A method for detecting damages of railway vehicle wheelsets based on sound

Abstract

The present invention discloses a method for detecting damages of railway vehicle wheelsets based on sound, belonging to the field of railway transportation safety; specifically, first, a speed measurement and positioning device, a sound acquisition device, and a control and processing device are installed outside the railway operation clearance; when a train passes by, the speed measurement and positioning device holographically records the speed and position of the wheels, and at the same time, the sound acquisition device holographically records the sound; based on the speed and position, a "time-position curve" S1 of each wheel is drawn; based on the position and sound, a "time-position curve" S2 corresponding to each wheel is drawn; finally, the distance D between the two "time-position curves" S1 and S2 is calculated, and it is determined whether the distance D is less than a set threshold. If so, it is determined that there is abnormal sound in the wheel, that is, the wheel is damaged or the running quality is poor; otherwise, the wheel is normal, and the detection of wheel damage, the detection of running quality, and the detection of overloading and offloading are completed. The present invention adopts a non-contact dynamic detection method and does not transform the railway line, with strong practicability.

Description

Technical Field

[0001] The present invention belongs to the field of railway transportation safety, and particularly relates to a method for detecting damages of railway vehicle wheelsets based on sound. Background Art

[0002] The railway transportation in China has developed rapidly. The total operating mileage of the national railway has reached 1.5 million kilometers, basically covering cities with a population of over 200,000. The construction of the "eight vertical and eight horizontal" railway transportation network has been completed, forming the world's most modern railway network and high-speed railway network with reasonable layout and wide coverage. However, the problem of railway transportation safety is still of vital importance.

[0003] The wheelset is a large component that bears railway vehicles and is one of the key components to ensure the safe operation of trains. The TPDS (Trackside Dynamic Monitoring System for Vehicle Running Quality), a subsystem of the railway vehicle operation safety monitoring system (5T), is currently the only system that can dynamically monitor components such as wheelsets. It uses contact sensors such as acceleration sensors, force sensors, or fiber optic sensors to identify the running state of vehicle wheels, wheel tread abrasions, and measure overloading and offloading.

[0004] The TPDS system is complex to install and has strict detection conditions. It requires installing fixtures on the rails, grinding the rails, replacing sleepers and track slabs, etc., which are direct transformation operations on the railway line. Moreover, the operation is within the railway operation clearance, with high risks in construction, operation, and maintenance, and high costs, resulting in limitations of the TPDS system. Due to the characteristic limitations of the above sensors, the TPDS also has relatively strict restrictions on the running speed of trains.

[0005] In addition to the TPDS, the existing means for detecting wheelsets is manual inspection after the vehicle enters the depot, with low detection efficiency, heavy workload, and high missed inspection rate. Therefore, the detection devices and methods for components such as wheelsets have fallen behind the development status of railway transportation. Summary of the Invention

[0006] In view of the above problems, the present invention proposes a method for detecting damages of railway vehicle wheelsets based on sound, which can be detected non-contact outside the railway operation clearance without transforming the railway line; can be detected dynamically without affecting the operation of railway vehicles; and can simultaneously complete the detection of wheelset damages such as defects, abrasions, peeling, out-of-roundness, etc., the detection of operation quality, and the detection of overloading and offloading.

[0007] The method for detecting damages of railway vehicle wheelsets based on sound specifically comprises the following steps:

[0008] Step 1: Install a speed measurement and positioning device, a sound acquisition device, and a control and processing device outside the railway operation clearance.

[0009] The control and processing device is simultaneously connected to the speed measurement and positioning device and the sound collection device, and receives the speed measurement and positioning information and the collected sound information.

[0010] The speed measurement and positioning device is installed on one side outside the track limit and consists of at least two sensors. The spatial positions between the sensors are relatively fixed; by recording the same wheel at different times with different sensors, the running speed of the wheel is calculated based on the positions of the sensors and the recorded times.

[0011] The sound collection device includes a first sound collection module and a second sound collection module, which are respectively located on the left and right symmetric sides of the track and are relatively fixed in spatial position with respect to the railway track. Each sound collection module consists of no less than 3 microphones, and the spatial positions of different microphones are relatively fixed;

[0012] The control and processing device is installed on the same side outside the track limit as the speed measurement and positioning device and has a time synchronization function and a signal processing function. It synchronizes the time of the speed measurement and positioning device and the sound collection device, and comprehensively processes the speed measurement signal, positioning signal, and sound signal.

[0013] Step 2: When the measured wheel pair of the train passes by, the speed measurement and positioning device holographically records the speed information and position information of the wheel, and at the same time, the sound collection device holographically records the sound information.

[0014] First, each sensor of the speed measurement and positioning device respectively records the time when the current measured wheel pair arrives at each sensor, and further uses the time difference and the spatial position of the sensor to measure the speed of the measured wheel pair;

[0015] The calculation formula for the speed v is: v = (P2 - P1) / (t2 - t1) * k.

