CN115931399A

CN115931399A - Method for real-time online detection of polygonal fault of high-speed train wheel

Info

Publication number: CN115931399A
Application number: CN202211450818.4A
Authority: CN
Inventors: 侯永强; 高岭松; 张起源
Original assignee: Dalian Baishengyuan Technology Co ltd
Current assignee: Dalian Baishengyuan Technology Co ltd
Priority date: 2022-11-18
Filing date: 2022-11-18
Publication date: 2023-04-07

Abstract

The invention belongs to the technical field of high-speed train fault detection, and discloses a method for detecting polygonal faults of wheels of a high-speed train on line in real time, which comprises the following steps: reading the vehicle speed and the diameter, calculating the wheel frequency conversion harmonic wave, inputting a vibration acceleration signal, performing low-pass filtering, intercepting a middle section, and performing FFTW frequency domain transformation; intercepting a signal obtained after the middle section, firstly calculating a time domain effective value RMS, and then calculating an effective value Rmsd; selecting the polygon characteristic with the maximum amplitude value to obtain a polygon characteristic value; extracting peak amplitude-frequency characteristics of signals obtained by FFTW frequency domain transformation, and then performing polygon characteristic coupling extraction to obtain a calculated polygon index Ad; and calculating a polygon evaluation index Kd, judging whether the polygon evaluation index Kd exceeds a set threshold value, and giving an alarm. The detection method of the invention adopts the total amount of the vibration signal effective value and the polygon characteristic index after weighting as the judgment standard, has stable performance and accurate monitoring, and can monitor medium and high speed trains.

Description

Method for real-time online detection of polygonal fault of high-speed train wheel

Technical Field

The invention belongs to the technical field of high-speed train fault detection, and relates to a method for detecting polygonal faults of a high-speed train wheel on line in real time.

Background

Through the development of decades, the national railway service mileage reaches 14.6 kilometers and the high-speed railway service mileage reaches 3.79 kilometers by the end of 2020, and the railway is stable in the first place in the world. By 30 days 12 months in 2021, the Chinese high-speed rail operation mileage breaks through 4 kilometers. As the 20 days of 6 months in 2022, nearly 3200 km of high-speed rail in China has been normalized and operated at the high standard of 350 km/h.

In the process of rapid operation of a high-speed train, a plurality of technical indexes need to be detected on line, and polygonal faults of wheels of the high-speed train are one of the technical indexes. The polygon of the wheel is also called as the corrugation of the wheel or the periodic non-rounding of the wheel, the polygon of the wheel can cause severe vibration noise of the vehicle and damage lines and vehicle parts, the polygon of the wheel can not only cause larger wheel-rail impact force and rolling noise, but also reduce the riding comfort by large amplitude vibration generated at high frequency, and can cause safety problem to derail in serious conditions.

Train wheel running state detection is mainly divided into two categories: manual detection and real-time online monitoring. At present, the wheel state detection of most train maintenance work sections in China mainly takes manual work as the main part. Manual detection requires that the vehicle is disassembled in routine inspection and maintenance in a work section and then inspected by means of naked eyes, key striking, ear listening and the like of workers or by using various manual caliper tools and the like. The real-time monitoring method mainly comprises an eddy current method, an ultrasonic remote measurement method, a laser sensor method, a vibration acceleration method, an image method and the like.

The existing manual method has the defects that the train turnover time is too long due to inspection, and the manual detection is influenced by factors such as artificial reasons, operation conditions and the like, so that the detection speed is low, the detection is inaccurate, the labor intensity is high, the method is only effective for the flat sore fault of the wheel, and the method is not effective for the polygonal fault of the wheel; at present, all equipment real-time monitoring methods have good identification on whether a wheel has a polygonal fault, but the identification on the order of the polygonal wheel has certain limitation, for example, some monitoring methods cannot quantitatively reflect the damage degree of the fault, some monitoring methods have poor stability, or the operation speed is low and the time consumption is long, or the monitoring result is inaccurate; in addition, when the sensor signal has a large impact, erroneous judgment is easily caused.

Disclosure of Invention

The invention provides a specific solution to the difficult problem. The method is realized by adopting an acceleration sensor real-time monitoring means, not only can reflect the defects, but also can accurately reflect the damage degree of the defects, and the monitoring and installation cost is low.

