JP3229119B2 - Train drive control device and train running resistance learning device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a train drive control device and a train drive resistance learning device for suppressing the influence of train drive fluctuation due to a change in running resistance and learning the running resistance as needed.

[0002]

2. Description of the Related Art As a conventional technique, when a train is automatically operated, a current command value is calculated by feedback control so that a speed command and an actual speed of the train are matched, and this current command value is used for driving a train. There is a method in which the electric motor is supplied to the electric motor for driving. FIG. 8 is a configuration diagram of a conventional train drive control device.

In FIG. 8, reference numeral 41 denotes a train speed command unit, 42 denotes a speed control unit, 43 denotes a current supply unit, 44 denotes a train driving motor, 45 denotes a train, 46 denotes a position / speed detection unit, and 47 denotes a subtraction unit. is there. The speed command section 41 outputs a speed command Vc. Subtraction unit 47
The speed command Vc and the position detected by the position / speed detector 46
The difference between the train speed and the actual speed V of the train is calculated, and the speed control unit 42 amplifies the difference to calculate the current command value Im. The current supply unit 43 supplies a current to the electric motor 44 according to the current command value Im, and causes the train 45 to run by the driving force FX of the electric motor 44. At this time, the running resistance Fd such as air resistance and friction
Is different between a tunnel T (hereinafter referred to as a tunnel section) and a section other than the tunnel T (hereinafter referred to as a light section). For example, when the train 45 enters or exits the tunnel T, the speed and acceleration of the train 45 Fluctuates.

FIGS. 9 and 10 show examples of train speeds and accelerations which fluctuate due to running resistance applied to the train 45. FIG. FIG.
9A shows a change in train acceleration, FIG. 9B shows a change in train speed, and FIG. 9C shows a change in air resistance coefficient. Air resistance is said to be approximately proportional to the velocity squared. On the other hand, since the air resistance coefficient is different between the tunnel section and the light section, when the train 45 enters or exits the tunnel T, the air resistance changes discontinuously.
(A), the speed and acceleration vary as shown in FIG. 9 (b).

FIG. 10 shows a change in running resistance due to a gradient.
FIG. 10A is a diagram showing a change in train acceleration, FIG. 10B is a diagram showing a change in train speed, and FIG. 10C is a diagram showing a relationship between a slope and a slope coefficient. On a slope, the resistance value is determined by the train weight M and the slope coefficient. For example, on a slope with a slope of Y% (inclination angle θ), a force of Fg = M · g · sin θ is received downward on an uphill slope. At this time, the speed and acceleration are shown in FIG.
As shown in FIG.

[0006]

As described above, the speed and acceleration of a train fluctuate between a tunnel section and a light section due to a change in an air resistance coefficient. Also, in the case of an uphill, the train receives a force Fg in the direction opposite to the traveling direction even on an uphill, so that the speed and the acceleration fluctuate. The actual speed of the train is detected by the position / speed detection unit, and the speed control unit controls the actual speed to follow the speed command value output from the speed command unit. The difference between the air resistance and the slope caused by the slope acts on the train as a disturbance between the place and the place with the slope, so that a sudden change occurs in the difference between the speed command value and the actual speed. This change had a negative effect on the riding comfort of the train.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and is intended to solve the problem of a train which suppresses a driving fluctuation of a train due to a disturbance of the running resistance by compensating a running resistance for a current command value. It is an object to provide a drive control device. It is another object of the present invention to provide a train running resistance learning device that learns running resistance at any time.

[0008]

[0009]

[0010]

According to the first aspect of the present invention, there is provided a position and speed detecting means for detecting a position and a speed of a train,
Speed command means for outputting a speed command value and an acceleration command value of the train; speed control for calculating a current command value from a speed detected value detected by the position / speed detecting means and a speed command value output from the speed command means Means, and recognizes a change in the road condition from the position detected by the position / speed detecting means, and learns each coefficient constituting the running resistance using the change in the road condition, the detected speed value, and the current command value. Learning means for calculating a compensation current command value using each coefficient constituting the running resistance,
Current supply means for supplying the compensation current command value calculated by the learning means to the electric motor.

According to a second aspect of the present invention, in addition to the first aspect, a coefficient learning means is provided in the learning means for learning each coefficient constituting the running resistance so that the current command value becomes zero. And a compensation value calculating means for calculating a compensation current command value from each coefficient learned by the coefficient learning means, the speed command value and the acceleration command value output from the speed command means, provided in the learning means. .