[0016] P1 is the spatial position of the first sensor, P2 is the spatial position of the second sensor, t1 is the time when the wheel is recorded by the first sensor, t2 is the time when the wheel is recorded by the second sensor, and k is the correction coefficient;

[0017] At the same time, the measured wheel pair is identified and positioned using the recorded information of the sensor to determine the position of the wheel;

[0018] Then, the sound collection device holographically collects the sound of the measured wheel pair, performs beam steering processing on the recorded holographic sound signal, enhances the directivity of the sound signal, and extracts the sound signal of the wheel according to the dynamic position of the measured wheel pair during operation;

[0019] The specific process is as follows:

[0020] First, for the current wheel position to be measured, calculate the phase information of each sound channel in the sound acquisition device relative to this wheel position, adjust each phase, and superimpose all the adjusted signals after increasing their weights;

[0021] The current wheel position to be measured is P w , and the position of channel 1 in the sound acquisition device is P s1 , and the phase adjustment amount is:

[0022]

[0023] where f is the sound frequency and V0 is the speed of sound;

[0024] Similarly, obtain the phase adjustment amounts of the other N sound channels

[0025] Superimposed signal:

[0026] where S1, S2... S N are the original sound signals of each channel respectively, h1, h2... h N are the position weights of each channel respectively, and f() is the transfer function of the phase weight.

[0027] Similarly, repeatedly calculate the superimposition of the weighted sound signals of the next wheel position until the signals of all positions are processed, and then perform splicing and combination followed by filtering and optimization;

[0028] Finally, utilize the positioning of the wheels by the sensors, and through pointing processing, point to the positions of the wheels, and the corresponding signals are the signals of the wheels;

[0029] The wheel sound signals include: the sound of the wheel contacting the track, the sound of the wheel vibrating itself, the sound of the track vibrating, the sound of the train body vibrating, the sound of the motor and the fan, etc.

[0030] Step 3: Draw the "time-position curve" S1 of each wheel based on the speed information and the position information;

[0031] In the "time-position curve" S1, the abscissa represents the time information and the ordinate represents the position information;

[0032] Step 4: Draw the "time-position curve" S2 corresponding to each wheel based on the position information and the sound information;

[0033] The specific process is as follows:

[0034] First, frame the holographic sound signals corresponding to each wheel collected and perform discretization.

[0035] Then, perform one-dimensional sound field positioning on each frame of sound signals along the track direction where the wheels pass, determine the sound field positions of each frame, and form a sound field position array;

[0036] Finally, based on the sub-frame information of the sound field position array, fit the discrete sound field positions and draw the "time-position curve" S2 of the sound field.

[0037] Step Five: Calculate the distance D between the two "time-position curves" S1 and S2, and determine whether the distance D is less than the set threshold. If so, it is determined that there is abnormal sound in the wheel, that is, the wheel is damaged or the running quality is poor; otherwise, the wheel is normal, and the detection of wheel damage, running quality detection, and overloading detection are completed.

[0038] When there is abnormal sound in the wheel, further determine the type and grade of the wheel damage; specifically:

[0039] First, extract the frequency characteristics of the sound signals of each wheel respectively, perform envelope extraction on the frequency characteristics to obtain the envelope signal, and further extract the characteristic frequency of the envelope signal;

[0040] Then, use the amplitude of the characteristic frequency to determine the grade of the wheel damage; use the distribution of the steepness of the impact and the characteristics of the steepness in the envelope signal to determine the type of the wheel damage; use the change of the characteristic frequency components of the envelope signal to assist in determining the defective type of wheel damage.

[0041] When there are damages on both wheels on both sides of the same wheel set, respectively take the envelope signals of the two wheels on the left and right sides, extract the impact positions of the envelope signals, obtain the moments of the impact positions on both sides respectively, and calculate the actual time difference of the impact positions; then, according to the spatial position of the sound acquisition device and the position of the wheels, calculate and obtain the recorded time difference of the moments of the impact positions of the wheels on both sides. Compare the actual time difference of the positions and the recorded time difference. When the difference between the two is greater than the allowable error, it is determined that there is abnormal sound in both wheels; otherwise, the wheel on the side with the earlier moment of the impact position is damaged, and the other side is caused by the transmission of interference from the opposite side and is excluded.

[0042] When the running quality of all wheel sets of the same vehicle is poor, it is determined that the running quality of the vehicle is poor; otherwise, it is determined that the wheels are out-of-round.

[0043] Extract the strength of each wheel of the same vehicle, and perform comparison calculations on the strength of different wheels between both sides of the same wheel set and between the front and rear bogies respectively. When the ratio exceeds the set threshold, it is determined as overloading.

