JP2022047094A - Method and program for determining loosening of rail joint plate bolt and information processing system - Google Patents

Abstract

To determine loosening of rail joint plate bolts by train running.SOLUTION: A method for determining loosening of rail joint plate bolts includes: (A) a step of acquiring vibration or acoustic data at rail normal joints through a vibration sensor that detects vibration of an axle box including bearings of the axle of the wheel of the train or an acoustic sensor installed on the train and detecting the sound from the rail; (B) a step of executing a noise removal process for the acquired vibration or acoustic data and a feature extraction process for extracting a predetermined type of feature amount that correlates with loosening of the joint plate bolts of the rail; and (C) a step of executing a process of determining whether or not the joint plate bolts of the rail are loose with respect to the extracted features of a predetermined type.SELECTED DRAWING: Figure 3

Description

The present invention relates to a technique for determining looseness of rail seam plate bolts.

In carrying out the railway business, it is important to maintain the rails in order to ensure the safety of railway operation. Even in the maintenance of rails, it is especially important to grasp the state of the ordinary joints of the rails because the ordinary joints of the rails are subject to various loads and are prone to deterioration.

At present, the track maintenance staff goes to the normal seam part of the rail, hits the seam plate bolt with a test hammer, and the track maintenance staff listens to the tapping sound of the seam plate bolt to determine whether the seam plate bolt is loose or not. There is. In this method using tapping sound, since the discrimination is made by the subjectivity of each track maintenance staff, the criteria for discrimination are slightly different depending on the track maintenance staff, and there is a problem in accuracy. In addition, since the track maintenance staff goes to the place of the normal seam of all the rails and inspects the tapping sound, there is also a problem that the work time and cost involved in the movement are enormous. In order to eliminate the subjectivity of the track maintenance staff, it is conceivable to introduce an automatic discriminator, but the time and effort required for the track maintenance staff to go to the normal joints of all rails remains the same.

Furthermore, it is conceivable to install a sensor for detecting looseness at each of the ordinary joints of the rail to detect looseness remotely, but a huge number of sensors must be installed and it is detected remotely. The system for doing this is also costly and impractical.

On the other hand, if the looseness of the seam plate bolts can be determined from the running train, the work time and cost of the track maintenance staff can be reduced. A method for detecting looseness or the like is known. However, these techniques do not cover rail seam plate bolts.

Japanese Unexamined Patent Publication No. 5-247903 Japanese Unexamined Patent Publication No. 9-315304

Therefore, an object of the present invention is, according to one aspect, to provide a novel technique for determining looseness of a rail seam plate bolt by running a train.

The method for determining the looseness of the joint plate bolt of the rail according to the present invention is provided in (A) a vibration sensor for detecting the vibration of the axle box including the bearing of the axle of the wheel of the train, or the sound from the rail. There is a correlation between the step of acquiring vibration or acoustic data at the normal joint of the rail by the acoustic sensor that detects, and (B) noise removal processing and loosening of the joint plate bolt of the rail for the acquired vibration or acoustic data. A step of executing a feature extraction process for extracting a predetermined type of feature amount and (C) a process of determining whether or not the rail joint plate bolt is loose with respect to the extracted predetermined type of feature amount are executed. Including steps.

According to one side, it becomes possible to determine the looseness of the joint plate bolt of the rail by running the train.

FIG. 1 is a diagram showing an outline of ordinary joints of rails. FIG. 2 is a diagram for explaining the arrangement of the vibration sensor according to the first embodiment. FIG. 3 is a diagram showing an outline of the information processing apparatus according to the first embodiment. FIG. 4 is a diagram showing a configuration example of the feature extraction unit. FIG. 5 is a diagram showing a processing flow according to the embodiment. FIG. 6 is a diagram for explaining the arrangement of the acoustic sensor according to the second embodiment. FIG. 7 is a diagram showing an outline of the information processing apparatus according to the second embodiment.

[Embodiment 1]
First, FIG. 1 shows an outline of ordinary rail seams. The rail 1001 and the rail 1002 on the front side are connected by the seam plate bolts 1004a to 1004d with the seam plate 1003 applied to the abdomen of the rails. In the rails 1001 and 1002 on the front side, only the outside is shown, but the inside is also connected by applying a seam plate. The same applies to the rail on the back side. Such looseness of the seam plate bolt is determined by running the train.

The arrangement of the vibration sensor used in this embodiment will be described with reference to FIGS. 2 (a) and 2 (b). In this embodiment, the vibration sensor 101 is installed inside or outside either (or both) the axle boxes 2004a and 2004b used to fix the axles 2003 of the train wheels 2005a and 2005b to the bogie frame 2001. .. The axle box 2004a receives the axle 2003 by the roller bearing 20044 and is suspended from the bogie frame 2001 by the spring 2002a. Similarly, the axle box 2004b is suspended from the bogie frame 2001 by a spring 2002b. The inside of the axle box 2004b is the same as the inside of the axle box 2004a.

