JP2018203024A - Derailment sign detection system, control device, derailment sign detection method, and derailment sign detection program - Google Patents

Abstract

A derailment sign detection system capable of increasing the accuracy of detection of a derailment sign of a train is provided. Wavelet analysis is applied to each of a pitch angular velocity θ (t) and a roll angular velocity φ (t) output from an angular velocity sensor 35 mounted on a carriage, and a wavelet coefficient 14 of a pitch angular velocity and a wavelet coefficient 15 of a roll angular velocity are applied. Is calculated. Each of the two wavelet coefficients 14 and 15 changing in time series is compared with a wavelet coefficient threshold 16 and a derailment sign is detected when both exceed the threshold. For example, a wavelet coefficient calculated for a low frequency region of 0.5 to 100 Hz is used. Derailment sign detection accuracy is improved by combining two types of derailment sign detection algorithms in the frequency domain and the time domain. By adopting wavelet analysis, real-time processing in the frequency domain becomes possible, and derailment prevention is realized. [Selection] Figure 4

Description

  The present invention relates to a derailment sign detection system, a control device, a derailment sign detection method, and a derailment sign detection program.

  As a conventional technique for detecting a sign of derailment of a railway train, techniques described in Patent Document 1 and Patent Document 2 by the present inventors are known.

  In the technique of Patent Document 1, the pitch angular velocity and the roll angular velocity of a traveling cart are detected by a sensor attached to the cart frame, and the detected value of the detected cart pitch angular velocity or the cart pitch angular velocity is larger than a preset threshold value. And that the detected bogie roll angular velocity or the integrated value of the bogie roll angular velocity is larger than a preset threshold value, it is determined as a derailment sign.

  In the technique of Patent Document 2, the roll angular velocity of the detected carriage is stored in a storage device, a moving average is calculated based on the history of the roll angular velocity, and a predicted value of the roll angular velocity is calculated from the moving average. . Then, when each of the detected pitch angular velocity of the carriage and the predicted value after a predetermined time of the roll angular velocity exceeds a predetermined threshold, it is determined that the train is derailed.

Japanese Patent No. 5468016 JP 2014-231308 A

  By using the prior arts of Patent Document 1 and Patent Document 2, it is possible to detect a sign of a derailment of a train. However, these conventional techniques may not always exhibit sufficient performance depending on conditions. For example, when parameters such as threshold values determined based on data measured on the test track are applied to trains traveling on actual business routes, disturbances that occur in the pitch angular velocity and roll angular velocity at locations such as rail joints In some cases, the detection of derailment signs may be erroneously detected due to the influence of the extreme value. In addition, since the threshold value used to detect the derailment sign depends largely on the traveling speed, it is difficult to determine an appropriate threshold value, and there is a tendency that false detection tends to occur particularly in a low speed range.

  In addition, various methods may be adopted for analysis of pitch angular velocity, roll angular velocity, etc., but when detecting a derailment sign of a train, it is detected within 0.2 seconds after the sign is generated. Since there is a restriction that it is difficult to avoid derailment unless measures are taken immediately, methods other than those that can be analyzed almost in real time cannot be adopted.

  The present invention has been made in view of the above-described circumstances, and a purpose thereof is a derailment sign detection system, a control device, a derailment sign detection method, and a derailment sign detection capable of improving the detection accuracy of a train derailment sign. To provide a program.

In order to achieve the above-described object, a derailment sign detection system, a control device, a derailment sign detection method, and a derailment sign detection program according to the present invention are characterized by the following (1) to (6).
(1) A detection unit that is provided in a train and detects a pitch angular velocity and a roll angular velocity of a running train;
The pitch angular velocity wavelet coefficient is calculated as a first wavelet coefficient, the roll angular velocity wavelet coefficient is calculated as a second wavelet coefficient, and the calculated first wavelet coefficient and second wavelet coefficient are calculated. When each exceeds a predetermined threshold value, a control device that determines a derailment sign of the train,
When the controller determines that it is a derailment sign, an output unit that notifies the derailment sign to the outside,
Derailment sign detection system with

  According to the derailment sign detection system configured as described in (1) above, it is possible to improve the detection accuracy of a train derailment sign. That is, the first wavelet coefficient obtained by wavelet analysis of each detected value of the pitch angular velocity and the roll angular velocity, and the second wavelet coefficient are compared with a threshold value to determine at a position such as a rail joint at a low speed. It is possible to reduce false detection due to the influence of the generated disturbance.

(2) The control device calculates the first wavelet coefficient and the second wavelet coefficient only within the low frequency region.
The derailment sign detection system according to (1) above.