[0044] Based on the strength and wheel speed of each wheel of the same vehicle, calculate the vehicle weight of the vehicle according to the transfer function. When the vehicle weight exceeds the set threshold, it is determined as overloading.

[0045] The advantages of the present invention are:

[0046] 1) A method for detecting damage to the wheelset of a railway vehicle based on sound, which adopts a non-contact detection method and can be carried out outside the railway operation limit without reconstructing the railway line; it is simple and convenient.

[0047] 2) A method for detecting damage to the wheelset of a railway vehicle based on sound, which adopts dynamic detection and does not affect the normal operation of the railway vehicle.

[0048] 3) A method for detecting damage to the wheelset of a railway vehicle based on sound. The detection method has no strict limit on the running speed of the train and can be carried out at speeds above 30 km / h, with strong practicability. Description of the Drawings

[0049] Figure 1 It is a flowchart of a method for detecting damage to the wheelset of a railway vehicle based on sound according to the present invention;

[0050] Figure 2 It is an overall structure diagram of a device for detecting damage to the wheelset of a railway vehicle based on sound according to the present invention;

[0051] 101 - Railway track, 102 - Railway limit, 103 - Speed measurement and positioning device, 104 - Sound acquisition device, 105 - Control and processing device.

[0052] Figure 3 It is a composition diagram of the speed measurement and positioning device described in the present invention;

[0053] 201 - Fixed bracket, 202 - Camera No. 1, 203 - Camera No. 2.

[0054] Figure 4 It is a composition diagram of the sound acquisition module described in the present invention;

[0055] 301 - Microphone fixed bracket, 302 - Microphone No. 1, 303 - Microphone No. 2, 304 - Microphone No. 3.

[0056] Figure 5 It is the time-position curve S1 of each wheel drawn according to the present invention.

[0057] Figure 6 It is the determination process of whether there is abnormal sound in the wheel according to the present invention;

[0058] 601 - Sound field position, 602 - Time-position curve S2, 603 - Time-position curve S1.

[0059] Figure 7 It is a schematic diagram of the beam channel according to the present invention;

[0060] 701 - Wheel to be tracked, 702 - Beam channel.

[0061] Figure 8 This is the diagnostic atlas of wheel damage of the present invention;

[0062] 801 - Wheel sound signal diagram, 802 - Spectrum diagram of the wheel, 803 - Extracted first - order envelope diagram, 804 - Normalized characteristic frequency diagram, 805 - Extracted second - order envelope diagram.

[0063] Figure 9 This is the diagnostic atlas of wheel damage type of the present invention;

[0064] 901 - Normalized characteristic frequency diagram, 902 - Envelope diagram, 903 - Level 1 damage threshold, 904 - Level 2 damage threshold, 905 - Level 3 damage threshold, 906 - Level 4 damage threshold, 907 - Peak value of characteristic frequency, 908 - Peak value of envelope, 909 - Bottom value of envelope.

[0065] Figure 10 This is the diagnostic atlas for discriminating interference of wheel damage of the present invention;

[0066] 1001 - Wheel envelope at the detection end, 1002 - Wheel envelope at the opposite end, 1003 and 1004 - A group of corresponding envelope peak values at both ends, 1005 and 1006 - Another group of corresponding envelope peak values at both ends. Detailed implementation manners

[0067] The following will detail the specific implementation method of the present invention in conjunction with the accompanying drawings.

[0068] The existing detection methods for damage of railway vehicle wheelsets have the following principles: First, the railway vehicle wheelset runs on the track in a rolling form. When the wheelset contacts the track, noise is generated, and the intensity of the noise is related to the contact time, contact area, and extrusion force between the wheelset and the track. Second, when the wheelset and the track are healthy, the railway vehicle wheelset is a regular circle, the tread of the wheelset is flat, the railway track is flat, the vehicle runs smoothly, and the contact between the wheelset and the track generates stable noise. Third, when there is damage to the wheelset, the shape of the wheelset is irregular, the rolling is not smooth, the frequency characteristics of the noise change, and the damaged wheelset contacts the track to generate impacts, exciting impact sounds. Fourth, when the vehicle runs unevenly and the operation quality is poor, the contact between the wheelset and the track changes periodically, and the intensity of the generated noise will change periodically. Fifth, when the loads of each wheelset of the vehicle are unbalanced, the wheelset contacts the track to generate stable noise, and due to different extrusion forces, the noise will have periodic differences.

[0069] Based on the above principles, the present invention proposes a method for detecting damage of railway vehicle wheelsets based on sound, which is based on a wheelset damage detection device installed outside the railway operation limit and does not require modification of the railway line.

