Train running gear axle box bearing fault monitoring method and system

Abstract

The invention belongs to the technical field of rail transit, and particularly discloses a train running gear axle box bearing fault monitoring method and system, and the method comprises the following steps: installing vibration temperature sensors on a plurality of axle box bearings of a train running gear to collect temperature and vibration data in real time, and extracting the characteristics of the collected data for cooperative processing; performing feature level fusion on the obtained corresponding collaborative feature values, and preliminarily diagnosing the health condition of the bearing; calculating a corresponding fault probability through the collaborative characteristic value, and secondarily diagnosing the health condition of the bearing; and according to the results of the primary diagnosis and the secondary diagnosis, whether early warning information is sent is judged. By adopting the technical scheme, potential faults can be found earlier and early warning can be performed in time through a coordination and fusion strategy, continuous and stable operation of the rail train is ensured, and a powerful guarantee is provided for the safety of train operation.

Description

Technical Field

[0001] The invention belongs to the technical field of rail transit, and relates to a method and system for monitoring axle box bearing faults of a train running gear. Background Art

[0002] The safe operation of rail trains depends on the stability of the running parts, and bearings, as important parts of the running frame, directly affect the operating efficiency and safety of the train. During long-term operation, bearings will suffer from plastic deformation, fatigue fracture, pitting and spalling due to alternating loads and complex environmental factors. Once these faults are ignored, they will not only lead to unplanned downtime and damage to train equipment, but may also cause safety accidents. Therefore, real-time monitoring of the health status of train running frame bearings and high-precision fault diagnosis have become the key to ensuring the safe operation of trains.

[0003] Due to the complex train operating environment, external interference (such as changes in road conditions) can easily affect the data obtained by sensors, resulting in an increase in false alarms, which poses a challenge to fault detection. Summary of the invention

[0004] The purpose of the present invention is to provide a train running gear axle box bearing fault monitoring method and system, which can detect potential faults earlier and issue warnings in time to ensure the continuous and stable operation of rail trains and provide strong protection for the safety of train operations.

[0005] In order to achieve the above object, the basic scheme of the present invention is: a method for monitoring axle box bearing faults of a train running gear, comprising the following steps:

[0006] Vibration temperature sensors are installed on multiple axle box bearings of the train running gear to collect temperature and vibration data in real time;

[0007] Extract the features of the collected temperature and vibration data and perform collaborative processing to obtain the corresponding collaborative feature values;

[0008] The collaborative eigenvalues ​​are fused at the feature level to make a preliminary diagnosis of the health status of the bearing;

[0009] Using the fusion diagnosis method based on DS evidence theory, the corresponding fault probability is calculated through the collaborative eigenvalue to diagnose the health status of the bearing;

[0010] According to the results of the preliminary diagnosis and the secondary diagnosis, if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is also in a faulty state, an early warning message is issued; if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is in a normal state, no early warning message is issued.

[0011] The working principle and beneficial effects of this basic solution are as follows: This technical solution overcomes the disadvantage that a single sensor is susceptible to external interference by coordinating the vibration and temperature characteristic data of multiple acquisition nodes, thereby achieving more accurate bearing fault diagnosis.

[0012] The diagnostic strategy combining feature-level fusion and decision-level fusion is adopted to avoid misdiagnosis and greatly improve the reliability and accuracy of diagnosis. The invention monitors the bearing status in real time, gives early warning, ensures the safe operation of the train, and facilitates efficient maintenance.

[0013] Furthermore, the method of extracting the features of the collected vibration data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is as follows:

[0014] Four corresponding synergistic features in the time and frequency domain are extracted from the vibration data, namely: synergistic RMS value, synergistic peak-to-peak value, synergistic kurtosis value and synergistic root mean square frequency;

[0015] Synergy RMS value X RMS for:

[0016]

[0017] Among them, RMSa represents the RMS value of the vibration of the axle box under test, RMS i represents the vibration RMS value of the i-th axle box, RMSa is not included in all RMSi, and N represents the total number of axle boxes;

[0018] Synergistic peak value X P for:

[0019]

[0020] Where Pa represents the peak-to-peak value of the vibration of the axle box under test, P i represents the peak-to-peak value of the vibration of the i-th axle box, and all Pi do not include Pa;

[0021] Co-kurtosis value X K for:

[0022]

[0023] Among them, Ka represents the vibration kurtosis of the axle box under test, K i represents the vibration kurtosis of the i-th axle box, all Ki excluding Ka;

[0024] Coordinated RMS frequency X RMSF for:

[0025]

[0026] Among them, RMSFa represents the root mean square frequency of the tested axle box, RMSFi represents the root mean square frequency of the i-th axle box, and all RMSFi do not include RMSFa.