In order to realize the purpose, the technical scheme adopted by the invention for solving the problem is as follows:

a real-time online detection method for polygon faults of high-speed train wheels comprises the following steps:

detecting the real-time running speed and the current wheel diameter of the high-speed train; calculating the current wheel rotating frequency according to the real-time vehicle speed and the wheel diameter, and calculating a corresponding wheel polygon order frequency value sequence;

detecting a vibration vertical acceleration signal of the axle box of the high-speed train; after the obtained data signal is subjected to low-pass filtering, intercepting a middle section, and simultaneously performing frequency domain transformation;

intercepting the data signal obtained after the middle section, firstly calculating a time domain effective value, and then calculating partial evaluation indexes of the effective value;

according to the frequency domain data obtained by frequency domain transformation, firstly extracting the maximum peak value and the corresponding frequency sequence in the frequency domain data to form a peak value sequence, adopting a difference threshold algorithm to extract a polygon order characteristic sequence from the peak value sequence, wherein the polygon order characteristic sequence comprises a polygon order and a frequency domain amplitude value corresponding to the order, completing the extraction of peak amplitude-frequency characteristics,

according to the extracted polygon order feature sequence, taking a maximum group of polygon features as a characteristic quantity, and calculating a polygon index and a polygon order;

calculating the polygon evaluation index through the obtained effective value part evaluation index, the polygon index value and the weighting coefficient;

when the vehicle speed TSpd is more than or equal to 100km/h, comparing the polygon evaluation index with an alarm threshold value, and judging whether the polygon evaluation index exceeds the set threshold value or not and alarming or not; and outputting the polygon index and the early warning state according to the comparison result.

A method for detecting polygonal faults of high-speed train wheels on line in real time comprises the following specific implementation steps:

step 1: detecting the real-time running speed TSpd of the high-speed train and the current wheel diameter TDia;

step 2: calculating the current wheel rotation frequency St = TSpd 1000/(3.6 TDia pi) according to the real-time vehicle speed TSpd and the wheel diameter TDia, and calculating the corresponding wheel polygon order frequency value sequence On (St);

and step 3: detecting a vibration vertical acceleration signal of the axle box of the high-speed train;

and 4, step 4: after the data signal obtained in the step 3 is subjected to low-pass filtering, intercepting a middle section, and simultaneously carrying out FFTW frequency domain transformation;

and 5: according to the data signals obtained after the middle section is intercepted in the step 4, firstly, a time domain effective value RMS is calculated, and then, an effective value part evaluation index Rmsd =20log (RMS) is calculated;

and 6: according to the frequency domain data obtained by the FFTW frequency domain transformation in the step 4, firstly extracting the maximum peak value and the corresponding frequency P (x, y) sequence in the frequency domain data to form a peak value sequence Px (x, y), and extracting a polygon order feature sequence OPn (x, y) from the peak value sequence by adopting a difference threshold algorithm, wherein the polygon order feature sequence OPn comprises polygon order frequencies and frequency domain amplitude values corresponding to the order frequencies, so as to finish extracting peak amplitude-frequency features;

and 7: according to the multi-order feature sequence OPn (x, y) extracted in the step 6, taking the largest group of polygon features OPmax (x, y) as the characteristic quantity, calculating a polygon index Ad =20log (OPmax (y)) and a polygon order Poly = OPmax (x)/St,

and step 8: obtaining an effective value partial evaluation index Rmsd through the step 5, obtaining a polygon index Ad value and a weighting coefficient a through the step 7, and calculating a polygon evaluation index Kd = a Ad + (1-a) Rmsd;

and step 9: when the vehicle speed TSpd is more than or equal to 100km/h, the polygon evaluation index Kd is compared with an alarm threshold value, and whether the polygon evaluation index Kd exceeds the set threshold value or not and whether the alarm is given or not is judged.

Step 10: and outputting the polygon index and the early warning state according to the comparison result in the step 9.

In step 2, the wheel polygon order frequency value sequence On (St) is a sequence formed by the product of the wheel rotation frequency St and the polygon fault order On, and the order On of the wheel with the polygon fault is preferably between 15 and 60.

Preferably, in step 4, when the evaluation index of the effective value part is calculated, the middle section is intercepted to be the data signal between 10% and 90% of the intercepted data section.

