JP7425430B2 - Method and program for specifying the time position of a normal joint of a rail, and an information processing system - Google Patents

Description

The present invention relates to a technique for specifying the temporal position of a regular rail joint from the sound of a train running.

In Patent Document 1, beams in at least two different directions are emitted through the rail of a railway track, and at least one ultrasonic beam is emitted into the air near the abdomen of the rail, and the received signals are compared. A technique for detecting discontinuities or abnormalities in rails is disclosed. Such technology requires a device that can transmit and receive two ultrasound beams.

In addition, in Patent Document 2, the train position is specified from a GPS (Global Positioning System) signal, etc., but waveform characteristics are extracted from vertical vibration signals from an acceleration sensor and acoustic signals from a sound collector, and a waveform pattern prepared in advance is used. A technique has been disclosed in which it is determined whether the train has passed through a feature point such as a rail joint or a branch by comparing the trains, and the train position for detecting the passing of the feature point is corrected. In such a technique, waveform patterns corresponding to feature points such as rail joints and branching parts must be prepared in advance. Also, various sensors must be prepared.

Japanese Unexamined Patent Publication No. 61-501724 Japanese Patent Application Publication No. 2011-225188

Therefore, according to one aspect, an object of the present invention is to provide a technique that enables the time position of a normal joint of a rail to be specified with high precision with a simple configuration.

The method for specifying the time position of a normal rail joint according to the present invention includes (A) performing an envelope extraction process on the running sound of a train after noise removal, and using the extracted envelope to determine the time position of a normal rail joint. (B) determining the hypothetical time position of the normal joint sound of the rail from the rail length data stored in the storage unit that stores the rail length data in the running section of the train; and (C) identifying the time position of the sound of the normal joint of the rail by comparing the candidate time position with the hypothetical time position.

According to one aspect, the time position of a normal joint of a rail can be accurately specified with a simple configuration.

FIG. 1 is a diagram showing an overview of an information processing apparatus according to an embodiment. FIG. 2 is a diagram illustrating a configuration example of a hypothesis generation unit according to the first embodiment. FIG. 3 is a diagram showing a processing flow according to the embodiment. FIG. 4 is a diagram showing an example of the process of processing sound data in this embodiment. FIG. 5 is a diagram showing a processing flow of hypothetical time position generation processing according to the first embodiment. FIG. 6 is a diagram showing an example of the process of processing sound data in this embodiment. FIG. 7 is a diagram showing a processing flow of matching processing according to the first embodiment. FIG. 8 is a diagram illustrating a configuration example of a hypothesis generating section and a matching section according to the second embodiment. FIG. 9 is a diagram showing a processing flow of hypothetical time position generation processing according to the second embodiment. FIG. 11 is a diagram showing a processing flow of matching processing according to the second embodiment.

[Embodiment 1]
FIG. 1 shows a configuration example of an information processing apparatus according to an embodiment of the present invention.

The information processing device 100 includes a microphone 101, an AD (Analog-to-Digital) conversion section 103, a first data storage section 105, a noise removal section 107, an envelope extraction section 109, a candidate extraction section 111, It has a second data storage section 113, a collation section 115, an output section 117, a rail length database (DB) 119, a hypothesis generation section 121, and a third data storage section 123.

The microphone 101 collects running sounds on the train while the train is running. The running noise referred to here is the combination of all the sounds that occur as the train runs, such as the sound of pantographs, the sound of bogies, the sound of the car body, wind noise, and the sound generated from bolts in ordinary joints. It refers to the sound that is observed. The AD conversion section 103 converts the sound input signal collected by the microphone 101 into digital sound data, and stores the digital sound data in the first data storage section 105 . Noise removal section 107 performs noise removal processing on the sound data stored in first data storage section 105 and outputs it to envelope extraction section 109 . The envelope extraction unit 109 performs envelope extraction processing on the noise-removed sound data and outputs it to the candidate extraction unit 111. The candidate extraction unit 111 extracts a candidate time position of the sound of the normal joint of the rail by comparing the sound data after the envelope extraction with a predetermined threshold, and stores the data of the candidate time position in the second data storage unit 113. Store. Note that a normal joint is a butt joint in which a pair of joint plates are placed against the rail abdomen, assembled with joint plate bolts, and the ends of the rails are connected.