[0027] The RMS value is used to measure the energy of the signal and can reflect the health of the bearing during operation; the peak-to-peak value helps capture the large instantaneous fluctuations caused by damage inside the bearing; the kurtosis value can identify the peak characteristics and timely warn of small impacts or early damage; the root mean square frequency extracts the frequency domain characteristics and reveals the frequency information of specific fault modes. Feature-level fusion of multiple features helps to comprehensively monitor the bearing status, thereby improving the accuracy and sensitivity of fault diagnosis.

[0028] Furthermore, the method of extracting the features of the collected temperature data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is as follows:

[0029] Two corresponding collaborative features are extracted from the temperature data, namely: collaborative absolute temperature X T and the coordinated temperature rise rate X VT =:

[0030]

[0031] Where Ta represents the temperature of the axle box under test, T i Indicates the temperature of the i-th axle box, all Ti does not include Ta, N indicates the total number of axle boxes; VTa indicates the temperature rise rate of the axle box under test, VT i Represents the temperature rise rate of the i-th axle box, excluding VTa from all VTi.

[0032] The feature information of multiple acquisition nodes is processed collaboratively to obtain corresponding collaborative feature values, which is beneficial for subsequent use.

[0033] Furthermore, the collaborative eigenvalues ​​are fused at the feature level to preliminarily diagnose the health status of the bearing. The specific steps are as follows:

[0034] Perform principal component analysis (PCA) on the extracted collaborative features and standardize the obtained collaborative feature quantities:

[0035]

[0036] in, is the standardized feature quantity, ξ j is the characteristic quantity, E(ξ j ) is the mean value of the characteristic quantity, S(ξ j ) is the standard deviation of the characteristic quantity;

[0037] All standardized features are combined into a sample matrix X n×m ,for:

[0038]

[0039] Among them, each row represents an observation sample, and each column represents a collaborative feature;

[0040] Through the sample matrix X n×m The covariance matrix M is calculated as:

[0041] M=cov(X n×m );

[0042] The eigenvalues ​​λ of the covariance matrix M i The values ​​are arranged in descending order, and their corresponding eigenvectors form the feature matrix K:

[0043]

[0044] Calculate the contribution rate G of the characteristic quantity i :

[0045]

[0046] Take the first j features whose cumulative contribution rate reaches 85% as the principal element P j ,for:

[0047]

[0048] The pivot element P j Input the trained support vector machine SVM to preliminarily determine whether the bearing is faulty.

[0049] The collaborative eigenvalues ​​are fused at the feature level to make a preliminary diagnosis of the health status of the bearing.

[0050] Furthermore, a fusion diagnosis method based on DS evidence theory is used to calculate the corresponding fault probability through collaborative eigenvalues ​​to perform secondary diagnosis of the health status of the bearing. The specific method is as follows:

[0051] The identification framework H for bearing fault diagnosis is determined as:

[0052] H={h1,h2}

[0053] Among them, h1 indicates a fault, and h2 indicates no fault;

[0054] Based on the collaborative eigenvalues ​​of temperature and vibration data, the fault probability value, no-fault probability value and fault uncertainty probability value corresponding to each collaborative eigenvalue are calculated respectively;

[0055] Normalizing the fault probability value, no-fault probability value and fault uncertainty probability value of each collaborative eigenvalue to obtain the basic probability function of each collaborative eigenvalue;

[0056] Based on the basic probability functions of each collaborative eigenvalue, decision fusion is performed to obtain the final basic probability function of bearing fault; wherein the basic probability function includes fault, no fault and fault uncertainty, and whether there is a fault is determined according to the probability size.

[0057] To avoid misdiagnosis, the characteristic information of multiple acquisition nodes is first processed collaboratively, and the fault probability function of the corresponding collaborative characteristic value is calculated. The fusion diagnosis method based on DS evidence theory is used to make a more accurate diagnosis of the operating status of the bearing.

[0058] Furthermore, based on the collaborative eigenvalues ​​of temperature and vibration data, the steps of respectively calculating the fault probability value, the non-fault probability value and the fault uncertainty probability value corresponding to each collaborative eigenvalue are as follows:

[0059] By cooperating RMS value X RMS Calculate the probability of failure m RMS , the probability of no failure n RMS , Failure uncertainty probability p RMS :

[0060]

[0061] Among them, X RMS represents the collaborative RMS value, α, β, σ represent the correction coefficients of each formula;

[0062] By the co-kurtosis value X K Calculate the probability of failure m K , the probability of no failure n K , Failure uncertainty probability p K :

[0063]

[0064] Among them, X K represents the co-kurtosis value;

[0065] By coordinating the absolute temperature X T Calculate the probability of failure m T , the probability of no failure n T , Failure uncertainty probability p T :

[0066]

[0067] Among them, X T represents the coordinated absolute temperature value;

[0068] Through the coordinated temperature rise rate X VT = Calculate the probability of failure m VT , the probability of no failure nVT , Failure uncertainty probability p VT :

[0069]

[0070] Among them, X VT represents the coordinated temperature rise rate.