Preferably, in step 6, the method for extracting the peak sequence Px (x, y) includes: and extracting the maximum peak value and the corresponding frequency P (x, y) sequence in the frequency domain data of a plurality of points by adopting a least square fitting algorithm, sequencing the P (x, y) sequences according to the peak amplitude after extraction to obtain a peak value sequence Px (x, y), wherein the sequence is a group of coordinate values, the x axis represents the frequency, and the y axis represents the corresponding peak value. Furthermore, in the P (x, y) sequence, 40-60 maximum peaks and corresponding frequencies are extracted.

Preferably, in step 6, the method for extracting the polygon order feature sequence OPn (x, y) from the peak sequence by using the difference threshold algorithm includes: setting a coupling coefficient, coupling the frequency values in the wheel polygon order frequency value sequence On (St) and the peak value sequence Px (x, y) one by one, when | On (St) -Px (x) | is less than or equal to the coupling coefficient, indicating that the frequency value is a fault feature, extracting the frequency value and outputting a corresponding peak value to form a polygon order feature sequence OPn (x, y), wherein the feature sequence OPn (x, y) is a group of coordinate values, an abscissa x represents the extracted order frequency value, and an ordinate y represents a corresponding peak value.

In step 8, the value range of the weighting coefficient a is preferably 0.2-0.8.

Preferably, in step 9, the method for determining whether to alarm includes: and comparing the polygon evaluation index Kd with an alarm threshold, and when the continuous display of the abnormity is performed and the polygon orders Poly are the same, taking the abnormity to perform alarm prompt, otherwise, not performing alarm prompt. In the actual running process of the train, due to the fluctuation of the rotating speed and the dynamic change characteristic of the vibration signal, the calculated polygon evaluation index can generate numerical value fluctuation, and in order to diagnose the rim polygon fault more stably and reliably, time dimension parameters are introduced, namely whether the polygon evaluation index exceeds an alarm threshold value (namely whether abnormal display is continuously performed) in a certain time period or not is continuously judged, and if the condition is met, the rim polygon fault is diagnosed; preferably, when abnormality is displayed for 10 seconds or more continuously and the polygon order Poly is the same, an alarm is given.

Further, the judgment criteria of whether the display abnormality is abnormal are as follows: the polygon evaluation index Kd is less than the lowest value of the alarm threshold value, the operation is normal, and the alarm state does not display abnormity; the situations in which the abnormality is displayed include: setting the alarm state as pre-judgment alarm when the lowest value of the alarm threshold is less than or equal to Kd and less than the middle value of the alarm threshold; setting the alarm state as early warning alarm when the Kd between the alarm threshold values is less than or equal to the Kd and less than the maximum alarm threshold value; kd is larger than or equal to the highest alarm threshold value, and the alarm state is set as alarm.

The invention has the beneficial effects that: the polygon fault monitoring method provided by the invention selects a weighting mode to judge the polygon fault, and takes the total amount of the vibration signal effective value and the polygon characteristic index after weighting as a judgment standard. The polygon fault detection method has the advantages of high operation speed in the detection process, accurate and stable monitoring result, small requirement on system performance, capability of monitoring medium and high speed trains (more than 100 km/h), and capability of meeting the requirement of a low-cost embedded device real-time online monitoring system; in addition, the invention introduces time dimension parameters, and ensures more stable and reliable alarm output through continuous time dimension.

Drawings

FIG. 1 is a flow chart of an example embodiment of the method of the present invention.

FIG. 2 is a flow chart of the detection method of the present invention.

Fig. 3 is a schematic view of communication monitoring.

FIG. 4 is a plot of the measured points waveform.

Fig. 5 is a frequency domain amplitude plot.

Detailed Description

The following detailed description of the invention refers to the accompanying drawings.

Example 1

As shown in figure 1, a vibration sensor is arranged above a bearing of a train wheel set, 8 wheel sets of each carriage are respectively provided with a vibration acceleration sensor, a data acquisition module acquires a radial vibration signal of a wheel and sends the radial vibration signal to a diagnosis and analysis algorithm processing module, the data processing module acquires the synchronous rotating speed and the current wheel diameter value of the train from a vehicle-mounted central control unit through an Ethernet and is used for calculating the real-time rotating frequency of the wheel, and a calculation result and alarm information are sent to the vehicle-mounted central control unit from an Ethernet port in real time. The invention specifically comprises the following steps:

step 1: detecting the real-time running speed TSpd and the current wheel diameter TDia of the high-speed train;