[0070] As Figure 1As shown in the figure, the method for detecting damage to the wheelset of a railway vehicle based on sound comprises the following specific steps:

[0071] Step 1: Install a railway vehicle wheelset damage detection device composed of a speed measurement and positioning device, a sound acquisition device, and a control and processing device outside the railway operation limit;

[0072] The control and processing device is connected to the speed measurement and positioning device and the sound acquisition device at the same time, and receives speed measurement and positioning information and the acquired sound information.

[0073] The speed measurement and positioning device is installed on one side outside the track limit and consists of at least two sensors. The sensors have a certain spacing in space and their positions are relatively fixed. Through this group of sensors, each sensor will record the same object (such as a wheel) at different times. According to the spatial positions of the sensors and the time difference recorded by multiple sensors, the running speed of the measured wheelset can be judged. Therefore, with the spatial positions relatively fixed and the sensor positions known, the speed can be measured by combining the time information.

[0074] The sound acquisition device includes a first sound acquisition module and a second sound acquisition module, which are used to acquire the holographic sound field, are respectively located on the left and right symmetrical sides of the track, and are relatively fixed in space with respect to the railway track. Each sound acquisition module consists of no less than 3 microphones, and the spatial positions of different microphones are relatively fixed; the sound acquisition device performs holographic acquisition of the sound of the measured wheel, and tracks, separates, and extracts it.

[0075] The control and processing device is installed on the same side as the speed measurement and positioning device outside the track limit, has a time synchronization function and a signal processing function, synchronizes the time of the speed measurement and positioning device and the sound acquisition device, and comprehensively processes the speed measurement signal, positioning signal, and sound signal.

[0076] Step 2: When the measured wheelset of the train passes by, the speed measurement and positioning device holographically records the speed information and position information of the wheel, and at the same time, the sound acquisition device holographically records the sound information.

[0077] First, each sensor of the speed measurement and positioning device respectively records the time when the current wheelset to be measured reaches its respective position, and further measures the speed of the measured wheelset by using the time difference and the spatial position of the sensors;

[0078] Taking two sensors as an example, the spatial position of sensor 1 is P1, the spatial position of sensor 2 is P2, the time when the wheel is recorded by sensor 1 is t1, the time when the wheel is recorded by sensor 2 is t2, and the correction coefficient is k; then the speed v = (P2 - P1) / (t2 - t1)*k.

[0079] At the same time, the measured wheelset is directly identified and positioned by using the recorded information of the sensors.

[0080] For example, a camera is used as a sensor to directly take pictures, and the wheels are identified by matching with the wheel pictures. If a laser is used, matching is performed according to the characteristics of the change in the measured distance and the actual lateral distance change of the train, so as to identify the wheels.

[0081] Then, the sound acquisition device performs holographic acquisition on the sound of the measured wheel set, and performs beam pointing processing on the recorded holographic sound signal to enhance the directivity of the sound signal, and extracts the sound signal of the wheel according to the dynamic position of the wheel during operation;

[0082] The specific process is as follows: First, for the current position of the wheel to be measured, calculate the phase information of each sound channel in the sound acquisition device relative to the position of the wheel at that time, and adjust the signal phase of each sound channel; then, add weights to the signals of all sound channels after phase adjustment and perform superposition; similarly, process the sound signals of the next wheel position, repeat processing the signals of all positions of the wheel, and then perform splicing and combination, and perform filtering and optimization;

[0083] The current position of the wheel to be measured is P w The position of channel 1 in the sound acquisition device is P s1 The phase adjustment amount is:

[0084]

[0085] where f is the sound frequency and V0 is the speed of sound;

[0086] Similarly, the phase adjustment amounts of the other N sound channels are obtained

[0087] Superimposed signal:

[0088] where, S1, S2... S N are the original sound signals of each channel respectively, and h1, h2... h N are the position weights of each channel respectively, and f() is the transfer function of the phase weight.

[0089] Since the position of the wheel has been located using the information of the sensor, the position of the wheel can be determined; according to the position of the wheel, through the pointing process, point to the position of the wheel, and the corresponding signal is the signal of the wheel; the wheel sound signal is all the sound signals related to the wheel, and also includes the external noise and interference sounds recorded at the same time.

[0090] The sounds mainly include: the sound of the wheel contacting the track, the sound of the wheel vibrating itself, the sound of the track vibrating, the sound of the train body vibrating, the sound of the motor and the fan, etc.

[0091] Step 3: Based on the speed information and position information, draw the "time-position curve" S1 for each wheel;

[0092] The abscissa of the "time-position curve" S1 represents the time information, and the ordinate represents the position information;

[0093] Step 4: Based on the position information and sound information, draw the "time-position curve" S2 corresponding to each wheel;

[0094] The specific process is as follows:

[0095] First, frame the holographic sound signals corresponding to each wheel collected and perform discretization.

