JPH09200910A - Automatic train operating apparatus - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a train operating apparatus which is capable of conducting deceleration control with good quick-response and following-up in a deceleration control stage prior to proceeding to positioning control to keep a stop position. SOLUTION: An estimating means 40 for a target system estimates the target system of a set position stop control system in consideration of the dead time which is held by the target system estimated by a dead time estimating means 50. A control parameter calculating means 70 calculates a control parameter which keeps the target system in consideration of the dead time. Using the control parameter, a principle control means 60 conducts deceleration control. It is thus possible to conduct a set position stop control in consideration of the dead time held by the target system.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic train operation device for automatically operating a train in a railway and a new transportation system.

[0002]

2. Description of the Related Art In general, in the field of railways and new transportation systems, in order to improve safety, reduce the skill level and burden required of a driver, and further increase the density of train operation by automating speed control. In addition, the development and introduction of an automatic train operation device for automatically operating a train are in progress.

The automatic train operation device automatically departs a train and operates the train automatically by operating a device for instructing departure such as a departure button or by remote control.
Then, in the speed limit section, in addition to an automatic train control device (called an ATC) that automatically brakes and slows down when the train speed becomes equal to or higher than the speed limit, a target speed following function for automatically operating the train, Fixed position stop control function,
It can be said that the scheduled operation function is added. Among them, the fixed position stop control function accurately stops the train at a predetermined stop target, and is an important function in terms of service improvement. The fixed position stop control function of the conventional automatic train operation device will be described with reference to FIGS. 9 and 10.

FIG. 9 is an explanatory diagram of a case where the train is controlled by the fixed position stop control function of the automatic train operation device. FIG. 9 shows a state where the train 200 is about to stop at the platform 210 of the station. Train 200
Runs on the rail 220, and the running position is the ground child P1.
Detected by P4. That is, the train 200 communicates with the ground terminals P1 to P4 in order to accurately stop at the preset stop position of the platform 210. Here, each ground element P1 to P4 is, for example, the ground element P1 for confirming a point for shifting to the fixed position stop control, the ground element P2 for confirming a point during the fixed position stop control, and the ground element P3 is stopped. The last one on the verge of applying the brake, the ground element P4 is used for confirming the stop.

FIG. 10 shows a running curve of a train from station A to station B. The train departs from station A, begins to power by the automatic train driving device, and after a while goes into row operation. As a result, the train speed does not exceed the speed limit even in the set speed limit section. Then, when the position of the ground element P1 is exceeded in order to re-power and stop at the B station, the fixed position stop control function works to start deceleration control. At the point of the ground element P2, the degree of deceleration is checked, and at the point of the ground element P3, the speed control is shifted to the position control. At the position of the ground child P4, confirmation is performed immediately before or immediately after the stop, and for example, when an abnormality occurs, an emergency brake is applied (Reference 1, Hitachi Review VOL76, No. 5,199).
4,5 "ATC / ATO system for high-speed and high-density operation").

In this fixed position stop control, in the case of a freight vehicle, the deceleration performance changes depending on the loading rate or the freight vehicle, and the train braking device has a dead time due to the mechanism, and the dead time changes depending on the situation. In addition, since the characteristics of the target system including dead time change depending on the weather, there are problems such as variations in stopping accuracy and sluggish driving for stopping position adjustment. On the other hand, a method applying the preview fuzzy control method
No. 117304, Japanese Patent Laid-Open No. 3-117305)
In addition, a method applying neuro-fuzzy control (reference document 2, Toshiba Review, 1995, VOL50, No. 2, “Fuzzy, neuro control and application to automatic train operation”) has been developed.

[0007]

However, in the method applying the predictive fuzzy control method and the method applying the neuro-fuzzy control, although the stop control accuracy can be improved, the characteristics (dead time element) of the target system are positively affected. However, there were problems such as spending a lot of time on adjustment work to ensure control performance due to differences in routes and whether the vehicle was old or new.

An object of the present invention is to obtain an automatic train operation device capable of performing deceleration control with good responsiveness and followability in the stage of deceleration control before shifting to position control for stop position adjustment. is there.

