Railway acoustic wave data stitching method

Abstract

A railway acoustic wave data stitching method, comprising the following steps: step 1: acquiring an acoustic wave signal of a railway; step 2: acquiring a set of valid acoustic wave signal fragments on the basis of the acoustic wave signal and an acoustic wave detection model, and arranging the valid acoustic wave signal fragments in the order of acquisition time to obtain acoustic wave signal fragments to be stitched; and step 3: stitching, in chronological order, the acoustic wave signal fragments to be stitched, to obtain an acoustic wave signal of a complete cycle. The method provides complete and clean acoustic wave data for railway condition diagnosis, thereby improving the recognition rate of railway condition diagnosis.

Description

A railway acoustic wave data splicing method Technical Field

[0001] The invention belongs to the field of railway detection, relates to an acoustic wave data splicing technology, and specifically is a railway acoustic wave data splicing method. Background Art

[0002] By using the acoustic wave test method to test the existing railway subgrade, and combining the results of on-site geological surveys and portable variable-energy dynamic penetration tests, a rapid evaluation of the railway subgrade bearing capacity can be made, providing a fast and effective testing method for existing line subgrade inspections.

[0003] When using the acoustic wave testing method to test railway subgrade, it is necessary to input pre-processed waveform data into the fault diagnosis program to make a comprehensive diagnosis of the railway condition. Whether complete and "clean" acoustic wave data can be provided for railway diagnosis is the key to ensuring the railway condition diagnosis recognition rate.

[0004] Therefore, a railway acoustic wave data splicing method is proposed.

[0005] Summary of the Invention

[0006] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a railway acoustic wave data splicing method, which solves the problem of providing complete and "clean" acoustic wave data for railway diagnosis and ensuring the recognition rate of railway status diagnosis.

[0007] To achieve the above-mentioned object, according to an embodiment of the first aspect of the present invention, a railway acoustic wave data splicing method is provided, comprising the following steps:

[0008] Step 1: collecting railway sound wave signals; wherein the sound wave signals include a first sound wave signal, a second sound wave signal, a third sound wave signal, and a fourth sound wave signal;

[0009] Step 2: obtaining a set of valid sound wave signal segments based on the sound wave signal and the sound wave detection model; wherein the sound wave detection model is established based on an artificial intelligence model;

[0010] Arrange the valid sound wave signal segments in the order of acquisition time to obtain the sound wave signal segments to be spliced;

[0011] Step 3: Splice the sound wave signal segments to be spliced ​​in chronological order to obtain a sound wave signal of a complete cycle.

[0012] Preferably, step one further includes:

[0013] a first microphone, a second microphone, a third microphone, and a fourth microphone;

[0014] The first microphone is used to collect a first sound wave signal;

[0015] The second microphone is used to collect a second sound wave signal;

[0016] The third microphone is used to collect a third sound wave signal;

[0017] The fourth microphone is used to collect a fourth sound wave signal.

[0018] Preferably, the first microphone is arranged on one side of the railway;

[0019] The second microphone is arranged on the same side as the first microphone, and the first microphone is separated by s centimeters;

[0020] The third microphone is arranged on the other side of the railway and is on the same horizontal line as the first microphone;

[0021] The fourth microphone is arranged on the other side of the railway and is on the same horizontal line as the second microphone.

[0022] Preferably, obtaining a set of valid sound wave signal segments according to the sound wave signal and the sound wave detection model comprises the following steps:

[0023] Acquire an acoustic wave interception model; wherein the acoustic wave interception model is established based on an artificial intelligence model;

[0024] The first sound wave signal, the second sound wave signal, the third sound wave signal and the fourth sound wave signal are respectively input into the sound wave interception model to obtain a valid sound wave signal segment set; wherein the valid sound wave signal segment set includes a first valid sound wave signal segment set, a second valid sound wave signal segment set, a third valid sound wave signal segment set and a fourth valid sound wave signal segment set.

[0025] Preferably, establishing the acoustic wave detection model based on the artificial intelligence model includes the following steps:

[0026] Obtain standard training data;

[0027] The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as an acoustic wave detection model;

[0028] Artificial intelligence models include deep convolutional neural network models and RBF neural network models.

[0029] Preferably, arranging the valid sound wave signal segments in the order of acquisition time to obtain the sound wave signal segments to be spliced ​​includes the following steps:

[0030] Calculating and obtaining the signal strength of each valid sound wave signal segment in the valid sound wave signal segment set;

[0031] Arrange the effective sound wave signal segments in the order of acquisition time;

[0032] Marking valid sound wave segments in the acquisition time period as sound wave signal segments to be spliced;

[0033] When two or more valid sound wave segments appear in the same acquisition time period, the valid sound wave segment with the largest signal strength is marked as the sound wave signal segment to be spliced;

[0034] If no valid sound wave segment appears during the collection period, the collection task fails and the sound wave signal is collected again.