[0096] Then, perform one-dimensional sound field positioning on each frame of the sound signal along the track direction where the wheel passes, determine the sound field position of each frame, and form a sound field position array.

[0097] Finally, based on the frame information of the sound field position array, fit the discrete sound field positions and draw the "time-position curve" S2 of the sound field.

[0098] The fitting methods include: least squares fitting, gradient descent method, Gauss-Newton method, or Levenberg-Marquardt method, etc.

[0099] Step 5: Calculate the distance D between the two "time-position curves" S1 and S2, and determine whether the distance D is less than the set threshold. If so, it is determined that there is abnormal sound in the wheel, that is, the wheel is damaged or the running quality is poor; otherwise, the wheel is normal, and the detection of wheel damage, the detection of running quality, and the detection of overloading are completed.

[0100] Use the Euclidean distance to calculate the distance D between the two curves. When there is abnormal sound in the wheel, further determine the type and grade of the wheel damage; specifically:

[0101] First, extract the frequency characteristics of the sound signal of each wheel respectively, perform envelope extraction on the frequency characteristics to obtain the envelope signal, and further extract the characteristic frequency of the envelope signal; at the same time, extract the time-domain characteristics of the sound signal and calculate the intensity of the sound signal.

[0102] Then, combine the sound signal, characteristic frequency, envelope signal, characteristic frequency of the envelope signal, time-domain characteristics of the sound signal, and parameters such as intensity with the established model for matching to determine the type of wheel damage; determine the grade of wheel damage according to the model threshold, and determine whether the wheel is out-of-round or the running quality is poor by matching with the model. Use the distribution and characteristics of the steepness of the impact in the envelope signal to determine the type of wheel damage; use the change of the characteristic frequency components of the envelope signal to assist in determining the defective wheel damage.

[0103] When there are damages on both wheels on both sides of the same wheel set, the envelope signals of the two wheels on the left and right sides are respectively taken, the impact positions of the envelope signals are extracted, the moments of the impact positions on both sides are respectively obtained, and the actual time difference of the impact positions is calculated; then, according to the spatial position of the sound acquisition device and the position of the wheels, the recorded time difference of the moments of the impact positions of the wheels on both sides is calculated. Compare the actual time difference of the positions with the recorded time difference. When the difference between the two is greater than the allowable error, it is determined that there are abnormal sounds on both wheels. Otherwise, the wheel on the side with the earlier moment of the impact position is damaged, and the other side is caused by the interference transmission of the opposite side and is excluded. The measurement of the difference can also be converted to the distance dimension for calculation.

[0104] Extract the out-of-roundness and poor operation quality results of each wheel of the same vehicle. When all the wheel sets of the same vehicle have poor operation quality, it is determined that the operation quality of the vehicle is poor. Otherwise, it is determined as out-of-roundness of the wheels.

[0105] Extract the strength of each wheel of the same vehicle, and respectively compare and calculate the strength between the two sides of the same wheel set and between different wheels of the front and rear bogies. When the ratio exceeds the set threshold, it is determined as partial load.

[0106] Based on the strength and wheel speed of each wheel of the same vehicle, calculate the vehicle weight of the vehicle according to the transfer function. When the vehicle weight exceeds the set threshold, it is determined as overloading.

[0107] The present invention includes: First, holographically record the speed information, position information, and sound information when the train passes by. Then, based on the speed information and position information, draw the "time-position curve" S1 of each wheel. At the same time, based on the position information and sound information, determine the "time-position curve" S2 corresponding to each wheel; Finally, calculate the distance D between the two "time-position curves" S1 and S2, and further compare it with the set threshold to determine whether there is an abnormal sound, that is, whether the wheel has damages (defects, scratches, peeling, out-of-roundness, etc.) or poor operation quality; or the wheel is normal, and the detection of wheel damages is completed.

[0108] Furthermore, the wheel damage model uses the characteristic frequency and the amplitude of the characteristic frequency to determine the level of damage, uses the distribution of the steepness of the envelope impact and the characteristics of the steepness to determine the type of wheel damage, and uses the change of the frequency components to assist in determining the defect-type wheel damage. The detection method uses methods such as comparing the damage types of the same vehicle for operation quality detection and assisting in the detection of out-of-roundness-type wheel damage. The detection method uses the correlation between the sound intensity and the load, and uses methods such as coaxial, same-side, and same-vehicle comparisons for overloading and partial load detection.

[0109] The extraction of the sound signal forms a directional beam, effectively reducing the interference of adjacent wheels on the same side. However, the wheels on both sides of the same wheel set are in the same line position in space, and the beam pointing suppression ability is weak. Therefore, it is necessary to exclude the interference from the opposite side of the source.

