Onboard equipment

Abstract

Improving the accuracy of detecting train speed on board the vehicle. [Solution] The onboard device 1A mounted on the train calculates the TG speed, which is a first candidate for the train's running speed, based on the output signal of the speed generator 10; the GNSS speed, which is a second candidate for the train's running speed, based on the positioning signal from a given positioning satellite; and the IMU speed, which is a third candidate for the train's running speed, based on the measurement value of the inertial sensor 14. The device determines whether the train is in a running state after a low-speed state, selects one of the first, second, and third candidate running speeds based on the determination of the running state, and determines the train's running speed based on the selected candidate running speed.

Description

Technical Field

[0001] The present invention relates to an on-vehicle device.

Background Art

[0002] As a method for detecting the running position of a train on a vehicle, a method of measuring the running speed of the train and calculating the running distance based on the running speed is well known. As a method for detecting the running speed, a method of calculating by counting the number of rotations of a wheel based on a pulse signal output from a tachogenerator attached to an axle, a method of calculating based on the Doppler frequency shift of a positioning signal from a positioning satellite in a GNSS (Global Navigation Satellite System) such as GPS (Global Positioning System), a method based on measurement values of an inertial sensor (IMU module) having an acceleration sensor and an angular velocity sensor (gyro sensor), etc. are known.

[0003] Since these methods may cause detection errors when used alone, various devices have been made to improve the detection accuracy. For example, since a train runs on a predetermined route called a track, a map-based method has been proposed to improve the detection accuracy of the running position by using a three-dimensional track map that defines the three-dimensional linearity of the track (see, for example, Patent Document 1).

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0005] However, the aforementioned map-based method requires a database such as a three-dimensional track map on board the train. Therefore, if track modifications or other changes occur, it becomes necessary to update the database (three-dimensional track map) for all trains, which places a significant burden on the system.

[0006] The problem that this invention aims to solve is to provide a new technology that improves the accuracy of detecting the train's speed while on board a vehicle. [Means for solving the problem]

[0007] The first invention for solving the above problem is: Onboard equipment installed on a train, A first running speed candidate calculation means (for example, the TG speed calculation unit 102 in Figure 1) calculates a first running speed candidate for the train based on the output signal of the speed generator, A second running speed candidate calculation means (for example, the GNSS speed calculation unit 104 in Figure 1) calculates a second running speed candidate for the train based on positioning signals from a given positioning satellite, A third running speed candidate calculation means (for example, the IMU speed calculation unit 106 in Figure 1) calculates a third running speed candidate for the train based on the measured values ​​of the inertial sensor, The aforementioned train's running state includes a running state determination means (for example, the running state determination unit 112A in Figure 1) that determines whether or not the train is in a running state after a low-speed state, A speed determination means (for example, speed determination unit 114A in Figure 1, speed determination unit 114B in Figure 7, speed determination unit 114C in Figure 10) selects one of the first running speed candidate, the second running speed candidate, and the third running speed candidate based on the determination result of the running state determination means, and determines the running speed of the train based on the selected running speed candidate, This is an on-board device equipped with [a specific feature / feature].

[0008] According to the first invention, the accuracy of detecting the train's running speed on board the vehicle can be improved. Specifically, on board the vehicle, a first running speed candidate based on the output signal of the speed generator, a second running speed candidate based on the positioning signal from the positioning satellite, and a third running speed candidate based on the measurement value of the inertial sensor are calculated as running speed candidates. The train's running speed is then determined based on the running speed candidate selected based on the determination result of whether or not the train is running. Since the accuracy of these running speed candidates may differ depending on the running conditions such as the running position and running time, the accuracy of detecting the running speed can be improved by selecting a running speed candidate with high accuracy.

[0009] The second invention is, in the above invention, The driving state determination means enables a state transition to the driving state only after passing through the low-speed state from the stationary state, and determines whether the driving state is the stationary state, the low-speed state, or the driving state. It is an on-board device.

[0010] According to the second invention, the state immediately after transitioning from a stationary state or immediately before transitioning from a moving state to a stationary state after decelerating can be determined to be a low-speed state. By defining the low-speed state as a different running state from the running state, even though the train is in motion, it is possible to improve the accuracy of detecting the train's speed.

[0011] The third invention is, in the above invention, The driving state determination means determines the driving state based on the average equivalent value of acceleration based on the measured value of the inertial sensor. It is an on-board device.

[0012] According to the third invention, if the running state is determined based on only one measurement value from an inertial sensor, there is a possibility of making an incorrect judgment due to errors in the measurement value. Therefore, by using the average equivalent value of acceleration based on multiple measurement values, the accuracy of adopting the measurement results from the inertial sensor can be improved, thereby improving the accuracy of determining the running state of the train.

[0013] The fourth invention is, in the above invention, The driving state determination means determines the driving state based on the history of the average equivalent value over the immediate vicinity period. It is an on-board device.

[0014] According to the fourth invention, the accuracy of determining the train's running state can be improved by using the history of the average equivalent value of acceleration based on the measurements of the inertial sensor over the immediate vicinity period.

[0015] The fifth invention is, in the above invention, A second reliability determination means (for example, the GNSS reliability determination unit 124A in Figure 1) for determining the reliability of the second candidate travel speed, Furthermore, The speed determination means includes a running speed determination means that, when the running state determination means determines that the train is running, selects either the first running speed candidate or the second running speed candidate based on the determination result of the second reliability determination means, and determines the running speed of the train based on the selected running speed candidate. It is an on-board device.

[0016] While the train is in motion, the error in the inertial sensor's measurement can accumulate and increase as the travel time progresses, thus reducing the accuracy of the third candidate speed. For this reason, as in the fifth invention, when the train is in motion, it is possible to improve the accuracy of detecting the train's speed by, for example, selecting the second candidate speed if its reliability is good, and selecting the first candidate speed if its reliability is not good.

[0017] The sixth invention is, in the above invention, A second reliability determination means (for example, the GNSS reliability determination unit 124A in Figure 7) for determining the reliability of the second candidate travel speed, A third reliability determination means (for example, the IMU reliability determination unit 126B in Figure 7) for determining the reliability of the third candidate travel speed, Furthermore, When the running state determination means determines that the vehicle is in the running state, if the determination result of the second reliability determination means is good, the second running speed candidate is selected; if not, based on the determination result of the third reliability determination means, either the first running speed candidate or the third running speed candidate is selected, and the running speed of the train is determined based on the selected running speed candidate. The vehicle-mounted device has a running determination means during running. It is an on-vehicle device.

