{x = + ε} , に お け_る Data of y / ∂x _{x = −ε} ) may be stored. The acceleration data file stores acceleration data of measurement points by the second sensor (accelerometer).

  In the present embodiment, when the value acquired from the sensor can be directly used for calculation, the value acquired from the sensor is stored in the storage device 22 as data of each file. On the other hand, when the value acquired from the sensor is subjected to a certain process and used for calculation, the data file generation unit 44 performs the necessary calculation on the value acquired from the sensor to generate file data. And stored in the storage device 22.

  For example, when data is acquired from a load cell arranged on a hanger, the data can be used as it is as a hanger force, and is stored in the storage device 22 as a hanger force data file 36. On the other hand, when the data is acquired from the strain gauge arranged on the hanger, the data file generation unit 44 generates the hanger force data file 36 including the data obtained by converting the output of the strain gauge into force, and the storage device 22.

  The data of the trolley wire tension data file 38 (vertical component data by the trolley wire) is generated by the data file generator 44 as follows and stored in the storage device 22.

In the second term on the right side of the equations (2) and (4), ∂y / ∂x _{x = + ε} can be expressed as the following equation (6).

In equation (6), y (ε + Δx / 2, t) is the displacement in the y direction at one end of the measurement interval, and y (ε + Δx / 2, t) is the displacement in the y direction at the other end of the measurement interval. In the present embodiment, first sensors (accelerometers) are arranged at both ends of the measurement section. The y (ε + Δx / 2, t) and y (ε + Δx / 2, t) can be obtained from the two accelerometers arranged at each end and the distance Δx of the measurement section. Similarly, ∂y / ∂x _{x = −ε} can be obtained. Based on these, the data file generation unit 44 generates a trolley line tension data file 38 including vertical component data based on the trolley line tension, and stores it in the storage device 22.

  In the first embodiment, instead of the acceleration data of the measurement point, the data file generation unit 44 generates data indicating an almost continuous acceleration distribution of the overhead line based on the use of the image data of the overhead line. Is stored in the storage device 22 as the acceleration data file 40.

Contact force calculating unit 28 obtains the hanger force h _i from the hanger force data file 36, also from the trolley wire tension data file 38, obtains the data of the vertical component due to the contact wire tension. Alternatively, the slopes (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ) at both ends of the measurement section may be taken out, and based on these, the vertical component due to the trolley line tension may be calculated.

  Further, the contact force calculation unit 28 extracts a substantially continuous acceleration distribution from the acceleration data file 40. Based on these, the contact force calculation unit 28 obtains the contact force f (t) at each time t according to the equation (2). A calculation result file 42 including the obtained f (t) is stored in the storage device 22.

Next, a second embodiment of the present invention will be described. In the second embodiment, the contact force is calculated according to the equation (3), and the inertia force correction coefficient is a value according to the uniform equivalent mass condition. In the second embodiment, in the acceleration data file 40, according to the first aspect, data of accelerometers provided on the trolley line and between the connection points with the hanger, respectively, or the second According to the third aspect and the third aspect, the data of the accelerometers provided at two places between the connection points with the hanger are stored. The correction coefficient identification processing unit 26 calculates w _j according to equation (4).

Contact force calculating unit 28 obtains the hanger force h _i from the hanger force data file 36, also from the trolley wire tension data file 38, obtains the data of the vertical component due to the contact wire tension. Alternatively, the slopes (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ) at both ends of the measurement section may be taken out, and based on these, the vertical component due to the trolley line tension may be calculated.

  Further, the contact force calculation unit 28 acquires accelerometer data from the acceleration data file 40, and obtains the contact force f (t) at each time t according to the equation (3) based on the acquired data. The calculation result file 42 including the obtained f (t) is stored in the storage device 22.

  Next, a third embodiment of the present invention will be described. In the second embodiment, the contact force is calculated according to the equation (3), and the value according to the equivalent mass identification condition is adopted as the inertia force correction coefficient. In the third embodiment, the data of various files in the storage device 22 is the same as in the second embodiment. Hereinafter, calculation of the inertia force correction coefficient will be described.

  Hereinafter, in the formulas (7) to (14), the contact force in the time domain is represented by a lower case letter “f” in order to distinguish between the time domain and the frequency domain. It is represented by an uppercase “F”. Thereby, contact force f (t) is expressed by the following formula (7).