[0110] Embodiment:

[0111] As Figure 2 shown, the detection device consists of a speed measurement and positioning device 103, a sound acquisition device 104, a control and processing device 105 and their connecting wires. Each part of the detection device is installed outside the railway operation limit 102, and there is no need to transform the railway track 101.

[0112] As Figure 3 shown, the speed measurement and positioning device consists of a No. 1 camera 202, a No. 2 camera 203 and a fixed bracket 201. The No. 1 camera 202 and the No. 2 camera 203 are installed at the same height and parallel to the plane where the railway track 101 is located, parallel to the railway track 101, and the center distance between the two is 10 cm. The No. 1 camera 202 and the No. 2 camera 203 simultaneously take pictures of the passing train wheelset, and determine the position of the wheelset through image recognition. The speed is measured through the time difference of the same position captured by the No. 1 camera 202 and the No. 2 camera 203 and the center distance between the two cameras on the fixed bracket 201.

[0113] The sound acquisition device 104 is as Figure 4 shown. The No. 1 microphone 302, the No. 2 microphone 303 and the No. 3 microphone 304 are fixed on the microphone fixed bracket 301. The central positions of the No. 1 microphone 302, the No. 2 microphone 303 and the No. 3 microphone 304 form an equilateral triangle with a side length of 5 cm. The No. 1 microphone 302 and the No. 3 microphone 304 are installed at the same height and parallel to the plane where the railway track 101 is located, parallel to the railway track 101. The plane where the No. 1 microphone 302, the No. 2 microphone 303 and the No. 3 microphone 304 are located is perpendicular to the plane where the railway track 101 is located. The connection line of the first sound acquisition module and the second sound acquisition module is parallel to the plane where the railway track 101 is located and perpendicular to the railway track 101. The No. 1 microphone 302, the No. 2 microphone 303 and the No. 3 microphone 304 are omnidirectional microphones of the same specification. The sound acquisition device composed of a total of six microphones and the fixed bracket records the holographic sound signal of the train, and performs tracking, separation and extraction of the wheel sound.

[0114] The control and processing device uses an industrial computer, and uses the operating system time as the synchronization time of the speed measurement and positioning device and the sound acquisition device. It has the ability of signal processing and operation, and comprehensively processes the speed measurement signal, positioning signal and sound signal.

[0115] The detection method includes steps such as recording information, information classification and extraction, abnormal sound detection, wheel damage classification and grading, interference information discrimination, out-of-round detection, vehicle operation quality detection, and overloading and offloading detection.

[0116] When the train runs on the railway track 101 and passes through the detection device, the speed measurement and positioning device 103 records the speed information and position information of the train, extracts the speed V of each wheel, and the 1st microphone 302, 2nd microphone 303, and 3rd microphone 304 of the sound acquisition device 104 record the holographic sound recording of the train.

[0117] Use the speed information and position information of the speed measurement and positioning device 103 to locate and extract the detection wheels, use the speed information and position information to determine the position relationship and draw the "time-position curve" S1, as Figure 5 shown;

[0118] Extract the information segment corresponding to the detection wheel in the holographic sound information according to the position relationship, frame the sound information of the processed wheel set, discretize the track direction, calculate the sound field intensity along the track direction for each frame to obtain the sound field curve, and extract the sound field position. By sequentially calculating each discrete frame, the sound field position array is finally obtained. As Figure 6 shown, fit the sound field position 601 in the sound field position array to obtain the "time-position curve" S2 602 of the strongest sound field.

[0119] Using the included angle between the time-position curve S1 603 and the time-position curve S2 602, and the 0-point position vector, calculate the distance D between the curves S1 and S2 according to the weight formula. If this distance is less than the set distance threshold, it is determined as abnormal sound, that is, initially judged as abnormal. The calculation method of the distance D is the curve included angle multiplied by the 0-point position vector.

[0120] For the wheel 701 initially judged as abnormal, establish a beam channel 702, as Figure 7 shown, continuously track the position of the wheel 701, perform beam pointing processing, enhance the sound information, and extract the wheel sound signal 801.

[0121] As Figure 8 shown, calculate the spectrum 802 for the wheel sound signal 801, perform Hilbert transform on the wheel sound signal 801 and extract the first-order envelope 803 and second-order envelope 804, take the maximum value of the first-order envelope 803 as the sound signal intensity, perform Fourier transform on the first-order envelope 803 and normalize the sound signal intensity to obtain the normalized characteristic frequency map 805.

[0122] As Figure 9 shown, when there is a characteristic frequency peak 907 of a specific frequency, it is determined that the wheel is damaged. The numerical value of the characteristic frequency peak 907 indicates the order type of the wheel damage, and the height of the characteristic frequency peak 907 is compared with the damage thresholds 903, 904, 905, and 906 to determine the wheel damage level.