[0018] During the running state, as time passes during running, the error of the measured value of the inertial sensor can accumulate and increase, so the accuracy of the third running speed candidate decreases. Therefore, as in the sixth invention, when the running state is the running state, if the reliability of the second running speed candidate is good, the second running speed candidate is selected; if not (not good), based on the reliability of the third running speed candidate, either the first running speed candidate or the third running speed candidate is selected. By doing so, it is possible to improve the detection accuracy of the running speed of the train.

[0019] The seventh invention is in the above-mentioned invention. When the determination result of the second reliability determination means is no, the running determination means during running selects either the first running speed candidate or the third running speed candidate according to the duration for which the determination result is no. It is an on-vehicle device.

[0020] When the reliability of the second running speed candidate based on the positioning signal is good, it is possible to correct the measured value of the inertial sensor based on the speed and acceleration calculated based on the positioning signal. By correcting the measured value of the angular velocity sensor, the accuracy of the third running speed candidate based on the measured value of the inertial sensor can be improved. Therefore, as in the seventh invention, for example, when the duration for which the reliability of the second running speed candidate is no (not good) is short, the measured value of the inertial sensor is corrected based on the positioning signal and then the third running speed candidate based on the measured value of the inertial sensor is selected; when the duration is long, the first running speed candidate is selected. By doing so, it is possible to further improve the detection accuracy of the running speed of the train.

[0021] The eighth invention is, in the above invention, A second reliability determination means (for example, the GNSS reliability determination unit 124C in Figure 10) for determining the reliability of the second candidate travel speed, Furthermore, The speed determination means includes a running speed determination means that, when the running state determination means determines that the train is running, selects either the second running speed candidate or the third running speed candidate based on the determination result of the second reliability determination means, and determines the running speed of the train based on the selected running speed candidate. It is an on-board device.

[0022] While the train is in motion, the error in the inertial sensor's measurement can accumulate and increase as the travel time progresses, thus reducing the accuracy of the third candidate travel speed. For this reason, as in the eighth invention, when the train is in motion, it is possible to improve the accuracy of detecting the train's travel speed by, for example, selecting the second candidate travel speed if the reliability of the second travel speed is good, and selecting the third candidate travel speed if it is not (not good).

[0023] The ninth invention is, in the above invention, The second reliability determination means determines the reliability of the second travel speed candidate based on the variance equivalent value of the second travel speed candidate. It is an on-board device.

[0024] According to the ninth invention, it is possible to determine the reliability of a second candidate for a second driving speed based on its equivalent variance value. Since the equivalent variance value represents the variability, if the equivalent variance value of the second candidate for a second driving speed is large, it is considered that the reliability of the second candidate for a second driving speed is poor (unreliable) due to errors in receiving positioning signals, etc.

[0025] The tenth invention is, in the above invention, The second reliability determination means determines the reliability of the second candidate travel speed based on the history of the equivalent variance value over the immediate vicinity period. It is an on-board device.

[0026] According to the tenth invention, by using the history of the variance equivalent value of the second travel speed candidate over the immediate vicinity period, it becomes possible to improve the accuracy of judging the reliability of the second travel speed candidate.

[0027] The eleventh invention is, in the above invention, The inertial sensor comprises an acceleration sensor and an angular velocity sensor. The third travel speed candidate calculation means calculates the third travel speed candidate by removing gravity error based on the measured value of the acceleration sensor and the corrected measured value obtained by correcting the measured value of the angular velocity sensor based on a given correction value, When the driving state determination means determines that the vehicle is in motion, a Kalman filter is used to estimate the measured value of the angular velocity sensor based on the speed and acceleration calculated based on the positioning signal, and an angle correction value calculation means (for example, the angle correction value calculation unit 116 in Figure 7) calculates the correction value by estimating the measured value of the angular velocity sensor using a Kalman filter with filter coefficients corresponding to the determination result of the second reliability determination means. This is an on-board device that further includes the following features.

[0028] According to the eleventh invention, the accuracy of the third candidate running speed based on the inertial sensor's measurement can be improved by correcting the measurement value of the angular velocity sensor. As a result, it becomes possible to improve the accuracy of detecting the train's running speed. Furthermore, by using a Kalman gain corresponding to the reliability of the second candidate running speed, it becomes possible to improve the accuracy of the estimation calculation of the angular velocity sensor's measurement value. [Brief explanation of the drawing]

[0029] [Figure 1] An example of the configuration of the on-board device of the first embodiment. [Figure 2] Diagram illustrating the driving conditions. [Figure 3] Flowchart for determining the driving status. [Figure 4] An explanatory diagram of the variance equivalent value of the TG velocity. [Figure 5]An explanatory diagram of the equivalent value of the variance of GNSS speed. [Figure 6] A flowchart for determining the driving speed. [Figure 7] An example configuration of the on-board device in the second embodiment. [Figure 8] Flowchart for IMU reliability assessment process. [Figure 9] A flowchart for determining the driving speed. [Figure 10] An example configuration of the on-board device of the third embodiment. [Figure 11] An explanatory diagram of the acceleration mean series. [Figure 12] Flowchart for determining the driving status. [Figure 13] A flowchart for determining the driving speed. [Modes for carrying out the invention]

[0030] Preferred embodiments of the present invention will be described below with reference to the drawings. However, the applicable forms of the present invention are not limited to the following embodiments. Furthermore, in the drawings, the same elements are denoted by the same reference numerals.

[0031] This embodiment describes an on-board device that is mounted on a train and determines the train's speed based on its running conditions. Three embodiments of such an on-board device will be described in order below.

[0032] [First Embodiment] Figure 1 is a block diagram showing the functional configuration of the on-board device 1A of the first embodiment. As shown in Figure 1, the on-board device 1A has the following functional units: a TG (Tacho Generator) speed calculation unit 102, a GNSS (Global Navigation Satellite System) speed calculation unit 104, an IMU (Inertial Measurement Unit) speed calculation unit 106, a driving state determination unit 112A, a speed determination unit 114A, a TG reliability determination unit 122, a GNSS reliability determination unit 124A, and an IMU reliability determination unit 126A. Although not shown, the on-board device 1A also has a display unit, an operation unit, a storage unit, etc.

[0033] The TG speed calculation unit 102 calculates the TG speed, which is a candidate for the first running speed of the train, based on the output signal of the speed generator (hereinafter referred to as "TG" as appropriate) 10. The speed generator 10 is attached to the axle of the train and outputs a signal corresponding to the rotation speed of the axle. The TG speed calculation unit 102 counts the rotation speed of the axle based on the signal output from the speed generator 10 and calculates the running speed of the train.

[0034] The GNSS speed calculation unit 104 calculates the GNSS speed, which is a candidate for the train's second running speed, based on positioning signals from a given positioning satellite. The train is equipped with a GNSS receiver 12 that receives positioning signals from GNSS satellites, which are positioning satellites of a satellite navigation system such as GPS (Global Positioning System). The GNSS receiver 12 outputs positioning information (GNSS sentences) based on the received positioning signals. The GNSS speed calculation unit 104 calculates the train's running speed based on the positioning information output from the GNSS receiver 12.