  According to the derailment sign detection system configured as described in (2) above, it is possible to further increase the accuracy of detection of a train derailment sign. In other words, when the train wheel flange rides on the rail and reaches derailment, it is confirmed that there is a high output of low frequency in both the pitch angular velocity and the roll angular velocity, so it is calculated within the low frequency range. By using the first wavelet coefficient and the second wavelet coefficient, the detection accuracy of the derailment sign is improved.

(3) The control device calculates a moving average based on a history of detection values for at least the roll angular velocity, predicts a change in the roll angular velocity from the moving average, and calculates a predicted value of the roll angular velocity. Determining a derailment sign of the train based on the first wavelet coefficient and the second wavelet coefficient;
The derailment sign detection system according to (1) or (2) above.

  According to the derailment sign detection system configured as described in (3) above, it is possible to determine a condition that combines the predicted roll angular velocity, the calculated first wavelet coefficient, and the previous two wavelet coefficients. Can be further improved.

(4) A wavelet coefficient of the pitch angular velocity is calculated as a first wavelet coefficient based on the pitch angular velocity and the roll angular velocity of the running train that is provided in the train and detected by a predetermined detector, and the roll angular velocity A wavelet coefficient is calculated as a second wavelet coefficient, and when the calculated first wavelet coefficient and the second wavelet coefficient exceed a predetermined threshold value, it is determined as a derailment sign of the train, and a derailment sign A control device that notifies the outside of the derailment sign detection when it is determined.

  According to the control device having the configuration (4), it is possible to improve the detection accuracy of the derailment sign detection system for the train as in the derailment sign detection system (1).

(5) Enter the detected values of the pitch angular velocity and roll angular velocity of the running train,
Calculating the wavelet coefficient of the pitch angular velocity as a first wavelet coefficient;
Calculating the wavelet coefficient of the roll angular velocity as a second wavelet coefficient;
Comparing each of the calculated first wavelet coefficient and the second wavelet coefficient with a predetermined threshold;
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold, it is determined as a derailment sign of the train, and derailment sign detection is notified to the outside.
Derailment sign detection method.

  According to the derailment sign detection method having the configuration of (5), it is possible to improve the accuracy of detection of a derailment sign detection system for a train, as in the derailment sign detection system of (1).

(6) a step of inputting detected values of the pitch angular velocity and the roll angular velocity of the running train;
Calculating the pitch angular velocity wavelet coefficient as a first wavelet coefficient;
Calculating the roll angular velocity wavelet coefficient as a second wavelet coefficient;
Comparing each of the calculated first wavelet coefficients and the second wavelet coefficients with a predetermined threshold;
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold, determining a derailment sign of the train, and notifying derailment sign detection to the outside,
A derailment sign detection program for causing a computer to execute processing including

  By executing the derailment sign detection program configured as described in (6) above with a predetermined computer, it is possible to improve the detection accuracy of the derailment sign detection system of the train as in the derailment sign detection system according to (1).

  According to the derailment sign detection system, the control device, the derailment sign detection method, and the derailment sign detection program of the present invention, it is possible to improve the detection accuracy of the train derailment sign detection. In other words, when the train wheel flange rides on the rail and reaches derailment, it is confirmed that there is a high output of low frequency in both the pitch angular velocity and the roll angular velocity, so it is calculated within the low frequency range. By using the first wavelet coefficient and the second wavelet coefficient, the detection accuracy of the derailment sign is improved.

  The present invention has been briefly described above. Further, the details of the present invention will be further clarified by reading through a mode for carrying out the invention described below (hereinafter referred to as “embodiment”) with reference to the accompanying drawings. .

FIG. 1 is a side view showing a configuration example of a train on which a derailment sign detection system according to an embodiment of the present invention is mounted. FIG. 2 is a plan view showing a bogie frame of a train carriage. FIG. 3 is a side view showing a bogie frame of a train carriage. FIG. 4 is a block diagram illustrating a configuration example of the derailment sign detection system. FIG. 5 is a block diagram illustrating a configuration example of the frequency domain derailment sign detection unit. FIG. 6 is a block diagram showing input / output and determination conditions in the time domain derailment sign detection unit. FIG. 7 is a waveform diagram showing examples of input signals and various wavelets in wavelet analysis. FIG. 8 is a waveform diagram showing the waveform of a Morley wavelet used as a mother wavelet. 9 (a), FIG. 9 (b), FIG. 9 (c), FIG. 9 (d), FIG. 9 (e), FIG. 9 (f), FIG. 9 (g), and FIG. It is a time chart showing the time transition of the wavelet coefficient of roll angular velocity and pitch angular velocity in each of eight types of test data obtained by experiment using a test orbit. FIG. 10A and FIG. 10B are three-dimensional graphs showing the distribution state of wavelet coefficients for one set of test data obtained by an experiment using a test trajectory, and FIG. 10A shows the roll angular velocity. FIG. 10B shows the pitch angular velocity. FIG. 11A and FIG. 11B are time charts showing time transitions of wavelet coefficients based on a set of test data obtained in an experiment using a test trajectory. FIG. 11A shows a roll angular velocity. FIG. 11B shows the pitch angular velocity.