Note that the equations (7) and (2) are the same.

Here, the acceleration ∂ ² y _j (t) / ∂t ² in the y direction of the trolley line at the measurement point j is set as α _j (t), and the first term on the right side and the second term on the right side are respectively ( When expressed as shown in equations (8) and (9), equation (7) is expressed as equation (10).

Here, when the Fourier transform is applied to the expression (10) and the discussion in the frequency domain is performed, the expression (10) becomes as shown in the expression (11).

When both sides of equation (11) are divided by F (ω), equation (12) is obtained.

Here, the condition for the correct estimation is that the real part is “1” and the imaginary part is “0”, as shown in the equations (13) and (14). This is a case where the gain is 1.0 times and there is no phase delay. The correction coefficient identification processing unit 26 according to the present embodiment obtains w _j that most satisfies the expressions (13) and (14) by the method of least squares.

The contact force calculation unit 28 obtains data from the data file in the storage device 22 and executes the calculation in the same manner as in the second embodiment except for w _j obtained by the correction coefficient identification processing unit 26. The contact force f (t) at time t is calculated, and the calculation result file 42 including the contact force f (t) is stored in the storage device 22.

Next, a fourth embodiment of the present invention will be described. In the fourth embodiment, a nonlinear function f _ine that estimates inertial force is identified using a neural network. As shown in FIG. 4, the contact force estimation system according to the fourth embodiment includes a neural network calculation unit 46 for identifying the nonlinear function using a neural network. The storage device 22 stores data used in the neural network calculation and a data group (NN calculation data group) 48 generated in the calculation process.

  FIG. 5 is a block diagram showing the configuration of the neural network calculation unit 46 in more detail. As illustrated in FIG. 5, the neural network calculation unit 46 learns a neural network model (NN model), and an estimated value calculation process that calculates an estimated value of inertial force based on the learned NN model. Part 52.

  Further, the NN calculation data group 48 stored in the storage device 22 includes NN model data 54, a measured acceleration measured value data group 56, an inertial force measured value data group 58, and an estimated acceleration measured value data group 60. Is included. In addition, the inertia force estimated value data 60 calculated by the estimated value calculation processing unit 52 is also stored in the storage device 22. For example, the acceleration data constituting the acceleration measured value data group can be obtained from simulation. Moreover, the inertial force actual measurement value data can be obtained from the vertical component based on the contact force, the hanger force, and the trolley wire tension. That is, it can be obtained from the measured values of the left side, the first term on the right side, and the second term on the right side in the equation (2).

  In addition, although the actual measurement value here includes what is obtained by simulation, of course, the actual measurement value, for example, the contact force measured by a pantograph for contact force measurement (when an overhead wire excitation test is performed, Vibration force), actually measured hanger axial force, actual acceleration of trolley line inclination, and the like are also included.

  In a neural network, a neuron is connected to one or more other neurons on the input side. Each of the connections between the neurons is assigned a weight, and the product of the weight corresponding to the connection through which the input passes is given to the neuron. In the neuron, the sum of products of the input and the corresponding weight is calculated and output by an output function based on a value obtained by subtracting the threshold. Therefore, the NN model data 54 includes information on connections between neurons, weights associated with the connections, and neuron threshold values. The weights associated with the connections and the neuron threshold values are updated as needed during learning to be described later.

FIG. 6 is a diagram schematically showing the structure of the NN model at the time of learning. As shown in FIG. 6, acceleration measured value data (y ″ _j (t _i )) (j = 1, 2,..., P) at a certain time t _i are respectively input to neurons in the input layer and input. component propagates the neurons of the intermediate layer reaches the neurons of the output layer. Meanwhile, as a teacher signal, the inertial force _f ine at time _{ti (y "1, y"} 2, ···, y "p, t i ) Is given to the neurons in the output layer, and the error component propagates in the opposite direction from the output layer to the input layer by the error propagation method (BP (Back Propagation) method), and is related to the connection between the neurons The given weight is corrected. Y ″ (t) is a second-order derivative of the trolley line height y with respect to t, and is the same as ∂ ² y / ∂t ² .