According to a third aspect of the present invention, in addition to the first aspect of the present invention, a current compensation value and a current for a traveling resistance which are provided in the learning means and are determined in accordance with a change in a road condition and a detected speed value. A coefficient learning means for learning each coefficient constituting the running resistance such that the difference from the command value becomes zero, and a coefficient provided in the learning means, each coefficient learned by the coefficient learning means and a speed outputted from the speed command means. A compensation value calculating means for calculating a compensation current command value from the command value and the acceleration command value.

According to a fourth aspect of the present invention, there is provided a position / speed detecting means for detecting a position and a speed of a train, a speed command means for outputting a speed command value and an acceleration command value of a train, and a position / speed detecting means. Speed control means for calculating a current command value from the detected speed detection value and the speed command value output from the speed command means, and coefficient learning for learning each coefficient constituting the running resistance so that the current command value becomes zero. Means for calculating a compensation current command value from each coefficient learned by the coefficient learning means, the speed command value and the acceleration command value output from the speed command means.

According to a fifth aspect of the present invention, there is provided a position / speed detecting means for detecting a position and a speed of a train, a speed command means for outputting a speed command value and an acceleration command value of the train, and a position / speed detecting means. Speed control means for calculating a current command value from the detected speed value and the speed command value output from the speed command means, and a current compensation value and a current corresponding to a running resistance determined according to a change in the road condition and the speed detection value. Coefficient learning means for learning each coefficient constituting the running resistance so that the difference from the command value becomes zero, each coefficient learned by the coefficient learning means, a speed command value output from the speed command means, and an acceleration command. And a compensation value calculating means for calculating a compensation current command value from the value.

[0015]

[0016]

[0017]

According to the first to third aspects of the present invention, the current command value is calculated by the speed control means so that the actual speed of the train follows the speed command value. Using the speed, acceleration, position, and current command value of the train, the learning means learns each coefficient constituting the running resistance of the train, and outputs a compensation current command value using each learned coefficient.

As described above, by reflecting the compensation current value corresponding to the running resistance value on the current command value, the drive control of the train can be prevented from being affected by the change in the running resistance due to a tunnel, a slope, or the like. An apparatus can be provided.

According to the invention described in claim 4 or 5,
The difference between the current command and the feedback current command output from the speed control, or the component of the current command such that the feedback current command itself becomes zero, is expressed by a neural network. We learn by technique of h. As a result, it is possible to provide a train running resistance learning device that requires the running resistance and weight of the train.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the present invention will be described in detail with reference to the drawings. FIGS. 1 and 2 show an embodiment of the invention described in claims 1 and 2, FIG. 1 is a configuration diagram of a drive control device of a train, and FIG. 2 is a configuration diagram of a compensation value calculation unit. .

In FIG. 1, 1 is a speed command unit, 2 is a speed control unit, 3 is a current supply unit, 4 is a train driving motor, 5 is a train, 6 is a position / speed detection unit, 7 is a compensation value calculation unit, 8 is a subtraction part, 9 is an addition part. The speed command section 1, the speed control section 2, the current supply section 3, the electric motor 4, the position / speed detection section 6, the compensation value calculation section 7, the subtraction section 8, and the addition section 9 are mounted on the train 5, It may be installed in the central control device of the company.

The speed command unit 1 outputs a speed command value Vc of the train 5. The subtraction unit 8 receives the speed command value Vc and the actual speed V of the train 5 detected by the position / speed detection unit 6, and calculates a speed difference Vd between the speed command value Vc and the actual speed V. Train 5
Is detected by the position / speed detecting unit 6, and the position / speed detecting unit 6 includes, for example, an electric motor 4 mounted on the train 5.
This is a rotation speed detector for detecting the rotation speed of. This electric motor 4
The actual speed V of the train 5 and the traveling distance (position) X of the train 5 are detected from the rotation speed of the train 5. If the train 5 does not function to rotate the electric motor 4 like a magnetic levitation railway, a sensor for detecting a ground coil installed on the ground side is provided as a position / speed detecting unit 6 on the train 5 to detect the ground coil. Train 5 by number
, The actual speed V of the train 5 is calculated by differentiating the travel distance. Speed control unit 2
Receives the speed difference Vd calculated by the subtraction unit 8 and calculates a current command value ImFB supplied to the electric motor 4 to make the speed difference Vd zero by proportional control or PI control. When the electric current is supplied by the current command value ImFB, the electric motor 4 causes the train 5 to run with the thrust FX. However, the running resistance Fd such as the air resistance and the resistance due to the gradient is applied to the train 5, so that the thrust of the entire train 5 is FX. −Fd. Therefore, the compensation value calculating section 7 searches the stored running resistance Fd based on the actual speed V and the position X detected by the position / speed detecting section 6, and obtains the current compensation value ImFF corresponding to the running resistance Fd. Is calculated. Then, the addition section 9 adds the current compensation value ImFF and the current command value ImFB to obtain a compensation current command value Im. The current supply unit 3 supplies a current to the electric motor 4 based on the compensation current command value Im. In other words, the motor 4 has the compensation current command value Im
When the current is supplied, the train 5 is caused to run with the thrust FX + Fd, and the thrust of the entire train 5 becomes FX due to the running resistance Fd applied to the train 5.