In this embodiment, the rail length DB 119 stores data on a series of rail lengths in the section where the train runs. For example, data indicating that a Ym rail is lined up next to an Xm rail is stored. The hypothesis generation unit 121 generates data on the hypothetical time position of the sound of the normal joint of the rail using the data stored in the rail length DB 119, and stores it in the third data storage unit 123. In the present embodiment, the hypothesis generation unit 121 further uses data on candidate time positions of the sounds of normal rail joints, which is stored in the second data storage unit 113. However, in the second embodiment, the data in the second data storage section 113 is not used.

The collation unit 115 compares the candidate time position data of the sound of a normal rail joint stored in the second data storage unit 113 and the data of the sound of a normal rail joint stored in the third data storage unit 123. The correct time position of the sound of the normal joint of the rail is determined by comparing the data with the hypothetical time position data. The output unit 117 performs a process of outputting the processing results of the collation unit 115 to a display device, a printing device, or another device.

FIG. 2 shows a configuration example of the hypothesis generation unit 121 according to the present embodiment. The hypothesis generating section 121 includes an interval estimating section 1211, a speed estimating section 1212, a rail length extracting section 1213, and a hypothesis calculating section 1214.

The interval estimating unit 1211 executes a process of estimating a representative time interval from the intervals of candidate time positions stored in the second data storage unit 113, and outputs it to the speed estimating unit 1212. The rail length extraction unit 1213 extracts a rail length representative of the travel section from the rail length DB 119 and outputs it to the speed estimation unit 1212. The speed estimation unit 1212 estimates the speed of the train from the representative time interval and the representative rail length, and outputs it to the hypothesis calculation unit 1214.

The hypothesis calculating unit 1214 calculates a hypothetical time position of the sound of the normal rail joint from the estimated train speed and the rail length data stored in the rail length DB 119, and stores the data in the third data storage unit 123. Store in.

Next, the processing contents of the information processing apparatus 100 will be explained using FIGS. 3 to 6.

First, the sound of a running train is recorded on the train by the microphone 101, and the output signal of the microphone 101 is converted into digital sound data by the AD conversion section 103 and stored in the first data storage section 105 (step S1). Note that in addition to A/D conversion, other processing such as amplification may be performed if necessary. Further, although it is necessary to record the running sound of a train on the train, the following processing may be performed at a location other than the train. Note that the sound data obtained in step S1 is, for example, data as shown in FIG. 4(a), and contains a lot of noise, making it impossible to find the sound waveform of the normal joint of the rail.

Next, the noise removal unit 107 performs noise removal processing on the sound data stored in the first data storage unit 105 (step S3). Specifically, low frequency noise is removed using a high pass filter having a cutoff frequency of 400 Hz, for example. For example, for sound data as shown in FIG. 4(a), sound data as shown in FIG. 4(b) is obtained by noise removal processing. In this way, the waveform of the sound at the normal joint of the rail is sharpened.

The envelope extraction unit 109 then performs envelope extraction processing on the noise-removed sound data (step S5). The envelope of a waveform is obtained by converting the waveform into a rectified waveform by taking the absolute value, and smoothing the rectified waveform by applying a low-pass filter to the rectified waveform. Through such envelope extraction processing, a waveform envelope as shown in FIG. 4(c) is obtained.

Thereafter, the candidate extracting unit 111 compares a preset threshold with the waveform envelope, and identifies a time position at which the threshold is greater than or equal to the threshold as a candidate time position for the sound of a normal rail joint (step S7). When a threshold value is set for the waveform envelope as shown in FIG. 4(d), the time position is specified by a pulse as shown in FIG. 4(e). This is a candidate time position, and as shown by the upward arrow in Fig. 4(e), even sounds caused by track materials other than ordinary joints of the rail (hereinafter referred to as sounds from other than ordinary joints) are candidates. It will be specified as a time position. Normally, sounds from outside the seam cannot be identified by adjusting the threshold.

Therefore, in the present embodiment, the hypothesis generating unit 121 uses the rail length DB 119 to execute a hypothetical time position generation process that generates data on the hypothetical time position of the sound of the normal joint of the rail (step S9).

The hypothetical time position generation process according to this embodiment will be explained using FIG. 5.