FIG. 7 is a flowchart schematically showing processing during learning. As shown in FIG. 7, the learning processing unit 50 reads the NN model data 54 from the storage device 22 (step 701), and first uses a random number as the initial value of the weight associated with the connection between the neurons. A value is assigned (step 702). Next, the learning processing unit 50 initializes a parameter _i for specifying the time t _i (step 703), and the actual measured acceleration value data y ″ _j (t _i ) (j = 1, 2,... The operation is executed in the form of giving p) (step 704).

  In the calculation, the learning processing unit 50 sequentially calculates the product of the input signal and the connection weight so that the input signal is propagated toward the output layer, calculates the sum of the products for each neuron, It is determined whether or not a threshold value has been exceeded. When it is determined that the threshold value is exceeded for a certain neuron, the neuron is fired and a signal having a predetermined value is output to the coupled neuron.

Also, the learning processing unit 50, the output layer, the inertial force measured value data _{f ine (y "1, y} " 2, ···, y "p, t i) in such a way as to cause, by the error backpropagation, The weight associated with the connection between each neuron is learned (step 705), the weight associated with the connection between each neuron and the threshold of each neuron are updated, and the updated weight and threshold are stored in the storage device 22. stored (step 706). the process of step 703 to 705 is executed for all _{t i} (see step 707 and 708).

  By repeating the processing shown in FIG. 7 using various types of actual acceleration data and inertial force data, the relationship between the input signal and the teacher signal is learned in the NN model and associated with the connection between neurons. The given weight and the threshold value of each neuron are appropriate values.

  By learning, the weights associated with the connections between the neurons and the threshold values of the neurons become appropriate, and these values are stored in the storage device 22 as the NN model data 54. In this state, the estimated value calculation processing unit 52 calculates inertial force estimated value data based on the NN model data 54 and the estimated acceleration measured value data 60.

FIG. 8 is a flowchart schematically showing the estimation process. As shown in FIG. 8, the estimated value calculation processing unit 52 reads the NN model data 54 from the storage device 22 (step 801), and initializes a parameter i that specifies the time t _i (step 802). Next, the estimated value calculation processing unit 52 executes the calculation in such a manner that the estimated acceleration measured value data y ″ _j (t _i ) (j = 1, 2,..., P) is given to the input layer. (Step 803) In the calculation, the estimated value calculation processing unit 52 sequentially calculates the product of the input signal and the connection weight so that the input signal is propagated toward the output layer, and the product for each neuron. If it is determined that the threshold is exceeded for a certain neuron, the neuron is fired and a signal having a predetermined value is output. toward the output layer from sequentially signal weighting, the calculation of the sum in neurons, by repeating the output of the comparison and the signal with the threshold value, from the output layer, the inertial force estimate data _{f ine (y "1, y} " _2, ···, y _"p t _i) it is output. This inertial force estimated value data is stored in the storage device 22 (step 804). Estimate calculation processing unit 52, the processing of steps 803 and 804, is performed for all the _{t i} (see step 805 and 806).

The contact force calculation unit 28 calculates the contact force f (t) using the inertial force estimated value data obtained as described above as the third term on the right side of the equation (3). As the inertia force correction coefficient w _{j in} the second term on the right side, a value according to the uniform equivalent mass condition may be used, or a value according to the equivalent mass identification condition may be used.

  Hereinafter, the estimated value of the contact force obtained by the calculations according to the second embodiment to the fourth embodiment will be described. Since it is not easy to directly measure the actual value of the contact force, the contact force obtained by the simulation is used as the actual value of the contact force. In the simulation, the tension of the trolley wire and the suspension wire was 9810 N, respectively, and the line densities of the trolley wire and the suspension wire were 0.99 kg / m and 1.09 kg / m, respectively. The pantograph that moves while contacting the trolley wire was a conventional pantograph, and the static lifting force was 54N. The traveling speed of the vehicle was 130 km / h.

  As shown in FIG. 1, one span is used as a measurement section. As for the arrangement of the second sensor, the first mode (5 sensors are arranged at intervals of 1 meter between the hangers), the second mode (sensors at positions of ± 1.3 m on both sides of the hanger connection point) ) And the third embodiment (sensors are arranged at +2.5 m and −1.3 m positions of the hanger connection points).