and step 3: detecting a vibration vertical acceleration signal of a high-speed train axle box;

and 4, step 4: after the data signal obtained in the step 3 is subjected to low-pass filtering at 2500Hz, intercepting a middle section, and simultaneously carrying out FFTW frequency domain transformation;

step 6: according to the frequency domain data obtained by the FFTW frequency domain transformation in the step 4, firstly extracting the maximum peak value and the corresponding frequency P (x, y) sequence in the frequency domain data to form a peak value sequence Px (x, y), extracting a polygon order feature sequence OPn (x, y) from the peak value sequence by adopting an interpolation threshold algorithm, wherein the polygon order and the frequency domain amplitude corresponding to the order are included to finish extracting peak amplitude-frequency features,

and 7: according to the multi-order feature sequence OPn (x, y) extracted in the step 6, taking the maximum group of polygon features OPmax (x, y) as a characterization quantity, calculating a polygon index Ad =20log (OPmax (y)) and a polygon order Poly = OPmax (x)/St,

and step 8: obtaining an effective value part evaluation index Rmsd through the step 5, obtaining a polygon characteristic value and a polygon index Ad value through the step 7, and calculating a polygon evaluation index Kd = a Ad + (1-a) Rmsd through a weighting coefficient a;

Example 2

FIG. 2 shows a flow chart of the present invention.

Step 1: the reading system issues a real-time train running speed TSpd, and reads the current wheel diameter TDia;

step 2: calculating the current wheel rotation frequency St = TSpd 1000/(3.6 TDia pi) according to the real-time vehicle speed and the wheel diameter, and calculating a corresponding wheel polygon order frequency value sequence On (St) as a polygon fault characteristic reference value; the wheel polygon order frequency value sequence On (St) is a sequence formed by the product of the wheel rotation frequency St and the polygon fault order On, and the polygon fault order On of the wheel is generally between 15 and 60;

and step 3: reading a vibration vertical acceleration signal of the axle box, performing 2500Hz low-pass filtering, selecting a 6-order Butterworth filter as polygonal evaluation data by the filter, wherein the transfer function form of the filter is as follows:

and 4, step 4: calculating an effective value RMS of the filtered vibration data, wherein the filtered data can generate an error caused by edge jitter, so that part of signals are intercepted to calculate the effective value of the vibration data, the value is generally between 10% and 90% of a data segment when the signals are intercepted, and then an effective value part evaluation index Rmsd =20log (RMS) is calculated;

and 5: performing Fourier transform on the filtered vibration data, taking frequency domain data, selecting an FFTW formula for calculation by using a Fourier transform formula, wherein the calculation formula is as follows:

wherein it is present>

According to the properties of the correlation function and by adopting recursion division, all coefficients, namely frequency domain amplitude values, can be rapidly calculated;

step 6: extracting maximum peak values and corresponding frequency P (x, y) sequences in the frequency domain data, wherein the extraction algorithm is a least square fitting algorithm, preferably 40-60 maximum peak value sequences are extracted, and the maximum peak value sequences are sequenced according to the peak value amplitude after extraction to obtain a peak value sequence Px (x, y), the sequence is a group of coordinate values, the abscissa x represents the frequency, and the ordinate y represents the corresponding peak value; in the step, the maximum peak sequence of a plurality of points is extracted to ensure that the extracted peak sequence is a possible fault part, and then a least square fitting algorithm is adopted to extract data, so that the method is stable, reliable, short in time consumption and high in speed;

and 7: adopting a difference threshold value algorithm to carry out fault feature coupling, coupling a wheel polygon order frequency value sequence On (St) with frequency values in a peak value sequence Px (x, y), wherein the coupling coefficient is 2.1, and the specific calculation mode is that the absolute value of the frequency value difference between the polygon order frequency value sequence and the peak value sequence is compared, when | On (St) -Px (x) | is less than or equal to 2.1, the frequency value is indicated as a fault feature, the frequency value is recorded with corresponding peak value output, all polygon order frequency values and peak value sequence frequency values are sequentially compared with one another, a polygon order feature sequence OPn (x, y) is extracted, the feature sequence OPn (x, y) is a group of coordinate values, an abscissa x represents the extracted order frequency, and an ordinate y represents a corresponding peak value; in the step, a difference threshold algorithm is adopted for fault characteristic coupling, so that floating point calculation or frequency errors caused by rotation speed fluctuation can be abandoned, and failure in characteristic extraction is avoided;

