Railway vehicle wheel polygon detection method and system based on multi-channel fusion

Abstract

The present application belongs to the technical field of rail vehicle detection, and discloses a rail vehicle wheel polygon detection method and system based on multi-channel fusion, which comprises collecting multi-channel waveforms and calculating the roughness spectrum of each channel by using each channel waveform; point-multiplying the roughness spectrum of all channels to obtain a roughness product spectrum; identifying the wave peak point from the roughness product spectrum, and judging whether the wheel has polygon by first-order difference discrimination of the identified wave peak point. The present application calculates the roughness spectrum of each channel, fuses the roughness spectrum of each channel by point multiplication, thereby amplifying the signal-to-noise ratio, making the wave peak representing the polygon more prominent, and facilitating identification. After identifying the wave peak, the present application judges whether the wheel has polygon by first-order difference discrimination, thereby filtering out the error caused by noise and making the final judgment of the wheel polygon more accurate.

Description

A method and system for polygon detection of rail vehicle wheels based on multi-channel fusion Technical Field

[0001] This invention belongs to the field of rail vehicle inspection technology, and particularly relates to a method and system for detecting polygonal wheels of rail vehicles based on multi-channel fusion. Background Technology

[0002] When a train is running at high speed, the wheels are prone to periodic wear along the circumference due to manufacturing defects, wheel-rail friction, and track excitation. This results in a polygonal wheel. The polygonal wheel causes abnormal wheel-rail vibration during train operation, leading to loosening of train components and premature fatigue damage, posing a safety hazard and causing discomfort to passengers. Current methods for detecting polygonal wheel wear at the trackside include directly measuring the displacement between the wheel tread and the flange, indirectly measuring the rail deformation caused by wheel-rail contact during train movement, and detecting rail vibration using accelerometers at the trackside.

[0003] Currently, trackside inspection of train wheel polygons (with the detection system relatively stationary and the vehicle body moving) requires covering the entire circumference of the wheel tread, but almost no mechanical structure can meet the requirement of continuous measurement of approximately 3 meters (circumference). Therefore, multi-channel methods are generally used for measurement. After mapping the physical positions of the multi-channels to the circumferential positions of the wheel, a continuous waveform of the wheel's circumference can be stitched together. For example, Chinese patent application CN202310797604.2 discloses a rail vehicle wheel tread damage detection system and method. The method includes: using displacement sensors to collect data on the downward displacement of a detection rod; dividing the detection rod along its length into N sampling segments and numbering the N sampling segments; the length of the sampling segment is the product of the interval between two sensor acquisitions and the current speed of the wheel; selecting any point on the detection rod between adjacent displacement sensors as a splicing point, and calculating the sequence number of the sampling segment where the splicing point is located; assigning the sampling segment between the splicing points before and after each displacement sensor to that displacement sensor; and stitching together the valid data collected by each displacement sensor.

[0004] After mapping the multi-channel physical positions to the circumferential positions of the wheel, a continuous waveform of one revolution of the wheel can be stitched together. This stitched continuous waveform can be used for polygon recognition (qualitative and quantitative), but it has the following drawbacks:

[0005] 1) When the signal-to-noise ratio of the collected data is low, the recognition accuracy decreases regardless of whether it is a single-channel waveform or a spliced ​​waveform, and there is a risk of missed detection.

[0006] 2) In dynamic measurement systems, regardless of the algorithm used for post-processing, there will inevitably be splicing errors at the splicing points of each channel, meaning that the waveforms of the same circumferential segment of the wheel detected by different channels will have certain differences.

[0007] 3) If some channels are abnormal, have foreign objects or interference, the spliced ​​polygon features will be contaminated by noise, reducing the accuracy of polygon feature recognition. Summary of the Invention

[0008] In view of this, the present invention provides a method and system for detecting polygonal shapes of rail vehicle wheels based on multi-channel fusion, which eliminates the need for signal splicing and noise point removal, and quickly and accurately determines whether a wheel has polygonal shapes.

[0009] To solve the above technical problems, the technical solution of the present invention is to adopt a polygon detection method for rail vehicle wheels based on multi-channel fusion, including:

[0010] Acquire multi-channel waveforms and calculate the roughness spectrum of each channel using the waveforms of each channel;

[0011] Multiply the roughness spectra of all channels to obtain the roughness product spectrum;

[0012] The peak points are identified from the roughness product spectrum, and the identified peak points are subjected to first-order difference discrimination to determine whether the wheel has polygons.

[0013] As an improvement, methods for calculating the roughness spectrum of each channel include:

[0014] Using formula

[0015]

[0016] Calculate the roughness spectrum, where, For roughness spectrum, It is the value quantized in 1 / 3 octave k after the wheel out-of-roundness waveform is converted to the frequency domain. This is a reference value for wheel roughness.