  Specific embodiments relating to the present invention will be described below with reference to the drawings.

<Example of train configuration>
The structural example of the train 20 carrying the derailment sign detection system of embodiment of this invention is shown in FIG. The state which looked at the bogie frame 25 of the trolley | bogie 22 of the train 20 from upper direction is shown in FIG. Moreover, the state which looked at this bogie frame 25 from the side is shown in FIG.

  The derailment sign detection system 50 of the present invention is used in a state of being mounted on the train 20 having the configuration as shown in FIG. In addition, although the train 20 shown in FIG. 1 is a case of two-car train, it can also be used for a train 20 of one train or three or more trains.

  As shown in FIG. 1, each vehicle body 21 of the train 20 includes a plurality of carriages 22 that are connected to the lower portion thereof and support the vehicle body 21. In the example of FIG. 1, two carriages 22 are provided before and after each vehicle body 21. Each carriage 22 includes a bogie frame 25 as shown in FIGS.

  The bogie frame 25 is provided with an axle 23 so as to straddle the wheel, and wheels 24 in contact with the rails are connected to both ends of the axle 23. As shown in FIGS. 2 and 3, an angular velocity sensor 35 is disposed near the center in the front-rear direction of the bogie frame 25 and on the left side in the width direction of the train 20. The carriage 22 is also provided with a speed sensor 34 (see FIG. 4) that detects the traveling speed of the train 20.

  The angular velocity sensor 35 detects the pitch angular velocity and the roll angular velocity. In the present embodiment, an angular velocity sensor 35 configured as a micro gyroscope having a tuning fork vibrator is used. The pitch angular velocity is an angular velocity of rotation (or inclination) around the pitch direction, that is, the width direction of the train 20, and the roll angular velocity is the rotation (or inclination) around the roll direction, ie, the front-rear direction of the train 20. Angular velocity.

<Configuration example of derailment sign detection system 50>
FIG. 4 is a configuration example of the derailment sign detection system 50.

  The derailment sign detection system 50 shown in FIG. 4 is prepared for each carriage 22 of the train 20 shown in FIG. 1, for example, and the control unit 51 of each derailment sign detection system 50 is installed on the vehicle body 21. Electric signals from the angular velocity sensor 35 and the velocity sensor 34 mounted on each carriage 22 are input to the control unit 51 of each derailment sign detection system 50. Of course, the plurality of carriages 22 of the plurality of vehicle bodies 21 can be centrally controlled by the single control unit 51.

  The control unit 51 is a computer and has hardware with sufficient performance so that the calculation process can be completed in a short time. Further, control software for realizing each function corresponding to the frequency domain derailment sign detection unit 10, the time domain derailment sign detection unit 30, and the final determination unit 40 is provided. A plurality of independent computers may be provided inside the control unit 51.

  The derailment sign detection system 50 shown in FIG. 4 is equipped with two types of derailment sign detection algorithms in order to detect a derailment sign of the train 20 and prevent derailment. One derailment sign detection algorithm is incorporated in the frequency domain derailment sign detection unit 10, and the other derailment sign detection algorithm is incorporated in the time domain derailment sign detection unit 30.

  Although details will be described later, the derailment sign detection algorithm incorporated in the frequency domain derailment sign detection unit 10 inputs the pitch angular velocity θ (t) and the roll angular velocity φ (t) output from the angular velocity sensor 35, and these A wavelet coefficient is calculated for each signal by applying wavelet analysis to each of the signals. “T” represents time. Then, the presence / absence of a derailment sign is identified by comparing each of the wavelet coefficient of the pitch angular velocity and the wavelet coefficient of the roll angular velocity that change in time series with a predetermined threshold value. Actually, when the wavelet coefficient of the pitch angular velocity exceeds the threshold value and at the same time the wavelet coefficient of the roll angular velocity exceeds the threshold value, it is detected as a derailment sign.

  On the other hand, the derailment sign detection algorithm incorporated in the time domain derailment sign detection unit 30 is basically the same as the algorithm disclosed in Patent Document 2. That is, the pitch angular velocity θ (t) and roll angular velocity φ (t) output from the angular velocity sensor 35 and the traveling speed V (t) output from the speed sensor 34 are input and processed. Then, a change in the roll angular velocity is predicted based on the calculation of the moving average. Furthermore, the pitch angular velocity is compared with a threshold value, or the roll angular velocity is predicted and compared with a threshold value, and a derailment sign is detected based on the result. About each threshold value which the time domain derailment sign detection unit 30 uses, a suitable value is employ | adopted according to driving speed V (t).