and 8: extracting a maximum set of polygon features OPmax (x, y) among OPn (x, y) as a token, the token being a set of coordinates, OPmax (x) representing order frequency, OPmax (y) representing corresponding magnitude, calculating a polygon index Ad =20log (OPmax (y)), calculating a polygon order Poly = OPmax (x)/St;

and step 9: calculating a polygon evaluation index Kd = a Ad + (1-a) Rmsd + Kfac, wherein the value range of the weighting coefficient a is 0.2-0.8, the optimal value is a =0.5, the Kfac is a correction coefficient, and the value range is 0-50, so that the method is suitable for the statistical type of alarm values of various vehicle types; a weight algorithm is introduced into the polygon evaluation index, and the effective value of the vibration data is used as a partial component of the judgment parameter, so that the monitoring result can be ensured to be more accurate;

step 10: the vehicle speed TSpd is more than or equal to 100km/h, and the polygon evaluation index Kd is compared with an alarm threshold value: when Kd is less than 44 (dB), the operation is normal, and the alarm state ALtst =0; when Kd is more than or equal to 44 (dB) and less than 49 (dB), setting the alarm state as the prejudgment ALST =1; when Kd is more than or equal to 49 (dB) and less than 54 (dB), the alarm state is the early warning ALST =2; and when Kd is larger than or equal to 54 (dB), the alarm state is the alarm ALtst =3. When continuous alarming is carried out and the levels Poly are the same, the state ALttr = ALtst is informed; otherwise, the message-taking alarm state ALTRr =0; due to the fluctuation of the rotating speed and the dynamic change characteristic of the vibration signal, the calculated polygon evaluation index can generate numerical value fluctuation, in order to stably and reliably diagnose the rim polygon fault, a time dimension parameter is introduced, namely whether the polygon evaluation index exceeds an alarm threshold value within a certain time period (more than 10 seconds) or not is continuously judged, and if the condition is met, the rim polygon fault is diagnosed;

step 11: and outputting a polygon evaluation index Kd, a polygon order Poly and an alarm state ALTRs.

Example 3

The method is applied to an intelligent dynamic high-speed rail safety monitoring system in China to carry out polygonal fault diagnosis, and specifically comprises the following steps:

step 1: fig. 3 shows that the high-speed train central control unit CCU sends part of common data information to the vibration data processing module unit, the information includes information such as train composition, car number, ambient temperature, train speed, wheel diameter, etc., the reading system sends real-time train running speed TSpd =344.34Km/h, and reads current wheel diameter TDia =920mm;

step 2: calculating the current wheel rotation frequency St = TSpd 1000/(3.6 TDia pi) =33.11Hz according to the real-time vehicle speed and the wheel diameter, and calculating the sequence of corresponding wheel polygon order frequency values On (St) =496.65Hz, 529.76Hz.. 662.2Hz.. 1986.6 Hz;

and 3, step 3: and (3) reading the vertical acceleration signal of the vibration of the axle box, performing 2500Hz low-pass filtering as shown in figure 4, and selecting a 6-order Butterworth filter as a filter.

And 4, step 4: calculating an effective value RMS of the filtered data, wherein the filtered data can generate edge jitter, so that the effective value is calculated to intercept partial signals, the values are generally between 10% and 90% of data segments, and then an effective value part evaluation index Rmsd =20log (RMS) =20log (5.6597 g) =15.055dB is calculated;

and 5: calculating the filtering data, performing Fourier transform to obtain frequency domain data, and selecting an FFTW formula for calculation by using a Fourier transform formula as shown in FIG. 5 to calculate a frequency domain amplitude;

step 6: extracting maximum peak values and corresponding frequency P (x, y) sequences in frequency domain data, wherein the extraction algorithm is a least square fitting algorithm, generally extracting 40-60 maximum peak value sequences, and sorting Px (x, y) according to peak amplitude values after extraction, as shown in Table 1, the peak value sequence of the top 10 in sorting is shown;

table 1 peak sequence ordering

Serial number	frequency/Hz	Peak value/g
			1	662.00	7.2542
2	628.97	1.2796
			3	595.98	0.8245
4	529.92	0.7041
			5	1324.25	0.7018
6	666.62	0.5412
			7	658.34	0.4571
8	430.26	0.4101
			9	648.66	0.4033
10	695.15	0.3688

And 7: according to the coupling of the frequency value of the polygon order On (St) sequence of the wheel and the frequency value in the peak value sequence Px (x, y) and the coupling coefficient of 2.1, the specific calculation mode is used for comparing the absolute value of the frequency value difference between the polygon order and the peak value sequence, when the On (St) -Px (x) is less than or equal to 2.1, the frequency value is recorded and corresponding peak value output is recorded, all the polygon order frequency values and the peak value sequence frequency values are compared with each other in sequence, the polygon order feature sequence OPn (x, y) is extracted, and a part of extraction results are shown in a table 2.