[0035] The GNSS speed calculation unit 104 may calculate the travel speed from the change in position based on positioning information, or it may calculate the travel speed using the Doppler frequency shift. Furthermore, in order to improve the accuracy of GNSS, the following methods may be used. The first method is to convert the GNSS positioning information into kilometers and calculate the travel speed from the change in those kilometers. The second method is to integrate the speed information calculated using the GNSS positioning information and the speed information calculated from the Doppler frequency shift using a Kalman filter. The third method is to use not only the Doppler frequency shift of the carrier wave of the positioning signal from the GNSS satellite, but also the phase of the carrier wave. That is, by using the phase of the carrier wave to accurately determine the distance from the receiving antenna of the GNSS receiver 12 to the received positioning satellite (receiving satellite), high-precision and high-frequency positioning information can be obtained. The fourth method uses positioning information acquired using a method that employs the phases of two consecutive carrier waves (TDCP: Time-Differenced Carrier Phases) rather than Doppler frequency shift.

[0036] The IMU speed calculation unit 106 calculates the IMU speed, which is a candidate for the third running speed of the train, based on the measured values ​​of the inertial sensor (IMU module) (hereinafter referred to as "IMU" as appropriate) 14. The inertial sensor 14 has an acceleration sensor 16 and an angular velocity sensor (gyro sensor) 18. The IMU speed calculation unit 106 calculates the IMU speed, which is a candidate for the third running speed, by removing gravity errors based on the measured values ​​of the acceleration sensor 16 and the angular velocity sensor 18.

[0037] The acceleration sensor 16 and angular velocity sensor 18 are 3-axis sensors, mounted on the train such that, when the train's gradient angle is 0 (zero), the positive X-axis coincides with the direction of the train's travel, and the positive Y-axis coincides with the direction of the vertical. Therefore, if the train's gradient angle is 0 (zero), the direction of gravity coincides with the positive Y-axis, so the X-axis measurement of the acceleration sensor 16 does not include the component of gravitational acceleration G, and this X-axis measurement is the train's acceleration itself. However, if the train's gradient angle is not 0 (zero), the X-axis measurement of the acceleration sensor 16 includes the X-axis component of gravitational acceleration G in addition to the train's acceleration. For this reason, the train's gradient angle is determined from the measurement of the angular velocity sensor 18, and the X-axis component of gravitational acceleration G obtained from this gradient angle is subtracted from the X-axis measurement of the acceleration sensor 16 to remove the gravity error and calculate the train's acceleration.

[0038] Furthermore, if the train's running state, as determined by the running state determination unit 112A (described later), is "stopped," the bias (offset) of the angular velocity sensor 18 is removed to correct the measured value. "Stopped" means the train is at rest. In this state, the measured value of each axis of the angular velocity sensor 18 is used as the bias value, and this bias value is subtracted as a correction value from the measured value of the angular velocity sensor 18 to correct the measured value.

[0039] The running state determination unit 112A determines the running state of its own train. The running state of the train is defined as shown in Figure 2. Figure 2 is a diagram showing an example of the transition of running states. In Figure 2, the horizontal axis is time and the vertical axis is running speed, showing an example of a speed curve when a train travels between stations.

[0040] As shown in Figure 2, there are three states of operation: "stopped," "low-speed," and "running." When a train travels between stations, it transitions through these states in the following order: stopped, low-speed, running, low-speed, and stopped. The stopped state is when the train is stopped and its speed is 0 (zero), such as when it is stopped at a station. The low-speed state corresponds to a state where the train is running at a low speed, such as immediately after departing a station or before arriving at a station. There is also an upper limit to the duration of the low-speed state. In other words, the low-speed state is a state in which the train's speed is within a predetermined low-speed range and the duration is below a predetermined upper limit. The low-speed range can be set as appropriate, but for example, it can be 10 km / h or less. The running state is a state in which the train is running at a certain speed and is a state that can only be reached by passing through the low-speed state from the stopped state. When the duration of the low-speed state reaches a predetermined upper limit, the train transitions from the low-speed state to the running state. If the vehicle speed based on the TG speed becomes 0 (zero) while in a low-speed or moving state, or if the acceleration based on the measurement value of the inertial sensor 14 can be considered to have become 0 (zero), the vehicle transitions to a stopped state.

[0041] The running state determination unit 112A determines the running state of the train based on the TG speed. Figure 3 is a flowchart showing the flow of the process performed by the running state determination unit 112A to determine the running state. This process is repeated at a predetermined determination cycle Δt.

[0042] First, assume the train is stopped at a station. The running state determination unit 112A initially determines the running state to be "stopped" (step S1). Next, if the TG speed is 0 (no) (step S3), set the running time RT to 0 (zero) (step S19), return to step S1, and continue to determine that the running state is "stopped". The running time RT is a value that represents the duration of the state in which the TG speed is not 0 (zero), that is, the elapsed time since the start of running.

[0043] If the TG speed is not 0 (step S3: YES), the train has started moving, and the running state is determined to be "low speed state" (step S5). Subsequently, if the TG speed is 0 (step S7: NO), the train has stopped, and after setting the running time RT to 0 (step S19), the process returns to step S1. If the TG speed is not 0 (step S7: YES), the running time RT is updated to a value obtained by adding the judgment period Δt (step S9).

[0044] Next, if the driving time RT is below a predetermined threshold (step S11: NO), the system returns to step S5 and continues to determine that the driving state is a "low-speed state". On the other hand, if the driving time RT exceeds a predetermined threshold (step S11: YES), the system determines that the driving state is a "driving state" (step S13). This threshold is the upper limit of the duration of the "low-speed state", and is approximately several tens of seconds.

[0045] Next, if the TG speed exceeds the threshold (step S15: YES), the travel time RT is updated to a value obtained by adding the judgment period Δt (step S16), and then the process returns to step S13 to continue determining that the travel state is "traveling". On the other hand, if the TG speed is below the threshold (step S15: NO), the travel state is determined to be "low speed" (step S17). This threshold is the threshold at which the travel speed is considered low.

[0046] Next, if the TG speed is not 0 (step S18: YES), return to step S17 and continue to determine that the running state is "low speed state". On the other hand, if the TG speed is 0 (step S18: NO), the train has stopped, and after setting the running time RT to 0 (step S19), return to step S1.

[0047] Returning to Figure 1, the TG reliability determination unit 122 determines the reliability of the TG speed based on the history of the variance equivalent value of the TG speed, which is the first candidate for running speed, over the immediate vicinity period. Specifically, the level of the variance equivalent value corresponds to whether or not there is a large change in the TG speed that suggests the occurrence of slippage or skidding. If the variance equivalent value is low, the reliability of the TG speed is judged as "good," and if the variance equivalent value is high, the reliability of the TG speed is judged as "bad (not good)."