Next, the configuration of the compensation value calculator 7 will be described with reference to FIG. The compensation value calculation unit 7 calculates the air resistance FV and the gradient resistance Fg of the running resistance Fd based on the actual speed V and the position X detected by the position / speed detection unit 6. Air resistance FV
Is said to be approximately proportional to the square of speed. Square operator
At 71, the actual speed V is input and the square value of the actual speed V is calculated. The section storage unit 72 inputs the position X detected by the position / speed detection unit 6. Section information indicating whether the section is a light section L or a tunnel section T is given and stored in the section storage section 72 based on the position X. When the position X is input, the position X
, The section information L (hereinafter, referred to as a light section L) or the section information T (hereinafter, referred to as a tunnel section T) is output. The air resistance coefficient switching unit 73 switches the air resistance coefficient based on the light section L or the tunnel section T output from the section storage unit 72. The air resistance coefficient differs between the light section L and the tunnel section T as shown in FIG. Therefore, the air resistance coefficient is changed by the air resistance coefficient switching unit 73 to the air resistance coefficient (K
1) Switching between 74a and air resistance coefficient (K2) 74b. That is, when the section currently running is determined to be the light section L by the section storage section 72, the air resistance coefficient (K1) 74a is connected by the air resistance coefficient switching section 73, and the square value of the actual speed V and the air resistance The product of the coefficient (K1) 74a is output as the air resistance FV. If the section currently running is determined to be the tunnel section T by the section storage section 72, the air resistance coefficient switching section 73 connects the air resistance coefficient (K2) 74b,
The product of the square value of the actual speed V and the air resistance coefficient (K2) 74b is output as the air resistance FV. Thus the air resistance FV is in lights section traveling FV = K1v ^2, during the tunnel section traveling is calculated as FV = K2V ^2. On the other hand, the position X detected by the position / velocity detector 6 is input to the gradient storage 75.
The gradient storage unit 75 stores the gradient Y% of the position X in accordance with the position X.
When a position X is input, a gradient Y is output according to the position X. The gradient coefficient conversion unit 76 calculates a gradient coefficient sinθ from the gradient Y output from the gradient storage unit 75 based on the following equation (1).

[0024]

(Equation 1) Then, the gradient coefficient sinθ calculated by the gradient coefficient conversion unit 76
And the gradient resistance coefficient (K3) 77 is output as the gradient resistance Fg. The gradient resistance coefficient (K3) 77 is a product of the weight M of the train 5 and the gravitational acceleration g. The adder 78 has an air resistance F
The running resistance Fd, which is the sum, is calculated by inputting V and the gradient Fg. From the position X detected by the position / speed detector 6, the traveling section and the gradient, which is the road condition, are obtained, the traveling resistance Fd is calculated from the route condition and the actual speed V, and the traveling is performed via the conversion coefficient 79. A current compensation value ImFF for the resistance Fd is obtained.

Therefore, when the current compensation value ImFF corresponding to the running resistance Fd is added to the current command value ImFB, and the current compensation value ImFF corresponding to the running resistance is calculated, the running resistance Fd that changes discontinuously according to the road conditions is varied. By doing
Drive control of the train without fluctuation can be performed.

FIGS. 4 and 5 show an embodiment of the invention according to claims 3 and 4, wherein FIG. 4 is a block diagram of a train drive control device, and FIG. 5 is a neural network learning. It is a block diagram of a part.