FIG. 9 is a graph showing the time history response of the contact force estimation result. In FIG. 9, the vertical axis represents contact force (N), and the horizontal axis represents time (seconds). Graphs (1) to (5) in the figure are under the following conditions.
(1) Contact force obtained by simulation (actual value of contact force)
(2) Contact force calculated according to the second embodiment (obtaining w _j according to the uniform equivalent mass condition) in the sensor arrangement according to the first aspect. (3) The sensor arrangement according to the second aspect. Contact force calculated according to the second embodiment (determining w _j according to the uniform equivalent mass condition) (4) In the sensor arrangement according to the second aspect, according to the third embodiment (w _j is equivalent mass) Contact force calculated according to the identification condition (5) Contact force calculated according to the third embodiment (determining w _j according to the equivalent mass identification condition) in the sensor arrangement according to the third aspect. These are the graphs which show the power spectrum of the contact force estimation result of said (1)-(5). In FIG. 10, the vertical axis represents PSD (power spectral density), and the horizontal axis represents frequency (Hz).

  FIG. 11 is a graph showing the contact force calculated in accordance with the fourth embodiment (estimating inertial force using NN) with the contact force of (1) and the sensor arrangement according to the third aspect. .

  From these results, it was confirmed that the contact force can be estimated by arranging the accelerometers at a sufficient number or at predetermined positions according to the required frequency range. It was also found that using a neural network, good results can be obtained with a small number of accelerometers.

  The present invention is not limited to the above embodiments, and various modifications can be made within the scope of the invention described in the claims, and these are also included in the scope of the present invention. Needless to say.

  For example, in the second mode, two second sensors (accelerometers) are disposed between the connection points of the trolley wire and the hanger, and the positions of the second sensors (accelerometers) are approximately ± 1/4 * (connections). It is a position separated by a distance between points) (+ is the traveling direction of the train). Further, in the third aspect, two places are arranged between the connection points of the trolley line and the hanger, and the arrangement positions are approximately + 1/2 * (distance between the connection points) from the connection point, and approximately −. It is a position separated by 1/4 * (distance between connecting points). This takes into consideration a wave having a half-wavelength between the connection points, and a wave having a half-wavelength half the distance between the connection points. However, the arrangement position of the sensor (position of the measurement point) is not limited to these.

  Moreover, in the said embodiment, although the measurement area is set as between the support points of an overhead wire, a measurement area is not limited to this.

  Furthermore, in the said embodiment, although based on the wave equation of the trolley line as shown to (1) Formula, it is not limited to this.

  The original wave equation may be considered as in equation (15), and may be approximated by leaving the term of bending stiffness EI as in equation (16).

In this case, when the measurement range is set to “−ε <x <ε” and the equation (16) is integrated, the equation (17) that takes into account the hanger force can be obtained.

The expression (17) is similar to the expression (2), and the first term, the second term, and the fourth term on the right side of the expression (17) are respectively the first term, the second term, and the second term on the right side of the expression (2). This corresponds to item 3. In the equation (17), (∂ ³ y / ∂x ³ ) _{x = + ε} and (∂ ³ y / ∂x ³ ) _{x = −ε} in the third term on the right side are the trolley at the end point and the start point of the measurement section, respectively. This is the third-order spatial derivative of the line height. Therefore, four accelerometers are arranged at both ends of the measurement section, and (∂ ² y / ∂x ² ) _{x = + ε} and (∂ ³ y / ∂x ³ ) _{x = -Ε} can be calculated by the following equations (18) and (19), respectively.

Therefore, as a modification of the first embodiment of the present invention, based on the data of the first embodiment and the height data of the trolley line at the end point and the start point of the measurement section, Based on this, f (t) can also be calculated.

  Furthermore, the expression (20) can be obtained by representing the fourth term on the right side of the expression (17) by the acceleration at a finite location (p location).

The expression (20) is similar to the expression (3), and the first term, the second term, and the fourth term on the right side of the equation (20) are respectively the first term, the second term, and the second term on the right side of the equation (3). This corresponds to item 3. Therefore, as a modification of the second embodiment or the third embodiment, the data of the second embodiment or the third embodiment and the height of the trolley line at the end point and the start point of the measurement section Based on the data, it is also possible to calculate f (t) based on equation (20).

Further, the nonlinear function _fine for estimating the inertial force can be identified by the neural network (NN) by transforming the equation (20) as the equation (21).