[0048] Figure 4 illustrates the determination of whether the variance equivalent value of the TG velocity is high or low. In the lower part of Figure 4, as a concrete example of the determination process, the downward direction is the time axis, and each TG velocity calculated by the TG velocity calculation unit 102 is shown as a rectangle along this time axis. The upper part of Figure 4 shows the processing procedure for the TG velocity shown in the lower part, moving from right to left.

[0049] First, the variance value VAR, which is the equivalent value of the variance of a predetermined number of consecutive TG velocities (4 in Figure 4), is calculated (step S143). Next, this variance value VAR is compared with a predetermined threshold (for example, "0.5"). If it exceeds the threshold (step S145: YES), the count value is updated (reset) to 0 (step S147). If it is below the threshold (step S145: NO), the count value is updated to a value that is 1 added to it (step S149). Then, the count value is compared with a predetermined threshold (for example, "15"). If it is less than the threshold (step S151: YES), the variance value of the TG velocity is judged to be "high" (step S153). If it exceeds the threshold (step S151: NO), it is judged to be "low" (step S155).

[0050] Returning to Figure 1, the GNSS reliability determination unit 124A determines the reliability of the GNSS speed based on the history of the variance equivalent value of the GNSS speed, which is the second candidate for the travel speed, over the immediate vicinity period. Specifically, if a predetermined number of the following six conditions (for example, one or more) are met, the reliability of the GNSS speed is determined to be "poor (not good)," and if none of the conditions are met, the reliability of the GNSS speed is determined to be "good."

[0051] The first condition is that the number of received satellites included in the GNSS sentence output as a positioning result from the GNSS receiver 12 is less than or equal to a predetermined number N (e.g., N=4). The second condition is that the number of GNSS sentences with a positioning status of "unavailable" in the most recent predetermined period (e.g., 1 second) is greater than or equal to a predetermined number (e.g., 1). The third condition is that the variance value of the GNSS speed is high. The fourth condition is that the reliability of the TG speed is "good," and the TG speed and IMU speed are approximately the same, but the difference in GNSS speed is large compared to the TG speed and IMU speed. The fifth condition is that the TG speed is above a predetermined speed, and the change in TG speed is small (low variance equivalent value), but the difference in GNSS speed is large compared to the TG speed. The sixth condition is that the reliability of the IMU speed is "good," and the difference in GNSS speed is large compared to the IMU speed.

[0052] Whether the third condition is met is determined by looking at the history of the variance equivalent value of the GNSS speed, which is the second candidate for the driving speed, over the immediate vicinity period, and whether a state in which the variance equivalent value is large and the variation in GNSS speed is considered to be large has continued. This is because if the variation in GNSS speed is large, there is a possibility that an error has occurred in the reception of the positioning signal.

[0053] Figure 5 illustrates the determination of a high variance in GNSS velocity. The lower part of Figure 5 shows a concrete example of the determination process, with the downward direction representing the time axis, and each GNSS velocity calculated by the GNSS velocity calculation unit 104 arranged as a rectangle along this time axis. The upper part of Figure 5 shows the processing procedure for the GNSS velocity shown in the lower part, moving from right to left.

[0054] First, the variance value VAR, which is the equivalent value of the variance of a predetermined number of consecutive GNSS speeds ("4" in Figure 5), is calculated (step S103). Next, this variance value VAR is compared with a predetermined threshold (for example, "0.5"). If it exceeds the threshold (step S105: YES), the count value is updated (reset) to 0 (step S107). If it is below the threshold (step S105: NO), the count value is updated to a value that is 1 added to it (step S109). Then, the count value is compared with a predetermined threshold (for example, "15"). If it is less than the threshold (step S111: YES), it is determined that the variance value of the GNSS speed is "high" (step S113). If it exceeds the threshold (step S111: NO), it is determined that it is "low (not high)" (step S115).

[0055] Returning to Figure 1, the IMU reliability determination unit 126A determines whether the IMU speed is reliable or not. Specifically, since the IMU speed includes accumulated errors in measurement values ​​due to continuous driving, if the driving state determination unit 112A determines that the driving state is a "low-speed state", the reliability of the IMU speed is determined to be "good", and if it is anything else ("driving state" or "stopped state"), the reliability of the IMU speed is determined to be "bad (not good)".

[0056] The speed determination unit 114A selects one of the first running speed candidate (TG speed), the second running speed candidate (GNSS speed), and the third running speed candidate (IMU speed) based on the judgment result of the running state determination unit 112A, and determines the running speed of the train based on the selected running speed candidate.

[0057] The accuracy of the speed information from the three sensors (TG, IMU, GNSS) is, in principle, in the order of highest accuracy: GNSS speed > IMU speed > TG speed. Therefore, the speed determination unit 114A basically uses the GNSS speed, but constantly checks the reliability of the GNSS speed and switches to the IMU speed when the reliability of the GNSS speed decreases. After that, it checks the reliability of the IMU speed and switches to the TG speed when the reliability of the IMU speed decreases.

[0058] Figure 6 is a flowchart showing the process flow for determining the travel speed performed by the speed determination unit 114A. This process is repeated at predetermined intervals.

[0059] First, assume the train is stopped at a station. If the running state is "stopped" (step S21: YES), the speed determination unit 114A determines the IMU speed as the running speed (step S23). If the running state is "low speed" (steps S21: NO to S25: YES), the running speed is determined according to the reliability of the GNSS speed. That is, if the reliability of the GNSS speed is "good" (step S27: YES), the GNSS speed is determined as the running speed (step S29), and if the reliability of the GNSS speed is "bad" (step S27: NO), the IMU speed is determined as the running speed (step S31).

[0060] Furthermore, if the driving state is "driving" (steps S25: NO to S33: YES), the driving speed is determined according to the reliability of the GNSS speed. That is, if the reliability of the GNSS speed is "good" (step S35: YES), the GNSS speed is determined as the driving speed (step S37), and if the reliability of the GNSS speed is "bad" (step S35: NO), the TG speed is determined as the driving speed (step S39). After performing the above processing, the process returns to step S21 and the same processing is repeated.

[0061] [Second Example] Next, a second embodiment will be described. The main difference between the second embodiment and the first embodiment is that the measured value of the angular velocity sensor 18 is corrected based on the positioning signal received by the GNSS receiver 12. In the second embodiment, the explanation of components that are the same as those in the first embodiment described above will be simplified or omitted.