In the present embodiment, a vibration sensor 101 is provided in the axle box 2004a or 2004b to measure the left-right vibration of the axle box 2004a or 2004b while the train is running, so that the vibration of the seam plate 1003 caused by the running of the train is indirectly measured. It has become possible to measure the looseness of the seam plate bolt, which is the cause of the measurement. The left-right vibration is the left-right vibration in FIG. 2A, and is a vibration in the lateral direction with respect to the traveling direction of the train. The vibration sensor 101 is, for example, an acceleration sensor, but may be a sensor that measures the vibration speed or displacement.

FIG. 3 shows a configuration example of the information processing apparatus according to the embodiment of the present invention.

The information processing apparatus 100 includes a vibration sensor 101, an AD (Analog-to-Digital) conversion unit 103, a first data storage unit 105, a noise removal unit 107, a feature extraction unit 109, a discrimination unit 111, and an output. It has a part 113.

The vibration sensor 101 measures the left-right vibration of the axle box 2004a or 2004b while the train is running. The AD conversion unit 103 converts the measured vibration signal of the vibration sensor 101 into digital vibration data and stores it in the first data storage unit 105. The noise reduction unit 107 executes noise removal processing (for example, removal of components of 20 Hz or less) on the vibration data at each normal joint of the rail among the vibration data stored in the first data storage unit 105 to extract features. Output to unit 109. The feature extraction unit 109 extracts a feature amount having a correlation with the looseness of the joint plate bolt of the rail from the vibration data after noise removal, and outputs the feature amount to the discrimination unit 111. The discrimination unit 111 determines whether or not the joint plate bolt of the rail is loose based on the feature amount from the feature extraction unit 109, and outputs the discrimination result to the output unit 113. The output unit 113 stores the discrimination result in a storage device, outputs it to an output device such as a display device, or outputs it to another computer or the like.

Regarding the vibration data of the rail seam plate 1003, for example, the time interval of the normal seam of the rail is obtained from the arrangement of the rail length [m] and the traveling speed [m / s] in the traveling section stored in the rail length database. By calculating [s], the time position of the normal seam of the rail is estimated, and the vibration data at the normal seam of the rail from the entire vibration data (for example, the data of the predetermined time width before and after the time position of the normal seam of the rail). Cut out. After removing noise, the data at the normal joint of the rail may be cut out.

From the experiment, as a non-trivial and new finding, as a feature quantity that correlates with the looseness of the joint plate bolt of the rail, 1) the maximum amplitude of the left-right vibration of the axle box at the normal joint of the rail, and 2) the signal component of 20 Hz or more and 1 kHz or less. Energy, 3) The energy of the signal component of 1 kHz or more and 2 kHz or less was found. More specifically, in the experiment, for 1), the correlation coefficient with the looseness of the seam plate bolt was 0.90, and for 2), the correlation coefficient with the looseness of the seam plate bolt was -0.86. Regarding 3), the result that the correlation coefficient with the looseness of the seam plate bolt is 0.93 is obtained. The presence or absence of looseness of the seam plate bolt can be determined by using at least one of such feature amounts. In addition, another feature amount having a correlation with the looseness of the joint plate bolt of the rail (for example, a feature amount having a correlation coefficient of more than or equal to a positive threshold value or less than or equal to a negative threshold value) may be found and used.

From the results of such an experiment, the feature extraction unit 109 has, for example, a configuration as shown in FIG. That is, the feature extraction unit 109 has a peak value detection unit 1091, a first BPF (Band Pass Filter) unit 1092, and a second BPF unit 1093. The peak value detection unit 1091 detects the maximum amplitude (that is, the peak value) in the time waveform of the vibration data after the noise removal processing. The first BPF unit 1092 is a bandpass filter having a pass band of 20 Hz or more and 1 kHz or less, and extracts energy of a signal component in this band. Further, the second BPF unit 1093 is a bandpass filter having a pass band of 1 kHz or more and 2 kHz or less, and extracts energy of a signal component in this band.

Further, the discriminant unit 111 uses the actually measured value of such a feature amount and the actual presence / absence of loosening of the seam plate bolt, for example, by using a method of statistical linear discriminant analysis (for example, a support vector machine) or a neuronet in advance. It is a machine-learned discriminant unit.

Next, the processing content of the information processing apparatus 100 will be described with reference to FIG.

First, the vibration sensor 101 measures the left-right vibration (for example, acceleration in the left-right direction) of the axle box 2004a while the train is running, and the AD conversion unit 103 converts the left-right vibration signal into digital vibration data, and the first data. It is stored in the storage unit 105 (step S1).