Equation (21) is similar to Equation (5), and the first term, the second term, and the fourth term on the right side of Equation (21) are the first term, the second term, and the second term on the right side of Equation (5), respectively. This corresponds to item 3. Therefore, as a modification of the fifth embodiment, based on the data of the fifth embodiment and the height data of the trolley line at the end point and the start point of the measurement section, based on the equation (20), It is also possible to calculate f (t).

  In the modification of the fourth embodiment, the inertial force actual measurement data used for the output layer of the NN model includes the vertical component based on the contact force, the hanger force, the trolley wire tension, the bending stiffness EI, and the trolley wire. It can obtain | require based on the term of the force by bending rigidity of.

FIG. 1 is a diagram schematically showing an example of the structure of an overhead wire. FIGS. 2A to 2C are diagrams showing the arrangement of the second sensors according to the present embodiment. FIG. 3 is a block diagram showing the configuration of the contact force estimation system according to the first to third embodiments of the present invention. FIG. 4 is a block diagram showing the configuration of the contact force estimation system according to the fourth embodiment of the present invention. FIG. 5 is a block diagram showing the configuration of the neural network calculation unit in more detail. FIG. 6 is a diagram schematically showing the structure of the NN model at the time of learning. FIG. 7 is a flowchart schematically showing processing at the time of learning according to the fourth embodiment. FIG. 8 is a flowchart schematically showing an estimation process according to the fourth embodiment. FIG. 9 is a graph showing the time history response of the contact force estimation result. FIG. 10 is a graph showing the power spectrum of the tactile force estimation result. FIG. 11 is a graph showing the contact force calculated in accordance with the fourth embodiment (estimating inertial force using NN) with the contact force of (1) and the sensor arrangement according to the third aspect. .

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Overhead line 2 Pantograph 10 Trolley line 20 Contact force estimation system 22 Memory | storage device 26 Correction coefficient identification process part 28 Contact force calculation part 36 Hanger force data file 38 Trolley line tension data file 40 Acceleration data file 42 Calculation result file 44 Data file generation part

Claims (16)