The interval estimation unit 1211 calculates the time intervals of the candidate time positions stored in the second data storage unit 113, and specifies the median value among them as a representative time interval (step S21). Moreover, the rail length extraction unit 1213 specifies the median value as a representative rail length among a series of rail lengths of travel sections stored in the rail length DB 119 (step S23).

Then, the speed estimation unit 1212 calculates the estimated train speed by dividing the representative rail length by the representative time interval (step S25). Further, the hypothesis calculation unit 1214 sequentially reads the rail length [m] of the running section from the rail length DB 119 and divides each by the estimated train speed [m/s], thereby calculating the running time [s] of each rail. By arranging the calculated data and arranging them, data on the hypothetical time position of the sound of the normal joint of the rail is generated and stored in the third data storage unit 123 (step S27).

FIG. 6(a) shows the candidate time position of the sound of the normal joint of the rail, which is the same as FIG. 4(e). FIG. 6(b) schematically shows the rail length of the traveling section at pulse intervals. By dividing each rail length by the estimated train speed, the traveling time for each rail is estimated, and by arranging them, a pulse train as shown in FIG. 6(c) is obtained. The position of this pulse represents the hypothetical time position of the sound of the normal joint of the rail.

Returning to the explanation of the process in FIG. 3, the collation unit 115 performs a collation process on the data stored in the second data storage unit 113 and the data stored in the third data storage unit 123 (step S11). FIG. 7 shows a processing flow of the matching process according to this embodiment.

The matching unit 115 compares each candidate time position stored in the second data storage unit 113 with the hypothetical time position stored in the third data storage unit 123 (step S31), and determines which corresponds to the candidate time position. If there is no hypothetical time position, the sound is assumed to be coming from a place other than the normal joint, and the candidate time position is removed, thereby specifying the time position of the sound at the normal joint of the rail from the candidate time position (step S35). Whether or not they correspond may be determined by, for example, whether the difference between each candidate time position and the hypothetical time position is within a predetermined time.

Comparing each candidate time position in Fig. 6(a) with the hypothetical time position in Fig. 6(c), it is found that the pulse at the hypothetical time position corresponding to the pulse at the candidate time position generated in the track material other than the normal joint of the rail is Since it does not exist, by removing the pulse at this candidate time position, the time position of the sound of the normal joint of the rail as shown in FIG. 6(d) can be specified.

Returning to the explanation of the processing in FIG. 3, the output unit 117 outputs the processing result of the matching unit 115 to the output device (step S13). For example, data as shown in FIG. 6(d) is output.

In this way, it is possible to remove sounds from areas other than normal joints that would otherwise be mixed in by simply comparing the waveform envelope and the threshold, making it possible to accurately identify the time position of the sound at normal rail joints. It becomes like this.

When implementing railway projects, rail maintenance is important to ensure the safety of railway operations. When maintaining rails, if the amount of play in regular rail joints is large, the impact that is applied to the rails when a train passes becomes large, so regular rail joints become weak points in the track. Therefore, it is especially important to understand the condition of normal rail joints. A method of diagnosing the condition of each rail joint from the passing sound from a running train is strongly desired, but according to the present embodiment, the time position of the sound of the normal joint of the rail can be correctly identified, which improves efficiency. This leads to good inspection implementation and appropriate inspection decisions.

[Embodiment 2]
In this embodiment, an example will be shown in which the hypothesis generation unit 121 performs processing without using the data of candidate time positions stored in the second data storage unit 113.

In this embodiment, a hypothesis generation section 121b is used instead of the hypothesis generation section 121, and a comparison section 115b is used instead of the comparison section 115.

FIG. 8 shows an example of the configuration of the hypothesis generation section 121b and the matching section 115b.

The hypothesis generation section 121b includes a speed candidate generation section 1215 and a hypothesis pattern generation section 1216. The speed candidate generation unit 1215 generates a plurality of speed candidates at predetermined speed ranges and intervals. The hypothesis pattern generation unit 1216 generates a hypothesis time position pattern for each speed candidate from a series of rail lengths stored in the rail length DB 119, and stores it in the third data storage unit 123.

The matching unit 115b includes a deviation calculating unit 1151 and a selecting unit 1152. The deviation calculation unit 1151 calculates the deviation between the candidate time position data stored in the second data storage unit 113 and the plurality of hypothetical time position patterns stored in the third data storage unit 123. The selection unit 1152 selects the hypothetical time position pattern corresponding to the smallest deviation among the plurality of deviations calculated by the deviation calculation unit 1151, and employs it as the time position of the sound of the normal joint of the rail.