[0017] As a further improvement, the method for performing dot multiplication on the roughness spectrum includes:

[0018] The roughness values ​​of all points with the same index in the roughness spectrum are multiplied together to obtain the roughness product value of that point index.

[0019] The roughness product values ​​obtained from all points are combined to form a roughness product spectrum.

[0020] As another further improvement, the method for identifying peak points from the roughness product spectrum includes:

[0021] Traverse the roughness product values ​​of all points in the roughness product spectrum. If the roughness product value of a point is greater than the roughness product values ​​of the points before and after it, then mark that point as a peak point.

[0022] As an improvement, when determining the peak at a certain point, a preset interval point is used, and points at the interval before and after the current point are used for judgment.

[0023] As an improvement, the method for first-order difference discrimination of the identified peaks includes:

[0024] The roughness product of the wave crest is subtracted from the adjacent point in front of it; if the difference is greater than a threshold, the wheel is considered to have a polygonal shape; or...

[0025] The roughness product of the wave crest is subtracted from the adjacent point. If the difference is greater than a threshold, the wheel is considered to have a polygonal shape; or...

[0026] The roughness product of the wave crest is subtracted from the adjacent points in front and behind it. If both differences are greater than a threshold, the wheel is considered to have a polygonal shape; or...

[0027] The roughness product of the crest point is subtracted from the points adjacent to it in front and behind. If the sum of the two differences is greater than a threshold, the wheel is considered to have a polygonal shape.

[0028] As an improvement, when performing first-order difference discrimination on the peak point, a preset interval point is used, and the points at the interval points before and / or after the peak point are used for first-order difference discrimination.

[0029] The present invention also provides a polygon detection system for rail vehicle wheels based on multi-channel fusion, comprising:

[0030] The roughness spectrum calculation module is used to calculate the roughness spectrum of each channel using the waveform of each channel;

[0031] The dot product module is used to multiply the roughness spectra of all channels to obtain the roughness product spectrum;

[0032] The polygon detection module is used to identify peak points from the roughness product spectrum and perform first-order difference discrimination on the identified peak points to determine whether the wheel has polygons.

[0033] The present invention also provides a storage medium storing a program that, when executed, implements the above-described method for detecting polygonal wheels of rail vehicles based on multi-channel fusion.

[0034] The present invention also provides a computer system, including a processor and a memory; the memory stores a computer program, such that when the computer program is executed by the processor, the above-mentioned method for detecting polygonal wheels of rail vehicles based on multi-channel fusion can be implemented.

[0035] The advantages of this invention are:

[0036] This invention calculates the roughness spectrum of each channel and fuses the roughness spectra of each channel by dot product, thereby amplifying the signal-to-noise ratio and making the peaks representing polygons more prominent and easier to identify.

[0037] After identifying the wave peak, this invention uses first-order difference discrimination to determine whether the wheel has a polygon, thereby filtering out errors caused by noise and making the final judgment of the wheel polygon more accurate. Attached Figure Description

[0038] Figure 1 is a flowchart of the present invention.

[0039] Figure 2 is a schematic diagram of the structure of the present invention. Detailed Implementation

[0040] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to specific embodiments.

[0041] Chinese patent application CN202011568335.5 discloses a signal processing method and system for polygon detection of train wheels, comprising: acquiring detection waveform data of one revolution of a train wheel; removing abnormal data points from the detection waveform data and segmenting the detection waveform data using the abnormal data points as segmentation points to generate detection waveform data before and after the abnormality; splicing the detection waveform data with the latter occurring first and the former occurring last to generate real detection waveform data; performing Fourier transform on the real detection waveform data to calculate the spectrum of the waveform data, and generating wheel roughness data using a roughness level calculation method.

[0042] As can be seen, in order to generate roughness data for the entire wheel, the data of each channel needs to be removed for outliers, and then the data needs to be spliced ​​together. This is inefficient and the final recognition accuracy is also low.

[0043] As shown in Figure 1, this invention provides a method for detecting polygonal wheels of rail vehicles based on multi-channel fusion, the specific steps of which include:

[0044] S1 acquires multi-channel waveforms and uses the waveforms of each channel to calculate the roughness spectrum of each channel.

[0045] Compared to damage such as bumps and collisions, wheel polygons have a circumferentially uniform distribution, meaning the wheel is close to a regular polygon. This ensures that the roughness of each channel used to acquire waveforms has the same frequency.

[0046] In this invention, the waveform obtained by the detection equipment, such as the tread damage detection system in the background art (where each detection rod is a channel), is a time-domain waveform. The roughness spectrum in this invention is equivalent to a frequency-domain waveform, except that the horizontal and vertical axes have coefficient transformations; the properties and characteristics of the spectrum remain unchanged. In the frequency domain, frequency is called the order in the roughness spectrum, and amplitude is called roughness.