  Basically, sufficient performance can be obtained for derailment sign detection only by the frequency domain derailment sign detection unit 10 incorporating the newly developed derailment sign detection algorithm. However, the algorithm of the frequency domain derailment sign detection unit 10 and the algorithm of the time domain derailment sign detection unit 30 have different advantages and disadvantages. Therefore, it is possible to avoid derailment more effectively by properly using the two types of derailment sign detection algorithms.

  Therefore, the final determination unit 40 receives the derailment sign determination output 17 output from the frequency domain derailment sign detection unit 10 and the derailment sign determination output 18 output from the time domain derailment sign detection unit 30, and is based on these signals. And implement optimal train control.

  A train brake device 38 and a train watering device 39 are connected to the output of the final determination unit 40. That is, derailment can be prevented by applying braking by the brake device 38 when a derailment sign is detected. The water sprinkler 39 is a device that sprinkles water on the contact portion between the rail and the wheel flange. That is, when the wheel flange is on the rail as a derailment sign, water is sprayed on the contact portion to reduce the friction coefficient. As a result, the wheel flange riding on the rail can be returned to the normal position, and derailment can be prevented. In this embodiment, the case where the derailment sign detection system 50 includes the watering device 39 is described as an example, but this is an example, and derailment is performed instead of or in addition to the watering device 39. You may make it provide another apparatus for preventing.

  As a typical control example of the final determination unit 40, when a derailment sign is recognized by the derailment sign determination output 17 output from the frequency domain derailment sign detection unit 10, the final determination unit 40 controls the brake device 38 of the train. And control to apply the brake. Moreover, when the derailment sign is recognized by the derailment sign determination output 18 output from the time domain derailment sign detection unit 30, the final determination unit 40 controls the train watering device 39 to sprinkle to reduce the friction coefficient.

  When a derailment sign is detected, an alarm may be output or a danger may be notified using a pseudo audio signal or the like. However, in practice, derailment cannot be avoided unless water spray and brake control are performed within 0.2 seconds after the derailment sign is detected. Therefore, it is desirable that the final determination unit 40 automatically controls watering and braking.

<Frequency domain derailment sign detection unit 10>
A configuration example of the frequency domain derailment sign detection unit 10 is shown in FIG. The frequency domain derailment sign detection unit 10 illustrated in FIG. 5 includes two wavelet transform processing units 11 and 12 and a derailment sign determination unit 13.

  The wavelet transformation processing unit 11 repeatedly samples and inputs a signal of the pitch angular velocity θ (t) output from the angular velocity sensor 35 at a constant period (for example, 1/200 [second]), and performs wavelet transformation processing on the signal. To implement. A wavelet coefficient 14 of pitch angular velocity is obtained as a time-series signal at the output of the wavelet transform processing unit 11. The wavelet transform will be described later.

  The wavelet transform processing unit 12 repeatedly samples and inputs the roll angular velocity φ (t) signal output from the angular velocity sensor 35 at a constant period (for example, 1/200 [second]), and outputs the wavelet to this signal. Perform the conversion process. The wavelet coefficient 15 of the roll angular velocity is obtained as a time-series signal at the output of the wavelet transform processing unit 12.

  According to the results of various experiments, it was found that there was a high output of low frequency in both the roll angular velocity and the pitch angular velocity when the derailment occurred on the wheel flange of the train. Therefore, in order to effectively detect a sign of derailment, each of the wavelet transform processing units 11 and 12 is limited to a predetermined low frequency region, for example, a range of 0.5 to 100 [Hz]. To calculate wavelet coefficients.

  The derailment sign determination unit 13 is based on the pitch angular velocity wavelet coefficient 14 output from the wavelet transform processing unit 11, the roll angular velocity wavelet coefficient 15 output from the wavelet transform processing unit 12, and the wavelet coefficient threshold 16. Perform the determination. As the wavelet coefficient threshold 16, independent thresholds may be assigned to both the pitch angular velocity wavelet coefficient 14 and the roll angular velocity wavelet coefficient 15, but a common threshold is used for both. In the present embodiment, the wavelet coefficient threshold 16 is a predetermined constant and does not change. However, depending on the case, for example, the value of the threshold 16 of the wavelet coefficient may be adjusted to be slightly changed according to the traveling speed of the train.