[0062] Figure 7 is a block diagram showing the functional configuration of the on-board device 1B in the second embodiment. As shown in Figure 7, the on-board device 1B has the following functional units: a TG speed calculation unit 102, a GNSS speed calculation unit 104, an IMU speed calculation unit 106, a driving state determination unit 112A, a speed determination unit 114B, an angle correction value calculation unit 116, a TG reliability determination unit 122, a GNSS reliability determination unit 124A, and an IMU reliability determination unit 126B. Although not shown, the on-board device 1B also has a display unit, an operation unit, a storage unit, etc.

[0063] The IMU reliability determination unit 126B determines whether the IMU speed is reliable or not. Specifically, since the IMU speed includes accumulated errors in the measured values ​​associated with the train's movement, it calculates the speed difference between the IMU speed and the TG speed. If this speed difference exceeds a threshold and the train's running speed is low, it determines that the IMU speed is unreliable.

[0064] Figure 8 is a flowchart illustrating the process by which the IMU reliability determination unit 126B determines the reliability of the IMU speed. This process is repeated at predetermined intervals. First, if the train's running speed is in the low-speed range, that is, if the TG speed is below a predetermined speed (Step 41: YES), the reliability of the IMU speed is determined to be "good" (Step S49). On the other hand, if the train's running speed is not in the low-speed range, that is, if the TG speed is above a predetermined speed (Step 41: NO), and the variance equivalent value of the TG speed is high (Step S43: NO), the reliability of the IMU speed is determined to be "good" (Step S49). On the other hand, if the variance equivalent value of the TG speed is low (Step S43: YES), and the speed difference between the IMU speed and the TG speed does not exceed a predetermined threshold (Step S45: NO), the reliability of the IMU speed is determined to be "good" (Step S49). On the other hand, if the speed difference between the IMU speed and the TG speed exceeds a predetermined threshold (step S47: YES), the reliability of the IMU speed is judged as "not" (step S47).

[0065] Returning to Figure 7, the angle correction value calculation unit 116 is a Kalman filter that estimates the measured value of the angular velocity sensor 18 based on the speed and acceleration calculated based on the positioning signal when the driving state determination unit 112A determines that the vehicle is in a driving state. It calculates a correction value for the measured value of the angular velocity sensor 18 by estimating the measured value of the angular velocity sensor 18 using a Kalman filter with filter coefficients corresponding to the determination result of the GNSS reliability determination unit 124A.

[0066] Specifically, by solving equation (1) using an estimation calculation with a Kalman filter, the roll angle (roll) and pitch angle (pitch), which are corrected values ​​obtained by correcting the measured values ​​of the angular velocity sensor, are calculated.

number

[0067] In this process, the covariance matrix R of the observation error, which indicates the uncertainty of the observed values, is selected according to the reliability of the GNSS speed determined by the GNSS reliability determination unit 124A, and an estimation calculation is performed using the Kalman gain K obtained using the selected covariance matrix R.

[0068] First, assuming the reliability of the GNSS speed is "good," and considering the train's speed GNSS_VEL and acceleration GNSS_ACCL, which are obtained from the positioning signal, as positive values, equations (2) and (3) hold true.

number

[0069] Furthermore, the IMU speed IMU_VELPRE at the previous time is expressed by the following equation (4).

number

[0070] Furthermore, the train's acceleration accl, corrected by removing the gravity error Gx from the X-axis measurement value IMU_ACCLx of the angular velocity sensor 18, is expressed by the following equation (6).

number

[0071] When equations (2) to (4) and (6) are expressed in matrix form, we get equation (1) above.

[0072] Furthermore, if the reliability of the GNSS speed is "not good," then assuming that the roll angle IMU_roll and pitch angle IMU_pitch, which are measured values ​​from the inertial sensor 14 (angular velocity sensor 18), are positive values, then the following equations (7) to (10) hold true.

number

[0073] In order to also represent the state where the reliability of the GNSS speed is "poor," equations (9) and (10) are added to equation (1), which represents the state where the reliability of the GNSS speed is "good," resulting in the following equation (12).

number

[0074] Equation (12) is the observation equation, and the state variable x, the observed value z, and the coefficient matrix H are given by equations (15) to (17), respectively.

number

[0075] In the estimation calculation, the initial values ​​are set to zero for the state variable x and the identity matrix P of the state error shown in equation (18).

number

[0076] The state equations representing the state transitions of the state variable x are given by equations (19) to (21), following the equations of motion.

number

[0077] The error y between the observed value z at time T and the prior predicted value x^ of the state variable x is given by equations (22) and (23).

number

[0078] The prior observation prediction z^, which is the prior prediction of the observed value z at time T, is obtained from the prior state prediction x^ at time T according to the observation equation shown in equation (12). The prior state prediction x^ at time T is obtained from the state variable x at the previous time T-1 according to the state equations shown in equations (19) to (21).

[0079] Then, the covariance matrix R of the observation errors is defined as shown in equation (24).

number

[0080] Then, the Kalman gain K is calculated by setting the value of the covariance matrix R to a value corresponding to the reliability of the GNSS speed, and the roll angle and pitch angle are calculated by obtaining the posterior predicted value of the state variable x according to equation (25).

number

[0081] The Kalman gain K is defined as shown in equation (26).

number

[0082] The prior predicted value P^ of the covariance matrix P of the state error is defined as shown in equation (27), and the posterior predicted value P is defined as shown in equation (28).

number

[0083] Equation (25) can be transformed from equation (23) as shown in equations (29) and (30). By substituting the value of the Kalman gain K into equations (29) and (30), the state variables A and B can be determined, and from equation (13), the roll angle and pitch angle can be calculated.

number

[0084] For example, if the reliability of the GNSS speed is "good," and the covariance matrix R is an example value shown in equation (31), then the value of the Kalman gain K obtained from equation (26) is given by equation (32).

number

[0085] Since state variable A is the element in the third row of the vector of state variable x, the calculation of state variable A uses the product of each element in the third row of the Kalman gain K and each element y1 to y6 of the error y. However, since the values ​​of the elements in the fifth and sixth columns of the third row of the Kalman gain K shown in equation (32) are almost zero, the calculation will be performed using only the elements y1 to y4 of the error y. Similarly, since state variable B is the element in the fifth row of the vector of state variable x, the calculation of state variable B uses the product of each element in the fifth row of the Kalman gain K and each element y1 to y6 of the error y. However, since the values ​​of the elements in the fifth and sixth columns of the fifth row of the Kalman gain K shown in equation (32) are almost zero, the calculation will be performed using only the elements y1 to y4 of the error y.