[0123] Furthermore, as Figure 9As shown in the figure, the steepness of the impact in the envelope 902 is calculated. The calculation method of the steepness is the envelope height divided by the envelope width. The envelope width is calculated by subtracting two envelope valleys 909. The envelope height is calculated as the maximum difference between the envelope peak 908 and two adjacent envelope valleys 909. The steepness calculations of all impacts in the envelope 902 are statistically analyzed. If the steepness distribution conforms to the distribution law, the damage types are determined as out-of-roundness and poor running quality. The maximum steepness outside the range of the distribution law is discriminated. When the steepness is between 0.5 and 0.7, the wheel damage is determined as abrasion. When the steepness is greater than 0.7, the wheel damage is determined as spalling. When the steepness is less than 0.5, it is determined as a defect.

[0124] Furthermore, for a wheel with a damage type of defect, the spectrum 802 of this wheel is compared with the standard spectrum. When the deviation value is greater than the threshold, the wheel damage level is increased by 1 level.

[0125] Interference is likely to occur between the wheels at both ends of the same wheel set. As Figure 10 shown in the figure, 1001 is the envelope signal of the damaged-end wheel, and 1002 is the envelope signal of the opposite-end wheel. For the corresponding envelope peaks 1003 and 1004, the time difference between them is 0.004 seconds. Calculated at the sound speed of 340 m / s, the corresponding distance difference is 1.36 meters. At the corresponding position, the distance difference between the wheel and the two modules of the sound acquisition device 104 is the spacing distance 1.435 meters of the railway track 101, and its deviation is 5%, which is less than the set threshold of 10%. It is determined that the envelope peak 1403 is an opposite-end interference envelope and is excluded. For the corresponding envelope peaks 1005 and 1006, the time difference between them is 0.1 seconds. Calculated at the sound speed of 340 m / s, the corresponding distance difference is 34 meters. At the corresponding position, the distance difference between the wheel and the two modules of the sound acquisition device 104 is the spacing distance 1.435 meters of the railway track 101, and its deviation is 24 times the normal distance, which is greater than the set threshold of 10%. It is determined that the envelope peaks 1005 and 1006 are independent envelopes and are retained.

[0126] The results of out-of-roundness and poor running quality of the wheels of the same vehicle are extracted. When all the wheels of a vehicle are determined to have out-of-roundness and poor running quality, it is determined that the running quality is poor. Otherwise, the results of out-of-roundness and poor running quality involved in this vehicle are determined as wheel out-of-roundness.

[0127] The strengths of the wheels of the same vehicle are extracted, and the strength comparison calculations are respectively carried out for different wheels on both sides of the same wheel set and between the front and rear bogies. When the ratio exceeds the set threshold of 20%, it is determined as uneven load. The weight of the vehicle is calculated based on the strengths and wheel speeds of the wheels of the same vehicle according to the transfer function. When the vehicle weight exceeds the set threshold, it is determined as overloading.

[0128] So far, the above detection method has completed the detection of wheel set damage (defects, abrasions, peeling, out-of-roundness, etc.), completed the operation quality detection, and completed the overloading and off-loading detection.

[0129] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, various changes and modifications can be made to the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included within the scope of the claims of the present application.

Claims

1. A method for detecting damages of railway vehicle wheelsets based on sound, characterized in that, The specific steps are as follows: First, install a speed measurement and positioning device, a sound acquisition device, and a control and processing device outside the railway operation clearance; when the measured wheel pair of the train passes by, the speed measurement and positioning device holographically records the speed information and position information of the wheel, and at the same time, the sound acquisition device holographically records the sound information; Then, based on the speed information and position information, draw the "time-position curve" S1 of each wheel; at the same time, based on the position information and sound information, draw the "time-position curve" S2 corresponding to each wheel; Finally, calculate the distance D between the two "time-position curves" S1 and S2. The calculation method of the distance D is the curve included angle multiplied by the position vector of the zero point, and determine whether the distance D is less than the set threshold. If so, it is determined that there is abnormal sound in the wheel, that is, the wheel is damaged or the running quality is poor; otherwise, the wheel is normal; complete the detection of wheel damage, running quality detection, and overloading detection; The sound acquisition device holographically acquires the sound of the measured wheel pair, performs beam pointing processing on the recorded holographic sound signal, enhances the directivity of the sound signal, and extracts the wheel sound signal corresponding to the current position according to the current position of the wheel pair to be measured during operation; the specific extraction process is: First, for the current position of the wheel to be measured, calculate the phase information of each sound channel in the sound acquisition device relative to the position of the wheel, adjust each phase, and superimpose all the adjusted signals after adding weights; The current position of the wheel to be measured is P w , and the position of Channel 1 in the sound acquisition device is P s1 , and the phase adjustment amount is: where f is the sound frequency and V0 is the speed of sound; Similarly, the phase adjustment amounts of the other N sound channels are obtained Superimposed signal: Among them, S1, S2... S N are the original sound signals of each channel respectively, h1, h2... h N are the position weights of each channel respectively, and f() is the transfer function of the phase weight; Similarly, repeat the calculation of the weighted superposition of the sound signals at the next wheel position until the signals at all positions are processed, and then perform splicing, combination, and filtering optimization; Finally, use the positioning of the wheel by the sensor, and through the pointing process, point to the position of the wheel, and the corresponding signal is the wheel sound signal.