  When the derailment sign determination unit 13 detects a state where the value of the wavelet coefficient 14 of the pitch angular velocity exceeds the threshold 16 and at the same time the value of the wavelet coefficient 15 of the roll angular velocity exceeds the threshold 16, the derailment sign determination output 17 outputs the derailment sign determination output 17. A signal indicating detection is output.

<Time domain derailment sign detection unit 30>
The input / output and determination conditions in the time domain derailment sign detection unit 30 are shown in FIG.

  As shown in FIGS. 6 and 4, signals of the pitch angular velocity θ (t), the roll angular velocity φ (t), and the traveling speed V (t) are input to the time domain derailment sign detection unit 30. Further, parameters used by the time domain derailment sign detection unit 30 for control include a pitch angular velocity threshold value, a roll angular velocity moving average predicted value threshold value, a detection time, an inclination measurement time, and a prediction time, as shown in FIG. Here, each value of the pitch angular velocity threshold value and the roll angular velocity moving average predicted value threshold value changes so as to use an optimum value for the traveling velocity V (t).

  As shown in the block of the time domain derailment sign detection unit 30 in FIG. 6, the time domain derailment sign detection unit 30 determines the detection of the derailment sign by comparing three conditions and outputs the result as a warning. To do. The three conditions to be compared are as follows, and these are determined by their logical product.

(1) The pitch angular velocity θ (t) is greater than or equal to the pitch angular velocity threshold (speed dependent).
(2) The moving average value Φ of the roll angular velocity φ (t) is equal to or greater than the threshold value (speed dependency).
(3) The predicted value Φp based on the moving average of the roll angular velocity φ (t) is equal to or greater than the threshold value (speed dependency).
Since the contents of the processing in the time domain derailment sign detection unit 30 are the same as those of the prior art disclosed in Patent Document 2, detailed description thereof is omitted.

<Description of wavelet transform>
FIG. 7 shows an input signal in the wavelet analysis and each waveform of various wavelets. FIG. 8 shows the waveform of a Morley wavelet used as a mother wavelet. However, various waveforms can be used for the mother wavelet according to the analysis target.

  Note that Fourier analysis is generally used as a technique for performing frequency analysis of signals. However, when the Fourier transform is performed, information about time is lost, so that necessary information cannot be obtained in an application for detecting an abrupt change such as a train derailment sign. Further, in the case of short-time Fourier analysis using a Hamming window, it is difficult to detect a low frequency in a short time, and therefore, a derailment sign cannot be detected within 0.2 seconds, for example.

  A wavelet is a temporary wave, and wavelet analysis is a technique for expressing arbitrary time-series data as a sum of wavelets. For example, as shown in FIG. 7, when a wave W1 is input as an arbitrary input signal, the wave W1 is decomposed into various wavelets. For example, the wave W1 is obtained as the sum of the wavelets W2, W3, W4, W5, and W6. Can be expressed.

  Wavelet transformation includes continuous wavelet transformation (CWT) and discrete wavelet transformation (DWT). In the continuous wavelet transform, a mother wavelet is used, and a copy that has been moved and enlarged / reduced with respect to the mother wavelet is compared with the wave of the input signal. In continuous wavelet transform, inner product is used to measure the similarity between signal and analytic function. Various shapes can be used for the mother wavelet. If the wavelet is a complex value, the CWT is a complex value function of scale and position, and if the signal is a real value, the CWT is a real value function of scale and position. The continuous wavelet transform can be expressed as:

However,
C (a, b; f (t), Ψ (t)): wavelet coefficient (CWT coefficient)
a: Scale parameter b: Position parameter f (t): Input signal (original signal)
Ψ (t): Mother wavelet t: Time

  That is, based on the mother wavelet Ψ (t) and the input signal f (t), wavelet coefficients are obtained by intermittent changes in the scale parameter (a) and the position parameter (b). Multiplying each coefficient by the appropriate scaled and moved wavelet (eg, W2-W6 shown in FIG. 7) produces the constituent wavelet of the original signal.

  In the wavelet transform processing units 11 and 12 of the present embodiment, the waveform of the Morley wavelet shown in FIG. 8 is adopted as the mother wavelet Ψ (t) of the wavelet transform. There are two types of Morray wavelets: a real value version and a complex value version. The real value version of the Morley wavelet is expressed by the following equation. Constants are used for normalization and reconstruction.

The continuous wavelet transform (CWT) of the first equation (1) can be rewritten as the following equation using inverse Fourier transform. This also allows continuous wavelet transforms to be interpreted as frequency-based filtering of signals.