[0086] Furthermore, if the reliability of the GNSS speed is "poor," and assuming that the value of the covariance matrix R is an example value shown in equation (33), the value of the Kalman gain K obtained from equation (26) is given by the following equation (34).

number

[0087] Since state variable A is the element in the third row of the vector of state variable x, the calculation of state variable A uses the product of each element in the third row of the Kalman gain K and each element y1 to y6 of the error y. However, since the values ​​of the elements in the first to fourth columns and the sixth column of the third row of the Kalman gain K shown in equation (34) are almost zero, the calculation will be performed using only the element y5 of the error y. Similarly, since state variable B is the element in the fifth row of the vector of state variable x, the calculation of state variable B uses the product of each element in the fifth row of the Kalman gain K and each element y1 to y6 of the error y. However, since the values ​​of the elements in the first to fifth columns of the fifth row of the Kalman gain K shown in equation (34) are almost zero, the calculation will be performed using only the element y6 of the error y.

[0088] Then, the IMU speed calculation unit 106 calculates the IMU speed, which is a third candidate for travel speed with gravity errors removed, based on the measured value of the acceleration sensor 16 and the corrected value obtained by correcting the measured value of the angular velocity sensor 18 calculated by the angle correction value calculation unit 116.

[0089] Figure 9 is a flowchart showing the process flow for determining the travel speed performed by the speed determination unit 114B. This process is repeated at predetermined intervals.

[0090] First, assume that the train is stopped at a station. If the running state is "stopped" (step S21: YES), the speed determination unit 114B determines the IMU speed as the running speed (step S23).

[0091] Furthermore, if the driving condition is "low speed" (steps S21: NO to S25: YES), the driving speed is determined according to the reliability of the GNSS speed and IMU speed. That is, if the reliability of the GNSS speed is "good" (step S52: YES), the GNSS reliability failure time is updated (reset) to 0 (zero) (step S53), and the GNSS speed is determined as the driving speed (step S54). The GNSS reliability failure time is a value that represents the duration of the state in which the reliability of the GNSS speed is judged to be "unreliable".

[0092] On the other hand, if the reliability of the GNSS speed is "unreliable" (step S52: NO), the GNSS unreliable time is updated to a value obtained by adding the judgment period Δt (step S55). If the updated GNSS unreliable time is above a predetermined threshold (step S56: NO), the TG speed is determined as the travel speed (step S59). On the other hand, if the GNSS unreliable time is less than the predetermined threshold (step S56: YES), the travel speed is determined according to the reliability of the IMU speed. That is, if the reliability of the IMU speed is "good" (step S57: YES), the IMU speed is determined as the travel speed (step S58), and if the reliability of the IMU speed is "unreliable" (step S57: NO), the TG speed is determined as the travel speed (step S59).

[0093] Furthermore, if the driving state is "driving state" (steps S25: NO to S33: YES), the angle correction value calculation unit 116 calculates a corrected value (corrected measurement value) by correcting the measurement value of the angular velocity sensor based on the positioning signal received by the GNSS receiver 12 (step S51). Then, as in the case of the driving state being "low speed state", the driving speed is determined according to the reliability of the GNSS speed and IMU speed (steps S52 to S59).

[0094] [Third Embodiment] Next, a third embodiment will be described. The third embodiment differs from the first and second embodiments mainly in that the speed generator 10 is unusable due to a malfunction or the like. In the third embodiment, the explanation of components that are the same as those in the first or second embodiment described above will be simplified or omitted.

[0095] Figure 10 is a block diagram showing the functional configuration of the on-board device 1C in the third embodiment. As shown in Figure 10, the on-board device 1C has the following functional units: a GNSS speed calculation unit 104, an IMU speed calculation unit 106, a driving state determination unit 112C, a speed determination unit 114C, an angle correction value calculation unit 116, and a GNSS reliability determination unit 124C. Although not shown, the on-board device 1C also has a display unit, an operation unit, a storage unit, etc.

[0096] The running state determination unit 112C determines the running state based on the history of the average equivalent value of acceleration based on the measured values ​​of the inertial sensor 14 over the immediate vicinity. Specifically, the running state of the train is determined by determining the acceleration / deceleration state of the train based on the acceleration average series based on the measured values ​​of the inertial sensor 14, as an example of the history of the average equivalent value shown in Figure 11.

[0097] Figure 11 illustrates the acceleration average series based on the measurements of the inertial sensor 14. The lower part of Figure 11 shows a concrete example of the acceleration average series calculation process, with the downward direction representing the time axis, and the X and Z axis measurements (Ax, Az) of the acceleration sensor 16 arranged as rectangles along this time axis. The upper part of Figure 11 shows the processing steps for the measurements (Ax, Az) shown in the lower part, moving from right to left.

[0098] First, the average value AVR of a predetermined number of consecutive (4 in Figure 11) measured values ​​(Ax, Az) is calculated according to the following formula (33) (step S123).

number

[0099] Next, this average AVR is compared with predetermined positive thresholds (e.g., "1.0") and negative thresholds (e.g., "-1.0"), and if it exceeds the positive threshold, "1" is added to the most significant bit of the acceleration average column (steps S125: YES~S127), if it falls below the negative threshold, "-1" is added (steps S129: YES~S131), and if it is less than or equal to the positive threshold and greater than or equal to the negative threshold, "0" is added (steps S129: NO~S133), thereby updating the acceleration average column in FIFO (First In First Out) format (step S135). Here, the positive threshold is a positive value that indicates the train is accelerating, and the negative threshold is a negative value that indicates the train is decelerating.

[0100] The acceleration average sequence is a bit sequence of a predetermined number of bits (8 in Figure 11) and is updated each time the acceleration sensor 16 outputs a measured value. A bit value of "1" in the acceleration average sequence indicates that the acceleration is positive (accelerating) and is relatively large, "-1" indicates that the acceleration is negative (deceleration) and is relatively large, and "0" indicates that the acceleration and deceleration are small. Therefore, the acceleration average sequence represents the change in acceleration over a short period of time. The more "1"s there are, the more the train is considered to be accelerating significantly for a prolonged period, and the more "-1"s there are, the more the train is considered to be decelerating significantly for a prolonged period. Initially, all bits of the acceleration average sequence are set to "0". The reason for using such an acceleration average sequence is that the measured values ​​from the inertial sensor 14 are unstable due to the effects of drift, and by basing the acceleration on multiple consecutive measurements rather than a single measurement, it is possible to determine the acceleration with greater accuracy.

[0101] Returning to Figure 10, the GNSS reliability determination unit 124C determines the reliability of the GNSS speed based on the history of the variance equivalent value of the GNSS speed, which is the second candidate for the driving speed, over the immediate vicinity period. Specifically, if a predetermined number (e.g., one or more) of the following four conditions are met, the reliability of the GNSS speed is determined to be "poor (not good)," and if none of the conditions are met, it is determined to be "good." The first condition is that the number of received satellites included in the GNSS sentence output as a positioning result from the GNSS receiver 12 is less than or equal to a predetermined number N (e.g., N=4), the second condition is that the number of GNSS sentences with a positioning status of "unable to position" in the most recent predetermined period (e.g., 1 second) is greater than or equal to a predetermined number (e.g., 1), the third condition is that the variance value of the GNSS speed is high, and the fourth condition is that, when the driving state is a "low speed state," the difference between the GNSS speed and the IMU speed is large.