A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval gradient _{(∂y / ∂x x = + ε} , ∂y / ∂x x = -ε), as well as data indicating continuous distribution of acceleration of the trolley wire in the measurement interval the (∂ ² y / ∂t ^2), respectively A storage device storing
From the data showing the continuous distribution of the hanger force, tilt and acceleration, the following equation
(T: trolley wire tension, ρ: trolley wire linear density)
A contact force calculation means for calculating a contact force f (t) based on
A contact force estimation system comprising: A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval gradient _{(∂y / ∂x x = + ε} , ∂y / ∂x x = -ε), a continuous distribution of acceleration of the trolley wire in the measurement period ^{(∂ 2 y / ∂t 2)} , as well as the measurement interval A storage device storing data indicating the force terms ( に よ _る ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ) due to the bending rigidity of the trolley wire at both ends ;
From the data on the term of force due to the continuous distribution of the hanger force, inclination and acceleration, and the bending stiffness of the trolley wire at both ends of the measurement section,
(T: Trolley wire tension, ρ: Trolley wire linear density, EI: Trolley wire bending stiffness)
Contact force calculation means for calculating the contact force f (t) based on
A contact force estimation system comprising: A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger on the trolley line in the measurement section by a predetermined distance A storage device storing data indicating (∂ ² y _j / ∂t ² : j = 1, 2,..., P),
From the data indicating the hanger force, tilt and acceleration, the following formula
(T: trolley wire tension, ρ: trolley wire linear density, w _j : inertia force correction coefficient)
A contact force calculation means for calculating a contact force f (t) based on
A contact force estimation system comprising: A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), acceleration of a plurality of measurement points separated by a predetermined distance from a connection point with the hanger on the trolley line in the measurement section (∂ ² y _j / ∂t ² : j = 1, 2,..., P), and terms of force due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), each of which stores data,
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the hanger force, inclination, acceleration, and measurement section, the following formula
(T: tension of trolley wire, ρ: linear density of trolley wire, w _j : inertia force correction coefficient, EI: bending rigidity of trolley wire)
Contact force calculation means for calculating the contact force f (t) based on
A contact force estimation system comprising: A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger on the trolley line in the measurement section by a predetermined distance A storage device storing data indicating (∂ ² y _i / ∂t ² : i = 1, 2,..., M),
From the data indicating the hanger force, tilt and acceleration, the following formula
(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force)
A contact force calculation means for calculating a contact force f (t) based on
A contact force estimation system comprising: A contact force estimation system for estimating a contact force between a pantograph boat hull and a trolley wire,
At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), at both ends of the measurement interval Inclination (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), acceleration of a plurality of measurement points separated by a predetermined distance from a connection point with the hanger on the trolley line in the measurement section (∂ ² y _i / ∂t ² : i = 1, 2,..., M) and a term of force due to the bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ), each of which stores data,
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the hanger force, inclination, acceleration, and measurement section, the following formula
(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force, EI: trolley wire bending stiffness)
Contact force calculation means for calculating the contact force f (t) based on
A contact force estimation system comprising: The storage device is
A neural network including the acceleration data for learning, the inertial force data for learning, and a neural network model including weights associated with connections between neurons constituting the neural network and threshold values of the neurons. Store network model data and acceleration data for estimation,
The contact force calculation means is
A learning means for making weights and thresholds included in neural network model data appropriate values,
Read the initial neural network model data stored in the storage device, give the learning acceleration data to the input layer of the neural network model, and from the input layer to the output layer through the intermediate layer In turn, according to the neuron connection in the neural network model data, an output function based on a value obtained by subtracting a threshold from the sum of the multiplication results of the signal from the neuron closer to the input layer and the weight associated with the connection. Computing means for repeatedly calculating an output value and outputting the output value to a neuron coupled to the output layer;
Inertial force data for learning is applied to the output layer of the neural network model, an error signal is propagated from the output layer to the input layer through the intermediate layer, and at least the coupling in the neural network data. Learning means comprising: error backpropagation means for modifying associated weights and thresholds for the neurons, and storing the weights associated with the modified connections and thresholds for the neurons as modified neural network data; ,
Read out the neural network model data corrected by learning by the learning means stored in the storage device, give the estimation acceleration data to the input layer of the neural network model, and from the input layer through the intermediate layer A value obtained by subtracting a threshold value from the sum of multiplication results of signals from neurons closer to the input layer and weights associated with the connection in accordance with the connection of neurons in the neural network model data sequentially toward the output layer The output function is calculated based on the output function, the output value is repeatedly output to the neuron coupled to the output layer, and the value from the output layer is used as the estimated inertia force _fine . Estimating means for storing in a storage device;
The contact force estimation system according to claim 5 or 6, characterized by comprising: There are two measurement points between the connection points, and the measurement points are separated from the connection points by ± 1/4 * (distance between the connection points) (+ is the traveling direction of the train). The contact force estimation system according to any one of claims 3 to 7. There are two measurement points between the connection points, and positions separated by + 1/2 * (distance between connection points) and -1 / 4 * (distance between connection points) from the connection point. The contact force estimation system according to any one of claims 3 to 7, wherein At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and data indicating the continuous distribution (分布² y / ∂t ² ) of the acceleration of the trolley line in the measurement section. In an information processing apparatus provided with a stored storage device, a method for estimating a contact force between a pantograph boat hull and a trolley wire,