In FIG. 4, 11 is a speed command unit, 12 is a speed control unit, 13 is a current supply unit, 14 is a motor for driving a train, 15 is a train, 16 is a position / speed detecting unit, and 17 is a neural network.
A learning unit (hereinafter, referred to as an NN learning unit) 18 is a subtraction unit. The speed command unit 11, the speed control unit 12, the current supply unit 13, the electric motor 14, the position / speed detection unit 16, the NN learning unit 17, and the subtraction unit 18 are mounted on the train 15, but are installed outside. There is also. The speed command unit 11 outputs a speed command value Vc and an acceleration command value ac of the train 15. The subtraction unit 18 receives the speed command value Vc and the actual speed V of the train 15 detected by the position / speed detection unit 16 and calculates a speed difference Vd between the speed command value Vc and the actual speed V. In the speed control unit 12, proportional control or PI control is performed,
A current command value ImFB that makes the speed difference Vd zero is calculated. When the motor 14 is supplied with a current by the current command value ImFB, the train 15 runs with a thrust FX. However, the running resistance Fd is applied to the train 15, so that the thrust of the entire train 15 is FX -F
becomes d. Therefore, the NN learning unit 17 calculates a current compensation value ImFF corresponding to the running resistance Fd with respect to the current command value ImFB, and outputs the compensation current command value Im to the current supply unit 13. The current supply unit 13 supplies a current to the electric motor 14 based on the compensation current command value Im. In other words, when the motor 14 is supplied with a current according to the compensation current command value Im, the thrust of the entire train 5 becomes F
X.

Next, the configuration of the NN learning unit 17 will be described with reference to FIG. The running resistance Fd depends on the speed. Further, as shown in FIG. 3, the coefficient of the air resistance of the running resistance differs between the tunnel section T and the light section L. The gradient resistance of the running resistance Fd varies depending on the train weight. That is, the running resistance Fd changes discontinuously when the train 15 enters the tunnel T even if the speed is constant, and the fluctuation is
Of the converter 13 and the current command value ImFB to the converter 13. Using the change in the current command value ImFB, the NN learning unit 17 learns the coefficient of the air resistance and the train weight of the running resistance of the train 15. By learning the air resistance coefficient and train weight of the train 15, if the initial value of the air resistance coefficient or the train weight differs from the actual value while the train 15 is running, the train weight may fluctuate due to passengers getting on and off. Therefore, the air resistance coefficient and the train weight are brought close to the constant values close to the true values by the NN learning unit 17, and the current compensation value ImFF calculated by the NN learning unit 17 is made more accurate.

In FIG. 5, reference numeral 171 denotes a coupling coefficient W to be described later.
1, a coupling coefficient learning section for learning W2 and W3; 172, a compensation value calculation section for calculating a current compensation value ImFF; 173, a compensation current command value Im which is the sum of a current command value ImFB and a current compensation value ImFF; This is an adding unit.

In the compensation value calculation section 172, the force (hereinafter referred to as acceleration resistance) Fa acting on the train 5 based on the air resistance FV, the gradient resistance Fg and the acceleration command value ac of the running resistance Fd.
Is calculated. The air resistance FV is said to be approximately proportional to the square of the speed. The square calculation unit 174 receives the speed command value Vc output from the speed command unit 11 and calculates the square value of the speed command value Vc. The section storage unit 175 is the position / speed detection unit 16
Input the position X detected in. The section storage unit 175 stores section information indicating whether the section is a light section L or a tunnel section T based on the position X. When the position X is input, the section information L (hereinafter referred to as “section information L”) is set based on the position X. , Light section L) or section information T (hereinafter referred to as tunnel section T). The coupling coefficient switching unit 176 is a section storage unit.
The coupling coefficient is switched based on the light section L or the tunnel section T output from 72. When the currently traveling section is determined to be the light section L by the section storage unit 175, the coupling coefficient (W1) 177a is connected by the coupling coefficient switching unit 176, and the square value of the speed command value Vc and the coupling coefficient (W1) are connected. ) 177a is output as the air resistance FV. Section storage unit 175
If the section currently running is determined to be a tunnel section T, the coupling coefficient switching unit 176 sets the coupling coefficient (W2).
177b is connected, and the product of the square value of the speed command value Vc and the coupling coefficient (W2) 177b is output as the air resistance FV. Therefore, the air resistance FV is FV = W1Vc during driving in the light section.
^2, in the tunnel section traveling is calculated as FV = W2Vc ^2. On the other hand, the position X detected by the position / speed detector 16 is input to the gradient storage 178. The position X is stored in the gradient storage unit 178.
The gradient Y% of the position X is given and stored according to
When the position X is input, a gradient Y is output according to the position X. The gradient coefficient conversion unit 179 calculates the gradient coefficient sinθ from the gradient Y output from the gradient storage unit 178 based on the equation (1). The multiplication unit 180 receives the gradient coefficient sinθ calculated by the gradient coefficient conversion unit 179 and the gravitational acceleration g, and calculates the product g · sinθ.