[0102] Figure 12 is a flowchart illustrating the process by which the driving state determination unit 112C determines the driving state. This process is repeated at a predetermined determination cycle Δt.

[0103] First, it is assumed that the train is stopped at a station. The running state determination unit 112C determines that the running state is "stopped" (step S61). Next, it determines whether the train's acceleration / deceleration state is "accelerating significantly". The acceleration / deceleration state is determined based on the number of "1"s and "-1"s included in the acceleration average column. If the acceleration is positive (accelerating) and the number of "1"s included in the acceleration average column is greater than or equal to a predetermined number that can be considered to be large, it is determined that the train is "accelerating significantly". If the train's acceleration / deceleration state is not "accelerating significantly" (step S63: NO), the running time RT and the time before stopping ST are set to 0 (zero) (step S89), and then the unit returns to step S61 to continue determining that the running state is "stopped". The running time RT is the duration of the state in which the IMU speed is not considered to be 0 (zero), that is, the elapsed time since the start of travel. The time before stopping ST is the duration of the state in which the train can be considered to be decelerating.

[0104] If the train's acceleration / deceleration state is "accelerating significantly" (Step S63: YES), the train has departed the station, and its running state is determined to be "low speed" (Step S65). Subsequently, if the average acceleration series is all "0" (Step S67: YES), the train has stopped, and after setting the running time RT and the time before stopping ST to 0 (zero) (Step S89 ), return to step S61.

[0105] On the other hand, if the average acceleration column is not all "0" (step S67: NO), the travel time RT is then updated to a value obtained by adding the judgment period Δt (step S69). If the travel time RT is below a predetermined threshold (step S71: NO), the system returns to step S65 and continues to judge that the travel state is a "low-speed state". On the other hand, if the travel time RT exceeds a predetermined threshold (step S71: YES), the travel state is judged to be a "traveling state" (step S73). The threshold is the upper limit of the duration of the low-speed state, which is about several tens of seconds.

[0106] Next, it is determined whether the train's acceleration / deceleration state is "significantly decelerating". If the acceleration is negative (decelerating) and the number of "-1"s in the acceleration average sequence is greater than a predetermined number, it is determined that the train is "significantly decelerating". Large If the train is not "decelerating" (Step S75: NO), the running time RT is updated by adding the judgment period Δt (Step S77), and the process returns to Step S73 to continue determining that the running state is "running". If the train's acceleration / deceleration state is "significantly decelerating" (Step S75: YES), the time before stopping ST is then determined. If the time before stopping ST is below the short-time threshold (Step S79: NO), the time before stopping ST is updated by adding the judgment period Δt (Step S81), and the process returns to Step S73 to continue determining that the running state is "running". If the time before stopping ST exceeds the short-time threshold (Step S79: YES), the running state is determined to be "low speed" (Step S83).

[0107] Next, if the time before stopping ST is less than or equal to the long-time threshold, which is slightly longer than the short-time threshold (step S85: NO), the time before stopping ST is updated by adding the judgment period Δt (step S87), and then the process returns to step S83 to continue the judgment that the driving state is a "low-speed state". If the time before stopping ST exceeds the long-time threshold (step S85 YES), after setting the running time RT and the time before stopping ST to 0 (zero) (step S89), return to step S61.

[0108] Figure 13 is a flowchart showing the process flow for determining the travel speed performed by the speed determination unit 114C. This process is repeated at predetermined intervals.

[0109] First, assume the train is stopped at a station. If the running state is "stopped" (step S21: YES), the speed determination unit 114C determines the IMU speed as the running speed (step S23). If the running state is "low speed" (steps S21: NO to S25: YES), the running speed is determined according to the reliability of the GNSS speed. That is, if the reliability of the GNSS speed is "good" (step S91: YES), the GNSS speed is determined as the running speed (step S93), and if the reliability of the GNSS speed is "bad" (step S91: NO), the IMU speed is determined as the running speed (step S95).

[0110] Furthermore, if the driving state is "driving state" (steps S25: NO to S33: YES), the angle correction value calculation unit 116 calculates a corrected value (corrected measurement value) by correcting the measurement value of the angular velocity sensor based on the positioning signal received by the GNSS receiver 12 (step S51). Next, similar to the case where the driving state is "low speed state," the driving speed is determined according to the reliability of the GNSS speed (steps S91 to S95).

[0111] [Effects and Effects] According to this embodiment, the accuracy of detecting the train's running speed on board the vehicle can be improved. Specifically, on board the vehicle, the following running speed candidates are calculated: a first running speed candidate, the TG speed, based on the output signal of the speed generator 10; a second running speed candidate, the GNSS speed, based on positioning signals from positioning satellites; and a third running speed candidate, the IMU speed, based on the measurement value of the inertial sensor 14. The train's running speed is then determined based on the running speed candidate selected based on the determination result of whether or not the train is running. Since the accuracy of these running speed candidates may differ depending on the running conditions such as the running position and running time, the accuracy of detecting the running speed can be improved by selecting a running speed candidate with high accuracy.

[0112] [Differentiation] It should be noted that the applicable embodiments of the present invention are not limited to those described above, and can be modified as appropriate without departing from the spirit of the invention.

[0113] (A) Use of digital maps A digital map (on-board database) may also be provided, and the gravitational acceleration applied to the IMU may be removed based on this digital map. The digital map includes data that associates kilometers with gradient angles.

[0114] Specifically, for example, when position information (kilometer information) is acquired from a ground sensor for position correction, the gradient angle of that location is obtained by referring to a digital map (onboard database), and the amount of gravity acceleration removal is calculated from the acquired gradient angle.

[0115] Alternatively, when the train is traveling at a high speed, when the difference between the IMU speed and the TG speed exceeds a predetermined threshold, the train's current position (kilometer mile) can be determined from the TG speed at that time, and the amount of gravity acceleration removed can be calculated from the gradient angle at that kilometer mile by referring to a digital map (onboard database).

[0116] (B) Kalman filter In the second embodiment, the Kalman filter used by the angle correction value calculation unit 116 may be improved as follows, taking advantage of the characteristics of the railway. Alternatively, a nonlinear Kalman filter may be used instead of a normal Kalman filter. Examples of nonlinear Kalman filters include unscented Kalman filters (UKF) and extended Kalman filters (EKF).