2. The method for detecting damage to the wheelset of a railway vehicle based on sound according to claim 1, wherein The control and processing device is connected to the speed measurement and positioning device and the sound acquisition device at the same time, and receives the speed measurement and positioning information and the acquired sound information; The speed measurement and positioning device is installed on one side outside the track limit and consists of at least two sensors. The same wheel is recorded by different sensors at different times. According to the positions of the sensors and the recorded time, the running speed of the wheel is calculated; The sound acquisition device includes two sound acquisition modules, which are respectively located on the left and right symmetric sides of the track and are relatively fixed in the spatial position with respect to the railway track; The control and processing device is installed on the same side outside the track limit as the speed measurement and positioning device, has a time synchronization function and a signal processing function, synchronizes the time of the speed measurement and positioning device and the sound acquisition device, and comprehensively processes the speed measurement signal, positioning signal, and sound signal.

3. The method for detecting damage of railway vehicle wheelsets based on sound according to claim 2, wherein, Each of the sound acquisition modules consists of no less than 3 microphones, and the spatial positions of different microphones are relatively fixed.

4. The method for detecting damage of railway vehicle wheelsets based on sound according to claim 1, characterized in that, Each sensor of the speed measurement and positioning device respectively records the time when the current measured wheel pair arrives at each sensor. Further, the time difference and the spatial position of the sensors are used to measure the speed of the measured wheel pair to obtain the speed information of the wheel; at the same time, the recorded information of the sensors is used to identify and position the measured wheel pair to determine the current position information of the wheel.

5. The method for detecting damage to railway vehicle wheelsets based on sound according to claim 1, wherein, In the "time-position curve" S1 of the wheel, the abscissa represents the time information and the ordinate represents the position information.

6. The method for detecting damage to the wheelset of a railway vehicle based on sound according to claim 1, wherein The specific drawing process of the "time-position curve" S2 is as follows: First, frame the holographic sound signals corresponding to each wheel collected and perform discretization; Then, perform one-dimensional sound field positioning on each frame of the sound signal along the track direction through which the wheel passes to determine the sound field position of each frame and form a sound field position array; Finally, based on the frame information of the sound field position array, fit the discrete sound field positions to draw the "time-position curve" S2 of the sound field.

7. The method for detecting damage of railway vehicle wheelsets based on sound according to claim 1, wherein, When there is abnormal noise in the wheel, further determine the type and grade of the wheel damage; specifically: First, extract the frequency characteristics of the sound signal of each wheel respectively, perform envelope extraction on the frequency characteristics to obtain an envelope signal, and further extract the characteristic frequency of the envelope signal; Then, use the amplitude of the characteristic frequency to determine the grade of the wheel damage; use the distribution and characteristics of the steepness of the impact in the envelope signal to determine the type of the wheel damage; use the change of the characteristic frequency components of the envelope signal to assist in determining the defective wheel damage; When there are damages on both wheels on both sides of the same wheel set, take the envelope signals of the two wheels on the left and right sides respectively, extract the impact positions of the envelope signals, obtain the moments of the impact positions on both sides respectively, and calculate the actual time difference of the impact positions; then, according to the spatial position of the sound acquisition device and the position of the wheel, calculate the recorded time difference of the moments of the impact positions of the two wheels on both sides; compare the actual time difference of the positions and the recorded time difference. When the difference between the two is greater than the allowable error, it is determined that there is abnormal noise on both wheels. Otherwise, the wheel on the side with the earlier impact position moment is damaged, and the other side is caused by the transmission of interference from the opposite side and is excluded.

8. The method for detecting damage of railway vehicle wheelsets based on sound according to claim 1, characterized in that, The judgment of the running quality detection and overloading and off-loading detection is as follows: when the running quality of all wheel sets of the same vehicle is poor, it is determined that the running quality of the vehicle is poor; otherwise, it is determined that the wheel is out of round; Extract the strength of each wheel of the same vehicle and perform comparison calculations on the strength of different wheels between the two sides of the same wheel set and between the front and rear bogies respectively. When the ratio exceeds the set threshold, it is determined as off-loading; Based on the strength and wheel speed of each wheel of the same vehicle, calculate the vehicle weight of the vehicle according to the transfer function. When the vehicle weight exceeds the set threshold, it is determined as overloading.

Images