[0047] Therefore, after acquiring the original time-domain waveform, it is necessary to transform it. Specifically, the method for calculating the roughness spectrum of each channel is as follows:

[0048] Using formula

[0049]

[0050] Calculate the roughness spectrum, where, For roughness spectrum, It is the value quantized in 1 / 3 octave k after the wheel out-of-roundness waveform is converted to the frequency domain. This is a reference value for wheel roughness.

[0051] S2 performs a dot product of the roughness spectra of all channels to obtain the roughness product spectrum.

[0052] Each channel's roughness spectrum contains several points R. ij Where i is the channel number, j is the point number, and each point R ij Each corresponds to a roughness value. In this step, it is necessary to perform a dot product between each point in each channel and the corresponding points in other channels, specifically:

[0053] S21 multiplies the roughness values ​​of all points with the same index in the roughness spectrum to obtain the roughness product value of that point index.

[0054] To simplify the model for example, assume there are three channels, each with four points, i.e., [R] 11 ,R 12 ,R 13 ,R 14 ]、[R 21 ,R 22 ,R 23 ,R 24 ]、[R 31 ,R 32 ,R 33 ,R 34 ], then the result of the dot product is [R] 11 *R 21 *R 31 R 12 *R 22 *R 32 R 13 *R23 *R 33 R 14 *R 24 *R 34 At that time, the actual situation was much more complex, involving far more numbers and points than the examples above.

[0055] S22 forms a roughness product spectrum from the roughness product values ​​obtained at all points.

[0056] The results in the above example [R] 11 *R 21 *R 31 R 12 *R 22 *R 32 R 13 *R 23 *R 33 R 14 *R 24 *R 34 This is the roughness product spectrum.

[0057] S3 identifies peak points from the roughness product spectrum and performs first-order difference discrimination on the identified peak points to determine whether the wheel has polygons.

[0058] In the prior art CN202011568335.5, a signal processing method and system for detecting polygonal train wheels is disclosed. Due to the low signal-to-noise ratio, abnormal data points in the waveform need to be removed in order to ensure the accuracy of the final detection result.

[0059] In this invention, the signal-to-noise ratio is increased through the dot product step, making the peaks representing polygons more prominent and easier to identify. In this embodiment, the method for identifying peak points from the roughness product spectrum is as follows: traverse the roughness product values ​​of all points in the roughness product spectrum; if the roughness product value of a point is greater than the roughness product values ​​of its preceding and following points, then that point is marked as a peak point. As a peak, its amplitude (roughness product) is certainly higher than that of its preceding and following points. Therefore, in this invention, by comparing the amplitude of a point with that of its preceding and following points, if it is greater than the amplitude of the preceding and following points, then that point is considered a peak.

[0060] However, in practice, waveforms may be interfered with by noise, resulting in a "peak cluster" phenomenon. Therefore, in this implementation, when identifying peaks, a certain width can be set, i.e., a preset interval point, and points at the interval before and after that point are used for judgment. Technicians can set the specific interval point according to the sampling rate and noise conditions; it is not a fixed value.

[0061] After identifying the peak point, a first-order difference discrimination is required. This is because the absolute height alone does not necessarily indicate a polygon; it could also be due to noise. Therefore, it is necessary to determine the height difference between the peak point and its surroundings. Only when the height difference reaches a certain threshold is it considered a polygon.

[0062] Specifically, methods for first-order difference discrimination include:

[0063] The roughness product of the peak point is subtracted from the adjacent point in front of it. If the difference is greater than a threshold, the wheel is considered to have a polygonal shape, expressed as Value_peak(peak point) – Value_former(front point) > Thr(threshold); or,

[0064] The roughness product of the peak point is subtracted from the adjacent point after it. If the difference is greater than a threshold, the wheel is considered to have a polygonal shape, expressed as Value_peak(peak point) – Value_latter(latter point) > Thr(threshold); or,

[0065] The roughness product of the peak point is subtracted from its two adjacent points in front and behind it. If both differences are greater than a threshold, the wheel is considered to have a polygonal shape, expressed as Value_peak – Value_former > Thr1 and Value_peak – Value_latter > Thr2; or,

[0066] The roughness product of the peak point is subtracted from the points adjacent to it in front and behind. If the sum of the two differences is greater than the threshold, the wheel is considered to have a polygon, expressed as (Value_peak–Value_former)+(Value_peak–Value_latter)>Thr3.

[0067] Of course, similar to the peak point determination steps, when performing first-order difference discrimination on the peak point, the interval points can be preset, and the points before or / and after the peak point can be used for first-order difference discrimination.