[0117] The first improvement involves taking into account the nonlinearity of the motion model of railway vehicles. Specifically, the Kalman filter uses equations (19) to (21) above to predict the prior state value x^ at time T. Equations (19) to (21) are linear systems, but “A” and “B” in equation (21) are sine and cosine functions, as shown in equation (13). Since the equations of motion for the vehicle become nonlinear when sine and cosine functions are introduced, the usual Kalman filter is not always suitable. However, railway vehicles do not accelerate or decelerate rapidly, and only move in an almost linear fashion along the track shape, and do not perform extreme lateral movements, sudden descents, or sharp turns. In other words, the roll angle and pitch angle are considered to be small, so it is possible to treat it as a linear system.

[0118] Furthermore, since the roll angle and pitch angle are considered to be small, if the roll angle and pitch angle estimated using equations (19) to (21) are large, it is possible to disregard these values ​​and instead use the results from the previous time T-1 or pre-prepared values.

[0119] Furthermore, the observed value z in equation (22) is a value based on GNSS and IMU measurements, as shown in equation (16). However, if the roll angle or pitch angle obtained as IMU measurements are large, those measurements are considered abnormal. In such cases, it is possible to use the measurement value from the previous time T-1 or a pre-prepared value as the observed value z.

[0120] (C) Integration of multiple speed information Furthermore, the accuracy may be improved by integrating the velocity information from each sensor (IMU, GNSS, TG) using a Kalman filter.

[0121] For example, the speed determination units 114A, 114B, and 114C select one of the following based on the judgment result of the running state: the TG speed (first running speed candidate), the GNSS speed (second running speed candidate), and the IMU speed (third running speed candidate), and determine the selected running speed candidate as the train's running speed. Alternatively, one or more running speed candidates may be selected based on the running state, and the value obtained by performing estimation calculations using a Kalman filter on the selected running speed candidate may be determined as the running speed.

[0122] In this case, when inputting velocity information from the IMU and velocity information from the GNSS into the Kalman filter, weighting based on the number of GNSS satellites that can be received makes it possible to improve the accuracy of the estimated velocity information.

[0123] Furthermore, each time an estimated value of velocity information is output from the Kalman filter, the difference between that estimated value and the velocity information obtained by the IMU is calculated, and based on that difference, the IMU... speed information You may also improve accuracy by correcting for this.

[0124] Alternatively, the difference in output cycles of measured values ​​between sensors (TG, GNSS, IMU) may be interpolated and compensated for. For example, since the output cycles of GNSS and IMU differ significantly (GNSS has a longer cycle), interpolating and increasing the measured value of the sensor with the longer output cycle (GNSS) can make the apparent output cycles of GNSS and IMU match, thereby stabilizing the estimated value obtained by the Kalman filter. [Explanation of Symbols]

[0125] 1A,1B,1C…Onboard equipment 102...TG speed calculation section 104...GNSS speed calculation section 106…IMU speed calculation unit 112A, 112C... Driving status determination unit 114A, 114B, 114C...Speed ​​determination section 116...Angle correction value calculation unit 122...TG Reliability Judgment Unit 124A, 124C…GNSS reliability determination unit 126A, 126B…IMU reliability determination unit 10… Speed ​​generator 12…GNSS receiver 14…Inertial sensor 16…Accelerometer 18…Angular velocity sensor

Claims

1. Onboard equipment installed on a train, A first running speed candidate calculation means calculates a first running speed candidate for the train based on the output signal of the speed generator, A second running speed candidate calculation means calculates a second running speed candidate for the train based on positioning signals from a given positioning satellite, A third running speed candidate calculation means calculates a third running speed candidate for the train based on the measured values ​​of the inertial sensor, A means for determining whether the train is in a running state after a low-speed state, A speed determination means that selects one of the first running speed candidate, the second running speed candidate, and the third running speed candidate based on the determination result of the running state determination means, and determines the running speed of the train based on the selected running speed candidate, An on-board device equipped with the following features.

2. The driving state determination means enables a state transition to the driving state only after passing through the low-speed state from the stationary state, and determines whether the driving state is the stationary state, the low-speed state, or the driving state. The on-board device according to claim 1.

3. The driving state determination means determines the driving state based on the average equivalent value of acceleration based on the measured value of the inertial sensor. The vehicle-mounted device according to claim 2.

4. The driving state determination means determines the driving state based on the history of the average equivalent value over the immediate vicinity period. The on-board device according to claim 3.

5. A second reliability determination means for determining the reliability of the second candidate travel speed, Furthermore, The speed determination means includes a running speed determination means that, when the running state determination means determines that the train is running, selects either the first running speed candidate or the second running speed candidate based on the determination result of the second reliability determination means, and determines the running speed of the train based on the selected running speed candidate. The on-board device according to claim 1.

6. A second reliability determination means for determining the reliability of the second candidate travel speed, A third reliability determination means for determining the reliability of the third candidate travel speed, Furthermore, The speed determination means includes a running speed determination means that, when the running state determination means determines that the train is running, selects the second running speed candidate if the determination result of the second reliability determination means is good, and, if not, selects either the first running speed candidate or the third running speed candidate based on the determination result of the third reliability determination means, and determines the running speed of the train based on the selected running speed candidate. The on-board device according to claim 1.

7. The in-driving determination means, when the determination result of the second reliability determination means is negative, selects either the first driving speed candidate or the third driving speed candidate according to the duration of the negative determination result. The vehicle-mounted device according to claim 6.

8. A second reliability determination means for determining the reliability of the second candidate travel speed, Furthermore, The speed determination means includes a running speed determination means that, when the running state determination means determines that the train is running, selects either the second running speed candidate or the third running speed candidate based on the determination result of the second reliability determination means, and determines the running speed of the train based on the selected running speed candidate. The on-board device according to claim 1.

9. The second reliability determination means determines the reliability of the second travel speed candidate based on the variance equivalent value of the second travel speed candidate. The on-board device according to any one of claims 5 to 8.

10. The second reliability determination means determines the reliability of the second candidate travel speed based on the history of the equivalent variance value over the immediate vicinity period. The on-board device according to claim 9.

11. The inertial sensor comprises an acceleration sensor and an angular velocity sensor. The third travel speed candidate calculation means calculates the third travel speed candidate by removing gravity error based on the measured value of the acceleration sensor and the corrected measured value obtained by correcting the measured value of the angular velocity sensor based on a given correction value, When the driving state determination means determines that the vehicle is in motion, a Kalman filter that estimates the measured value of the angular velocity sensor based on the speed and acceleration calculated based on the positioning signal, wherein an angle correction value calculation means calculates the correction value by estimating the measured value of the angular velocity sensor using a Kalman filter with filter coefficients corresponding to the determination result of the second reliability determination means. The vehicle-mounted device according to any one of claims 5 to 8, further comprising:

Images