Next, the noise removing unit 107 executes a noise removing process for removing, for example, a frequency component of 20 Hz or less with respect to the vibration data at each normal joint of the rail among the vibration data stored in the first data storage unit 105. (Step S3).

After that, the feature extraction unit 109 executes the feature extraction process for each vibration data after the noise removal process (step S5). Specifically, the peak value detection unit 1091 extracts the maximum amplitude value from each vibration data. Further, the first BPF unit 1092 extracts the energy of the signal component in the frequency band of 20 Hz or more and 1 kHz or less from each vibration data. The second BPF unit 1093 extracts the energy of the signal component in the frequency band of 1 kHz or more and 2 kHz or less for each vibration data. Then, these feature quantities for each vibration data are output to the discrimination unit 111.

Then, the discriminating unit 111 executes a discriminating process as to whether or not the seam plate bolt is loose based on the feature amount for each vibration data (that is, each normal seam of the rail) (step S7).

Then, the output unit 113 outputs the discrimination result of the discrimination unit 111 to the output device (step S9). As a result, for example, a sequence of discrimination results indicating whether or not the seam plate bolt is loose can be obtained. The determination result may be stored in the data storage unit or may be transmitted to another computer.

Looking at such a determination result, it becomes possible to identify a normal joint of a rail in which a loosening of a seam plate bolt occurs in a traveling section of a train. Therefore, it is only necessary to dispatch a track maintenance worker to the place to retighten the seam plate bolts, so that the maintenance cost can be reduced. It is also possible to eliminate the subjectivity of track maintenance personnel. Furthermore, since the vibration sensor 101 may be installed in the axle box 2004a (or 2004b or both) of the train to run the train, it is not necessary to install a sensor at each normal joint of the rail and make a determination by a remote system. System costs can also be reduced. An example is shown in which the vibration sensor 101 measures the left-right vibration of the axle box, but in some cases, the vertical vibration (vertical vibration with respect to the traveling direction of the train) in FIG. 2 (a) and FIG. 2 ( In a), front-back vibration (vibration parallel to the traveling direction of the train), vibration along a specific axis, etc. may become vibrations that are more closely correlated with loosening of the seam plate bolts, and when measuring them. There is also.

[Embodiment 2]
In the first embodiment, an example in which the vibration sensor 101 is used is shown, but the vibration of the joint plate 1003 of the rail may be directly obtained as an acoustic signal by an acoustic sensor (that is, a microphone).

As shown in FIG. 6, for example, an acoustic sensor 151 having unidirectionality in the rail direction (more specifically, the direction of the inner rail abdomen) is provided at an appropriate position under the floor of the train. Then, the acoustic sensor 151 collects an acoustic signal generated by the vibration of the rail seam plate 1003.

The information processing apparatus 100b according to the present embodiment has a configuration as shown in FIG. 7. The basic configuration is the same as that of the information processing apparatus 100 according to the first embodiment, except that the acoustic sensor 151 is adopted instead of the vibration sensor 101 and the noise is removed instead of the noise removing unit 107. The point is that the portion 107b is adopted.

The noise removing unit 107b is, for example, a high-pass filter having a cutoff frequency of 400 Hz.

Since the feature amount to be extracted from the vibration data by the feature extraction unit 109 is the same in that the vibration of the rail seam plate 1003 is detected, the feature amount is similar to that of the first embodiment. From the experiment, as for the peak value, the correlation coefficient with the looseness of the seam plate bolt is 0.75, so the peak value detection unit 1091 is adopted as it is. In addition to this, a feature amount that correlates with the looseness of the seam plate bolt can be adopted. For example, since the noise removing unit 107b is a 400 Hz high-pass filter, the frequency band may differ, but the first BPF unit 1092 and the second BPF unit 1093 may be adopted.

Although the embodiments of the present invention have been described above, the present invention is not limited thereto. For example, the functional block configuration of FIGS. 3, 4, and 7 is an example and may not match the program module configuration. The processing flow shown in FIG. 5 and the like may also include steps that may be executed in parallel or the order may be changed as long as the processing results are the same.

Further, the information processing device 100 may be implemented as one device or as a plurality of devices. For example, the vibration sensor 101 or the acoustic sensor 151, the AD conversion unit 103, and the first data storage unit 105 may be mounted on the train, and other components may be arranged in other places. As described above, the information processing device 100 or 100b may be constructed as an information processing system realized by one or a plurality of devices, and may be referred to as an information processing system including the case of one device.