[0009]

The invention according to claim 1 is a dead time estimating means for estimating a dead time of a fixed position stop control system (target system) for stopping a train at a fixed position, and this dead time. Target system estimating means for modeling the target system in consideration of the dead time estimated by the estimating means, control parameter calculating means for calculating control parameters based on the model estimated by the target system estimating means, and control The control parameter calculated by the parameter calculation means is used, and a deceleration control means for calculating the operation amount for decelerating the train in consideration of the dead time and outputting it to the train braking device is provided.

According to the first aspect of the present invention, when estimating the target system of the fixed-position stop control system, the dead time of the target system is estimated according to the situation, and the dead time is taken into consideration to match the target system. Calculate control parameters. Then, deceleration control is performed using the control parameter. Therefore, the fixed position stop control in consideration of the dead time of the target system becomes possible.

According to a second aspect of the present invention, in the first aspect of the present invention, a deceleration pattern creating means for creating a target deceleration pattern which is a target deceleration value for deceleration control of the train and outputting it to the deceleration control means is added. It is provided.

According to the second aspect of the present invention, the target deceleration pattern of the horizontal axis position or the time-vertical vertical target speed up to the set target speed (0 km / h) for stopping is created toward the stop of the train. It is possible to perform deceleration control that is suited to the route situation, passenger class and the intention of the railway company.

According to the invention of claim 3, in the invention of claim 1 or claim 2 after claim, the dead time estimation means is divided into when the train is braking and when the train is not braking. , The dead time is estimated.

According to the third aspect of the present invention, the dead time estimation by the dead time estimation means can be performed separately while the train is braking and when the train is not braking. Estimated dead time.

According to a fourth aspect of the present invention, in the third aspect of the present invention, the dead time estimation means separately estimates the dead time when the train starts to brake and when the train shifts to dead or power running. Is to do.

In the invention of claim 4, the value of the dead time is changed between when the train starts to brake and when the train shifts to running or power running. Therefore, the dead time can be estimated with higher accuracy.

According to a fifth aspect of the present invention, in the second aspect of the present invention, weather mode input means for inputting a weather mode is provided, and the deceleration pattern creating means creates a target deceleration pattern based on the weather mode. The target deceleration pattern that conforms to is output to the deceleration control means.

According to the fifth aspect of the present invention, since the weather mode can be input, it is possible to estimate the target system that reflects the braking effectiveness. Further, even when it is necessary to start braking early because the braking force is saturated, the deceleration pattern can be appropriately switched or changed.

According to a sixth aspect of the present invention, in the first aspect of the present invention, target system estimation learning means for tracking the seasonal variation and the secular change of the target system is provided, and the learning result is used for the estimation calculation of the target system estimation means. It is intended to be reflected.

According to the invention of claim 6, the target system is automatically learned at any time by the target system estimation learning means for keeping track of seasonal fluctuations and secular changes of the target system. Becomes

According to a seventh aspect of the present invention, in the sixth aspect of the present invention, a dead time estimation learning means is provided for keeping track of the seasonal variation and the secular change of the dead time of the target system, and the learning result is the dead time estimation means. This is reflected in the estimation calculation.

In the invention of claim 7, the dead time is automatically learned at any time by the dead time estimation learning means for keeping track of the seasonal variation and the secular change of the dead time of the target system. Maintenance free.

The invention according to claim 8 is claim 6 or claim 7.
In the invention, the parameter recording means for extracting the learning result of the target system estimation learning means or the learning result of the dead time estimation learning means and storing it in a recording medium is provided.

According to the invention of claim 8, since there is a parameter recording means for taking out the learning result (parameter) of the learned target system, the train data can be taken out and provided to another train.

[0025]

Embodiments of the present invention will be described below. FIG. 1 is a block diagram showing a first embodiment of the present invention.

The dead time estimating means 50 estimates the dead time of the fixed position stop control system (target system) for stopping the train at a fixed position, and the dead time estimated here is the target system estimation. It is input to the means 40. The target system estimating means 40 models the target system in consideration of the dead time estimated by the dead time estimating means 50. The parameter representing the target system of the model estimated by the target system estimation means 40 is input to the deceleration control means 60 and the control parameter calculation means 70.

The control parameter calculation means 70 calculates control parameters based on the parameters representing the target system. The deceleration control means 60 calculates an operation amount for decelerating the train in consideration of the dead time, based on the parameter representing the target system and the control parameter. Then, it outputs to the train brake device 15 to perform fixed position stop control of the train.