Then, g · sin calculated by the multiplication unit 180
The product of θ and the coupling coefficient (W3) 177c is output as the gradient resistance Fg. The product of the acceleration command value ac output from the speed command unit 11 and the coupling coefficient (W3) 177c is output as the acceleration resistance Fa. The adder 181 inputs the air resistance FV, the gradient resistance Fg, and the acceleration resistance Fa and calculates the running resistance Fd which is the sum. This running resistance Fd is determined by the conversion coefficient (K
i) It is converted into a current compensation value ImFF for the running resistance Fd through 182. Then, the current compensation value ImFF and the current command value ImFB are added by the adder 173 to obtain the compensation current command value IFB.
m is obtained. Now, when the compensation current command value Im is in the light section L,

[0032]

Im = (W1 · V ² + W3 · a + W3 · g · sin θ) · Ki (2) In the tunnel section T

[0033]

It is assumed that Im = (W2 · V ² + W3 · a + W3 · g · sin θ) · Ki (3) The coupling coefficient learning unit 171 learns each coupling coefficient W1, W2, W3. Based on the compensation current command value Im, which is the sum of the current compensation value ImFF calculated by the compensation value calculation unit 172 using the speed command value Vc and the acceleration command value ac output from the speed command unit 11 and the current command value ImFB. When the current supply unit 13 supplies a current to the electric motor 14 to drive the train 15, if the values of the coupling coefficients W1, W2, and W3 take a true value, the actual speed detected by the position / speed detection unit 16 is used. V substantially follows the speed command value Vc output from the speed command unit 11, and the current command value ImFB output from the speed control unit 12 becomes almost zero. That is, the motor 14 is controlled by the compensation current command value Im calculated by the equation (2) or (3), and
15 drives. When the current command value ImFB does not become zero, the coupling coefficients W1, W2, and W3 do not take true values.
The coupling coefficient learning unit 171 inputs the actual speed V detected by the position / velocity detecting unit 16 and the current command value ImFB, and outputs the current command value ImF.
In the direction in which the mean square value J of B decreases, each coupling coefficient W1,
Learn W2 and W3. The gradient of the cost function is

[0034]

(Equation 4) And the learning formula is

[0035]

[Number 5] Wi (K + 1) = Wi (K) + Kd · δZi, i = 1,2,3 ... (5) where Zi is a variable ^{Z1 = V 2, Z2 = V} 2, Z3
= A = dV / dt. Kd is a gain.

By repeating the equation (5), the coupling coefficients W1, W2, W3 converge to a constant value. This coupling coefficient is the coefficient of the running resistance of the train. The air resistance FV is said to be approximately proportional to the square of the speed. In a place such as a tunnel where the coefficient of the air resistance FV is determined to be different depending on the position, when the train 15 comes to that position, the switch of the learning of the running resistance is switched, and the learning of the coefficients of W1 and W2 is performed, respectively. The learning of the coefficient of the running resistance in the section T and the lighting section L can be performed. The switching is determined based on the position X input to the coupling coefficient learning unit 171.

Since the gradient resistance Fg takes the value of M · g · sin θ with respect to the weight M of the train 15, the actual speed V of the train 15
Is learned using the acceleration a obtained by differentiating the equation (5), and the obtained coupling coefficient W3 corresponds to the weight M of the train 15.

By outputting the compensation current command value Im to the current supply unit using the learning value while learning each coefficient of the running resistance as described above, the initial value of each coefficient of the running resistance becomes true. Even if the value deviates from the value, it can be made to converge to a substantially true value by performing learning. In addition, even when the train weight changes, especially when passengers and luggage get on and off, the train drive control that does not fluctuate is calculated by calculating the coupling coefficient corresponding to the train weight and compensating the current value based on the coupling coefficient. Can be realized.

Next, an embodiment of the present invention will be described with reference to FIG. FIG. 6 is a configuration diagram of the neural network learning unit. In FIG. 6, 183 is a coupling coefficient learning unit, 184 is a compensation value calculation unit that calculates a current compensation value ImFF,
185 is an output switching unit, 186 is a subtraction unit that calculates the difference between the current command value ImFB and the current compensation value ImFF, 187 is an output switching unit 185 according to the output of the subtraction unit 186, and instructs switching of each switching unit described later. This is a switching command section.

First, the compensation value calculator 184 will be described. The square calculator 188 receives the actual speed V as input and calculates the square value of the actual speed V. The square operation unit 189 receives the speed command value Vc and calculates the square value of the speed command value Vc. The differential value calculator 190 receives the actual speed V as input and calculates an acceleration a, which is a differential value of the actual speed V. The section storage unit 191 receives the position X, retrieves the section information for the stored position X, and outputs the light section L or the tunnel section T.