[0035] Preferably, in step three, a time domain smoothing method is used to perform splicing processing on the sound wave signal segments to be spliced.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] The present invention collects sound wave signals through a microphone; obtains a sound wave interception model; inputs the sound wave signal into the sound wave interception model to obtain a set of valid sound wave signal segments; calculates the signal strength of each valid sound wave signal segment in the set of valid sound wave signal segments; arranges the valid sound wave signal segments in the order of acquisition time; marks the valid sound wave segments in the acquisition time period as sound wave signal segments to be spliced; when two or more valid sound wave segments appear in the same acquisition time period, marks the valid sound wave segment with the largest signal strength as the sound wave signal segment to be spliced; if no valid sound wave segment appears in the acquisition time period, re-collects the sound wave signal; uses a time domain smoothing method to splice the sound wave signal segments to obtain a complete cycle of sound wave signals; provides complete and clean sound wave data for railway status diagnosis, and improves the recognition rate of railway status diagnosis. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] FIG1 is a flow chart of the present invention. DETAILED DESCRIPTION

[0039] The following is a clear and complete description of the technical solutions of the present invention in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0040] As shown in FIG1 , a railway acoustic wave data splicing method includes the following steps:

[0041] Step 1: collecting four acoustic wave signals of the railway; wherein the acoustic wave signals include a first acoustic wave signal, a second acoustic wave signal, a third acoustic wave signal, and a fourth acoustic wave signal;

[0042] In this embodiment, step one further includes four microphones; wherein the microphones include a first microphone, a second microphone, a third microphone, and a fourth microphone;

[0043] The first microphone is used to collect a first sound wave signal;

[0044] The second microphone is used to collect a second sound wave signal;

[0045] The third microphone is used to collect a third sound wave signal;

[0046] The fourth microphone is used to collect a fourth sound wave signal;

[0047] Specifically, the first microphone is arranged on one side of the railway;

[0048] The second microphone is placed on the same side as the first microphone, and the first microphone is separated by s centimeters. It has been verified that when s is 50, more accurate experimental results can be obtained.

[0049] The third microphone is arranged on the other side of the railway and is on the same horizontal line as the first microphone;

[0050] The fourth microphone is set on the other side of the railway and is on the same horizontal line as the second microphone. It should be further explained that the third microphone and the fourth microphone are also separated by s centimeters.

[0051] Step 2: Processing the sound wave signal to obtain valid sound wave signal segments;

[0052] In this embodiment, processing the sound wave signal to obtain a valid sound wave signal segment includes the following steps:

[0053] Acquire an acoustic wave interception model; wherein the acoustic wave interception model is established based on an artificial intelligence model;

[0054] The first sound wave signal, the second sound wave signal, the third sound wave signal and the fourth sound wave signal are respectively input into the sound wave interception model to obtain a valid sound wave signal segment set; wherein the valid sound wave signal segment set includes a first valid sound wave signal segment set, a second valid sound wave signal segment set, a third valid sound wave signal segment set and a fourth valid sound wave signal segment set; it should be further explained that the sound wave interception model is used to intercept valid sound wave segments in the sound wave signal and filter out invalid sound wave segments;

[0055] Calculate and obtain the signal strength of each valid sound wave signal segment in the valid sound wave signal segment set; it should be further explained that calculating the sound wave signal strength is a prior art and will not be described in detail;

[0056] Arrange the effective sound wave signal segments in the order of acquisition time;

[0057] Marking valid sound wave segments in the acquisition time period as sound wave signal segments to be spliced;

[0058] When two or more valid sound wave segments appear in the same acquisition time period, the valid sound wave segment with the largest signal strength is marked as the sound wave signal segment to be spliced;

[0059] If no valid sound wave segment appears during the collection period, the collection task fails and the sound wave signal is collected again.

[0060] In an optional embodiment, establishing the acoustic wave detection model based on an artificial intelligence model includes the following steps:

[0061] Obtain standard training data;

[0062] The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as an acoustic wave detection model.

[0063] In this embodiment, the standard training data includes several groups of input sound waves and corresponding sets of valid sound wave signal segments.

[0064] In this embodiment, the artificial intelligence model includes a deep convolutional neural network model or an RBF neural network model, etc., which have powerful nonlinear fitting capabilities.

[0065] Step 3: Splice the sound wave signal segments to be spliced ​​in chronological order to obtain a complete cycle of the sound wave signal;

[0066] In this embodiment, a time domain smoothing method is used to perform splicing processing on the sound wave signal segments to be spliced;

[0067] The acoustic signal segments to be spliced ​​are spliced ​​together in chronological order to form a complete periodic waveform for output. If they are spliced ​​together directly without any processing, some noise will be generated at the splicing boundary due to data discontinuity; there will also be mutations in the frequency domain; in addition, the overlapping waveforms at the connection will produce redundancy, which is not conducive to the final diagnosis.

[0068] Therefore, multi-channel sound wave signal segments should be smoothed when splicing to effectively eliminate discontinuities at the boundaries, which plays an important role in improving the naturalness of the multi-channel synthesized waveform while avoiding the introduction of false frequency components.

[0069] In order to reduce distortion and mutation at the splicing point, this embodiment uses a time domain smoothing method to process the spliced ​​sound wave signal segments. The specific algorithm is as follows: S(n) = λS1(n) + (1-λ)S2(n) n = 0, 1...R-1

[0070] in,

[0071] The acoustic signal segments to be spliced ​​become coherent and smooth after time domain smoothing, and the distortion and mutation at the splicing seams are eliminated.