The information processing devices 100 and 100b described above are computer devices, and include a memory, a CPU (Central Processing Unit), a hard disk drive (HDD), and a display control unit connected to the display device. The drive device for the removable disk, the input device, and the communication control unit for connecting to the network are connected by a bus. The operating system (OS: Operating System) and the application program for performing the processing in this embodiment are stored in the HDD, and are read from the HDD to the memory when executed by the CPU. The CPU controls the display control unit, the communication control unit, and the drive device according to the processing contents of the application program to perform a predetermined operation. Further, the data in the process of being processed is mainly stored in the memory, but may be stored in the HDD. In the embodiment of the present invention, the application program for performing the above-described processing is stored in a computer-readable removable disk, distributed, and installed from the drive device to the HDD. It may be installed in the HDD via a network such as the Internet and a communication control unit. Such a computer device realizes various functions as described above by organically cooperating with the hardware such as the CPU and the memory described above and the program such as the OS and the application program.

The above-described embodiments of the present embodiment can be summarized as follows.

The method for determining the looseness of the joint plate bolt of the rail according to the present embodiment is provided in (A) a vibration sensor for detecting the vibration of the axle box including the bearing of the axle of the wheel of the train, or from the rail. The step of acquiring vibration or acoustic data at the normal joint of the rail by the acoustic sensor that detects the sound of (B) and the noise removal processing and loosening of the joint plate bolt of the rail for the acquired vibration or acoustic data. A step of executing a feature extraction process for extracting a certain type of feature amount, and (C) a process of determining whether or not the rail joint plate bolt is loose with respect to the extracted predetermined type of feature amount. Includes steps to perform.

If the train is driven by adopting such a vibration sensor or an acoustic sensor, the above-mentioned feature amount can be obtained, and it becomes possible to determine whether or not the seam plate bolt is loose based on the feature amount. .. In other words, it is possible to eliminate the subjectivity of the track maintenance committee and reduce costs.

The predetermined type of features described above may be at least one of the maximum amplitude of vibration or acoustic data after the denoising process and the energy of the signal component in the predetermined frequency band. The effectiveness of such features has been non-trivially and newly discovered by the inventor of the present application.

The energy of the signal component in the predetermined frequency band may include at least one of the energy of the signal component of 20 Hz or more and 1 kHz or less and the energy of the signal component of 1 kHz or more and 2 kHz or less. The effectiveness of such features is also non-trivial and novelly discovered by the inventor of the present application.

A program for executing the above processing can be created, and the program can be read by a computer such as a flexible disk, an optical disk (CD-ROM, DVD-ROM, etc.), a magneto-optical disk, a semiconductor memory, a hard disk, or the like. It is stored in a storage medium or storage device. The intermediate processing result is temporarily stored in a storage device such as a main memory.

100,100b Information processing device 101 Vibration sensor 103 AD conversion unit 105 First data storage unit 107, 107b Noise removal unit 109 Feature extraction unit 111 Output unit 151 Acoustic sensor

Claims (5)

A step of acquiring vibration or acoustic data at a normal seam of a rail by a vibration sensor for detecting the vibration of a shaft box including a bearing of an axle of a wheel of a train or an acoustic sensor provided on the train and detecting a sound from a rail. When,
A step of executing a noise removal process for the acquired vibration or acoustic data and a feature extraction process for extracting a predetermined type of feature amount having a correlation with the looseness of the joint plate bolt of the rail.
A step of executing a process of determining whether or not the joint plate bolt of the rail is loose with respect to the extracted feature amount of the predetermined type, and
A program to make a computer execute. The predetermined type of feature amount is
The program according to claim 1, wherein at least one of the maximum amplitude of the vibration or acoustic data after the noise removal processing and the energy of the signal component in a predetermined frequency band. The energy of the signal component in the predetermined frequency band
The program according to claim 2, wherein the program includes at least one of the energy of a signal component of 20 Hz or more and 1 kHz or less and the energy of a signal component of 1 kHz or more and 2 kHz or less. A step of acquiring vibration or acoustic data at a normal seam of a rail by a vibration sensor for detecting the vibration of a shaft box including a bearing of an axle of a wheel of a train or an acoustic sensor provided on the train and detecting a sound from a rail. When,
A step of executing a noise removal process for the acquired vibration or acoustic data and a feature extraction process for extracting a predetermined type of feature amount having a correlation with the looseness of the joint plate bolt of the rail.
A step of executing a process of determining whether or not the joint plate bolt of the rail is loose with respect to the extracted feature amount of the predetermined type, and
A determination method performed by a computer, including. A means for acquiring vibration or acoustic data at a normal seam of a rail by a vibration sensor for detecting the vibration of a shaft box including a bearing of an axle of a wheel of a train or an acoustic sensor provided on the train and detecting a sound from a rail. When,
A means for performing noise removal processing on the acquired vibration or acoustic data and feature extraction processing for extracting a predetermined type of feature amount having a correlation with looseness of the joint plate bolt of the rail.
A means for executing a process of determining whether or not the joint plate bolt of the rail is loose with respect to the extracted feature amount of the predetermined type.
Information processing system with.

Images