The position X is input to the gradient storage section 192.
The gradient storage unit 192 stores the gradient Y of the position X in accordance with the position X.
When the position X is input, the gradient Y is output according to the position X. The gradient coefficient conversion unit 193 calculates the gradient coefficient s from the gradient Y output from the gradient storage unit 192.
inθ is calculated based on equation (1). The multiplication unit 194 inputs the gradient coefficient sinθ and the gravitational acceleration g, and multiplies them by g · s
Calculate inθ.

The outputs of the square calculation units 188 and 189 are input to the coupling coefficient switching unit 195. The coupling coefficient switching unit 195 switches input and output according to the light section L and the tunnel section T output from the section storage unit 191 and according to the output from the switching command unit 187 described later. Differential value calculation unit
The output of 190 and the acceleration command value ac are input to the acceleration switching unit 196. The acceleration switching unit 196 switches the input according to the output from the switching command unit 187. Outputs of the coupling coefficient switching unit 195 and the acceleration switching unit 196 are coupling coefficients (W1) 197a and (W2)
197b and (W3) 197c are converted into air resistance FV and acceleration resistance Fa. The output g · sin θ of the multiplication unit 194 is also converted to a gradient resistance Fg via the coupling coefficient (W3) 197c. The adder 198 inputs the air resistance FV, the gradient resistance Fg, and the acceleration resistance Fa and inputs a running resistance Fd which is a sum.
Is calculated. This running resistance Fd has a conversion coefficient (Ki) of 199.
Is converted to a current compensation value ImFF for the running resistance Fd. The output switching unit 185 inputs the current compensation value ImFF to the subtraction unit 186 according to the switching instruction of the switching command unit 187, or outputs the current compensation value ImFF as the compensation current command value Im.

The switching command section 187 includes coupling coefficients W1, W2,
When learning W3, the signal “1” is changed to the coupling coefficient switching unit 195,
Output to acceleration switching section 196 and output switching section 185. Now, in the light section L where the current command value ImFB is

[0044]

ImFB = (W1 · V ² + W3 · a + W3 · g · sin θ) · Ki (6) In the tunnel section T

[0045]

It is assumed that ImFB = (W2 · V ² + W3 · a + W3 · g · sin θ) · Ki (7) The coupling coefficient learning unit 183 learns each coupling coefficient W1, W2, W3. Position / speed detector 16
The actual speed V detected in the step (1).
Compensation value calculator 1 using acceleration a obtained by differentiation by 90
The current compensation value ImFF calculated at 84 is calculated based on the coupling coefficients W1, W
2. If the value of W3 is a true value, the current command value ImFB
And the current compensation value ImFF are equal. Therefore subtraction unit 1
The output of 86 becomes zero. In this case, the coupling coefficients W1, W2, W
3, the coupling coefficients W1, W
2, learning of W3 is completed, and the switching command section 187 outputs the signal "2".
The coefficient switching section 195, acceleration switching section 196, output switching section 185
To enter. Then, the learned coupling coefficients W1, W2,
The current compensation value ImFF calculated by the compensation value calculation unit 184 using the speed command value Vc output from the speed command unit 11 and the acceleration command value ac is used as the compensation power command value Im.
13 will be output. In this manner, the learning and the calculation of the compensation current command value Im are alternately performed by the switching command section 187. If the current command value ImFB is not equal to the current compensation value ImFF, the output of the subtraction unit 186 is ImFB−ImF
F, and the coupling coefficient learning unit 183 calculates the mean square value J of the difference ImFB−ImFF between the current command value ImFB and the current compensation value ImFF.
The coupling coefficients W1, W2, and W3 are learned in such a direction as to decrease. The gradient of the cost function is

[0046]

(Equation 8) And the learning equation is the same as equation (5). Next, the switching command section 187 outputs the signal “2” to the coefficient switching section 195 and the acceleration switching section 1.
96, Input to output switching unit 185. Then, the current compensation value ImFF calculated by the compensation value calculation unit 184 using the learned coupling coefficients W1, W2, W3, the speed command value Vc output from the speed command unit 11, and the acceleration command value ac is used as the compensation command. Value I
m is output to the current supply unit 13. In this manner, the learning and the calculation of the compensation current command value Im are alternately performed by the switching command section 187. Then, at the time of learning, the equation (5) is repeated, so that the coupling coefficients W1, W2, W3 converge to a constant value. This coupling coefficient is the coefficient of the running resistance of the train.