TABLE 2 order index extraction results

Serial number	frequency/Hz	Peak value/g	Polygon order sequence frequency
				1	662.00	7.2542	662.2
	628.97	1.2796	629.09
				3	595.98	0.8245	595.98
4	529.92	0.7041	529.76
				5	1324.25	0.7018	1324.4
6	430.26	0.4101	430.43
				7	695.15	0.3688	695.31
8	1125.68	0.2466	1125.74
				9	728.98	0.2336	728.42
10	828.66	0.1923	827.75

And 8: taking a maximum set of polygon features OPmax (x, y) in the OPn (x, y) as a token, calculating a polygon index Ad =20log (OPmax (y)) =20log (7.2542 g) =17.212dB, calculating a polygon order Poly = OPmax (x)/St =662.00/33.11=19.99, and rounding to 20 orders;

and step 9: calculating polygon evaluation index

Kd＝0.5*17.212dB+0.5*15.055dB＝16.134dB。

High-speed rail dynamic safety monitoring systems typically set uniform alarm thresholds for regulatory management. The evaluation values obtained based on different algorithms are different, but the preset alarm threshold value cannot be correspondingly adjusted according to different algorithms, and for this reason, an algorithm provider can set a corresponding correction coefficient according to the alarm threshold value preset by the system so as to match the use requirements of the system. In this embodiment, the correction coefficient is 30dB, and the corrected polygon evaluation index has a value:

Kd＝0.5*17.212dB+0.5*15.055dB+30dB＝46.134dB

step 10: the vehicle speed TSpd is more than or equal to 100km/h, and the polygon evaluation index Kd is compared with an alarm threshold value: when Kd is less than 44 (dB), the operation is normal, and the alarm state ALtst =0; when Kd is more than or equal to 44 (dB) and less than 49 (dB), setting the alarm state as the prejudgment ALtst =1; when Kd is more than or equal to 49 (dB) and less than 54 (dB), the alarm state is the early warning ALtst =2; and when Kd is larger than or equal to 54 (dB), the alarm state is the alarm ALtst =3. When the alarm is given for more than 10 seconds continuously and the Poly is the same in order, the alarm state ALttr = ALtst is taken; otherwise, the message-taking alarm state ALTRr =0;

because the polygon evaluation index is 46.134dB, a prejudgment alarm is achieved, and an alarm is continuously given for 10 seconds, ALttr =1.

Step 11: outputting polygon evaluation index Kd =46.134, polygon order Poly =20, and alarm state ALTtr =1.

The intelligent high-speed rail dynamic safety monitoring system outputs a detection result every 0.5s, the calculation time of the algorithm is 50ms, and the algorithm is superior to the 0.1s operation speed of other algorithms. The detection result is consistent with the actual state of the wheel through the verification of the wheel polygon bench test carried out in the national engineering laboratory of the client. And the intelligent high-speed rail dynamic safety monitoring system adopting the algorithm has been continuously operated for 15 months, the prompting and alarming are carried out for 2 times, and the wheel roundness measurement is carried out on the wheel at the alarming position manually, so that the polygonal defect identical to the monitoring result is really existed, and the accuracy rate is 100%. The method has accurate monitoring result and extremely high monitoring stability.

Claims

1. A method for detecting polygonal faults of high-speed train wheels in real time on line is characterized by comprising the following steps:

when the vehicle speed TSpd is more than or equal to 100km/h, the polygon evaluation index is compared with an alarm threshold value, whether the polygon evaluation index exceeds the set threshold value or not is judged, and whether an alarm is given or not is judged; and outputting the polygon index and the early warning state according to the comparison result.