[0072] The above formulas are all calculated by removing dimensions and taking their numerical values. The formula is a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and preset thresholds in the formula are set by technicians in this field according to actual conditions or obtained by simulating a large amount of data.

[0073] Working principle of the present invention:

[0074] The first microphone is set on one side of the railway; the second microphone is set on the same side as the first microphone, and the first microphone is separated by s centimeters; the third microphone is set on the other side of the railway and is on the same horizontal line as the first microphone; the fourth microphone is set on the other side of the railway and is on the same horizontal line as the second microphone;

[0075] The first microphone collects a first sound wave signal; the second microphone collects a second sound wave signal; the third microphone collects a third sound wave signal; and the fourth microphone collects a fourth sound wave signal;

[0076] Obtaining a sound wave interception model; inputting the first sound wave signal, the second sound wave signal, the third sound wave signal, and the fourth sound wave signal into the sound wave interception model respectively to obtain a set of valid sound wave signal segments;

[0077] Calculate and obtain the signal strength of each valid sound wave signal segment in the valid sound wave signal segment set;

[0078] Arrange the effective sound wave signal segments in the order of acquisition time;

[0079] Mark the valid sound wave segments in the acquisition time period as the sound wave signal segments to be spliced; when two or more valid sound wave segments appear in the same acquisition time period, mark the valid sound wave segment with the largest signal strength as the sound wave signal segment to be spliced; if no valid sound wave segment appears in the acquisition time period, the acquisition task fails and the sound wave signal is collected again;

[0080] The time domain smoothing method is used to splice the acoustic wave signal segments to obtain the acoustic wave signal of a complete cycle.

[0081] The above embodiments are only used to illustrate the technical method of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A railway acoustic wave data splicing method, characterized in that: The following steps are involved: Step 1: collecting sound wave signals of the railway; wherein the sound wave signals include a first sound wave signal, a second sound wave signal, a third sound wave signal and a fourth sound wave signal; Step 2: According to the sound wave signal and the sound wave detection model, a set of valid sound wave signal segments is obtained; wherein the sound wave detection model is established based on an artificial intelligence model; Arrange the valid sound wave signal segments in the order of acquisition time to obtain the sound wave signal segments to be spliced; Step 3: Splice the sound wave signal segments to be spliced ​​in chronological order to obtain a sound wave signal of a complete cycle.

2. A railway acoustic wave data splicing method according to claim 1, characterized in that: Step 1 also includes: a first microphone, a second microphone, a third microphone, and a fourth microphone; The first microphone is used to collect a first sound wave signal; The second microphone is used to collect a second sound wave signal; The third microphone is used to collect a third sound wave signal; The fourth microphone is used to collect a fourth sound wave signal.

3. A railway acoustic wave data splicing method according to claim 2, characterized in that: The first microphone is arranged on one side of the railway; The second microphone is arranged on the same side as the first microphone, and the first microphone is separated by s centimeters; The third microphone is arranged on the other side of the railway and is on the same horizontal line as the first microphone; The fourth microphone is arranged on the other side of the railway and is on the same horizontal line as the second microphone.

4. A railway acoustic wave data splicing method according to claim 1, characterized in that: According to the sound wave signal and the sound wave detection model, obtaining a valid sound wave signal segment set includes the following steps: Acquire a sound wave interception model; wherein the sound wave interception model is established based on an artificial intelligence model; The first sound wave signal, the second sound wave signal, the third sound wave signal and the fourth sound wave signal are respectively input into the sound wave interception model to obtain a valid sound wave signal segment set; wherein the valid sound wave signal segment set includes a first valid sound wave signal segment set, a second valid sound wave signal segment set, a third valid sound wave signal segment set and a fourth valid sound wave signal segment set.

5. The railway acoustic wave data splicing method according to claim 1, characterized in that: Establishing the acoustic wave detection model based on the artificial intelligence model includes the following steps: Obtain standard training data; The artificial intelligence model is trained using standard training data, and the trained artificial intelligence model is marked as an acoustic wave detection model; Artificial intelligence models include deep convolutional neural network models and RBF neural network models.

6. A railway acoustic wave data splicing method according to claim 4, characterized in that: Arranging the valid sound wave signal segments in the order of acquisition time to obtain the sound wave signal segments to be spliced ​​includes the following steps: Calculate and obtain the signal strength of each valid sound wave signal segment in the valid sound wave signal segment set; Arrange the effective sound wave signal segments in the order of acquisition time; Marking valid sound wave segments in the acquisition time period as sound wave signal segments to be spliced; When two or more valid sound wave segments appear in the same acquisition time period, the valid sound wave segment with the largest signal strength is marked as the sound wave signal segment to be spliced; If no valid sound wave segment appears in the collection time period, the collection task fails and the sound wave signal is collected again.

7. The railway acoustic wave data splicing method according to claim 1, characterized in that: In step three, the time domain smoothing method is used to splice the sound wave signal segments to be spliced.

Images