The air resistance FV is said to be approximately proportional to the square of the speed. Air resistance F depending on the position of tunnel
Where the coefficient of V is determined to be different, the train resistance learning switch is switched when the train 15 comes to that position, and W
By learning the coefficients 1 and W2, for example, it is possible to learn the coefficients of the running resistance in the tunnel section T and the light section L, respectively. The switching is determined based on the position X input to the coupling coefficient learning unit 183. The gradient resistance Fg is given by M · g · sin θ with respect to the weight M of the train 15.
Is calculated using the acceleration a obtained by differentiating the actual speed V of the train 15, and the obtained coupling coefficient W3 corresponds to the weight M of the train.

By outputting the compensation current command value to the current supply unit using the learning value while learning the running resistance coefficient as described above, the initial value of each coefficient of the running resistance deviates from the true value. In this case, it is possible to converge to a substantially true value by performing learning. In particular, even when the train weight changes due to passengers and luggage getting on and off, by calculating the coupling coefficient corresponding to the train weight and compensating the current value based on the coupling coefficient,
The drive control of the train without fluctuation can be realized.

FIG. 7 is a block diagram of a running resistance learning apparatus according to an embodiment of the present invention. In FIG. 7, 21
Is a coupling coefficient learning unit, 22 is a compensation value calculation unit that calculates a current compensation value ImFF, and 23 is a current command value ImFB and a current compensation value ImFF.
Is a subtraction unit that calculates the difference between.

First, the compensation value calculator 22 will be described. The square calculator 24 receives the actual speed V as input and calculates the square value of the actual speed V. The differential value calculation unit 33 receives the actual speed V as input and calculates an acceleration a which is a differential value of the actual speed V. The section storage unit 25 stores the position X
, The section information for the stored position X is retrieved, and the light section L or the tunnel section T is output.

The position X is input to the gradient storage unit 27. The gradient storage unit 27 stores a gradient Y% of the position X in accordance with the position X, and when the position X is input, the position X
Output the gradient Y in accordance with. The gradient coefficient conversion unit 28 calculates the gradient coefficient sinθ from the gradient Y output from the gradient storage unit 27 based on the equation (1). The multiplier 29 has a gradient coefficient sin
θ and the gravitational acceleration g are input, and the product g · sin θ is calculated.

The output of the square operation unit 24 is input to the coupling coefficient switching unit 26. The coupling coefficient switching unit 26 switches the output according to the light section L and the tunnel section T output from the section storage unit 25. The output of the coupling coefficient switching unit 26 is converted into the air resistance FV via the coupling coefficients (W1) 30a and (W2) 30b, and the output a of the differential value calculating unit 33 is the coupling coefficient (W3)
Is converted to an acceleration resistance Fa. The output g · sin θ of the multiplying unit 29 is also converted to a gradient resistance Fg through the coupling coefficient (W3) 30c. The adder 198 inputs the air resistance FV, the gradient resistance Fg, and the acceleration resistance Fa, and calculates the running resistance Fd which is the sum. The running resistance Fd is converted into a current compensation value ImFF for the running resistance Fd via a conversion coefficient (Ki) 199.

It is assumed that the current command value ImFB is constituted by the equation (6) in the light section L and by the equation (7) in the tunnel section T. In the coupling coefficient learning unit 21, each coupling coefficient W1, W
2. Learn W3. The actual speed V and the current compensation value ImFF calculated by the compensation value calculating unit 22 using the acceleration a obtained by differentiating the actual speed V by the differential value calculating unit 33 are represented by a coupling coefficient W
If the values of 1, W2 and W3 are true values, the current command value ImFB and the current compensation value ImFF will be equal. Therefore, the output of the subtraction unit 23 becomes zero. If the current command value ImFB is not equal to the current compensation value ImFF, the output of the subtraction unit 23 is ImFB−ImFF, and the coupling coefficient learning unit 21 outputs the difference ImFB− between the current command value ImFB and the current compensation value ImFF−. ImF
In the direction in which the mean square value J of F decreases, each coupling coefficient W1,
Learn W2 and W3. The gradient of the evaluation function is expressed by equation (8), and the learning equation using Kd as the gain is the same as equation (5). By repeating the equation (5), each coupling coefficient W1,
W2 and W3 converge to a constant value. This coupling coefficient becomes the coefficient of the running resistance of the train, and the coefficient of the running resistance is learned by the above-described method.

By learning the running resistance coefficient as described above, the result can be used for correction of train speed control, examination of running characteristics, and the like. Further, the running resistance learning device is not limited to the present embodiment, and the learning may be performed using the speed command value and the acceleration command value of the train as the same configuration as the NN learning unit of the drive control device illustrated in FIG. Similar effects can be obtained.