That is, in performing the fixed position stop control, the target system estimation means 40 models the fixed position stop control system (brake system) which is the target system. Since the target system can be approximated by including the dead time, the dead time estimation means 5
0, and the estimation result is output to the target system estimation means 40.

In the control parameter calculation means 70, the parameter representing the target system output from the target system estimation means 40 is input, the control parameter for performing the deceleration control is calculated, and this control parameter is input to the deceleration control means 60. To be done. The deceleration control means 60 calculates the operation amount for decelerating the train using the parameters and control parameters of the target system, and gives the operation amount to the train braking device 15.

Here, in the target system estimating means 40,
The transfer function Gr (s) of the target system is approximated by the equation (1).

[0031]

[Equation 1]

E-SL on the right side of the equation (1) means that there is a dead time L [seconds]. In the first embodiment, G (s) is approximated by a first-order lag system, that is, one having a gain of Kc and a time constant of Tc seconds, as shown in equation (2).

G (s) = Kc / (1 + Tc · s) (2) Next, the deceleration control means 60 and the control parameter calculation means 7
0 will be described. FIG. 2 is a block diagram of the deceleration control means 60. The deceleration control means 60 is a feedback control system that feeds back the actual speed of the train and controls it based on the deviation from the target value. As the target value of the train, a fixed value or a target deceleration pattern created by the deceleration pattern creating means 80 described later is used.

To detect the actual speed of the train, a method of detecting and calculating the rotation amount of the train driving motor is adopted. Since the target system has dead time, the Smith compensation type controller that can design the control system by subtracting the dead time is adopted. The transfer function of the main controller 61 of the Smith compensation type controller is Gc (s). In the Smith compensation type controller, the transfer function G (s) · es of the target system
If L can be estimated, by constructing it as shown in Fig. 2,
The response of the target system excluding the dead time can be adjusted with Gc (s). It is also known that the estimation error is robust. 2 in FIG. 2 indicates that it is an estimated value.

Transfer function Gc of main controller 61
(S) uses PI control, and its control parameter is determined by using a general method in control theory, such as the Ziegler Nicholas method, in consideration of the quick response and the settling property.

The parameters and control parameters of the target system determined as described above are input to the deceleration control means 60, and the operation amount (for example, brake notch) for the train braking device 15 is calculated according to the block diagram of FIG.

As described above, in the first embodiment, the target system is estimated by explicitly considering the dead time element, and the deceleration control system is configured to perform the deceleration control in consideration of the dead time. Is improved.

Next, a second embodiment of the present invention will be described. FIG. 3 is a block diagram showing a second embodiment of the present invention. The second embodiment is different from the first embodiment shown in FIG. 1 in that a target deceleration pattern that is a target deceleration value when performing deceleration control of a train is created and output to a deceleration control means. The creation means 80 is additionally provided. That is, the deceleration pattern creating means 80 sets the target speed (0 km / km) for stopping the train.
A target deceleration pattern of the horizontal axis position or time vs. vertical axis target speed up to h) is created. Therefore, it is possible to perform deceleration control suitable for the route situation, the passenger class, and the intention of the railway company.

In FIG. 3, deceleration pattern creating means 80
Creates a target train speed in deceleration control of the train and inputs it to the deceleration control means 40. Regarding the braking performance of trains, there are also operational conditions such as different mechanical structures depending on the type of vehicle, and that the deceleration should not be increased in the high-speed range and that the deceleration should be decreased on rainy days. Further, in order to control train operation intervals and the like, in order to adjust the running time between stations of the train, the brake start position at the time of stopping the station may be advanced and the speed may be slowly reduced.

A target deceleration pattern is created in consideration of such various conditions. In the second embodiment, from the maximum speed to 0 km / h, 85% of the normal maximum braking speed is used.
Create a target deceleration pattern that decelerates at a constant deceleration such as. By doing so, the target deceleration pattern can be easily created.

Next, the dead time estimation means 50 estimates the dead time by dividing the train into the period when the train is braking and the time when the train is not braking. How to use the dead time, gain, and time constant of the target system in the control parameter calculation means 70 in this case will be described. When you start deceleration to stop the station, the brakes start to be applied, so the dead time before the brakes start to take effect is large. Therefore, the operation amount of the train braking device 15 is calculated using the value of the dead time when the brake starts to be broken, which is determined by the dead time estimation means 50.