More specifically, the hypothetical time position generation process according to this embodiment is a process as shown in FIG. That is, the speed candidate generation unit 1215 generates a plurality of speed candidates based on the settings (step S41). For example, if a speed of 30 km/h to 45 km/h is assumed in the travel section, speed candidates are generated at 1 km/h intervals. That is, a plurality of speed candidates such as 30 km/h, 31 km/h, 32 km/h, . . . 44 km/h, 45 km/h are generated. Ranges and intervals are based on settings.

Then, the hypothesis pattern generation unit 1216 generates a plurality of hypothesis time position patterns from a series of rail lengths and a plurality of speed candidates in the travel section stored in the rail length DB 119, and stores them in the third data storage unit 123. (Step S43). That is, by dividing a series of rail lengths by a certain speed candidate, the traveling time of each rail can be obtained, and by arranging them, one hypothetical time-position pattern can be obtained. By performing this for a plurality of velocity candidates, a plurality of hypothetical time-position patterns are generated. A pulse train as shown in FIG. 6(c) is generated for each speed candidate. Note that the larger the value of the speed candidate, the narrower the pulse interval is generated, and the smaller the value of the speed candidate, the wider the pulse train is generated.

Further, the matching process according to this embodiment is a process as shown in FIG.

The deviation calculation unit 1151 uses dynamic programming for the candidate time position data stored in the second data storage unit 113 and each hypothetical time position pattern stored in the third data storage unit 123. By performing nonlinear expansion/contraction matching, a deviation from the candidate time position is calculated for each hypothetical time position pattern (step S51). Since nonlinear expansion/contraction matching is well known, a detailed explanation of the algorithm will be omitted.

The selection unit 1152 specifies the hypothetical time position pattern that causes the smallest deviation by comparing the deviations for each hypothetical time position pattern calculated by the deviation calculation unit 1151, and selects the hypothetical time position pattern from the rail. It is selected as data representing the time position of the normal joint sound (step S53).

Such processing also makes it possible to specify the time position of the sound at the normal joint of the rail with high accuracy.

Although the embodiments of the present invention have been described above, the present invention is not limited thereto. For example, the functional block configurations in FIGS. 1, 2, and 7 are just examples, and may not match the program module configurations. The processing flow shown in FIG. 3 and the like may also include steps that may be executed in parallel or whose order may be changed as long as the processing results are the same.

Further, the information processing device 100 may be implemented as a single device, or may be implemented as a plurality of devices. For example, the microphone 101, the AD conversion unit 103, and the first data storage unit 105 may be mounted on the train, and the other components may be placed elsewhere. In this way, the information processing device 100 may be constructed as an information processing system realized by one or more devices, and may be referred to as an information processing system including the case of one device.

The information processing device 100 described above is a computer device that includes a memory, a CPU (Central Processing Unit), a hard disk drive (HDD), a display control unit connected to a display device, and a removable computer. A disk drive device, an input device, and a communication control unit for connecting to a network are connected via a bus. An operating system (OS) and application programs for implementing 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 section, communication control section, and drive device to perform predetermined operations according to the processing content of the application program. Furthermore, data that is being processed is mainly stored in the memory, but may also be stored in the HDD. In an embodiment of the present invention, an application program for implementing the above-described processing is stored and distributed on a computer-readable removable disk, and is installed on the HDD from the drive device. It may also be installed on the HDD via a network such as the Internet and a communication control unit. Such a computer device realizes the various functions described above through organic cooperation between hardware such as the CPU and memory described above, and programs such as the OS and application programs.

The present embodiment described above can be summarized as follows.

The method of specifying the time position of a normal joint of a rail according to the present embodiment is as follows: (A) Execute an envelope extraction process for the running sound of a train after noise removal, and use the extracted envelope to (B) Identifying the candidate time position of the sound of the normal joint of the rail from the rail length data stored in the storage unit that stores the data of the rail length in the running section of the train. The method includes a generation step of generating a hypothetical time position of the sound, and (C) a second specifying step of comparing the candidate time position and the hypothetical time position to specify the time position of the sound of the normal joint of the rail.