[0068] As shown in Figure 2, the present invention also provides a polygon detection system for rail vehicle wheels based on multi-channel fusion, comprising:

[0069] The roughness spectrum calculation module is used to calculate the roughness spectrum of each channel using the waveform of each channel;

[0070] The dot product module is used to multiply the roughness spectra of all channels to obtain the roughness product spectrum;

[0071] The polygon detection module is used to identify peak points from the roughness product spectrum and perform first-order difference discrimination on the identified peak points to determine whether the wheel has polygons.

[0072] The present invention also provides a storage medium storing a program that, when executed, implements the above-described method for detecting polygonal wheels of rail vehicles based on multi-channel fusion.

[0073] The present invention also provides a computer system, including a processor and a memory; the memory stores a computer program, such that when the computer program is executed by the processor, the above-mentioned method for detecting polygonal wheels of rail vehicles based on multi-channel fusion can be implemented.

[0074] The above are merely preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be considered as limitations on the present invention, and the scope of protection of the present invention should be determined by the scope defined in the claims. For those skilled in the art, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting polygonal shapes of rail vehicle wheels based on multi-channel fusion, characterized in that... include: Acquire multi-channel waveforms and calculate the roughness spectrum of each channel using the waveforms of each channel; Multiply the roughness spectra of all channels to obtain the roughness product spectrum; The peak points are identified from the roughness product spectrum, and the identified peak points are subjected to first-order difference discrimination to determine whether the wheel has polygons.

2. The method for detecting polygonal shapes of rail vehicle wheels based on multi-channel fusion according to claim 1, characterized in that... Methods for calculating the roughness spectrum of each channel include: using formulas Calculate the roughness spectrum, where, For roughness spectrum, It is the value quantized in 1 / 3 octave k after the wheel out-of-roundness waveform is converted to the frequency domain. This is a reference value for wheel roughness.

3. The method for polygon detection of rail vehicle wheels based on multi-channel fusion according to claim 1, characterized in that... The method for performing dot product on the roughness spectrum includes: multiplying the roughness values ​​of all points with the same index in the roughness spectrum to obtain the roughness product value of that point index; and forming a roughness product spectrum by combining the roughness product values ​​obtained from all points.

4. The method for detecting polygonal shapes of rail vehicle wheels based on multi-channel fusion according to claim 1, characterized in that... The method for identifying peak points from the roughness product spectrum includes: traversing the roughness product values ​​of all points in the roughness product spectrum; if the roughness product value of a certain point is greater than the roughness product values ​​of the points before and after it, then the point is marked as a peak point.

5. The method for detecting polygonal wheels of rail vehicles based on multi-channel fusion according to claim 4, characterized in that: When determining the peak at a certain point, preset interval points are used, and points at intervals before and after the current point are used for judgment.

6. The method for detecting polygonal shapes of rail vehicle wheels based on multi-channel fusion according to claim 1, characterized in that... The method for first-order difference discrimination of the identified peaks includes: subtracting the roughness product value of the peak point from the point adjacent to it in front; if the difference is greater than a threshold, the wheel is considered to have a polygonal shape; or, subtracting the roughness product value of the peak point from the point adjacent to it behind; if the difference is greater than a threshold, the wheel is considered to have a polygonal shape; or, subtracting the roughness product value of the peak point from the points adjacent to it in front and behind, respectively; if both differences are greater than a threshold, the wheel is considered to have a polygonal shape; or, subtracting the roughness product value of the peak point from the points adjacent to it in front and behind, respectively; if the sum of the two differences is greater than a threshold, the wheel is considered to have a polygonal shape.

7. The method for detecting polygonal wheels of rail vehicles based on multi-channel fusion according to claim 1, characterized in that: When performing first-order difference discrimination on the peak point, a preset interval point is used, and the points at the interval points before and / or after the peak point are used for first-order difference discrimination.

8. A polygon detection system for rail vehicle wheels based on multi-channel fusion, characterized in that... include: The roughness spectrum calculation module is used to calculate the roughness spectrum of each channel using the waveform of each channel; The dot product module is used to multiply the roughness spectra of all channels to obtain the roughness product spectrum; The polygon detection module is used to identify peak points from the roughness product spectrum and perform first-order difference discrimination on the identified peak points to determine whether the wheel has polygons.

9. A storage medium, characterized in that: The storage medium stores a program that, when executed, implements the multi-channel fusion-based polygon detection method for rail vehicle wheels as described in any one of claims 1 to 7.

10. A computer system, characterized in that: It includes a processor and a memory; the memory stores a computer program such that when the computer program is executed by the processor, it can implement the multi-channel fusion-based polygon detection method for rail vehicle wheels as described in any one of claims 1 to 7.

Images