After that, during a fixed time including the dead time,
Hold and use the value of the dead time when starting to apply the brake. After that, if the vehicle is traveling in the brake, the operation amount is calculated using the dead time during the traveling in the brake. In this way, the value of the dead time is changed according to the situation of the target system. This improves the deceleration control performance.

The dead time estimating means 50 estimates the dead time separately when the train starts to brake and when the train shifts to dead running or power running. In the case of a train with a brake disc as the braking device, this is a large time factor for completely sandwiching the brake disc when starting to apply the brake, and also when the brake is removed. This is because there is a dead time before they can be separated. Therefore,
These dead times are set separately. This improves the modeling accuracy of the target system.

Next, FIG. 4 shows a third embodiment of the present invention. The third embodiment is different from the second embodiment shown in FIG. 3 in that a weather mode input means 90 for inputting a weather mode is provided, and the deceleration pattern creating means 80 is
A target deceleration pattern is created based on the input weather mode, and the target deceleration pattern that matches the weather mode is output to the deceleration control means 60. As a result, the weather mode can be input, and the target system that reflects the braking effectiveness can be estimated. Further, even when it is necessary to start braking early because the braking force is saturated, the deceleration pattern can be appropriately switched or changed.

FIG. 5 is a flow chart showing the operation of the target system estimating means 40 and the dead time estimating means 50 in the third embodiment. The weather mode is set by a train crew member or a commander of the train operation management system from the weather mode input means 90 by manual input such as a push button. The weather mode set by the weather mode input means 90 is input to the target system estimating means 40, the dead time estimating means 50, and the deceleration pattern creating means 80.

First, the target system estimating means 40 reads the weather mode set by the weather mode input means 90 (S41). Next, the load weight of the train is read (S42), and the load weight level is determined (S43). For this load weight, if the train has a means for detecting the load weight such as a load-bearing device, that value is used. Alternatively, the load weight may be set in advance according to the time zone and that value may be used.

The target system estimating means 40 stores a table as shown in FIG. 6 in advance, and the weather mode is as shown in FIG.
As shown in (4), it has four modes of "beginning of rain", "rain", "after rain", and "cloudy to fine". If the weather mode is not set, select "Cloudy to Sunny" as the default.

Regarding the load weight level,
The load weight level is determined from four load weight values of "light", "normal", "heavy", and "extra heavy". Using the weather mode and load weight level set as described above, the gain and time constant of the target system are selected from the table shown in FIG. 6 (S4).
4). In FIG. 6, the column shows the load weight level, the row shows the weather mode, and the upper part of the table shows the value of the gain Kc and the lower part shows the value of the time constant Tc. Thereby, (2)
The transfer function G (s) shown in the equation is determined, and the transfer function Gt (s) of the target system shown in the equation (1) is determined.

Next, the dead time estimating means 50 determines whether the current operating state of the train is power running, running or braking (S45), and determines the dead time according to these states. In the case of power running or running, the dead time at which the braking is started is selected (S46). For example, 2 seconds. Then, if necessary, the value of the dead time is finely adjusted according to the weather mode and the load weight level (S47).

On the other hand, when the vehicle is running in the brake mode, the dead time for applying the brake is selected after applying the brake once (S48). For example, it is determined as 0 seconds during braking. Then, if necessary, the value of the dead time is finely adjusted according to the weather mode and the load weight level (S49).

As described above, in the third embodiment,
By setting the weather mode, it is possible to compensate for slipping and gliding in the rain. Also, by changing the target deceleration pattern depending on the weather mode and lowering the target speed with the deceleration start position facing you, you can decelerate reasonably,
As a result, the stop position accuracy is improved.

Next, a fourth embodiment of the present invention will be described. FIG. 7 shows a fourth embodiment of the present invention. This fourth
The embodiment of the present invention is different from the first embodiment shown in FIG. 1 in that a target system estimation learning means 100 is provided to follow the seasonal variation and the secular change of the target system, and the target system estimation learning means 100 is also provided. The parameter recording means 120 for storing the learning result of 1. in the recording medium is provided.

As a result, the learning result of the target system estimation learning means 100 is reflected in the estimation calculation of the target system estimation means 40, and the learning result (parameter) of the learned target system is taken out by the parameter recording means 120. Offer for another train. In this case, as the parameter recording means 120, a floppy disk drive device or an IC memory installed as an input / output device of an automatic train operation device installed in a train is used.