[0055]

As described above, according to the first or third aspect of the present invention, the driving fluctuation of the train is corrected by correcting the current corresponding to the running resistance of the train with respect to the current command value. Can be suppressed, and the riding comfort can be improved. Further, by providing a function of learning each coefficient used for calculating the running resistance as needed, the current for the running resistance can be calculated more accurately. According to the invention described in claim 4 or 5, each coefficient of the running resistance can be learned at any time, and the learning result can be applied to correction of a current command value of a train, examination of running characteristics, and the like. Can be.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of a drive control device for a train, showing one embodiment of the invention described in claim 1 or 2;

FIG. 2 is a diagram showing one embodiment of a compensation value calculation unit in FIG. 1;

FIG. 3 is a characteristic diagram of an air resistance coefficient.

FIG. 4 is a configuration diagram of a drive control device for a train showing one embodiment of the invention according to claims 3 to 5;

FIG. 5 is a diagram illustrating an embodiment of an NN learning unit in FIG. 4;

FIG. 6 is a diagram illustrating another embodiment of the NN learning unit in FIG. 4;

FIG. 7 is a configuration diagram of a train running resistance learning apparatus according to an embodiment of the present invention.

FIG. 8 is a configuration diagram of a conventional train drive control device.

FIG. 9 is a diagram showing characteristics of a running resistance applied to a train, a train speed, and an acceleration.

FIG. 10 is a diagram showing a change in running resistance due to a gradient.

[Explanation of symbols]

 1, 11: speed command unit 2, 12, speed controller 3, 13, current supply unit 4, 14, electric motor 5, 15, train 6, 16, position / speed detector 7, compensation value calculator 17, NN learning unit 171, 183, 21 ... coupling coefficient learning unit 172, 184, 22 ... compensation value calculation unit

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) B60L 1/00-3/12 B60L 7/00-13/00 B60L 15/00-15/42 G05B 13 / 02

Claims (5)

(57) [Claims] 1. A position / speed detecting means for detecting a position and a speed of a train, a speed command means for outputting a speed command value and an acceleration command value of the train, and a speed detected value detected by the position / speed detecting means. Speed control means for calculating a current command value from the speed command value output from the speed command means, and recognizes a change in the route condition from the position detected by the position and speed detection means, and Learning means for learning each coefficient constituting the running resistance using the detected speed value and the current command value, and calculating a compensation current command value using each coefficient constituting the learned running resistance; And a current supply means for supplying a compensation current command value calculated in (1) to the electric motor. 2. The train drive control device according to claim 1, wherein the current command value is provided in the learning means, and the current command value is zero.
Coefficient learning means for learning each coefficient that constitutes the running resistance so as to be provided, and each coefficient learned by the coefficient learning means and the speed command value output from the speed command means. And a compensation value calculating means for calculating the compensation current command value from the acceleration command value. 3. The train drive control device according to claim 1, wherein a current compensation value for a running resistance, which is provided in the learning means and is determined according to the change in the route condition and the speed detection value, and the current command. A coefficient learning means for learning each coefficient constituting the running resistance so that the difference between the coefficient and the value becomes zero; and a coefficient provided in the learning means and output from each coefficient learned by the coefficient learning means and the speed command means. A drive control device for a train, comprising: a compensation value calculation unit configured to calculate the compensation current command value from the speed command value and the acceleration command value. 4. A position / speed detecting means for detecting a position and a speed of a train, a speed command means for outputting a speed command value and an acceleration command value of the train, and a speed detected value detected by the position / speed detecting means. Speed control means for calculating a current command value from the speed command value output from the speed command means, and coefficient learning means for learning each coefficient constituting a running resistance so that the current command value becomes zero , A train running resistance learning device having a compensation value calculating means for calculating a compensation current command value from each coefficient learned by the coefficient learning means and the speed command value and the acceleration command value output from the speed command means. . 5. A position and speed detecting means for detecting a position and a speed of a train, a speed commanding means for outputting a speed command value and an acceleration command value of the train, and a speed detected value detected by the position and speed detecting means. Speed control means for calculating a current command value from the speed command value output from the speed command means, a current compensation value for a running resistance determined according to a change in road condition and the speed detection value, and the current command value. So that the difference with
A coefficient learning means for learning the respective coefficients constituting the running resistance to the compensation current command from the velocity command value output coefficients learned from the speed command means for this coefficient learning means and said acceleration command value A train running resistance learning device having a compensation value calculating means for calculating a value.