  Equation (3) above shows that the spread of the wavelet after time causes the support to contract in the frequency domain. The center frequency moves toward lower frequencies due to the spread. In the wavelet transform, the expansion of the wavelet operation is defined as energy conservation. In order to conserve energy during contraction, frequency support requires an increase in peak energy levels. The quality factor (or sometimes called the filter factor) is the ratio of peak energy to bandwidth. Therefore, the wavelet is sometimes called a constant Q filter. This is because the contraction and expansion of the wavelet frequency support results in a proportional increase or decrease in peak energy. This is an important characteristic of a wavelet that has utility value in detecting a derailment sign of a train.

The above equation (3) basically defines CWT as the inverse Fourier transform of the product of the Fourier transform. This means that CWT can be calculated using inverse Fourier transform.
There is an efficient algorithm for the calculation of the discrete Fourier transform. Therefore, in order to enable the adoption of an efficient algorithm in CWT, the concept of CWT is applied to the Fourier domain, and a calculation formula for CWT is constructed.

The above expression (4) can be rewritten in the form of the following expression.
The above equation (5) expresses CWT explicitly as convolution.

When executing the discrete version of CWT, the input sequence is represented as an N-length vector (x [n]). A discrete version of the CWT convolution is expressed as:

  In order to obtain CWT in the form of equation (6) above, it is necessary to calculate the convolution of each value of the shifted parameter (b) and repeat this process for each scale (a). However, when the two sequences spread in a ring and are divided into lengths N, the circular convolution can be expressed by the following equation (7) as the product of the discrete Fourier transform. CWT is the inverse Fourier transform of the product.

  As described above, by expressing the CWT as the inverse Fourier transform, it is possible to calculate the CWT using the fast Fourier algorithm. As a result, the calculation efficiency can be greatly increased, and the calculation cost of convolution can be reduced.

  Using the method described above, the wavelet transform processing units 11 and 12 shown in FIG. 5 each perform a wavelet transform calculation process. The wavelet transform processing unit 11 processes the pitch angular velocity θ (t) to calculate the pitch angular velocity wavelet coefficient 14. The wavelet transform processing unit 12 processes the roll angular velocity φ (t) and calculates the roll angular velocity wavelet coefficient 15. The calculated pitch angular velocity wavelet coefficient 14 and roll angular velocity wavelet coefficient 15 are both time-series signals that change.

<Explanation of experimental data>
Parameters such as various thresholds are appropriately determined in advance so that each of the frequency domain derailment sign detection unit 10 and the time domain derailment sign detection unit 30 of the derailment sign detection system 50 shown in FIG. 4 can correctly detect the train derailment sign. There is a need to. For this purpose, an experiment was performed and the data of the result was used to adjust the derailment sign detection system 50.

  On the test track of the University of Tokyo Chiba Laboratory, a full-scale trolley was used to carry out flange derailment tests at various running speeds. In this test, various measures were taken so that the derailment test can be performed safely. Further, as in the case of the train 20 shown in FIG. 1, the angular velocity sensor 35, the speed sensor 34, and other sensors were attached on the carriage, and the test was performed. That is, time series data including a pitch angular velocity θ (t), a roll angular velocity φ (t), and a traveling speed V (t) in the carriage was measured and recorded using a data logger.

  As a result of this experiment, “Test 1”, “Test 2”, “Test 3”, “Test 4”, “Test 5”, “Test 6”, “Test 7”, and “Test 7” shown in Table 1 below are shown. Eight sets of data of “Test 8” were extracted and analyzed.

  FIG. 9A, FIG. 9B, and FIG. 9 show time transitions of the wavelet coefficients of the roll angular velocity (Roll) and the pitch angular velocity (Pitch) in each of the eight types of test data obtained in the experiment using the test trajectory. (C), FIG. 9 (d), FIG. 9 (e), FIG. 9 (f), FIG. 9 (g), and FIG. 9A to 9H, the horizontal axis of each graph represents time (seconds), and the vertical axis represents the value of the wavelet coefficient.

  9A to 9H, a portion surrounded by a square represents a portion where derailment or a sign thereof has occurred. That is, when the wheel flange of the bogie rides on the rail, both the wavelet coefficient of the roll angular velocity (Roll) and the wavelet coefficient of the pitch angular velocity (Pitch) increase at the same time. In other words, in a situation where derailment or a sign thereof is occurring, the wavelet coefficient of roll angular velocity (Roll) and the wavelet coefficient of pitch angular velocity (Pitch) are both increased to a predetermined value or more.

  Further, as shown in Table 1, “Test 1” to “Test 8” have different traveling speeds immediately before starting the wheel flange climbing, but each of FIGS. 9A to 9H. There is no big difference in the size of the wavelet coefficient at the occurrence of derailment sign.