By utilizing the rail length data in this way, it is possible to effectively exclude sounds from areas other than the normal joints of the rails, and it becomes possible to accurately identify the time position of the sound at the normal joints of the rails.

The generation step described above consists of (b1) identifying a train speed representative of the running section from the candidate time position data and rail length data, and (b2) identifying the normal rail speed from the rail length data and train speed. and generating a hypothetical temporal position of the sound of the seam.

If we can identify the train speed that is representative of the traveling section in this way, we can calculate the time interval of the sound at the ordinary joint of the rail from a series of rail lengths, so we can calculate the time position of the sound at the ordinary joint of the rail in an ideal state. Be able to identify. Note that from the candidate time position data, an interval representing the candidate time position (for example, the median value; however, other statistics may be used), and from the rail length data, a length representing the rail length (for example, the median value). However, the train speed may be calculated from other statistics.

Further, the generation step described above may include (b3) a step of generating a plurality of patterns of hypothetical time positions of the sound of the normal joint of the rail from the rail length data and the plurality of candidate speeds. . In that case, the second specifying step described above (c1) calculates the deviation between each of the plurality of patterns of the candidate time position and the hypothetical time position, and selects one of the plurality of patterns based on the deviation of the plurality of patterns. It may also include a step of selecting.

In this way, it is possible to generate multiple patterns of hypothetical time positions based on multiple possible candidate speeds without using candidate time positions to specify the train speed that represents the running section. Be able to identify the time position of sounds.

Note that in the first specifying step described above, the candidate time position may be specified by comparing the envelope with a predetermined threshold. Since comparison with such a simple threshold value cannot accurately extract only the time position of the sound at the normal joint of the rail, the above-described processing is performed.

Note that a program can be created to execute the above processing, and the program can be read by a computer such as a flexible disk, optical disk (CD-ROM, DVD-ROM, etc.), magneto-optical disk, semiconductor memory, hard disk, etc. It is stored in a storage medium or storage device. Note that intermediate processing results are temporarily stored in a storage device such as a main memory.

100 Information processing device 101 Microphone 103 AD conversion section 105 First data storage section 107 Noise removal section 109 Envelope extraction section 111 Candidate extraction section 113 Second data storage section 115, 115b Collation section 117 Output section 119 Rail length DB
121, 121b Hypothesis generation unit 123 Third data storage unit

Claims (5)

a first identifying step of performing an envelope extraction process on the running sound of the train after noise removal, and identifying a candidate time position of the sound of the normal joint of the rail from the extracted envelope;
a generation step of generating a hypothetical time position of a sound at a normal joint of the rail from data on the rail length stored in a storage unit that stores data on the rail length in the section where the train runs;
a second identifying step of identifying the time position of the sound of the normal joint of the rail by comparing the candidate time position and the hypothetical time position;
A program that causes a computer to execute. The generation step includes:
identifying a train speed representative of the traveling section from the candidate time position data and the rail length data;
generating from the rail length data and the train speed a hypothetical time position of a sound at a normal joint of the rail;
The program according to claim 1, comprising: The generation step includes:
generating a plurality of patterns of hypothetical time positions of sounds at normal joints of the rail from the rail length data and a plurality of candidate speeds;
The second specifying step is
2. The program according to claim 1, comprising: calculating a deviation between each of the plurality of patterns of the candidate time position and the hypothetical time position, and selecting one of the plurality of patterns based on the deviation between the plurality of patterns. . Executing an envelope extraction process on the running sound of the train after noise removal, and identifying a candidate time position of the sound of the normal joint of the rail from the extracted envelope;
Generating a hypothetical time position of the sound of the normal joint of the rail from the rail length data stored in a storage unit that stores rail length data in the running section of the train;
identifying the time position of the sound of the normal joint of the rail by comparing the candidate time position and the hypothetical time position;
A computer-implemented method for determining the temporal position of a common rail joint, comprising: Executing an envelope extraction process on the running sound of the train after noise removal, and specifying a candidate time position of the sound of a normal rail joint from the extracted envelope;
Means for generating a hypothetical time position of a sound at a normal joint of the rail from data on the rail length stored in a storage unit that stores data on the rail length in the section where the train runs;
means for identifying the time position of the sound of the normal joint of the rail by comparing the candidate time position and the hypothetical time position;
An information processing system with

Images