2. The method for detecting the polygonal fault of the high-speed train wheel on line in real time according to claim 1, which is characterized by comprising the following steps:

step 6: according to the frequency domain data obtained by the FFTW frequency domain transformation in the step 4, firstly extracting the maximum peak value and the corresponding frequency P (x, y) sequence in the frequency domain data to form a peak value sequence Px (x, y), and extracting a polygon order feature sequence OPn (x, y) from the peak value sequence by adopting a difference threshold algorithm, wherein the polygon order feature sequence OPn comprises polygon order frequencies and frequency domain amplitude values corresponding to the order frequencies, so as to finish extracting peak amplitude-frequency features;

and 8: obtaining an effective value part evaluation index Rmsd through the step 5, obtaining a polygon index Ad value and a weighting coefficient a through the step 7, and calculating a polygon evaluation index Kd = a Ad + (1-a) Rmsd;

and step 9: when the vehicle speed TSpd is more than or equal to 100km/h, comparing the polygon evaluation index Kd with an alarm threshold value, and judging whether the polygon evaluation index Kd exceeds the set threshold value or not and alarming or not;

3. The method for real-time online detection of polygon faults of high-speed train wheels according to claim 2, wherein the sequence of polygon order frequency values On (St) of the wheels is a sequence formed by multiplying a wheel rotation frequency St by a polygon fault order On, and the order On of the polygon fault of the wheels is between 15 and 60.

4. The method for real-time online detection of the polygonal fault of the high-speed train wheel according to claim 2 or 3, further characterized in that when the effective value part evaluation index is calculated, the middle section is intercepted as a data signal between 10% and 90% of the intercepted data section.

5. The method for real-time online detection of the polygon fault of the high-speed train wheel according to claim 2, wherein the method for extracting the peak sequence Px (x, y) comprises: and extracting the maximum peak value and the corresponding frequency P (x, y) sequence in the frequency domain data of the points by adopting a least square fitting algorithm, and sequencing the maximum peak value and the corresponding frequency P (x, y) sequence according to the magnitude of the peak value after extraction to obtain a peak value sequence Px (x, y), wherein the sequence is a group of coordinate values, the abscissa x represents the frequency, and the ordinate y represents the corresponding peak value.

6. The method for real-time online detection of polygonal faults of high-speed train wheels according to claim 5, wherein 40-60 maximum peaks and corresponding frequencies are extracted to form a P (x, y) sequence.

7. The method for detecting the polygon fault of the high-speed train wheel on line in real time according to the claim 2, 3 or 5, characterized in that the method for extracting the polygon order feature sequence OPn (x, y) from the peak value sequence by adopting the difference threshold algorithm comprises the following steps: setting a coupling coefficient, coupling the frequency values of a wheel polygon order frequency value sequence On (St) and a peak value sequence Px (x, y) one by one, when | On (St) -Px (x) | is less than or equal to the coupling coefficient, indicating that the frequency value is a fault feature, extracting the frequency value and outputting a corresponding peak value to form a polygon order feature sequence OPn (x, y), wherein the feature sequence OPn (x, y) is a group of coordinate values, an abscissa x represents the extracted order frequency value, and an ordinate y represents a corresponding peak value.

8. The method for detecting the polygonal fault of the high-speed train wheel on line in real time according to the claims 2, 3 or 5, wherein the value range of the weighting coefficient a is 0.2-0.8.

9. The method for detecting the polygonal fault of the high-speed train wheel on line in real time according to the claim 2, 3 or 5, characterized in that the method for judging whether to alarm is as follows: and comparing the polygon evaluation index Kd with an alarm threshold value, and when the abnormality is displayed for a continuous period of time and the polygon order Poly is the same, taking the confidence that the abnormality is subjected to alarm prompt, otherwise, not performing the alarm prompt.

10. The method for detecting the polygonal fault of the high-speed train wheel on line in real time according to claim 9, wherein when a polygonal evaluation index Kd is smaller than a lowest value of an alarm threshold, the operation is normal, and an alarm state does not show abnormality; the situations in which the abnormality is displayed include: when the lowest value of the alarm threshold value is less than or equal to Kd and less than the middle value of the alarm threshold value, setting the alarm state as pre-judgment alarm; when the Kd between the alarm threshold values is less than or equal to the Kd and less than the maximum alarm threshold value, setting the alarm state as early warning alarm; when Kd is larger than or equal to the highest alarm threshold value, the alarm state is set as alarm.