  Therefore, in the present embodiment, a constant constant independent of the traveling speed is given as the wavelet coefficient threshold value 16 to be given to the derailment sign determination unit 13 of the frequency domain derailment sign detection unit 10. Specifically, in the same environment as the data in FIGS. 9A to 9H, it is assumed that a value of about “0.08” is assigned as the threshold 16 of the wavelet coefficient. Therefore, when the derailment sign determination unit 13 shown in FIG. 5 detects that the wavelet coefficient 14 of the pitch angular velocity exceeds “0.08” and at the same time the wavelet coefficient 15 of the roll angular velocity exceeds “0.08”. Then, it is determined that “there is a sign of derailment”.

  10A and 10B show a three-dimensional graph representing the distribution state of the wavelet coefficients in the time axis direction and the frequency axis direction with respect to one set of test data obtained in the experiment using the test trajectory. FIG. 10A shows the roll angular velocity, and FIG. 10B shows the pitch angular velocity.

  Moreover, the time transition of the wavelet coefficient based on one set of test data obtained by the experiment using the test trajectory is shown in FIGS. FIG. 11A shows the roll angular velocity, and FIG. 11B shows the pitch angular velocity.

  In the derailment sign detection algorithm employed by the frequency domain derailment sign detection unit 10, the temporal transition of the power spectral density of the pitch angular velocity and roll angular velocity in the low frequency region (for example, 0.5 to 100 [Hz]) is monitored. Can do. That is, a derailment sign is detected by monitoring changes as shown in the graphs of FIGS. 10 (a) and 10 (b).

  By analyzing the experimental data obtained, the frequency components of the pitch angular velocity and the roll angular velocity distinguish between the normal traveling state of the carriage (including disturbances generated at rail joints) and the wheel flange climbing derailment state. It was found to be very effective above. Further, by adopting wavelet analysis of the pitch angular velocity and roll angular velocity, the time transition of the frequency component can be monitored in real time, so that it can be detected within 0.2 seconds after the occurrence of the derailment sign.

  Further, as in the derailment sign detection system 50 shown in FIG. 4, the frequency domain derailment sign detection unit 10 and the time domain derailment sign detection unit 30 are employed, and two types of derailment sign detection algorithms are used in combination. The sign detection accuracy can be further improved, and effective derailment prevention control can be performed. For example, in the case of the derailment sign detection algorithm alone of the time domain derailment sign detection unit 30, an erroneous detection of a derailment sign may occur due to the influence of a significant value due to a disturbance generated at the rail joint. However, the occurrence of false detection can be suppressed by combining algorithms that employ wavelet analysis.

  The functions of each part of the derailment sign detection system 50 can be realized as a dedicated control device mounted on a train, or can be realized by being incorporated as an application program in a general-purpose device such as a personal computer.

Here, the features of the embodiment of the derailment sign detection system, the control device, the derailment sign detection method, and the derailment sign detection program according to the present invention described above are briefly summarized and listed in the following [1] to [6], respectively.
[1] A detection unit (angular velocity sensor 35) that is provided in the train and detects a pitch angular velocity and a roll angular velocity of the running train,
The pitch angular velocity wavelet coefficient is calculated as a first wavelet coefficient, the roll angular velocity wavelet coefficient is calculated as a second wavelet coefficient, and the calculated first wavelet coefficient and second wavelet coefficient are calculated. When each exceeds a predetermined threshold value, a control device (frequency domain derailment sign detection unit 10) that determines that the train is a derailment sign,
When the control device determines that it is a derailment sign, an output unit (final determination unit 40) that notifies the derailment sign to the outside;
A derailment sign detection system (50) comprising:

[2] The control device (wavelet transform processing units 11 and 12) calculates the first wavelet coefficient and the second wavelet coefficient only within a low frequency region.
The derailment sign detection system according to [1] above.

[3] The control device (control unit 51) calculates a moving average based on a history of detection values for at least the roll angular velocity, predicts a change in the roll angular velocity from the moving average, and predicts the roll angular velocity. And determining a derailment sign of the train based on the calculated first wavelet coefficient and the second wavelet coefficient.
The derailment sign detection system according to the above [1] or [2].

[4] A wavelet coefficient of the pitch angular velocity is calculated as a first wavelet coefficient based on the pitch angular velocity and the roll angular velocity of the running train that is provided in the train and detected by a predetermined detection unit, and the roll angular velocity A wavelet coefficient is calculated as a second wavelet coefficient, and when the calculated first wavelet coefficient and the second wavelet coefficient exceed a predetermined threshold value, it is determined as a derailment sign of the train, and a derailment sign If it is determined, the control device (frequency domain derailment sign detection unit 10) for notifying the outside of the derailment sign detection.