From the data showing the continuous distribution of the hanger force, tilt and acceleration, the following equation
(T: trolley wire tension, ρ: trolley wire linear density)
The contact force estimation method characterized in that the contact force f (t) is calculated on the basis of the above. At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval _{(∂y / ∂x x = + ε} , ∂y / ∂x x = -ε), the continuous distribution of the acceleration of the trolley wire in the measurement period ^{(∂ 2 y / ∂t 2)} , and both ends of the measurement interval In an information processing apparatus including a storage device storing data indicating the force terms (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ) due to the bending rigidity of the trolley wire , A method for estimating the contact force between a pantograph boat hull and a trolley wire,
From the data on the term of force due to the continuous distribution of the hanger force, inclination and acceleration, and the bending stiffness of the trolley wire at both ends of the measurement section,
(T: Trolley wire tension, ρ: Trolley wire linear density, EI: Trolley wire bending stiffness)
The contact force estimation method characterized in that the contact force f (t) is calculated on the basis of the above . At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger by a predetermined distance on the trolley line in the measurement section ( In an information processing apparatus including a storage device that stores data indicating ∂ ² y _j / ∂ t ² : j = 1, 2,..., P), the pantograph boat hull and the trolley line A method for estimating contact force,
From the data indicating the hanger force, tilt and acceleration, the following formula
(T: trolley wire tension, ρ: trolley wire linear density)
The contact force estimation method characterized in that the contact force f (t) is calculated on the basis of the above. At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), accelerations of a plurality of measurement points (∂ ²⁾ separated from the connection point with the hanger on the trolley line in the measurement section. y _j / ∂t ² : j = 1, 2,..., p), and terms of force due to the bending rigidity of the trolley wire at both ends of the measurement section ( ^{３ 3} y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ) In an information processing apparatus including a storage device that stores data indicating each, a method for estimating a contact force between a pantograph boat hull and a trolley wire,
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the hanger force, inclination, acceleration, and measurement section, the following formula
(T: tension of trolley wire, ρ: linear density of trolley wire, w _j : inertia force correction coefficient, EI: bending rigidity of trolley wire)
The contact force f (t) is calculated on the basis of the contact force estimation method. At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), and accelerations of a plurality of measurement points separated from the connection point with the hanger by a predetermined distance on the trolley line in the measurement section (情報 処理² y _i / ∂ t ² : i = 1, 2,..., M), in an information processing device having a storage device, each of the pantograph boat hull and the trolley wire A method for estimating contact force,
From the data indicating the hanger force, tilt and acceleration, the following formula
(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force)
The contact force estimation method characterized in that the contact force f (t) is calculated on the basis of the above. At least the hanger force h _i (i = 1, 2,..., N) of the hanger connected to the trolley wire, which exists during the measurement interval (−ε <x <ε), and the inclination at both ends of the measurement interval (∂y / ∂x _{x = + ε} , ∂y / ∂x _{x = -ε} ), accelerations of a plurality of measurement points (∂ ²⁾ separated from the connection point with the hanger on the trolley line in the measurement section. y _i / ∂t ² : i = 1, 2,..., m), and terms of force due to bending rigidity of the trolley wire at both ends of the measurement section (∂ ³ y / ∂x ³ _{x = + ε} , ∂ ³ y / ∂x ³ _{x = −ε} ) In an information processing apparatus including a storage device that stores data indicating each, a method for estimating a contact force between a pantograph boat hull and a trolley wire,
From the data of the term of force due to the bending rigidity of the trolley wire at both ends of the hanger force, inclination, acceleration, and measurement section, the following formula
(T: trolley wire tension, ρ: trolley wire linear density, _fine : nonlinear function for estimating inertial force, EI: trolley wire bending stiffness)
The contact force f (t) is calculated on the basis of the contact force estimation method. The storage device is
A neural network including the acceleration data for learning, the inertial force data for learning, and a neural network model including weights associated with connections between neurons constituting the neural network and threshold values of the neurons. Store network model data and acceleration data for estimation,
Calculating the inertial force when estimating the contact force,
Read the initial neural network model data stored in the storage device, give the learning acceleration data to the input layer of the neural network model, and from the input layer to the output layer through the intermediate layer In turn, according to the neuron connection in the neural network model data, an output function based on a value obtained by subtracting a threshold from the sum of the multiplication results of the signal from the neuron closer to the input layer and the weight associated with the connection. A calculation step of repeatedly calculating the output value and outputting the output value to a neuron coupled to the output layer side; and
Inertial force data for learning is applied to the output layer of the neural network model, an error signal is propagated from the output layer to the input layer through the intermediate layer, and at least the coupling in the neural network data. A learning step comprising correcting an associated weight and a threshold for the neuron, and storing a weight associated with the modified connection and the threshold for the neuron as modified neural network data;
Read out the neural network model data corrected by learning in the learning step stored in the storage device, give the estimation acceleration data to the input layer of the neural network model, and from the input layer through the intermediate layer A value obtained by subtracting a threshold value from the sum of multiplication results of signals from neurons closer to the input layer and weights associated with the connection in accordance with the connection of neurons in the neural network model data sequentially toward the output layer the output function based on, and calculates an output value, repeatedly to output the neuron coupled the output value on the side of the output layer, the value from the output layer, as the estimated inertial force, the The estimation step of storing in a storage device, comprising: Tactile estimation method.