Here, the third embodiment of the above-mentioned first embodiment
In the embodiment of the present invention, the target system estimating means 40 is adapted to select a parameter representing the target system from a previously stored value, for example, a table as shown in FIG. In the fourth embodiment, the parameter representing the target system is learned as needed using control input and control result data.

The parameters representing the target system depend on the weather mode and the load weight level. Therefore, learning is performed by classifying the weather mode set value and the load weight level. In the fourth embodiment, the load weight level is learned by focusing on the route that can be roughly divided according to the time zone of operation, and is divided into the time zone and the weather mode.

Further, the data during the brake running which is not the beginning of the braking which is relatively less influenced by the dead time is used.

For example, the target system is converted from the transfer function G (s) of the continuous system to the discrete system model G (z) by using the Z transform.

[0058]

[Equation 2]

Here, (1-e-Ts) / s represents a zero-order hold. In general, the discrete system model can be expressed by the following equation (4).

[0060]

(Equation 3)

By processing this model parameter using the method of least squares and calculating the step response based on this parameter, the model parameter of the continuous target system can be estimated. In this way, each train learns the result of calculating the model parameters in daily running by using exponential smoothing and average.

According to the fourth embodiment, the model parameters of the own train can be learned inside the automatic train operation device installed in each train. As a result, the target system estimation learning means 100 for keeping track of seasonal changes and aging changes of the target system is automatically learning the target system at any time, so that the target system is estimated to be maintenance-free. Further, since the learning result is stored in the storage medium by the parameter recording means 120,
It can be shared with inspection data recording such as pre-work inspection, and the device can be simplified.

Next, FIG. 8 shows a fifth embodiment of the present invention. The fifth embodiment is different from the fourth embodiment shown in FIG. 7 in that a dead time estimating means 110 is provided. That is, a dead time estimation learning means is provided for keeping track of seasonal changes and aging changes in the dead time of the target system, the learning result is reflected in the estimation calculation of the dead time estimation means 50, and the learning result is recorded in the parameter recording means 120.
To be recorded on the recording medium.

The dead time estimation / learning means 110 sequentially learns the dead time of the target system by using the data of the control input and the control result. Learning with if-then-rule using the control area, which is an index of control response at the time when braking starts, that is, the value obtained by integrating the deviation (difference) of the control response from the target value and the peak value. It shall be.

Here, due to the nature of overshooting when the dead time is expected to be small and prolonging the time to reach the target value when the dead time is expected to be large, the dead time is changed using fuzzy inference according to the following rules. (Rule 1) If the peak value is large and the control area is large, the dead time is reduced. (Rule 2) If the peak value is less than the target value and the control area is small, the dead time is increased a little.

In the fifth embodiment, the dead time is automatically learned from time to time by the dead time estimation learning means 110 for keeping track of seasonal changes and aging changes in the dead time of the target system. Estimated to be maintenance-free. Further, the learning result (parameter) of the learned target system can be taken out by the parameter recording means 120, and the data of the train can be provided to another train.

As described above, the rules can be easily changed by incorporating a fuzzy inference function.

Here, in the above description, the target system estimating means 40 has shown the form of table lookup for extracting a predetermined value from the table as an example of setting the parameter according to the load weight, but the parameter load is not limited to this. It can also be in the form of a function of weight. In this case, the target system can be estimated in more detail.

Further, when the table is looked up, the load weight level can be regarded as a fuzzy variable and the parameter value can be determined by fuzzy inference. In this case as well, the target system can be estimated in more detail.

[0070]

As described above, according to the present invention, at the stage of deceleration control before shifting to the position control for stop position adjustment, the deceleration control having a good quick response and good followability is performed by a simple design method. As a result, it is possible to perform automatic train operation with good ride comfort and high stop position accuracy.

According to the first aspect of the invention, the dead time of the target system can be estimated according to the situation and reflected in the control system design, so that the deceleration control performance can be improved and a simple method can be used. The fixed position stop accuracy can be improved.

According to the second aspect of the present invention, it is possible to perform deceleration suitable for the route situation, the passenger class and the intention of the railway company, and it is possible to perform the fixed position stop control that is comfortable to ride.

According to the invention of claim 3, since the dead time is changed between when the train is braking and when the train is not braking, the dead time can be estimated more accurately and the deceleration control can be performed. The performance can be improved and the fixed position stop accuracy can be improved.