[5] Input the detected values of the pitch angular velocity and roll angular velocity of the running train,
Calculating the wavelet coefficient of the pitch angular velocity as a first wavelet coefficient (corresponding to the function of the wavelet transform processing unit 11);
The wavelet coefficient of the roll angular velocity is calculated as a second wavelet coefficient (corresponding to the function of the wavelet transform processing unit 12),
Each of the calculated first wavelet coefficient and the second wavelet coefficient is compared with a predetermined threshold (corresponding to the function of the derailment sign determination unit 13),
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold value, it is determined as a derailment sign of the train, and the derailment sign detection is notified to the outside (corresponding to the function of the final determination unit 40). ,
Derailment sign detection method.

[6] A step of inputting detected values of the pitch angular velocity and the roll angular velocity of the running train;
Calculating the wavelet coefficient of the pitch angular velocity as a first wavelet coefficient (corresponding to the function of the wavelet transform processing unit 11);
Calculating the wavelet coefficient of the roll angular velocity as a second wavelet coefficient (corresponding to the function of the function of the wavelet transform processing unit 12);
Comparing each of the calculated first wavelet coefficient and the second wavelet coefficient with a predetermined threshold (corresponding to the function of the derailment sign determination unit 13);
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold value, a step of determining a derailment sign of the train and notifying a derailment sign detection to the outside (corresponding to the function of the final determination unit 40) )When,
A derailment sign detection program for causing a computer to execute processing including

DESCRIPTION OF SYMBOLS 10 Frequency domain derailment sign detection unit 11,12 Wavelet transformation process part 13 Derailment sign determination part 14 Pitch angular velocity wavelet coefficient 15 Roll angular velocity wavelet coefficient 16 Wavelet coefficient threshold 17,18 Derailment sign judgment output 20 Train 21 Car body 22 Cart 23 Axle 24 Wheel 25 Bogie frame 30 Time domain derailment sign detection unit 31, 32 Train control signal 34 Speed sensor 35 Angular speed sensor 38 Train brake device 39 Train sprinkler 40 Final decision unit 50 Derailment sign detection system 51 Control unit

Claims (6)

A detector that is provided in the train and detects the pitch angular velocity and roll angular velocity of the running train, and
The pitch angular velocity wavelet coefficient is calculated as a first wavelet coefficient, the roll angular velocity wavelet coefficient is calculated as a second wavelet coefficient, and the calculated first wavelet coefficient and second wavelet coefficient are calculated. When each exceeds a predetermined threshold value, a control device that determines a derailment sign of the train,
When the controller determines that it is a derailment sign, an output unit that notifies the derailment sign to the outside,
Derailment sign detection system with The control device calculates the first wavelet coefficient and the second wavelet coefficient only within the low frequency region.
The derailment sign detection system according to claim 1. The control device calculates a moving average based on a history of detected values for at least the roll angular velocity, predicts a change in the roll angular velocity from the moving average, and calculates the predicted value of the roll angular velocity and the calculated first Determining a derailment sign of the train based on the wavelet coefficient of the second wavelet coefficient and the second wavelet coefficient;
The derailment sign detection system according to claim 1 or 2.   A wavelet coefficient of the pitch angular velocity is calculated as a first wavelet coefficient based on a pitch angular velocity and a roll angular velocity of a running train that is provided in the train and detected by a predetermined detection unit, and the wavelet coefficient of the roll angular velocity is calculated as the first wavelet coefficient. When it is calculated as a second wavelet coefficient, and the calculated first wavelet coefficient and the second wavelet coefficient exceed a predetermined threshold value, it is determined as a derailment sign of the train, and is determined as a derailment sign A control device that notifies the derailment sign detection to the outside in the event of a failure. Input the detected value of the pitch angular velocity and roll angular velocity of the running train,
Calculating the wavelet coefficient of the pitch angular velocity as a first wavelet coefficient;
Calculating the wavelet coefficient of the roll angular velocity as a second wavelet coefficient;
Comparing each of the calculated first wavelet coefficient and the second wavelet coefficient with a predetermined threshold;
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold, it is determined as a derailment sign of the train, and derailment sign detection is notified to the outside.
Derailment sign detection method. Inputting the detected values of the pitch angular velocity and the roll angular velocity of the running train;
Calculating the pitch angular velocity wavelet coefficient as a first wavelet coefficient;
Calculating the roll angular velocity wavelet coefficient as a second wavelet coefficient;
Comparing each of the calculated first wavelet coefficients and the second wavelet coefficients with a predetermined threshold;
When the first wavelet coefficient and the second wavelet coefficient exceed the threshold, determining a derailment sign of the train, and notifying derailment sign detection to the outside,
A derailment sign detection program for causing a computer to execute processing including