According to the fourth aspect of the present invention, the dead time is changed at the start of braking and at the time of shifting to power running or power running, so that the dead time can be more accurately estimated.
The deceleration control performance can be improved, and the fixed position stop accuracy can be improved.

According to the fifth aspect of the invention, since the target system is modeled in which the braking effectiveness is reflected according to the weather mode, the deceleration control performance is improved and the fixed position stop accuracy can be improved.

According to the invention of claim 6, the target system is automatically learned at any time by the target system estimation learning means for keeping track of seasonal fluctuations and secular changes of the target system, so that no adjustment work is required. Fixed position stop control can be easily realized.

According to the invention of claim 7, the dead time estimation learning means for keeping track of the seasonal variation and the secular change of the dead time of the target system automatically learns the dead time at any time. Fixed position stop control can be realized without performing the operation.

According to the invention of claim 8, since there is a parameter recording means for taking out the learned parameters of the target system, it is possible to take out the data of the present rolling-stock set and use it for another rolling-stock set. It is possible to shorten the adjustment engine when installing the device.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a block configuration diagram of deceleration control means of the present invention.

FIG. 3 is a block diagram showing a second embodiment of the present invention.

FIG. 4 is a block diagram showing a third embodiment of the present invention.

FIG. 5 is a flowchart showing operations of a target system estimating means and a dead time estimating means in the third exemplary embodiment of the present invention.

FIG. 6 is an explanatory diagram of a table for selecting parameters of a target system.

FIG. 7 is a block diagram showing a fourth embodiment of the present invention.

FIG. 8 is a block diagram showing a fifth embodiment of the present invention.

FIG. 9 is an explanatory diagram of a case where the train is controlled by the fixed position stop control function of the automatic train operation device.

FIG. 10 is an explanatory diagram of a train operation curve.

[Explanation of symbols]

 15 Train Brake Device 40 Target System Estimating Means 50 Dead Time Estimating Means 60 Deceleration Controlling Means 61 Main Controller 70 Control Parameter Computing Means 80 Deceleration Pattern Creating Means 90 Weather Mode Input Means 100 Target System Learning Means 110 Dead Time Estimating Learning Means 120 Parameter Recording Means 200 Trains 210 Platforms 220 Rails

Claims (8)

[Claims] 1. In an automatic train operation device for automatically running a train and automatically stopping it at a fixed position, a fixed position stop control system for stopping the train at a fixed position is set as a target system and has the target system. The dead time estimating means for estimating the dead time, the target system estimating means for modeling the target system in consideration of the dead time estimated by the dead time estimating means, and the target system estimating means for estimating the dead time A train brake device that calculates a control parameter calculating unit that calculates a control parameter based on a model and a control parameter that is calculated by the control parameter calculating unit to calculate an operation amount for decelerating the train in consideration of the dead time. An automatic train operation device, comprising: a deceleration control means for outputting to the. 2. The deceleration pattern creating means for creating a target deceleration pattern, which is a target deceleration value when performing deceleration control of the train, and outputting the deceleration pattern to the deceleration control means. Automatic train driving device. 3. The dead time estimation means separately estimates the dead time when the train is braking and when the train is not braking. The automatic train operation device according to claim 1 or 2. 4. The dead time estimating means estimates the dead time separately when the train starts to brake and when the train shifts to dead running or power running. Item 3. The automatic train operation device according to item 3. 5. A weather mode input means for inputting a weather mode is provided, the deceleration pattern creating means creates a target deceleration pattern based on the weather mode, and the target deceleration pattern matching the weather mode is decelerated. The automatic train operation device according to claim 2, wherein the automatic train operation device outputs the output to a control means. 6. A target system estimation learning means for keeping track of seasonal changes and secular changes of the target system,
The automatic train operation device according to claim 1, wherein the learning result is reflected in the estimation calculation of the target system estimation means. 7. A dead time estimation learning means is provided for keeping track of seasonal changes and aging changes in the dead time of the target system, and the learning result is reflected in the estimation calculation of the dead time estimation means. The automatic train operation device according to claim 6, which is characterized in that. 8. The parameter recording means for extracting and storing the learning result of the target system estimation learning means or the learning result of the dead time estimation learning means, according to claim 6 or claim 7. Automatic train driving device.