[0071] Determine the respective failure probability for subsequent accurate diagnosis.

[0072] Furthermore, the fault probability value, no-fault probability value and fault uncertainty probability value of each collaborative eigenvalue are normalized to obtain the basic probability function of each collaborative eigenvalue as follows:

[0073] By cooperating RMS value X RMS Calculate the basic probability function m of fault 11 (h1), basic probability function of no fault m 11 (h2), basic probability function of failure uncertainty m 11 (H):

[0074]

[0075] By the co-kurtosis value X K Calculate the basic probability function m of fault 12 (h1), basic probability function of no fault m 12 (h2), basic probability function of failure uncertainty m 12 (H):

[0076]

[0077] By coordinating the absolute temperature X T Calculate the basic probability function m of fault 21 (h1), basic probability function of no fault m 21 (h2), basic probability function of failure uncertainty m 21 (H):

[0078] Through the coordinated temperature rise rate X VT = Calculate the basic probability function m of fault 22 (h1), basic probability function of no fault m 22 (h2), basic probability function of failure uncertainty m 22 (H):

[0079]

[0080] The probability of each collaborative eigenvalue is processed to facilitate data calculation.

[0081] Furthermore, based on the basic probability function of each collaborative eigenvalue, decision fusion is performed. The specific steps are as follows:

[0082] The synthesis formula of DS evidence theory is:

[0083]

[0084] Among them, under the definition of DS evidence theory, C is the fused target set, m(C) is the probability function of the target set, A and B represent two sets of different evidence supporting C, m1 and m2 are two basic probability functions under the same recognition framework, and K is the trust conflict degree that describes the conflict between the two, which is:

[0085] The synthesis formula can be written as:

[0086]

[0087] The basic probability function m for the cooperative RMS value 11 and the basic probability function m of the co-kurtosis 12 The basic probability function m1 of vibration is obtained by fusion, namely:

[0088]

[0089] The basic probability function m for the coordinated absolute temperature 21 and the basic probability function m of the coordinated temperature rise rate 22 The basic probability function m2 of temperature is obtained by fusion:

[0090]

[0091] The basic probability function m1 of vibration and the basic probability function m2 of temperature are combined to obtain the final basic probability function m of bearing failure:

[0092]

[0093] The various probability data are fused to reduce the errors that lead to fault probability calculation and improve the accuracy of diagnosis.

[0094] The present invention also provides a train running gear axle box bearing fault monitoring system, comprising a data acquisition module and a processing module, wherein the data acquisition module is used to collect temperature and vibration data in real time, and the output end of the data acquisition module is connected to the input end of the processing module;

[0095] The processing module executes the method of the present invention to monitor the axle box bearing faults of the train running gear.

[0096] This system uses data acquisition modules and processing modules to monitor the axle box bearing faults of the train running gear. It can detect potential faults earlier and issue timely warnings to ensure the continuous and stable operation of rail trains, providing a strong guarantee for the safety of train operations.

[0097] Further, the data acquisition module includes a vibration temperature sensor and a preprocessor;

[0098] The vibration temperature sensor can be detachably mounted on the axle box of the running gear of the rail train, and is used to obtain the temperature information and vibration information of the axle box bearing of the running gear during the operation of the rail train;

[0099] The preprocessor is connected to the output end of the vibration and temperature sensor, and is used to convert the collected vibration and temperature information into a digital signal, and transmit it to the processing module through the mesh network.

[0100] The data acquisition module has a simple structure and uses a wireless mesh network to transmit data, reducing wiring requirements and simplifying installation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 It is a flow chart of a train running gear axle box bearing fault monitoring method of the present invention;

[0102] Figure 2 It is a schematic flow chart of the initial diagnosis of the health status of the bearings in the train running gear axle box bearing fault monitoring method of the present invention;

[0103] Figure 3 It is a schematic flow chart of secondary diagnosis of the health status of the bearing of the train running gear axle box bearing fault monitoring method of the present invention;

[0104] Figure 4 The present invention is a schematic diagram of the installation structure of the vibration temperature sensor of the train running gear axle box bearing fault monitoring system. DETAILED DESCRIPTION

[0105] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0106] In the description of the present invention, it is necessary to understand that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0107] In the description of the present invention, unless otherwise specified and limited, it should be noted that the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a mechanical connection or an electrical connection, or it can be the internal connection between two components. It can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to the specific circumstances.

[0108] The present invention discloses a method for monitoring the fault of axle box bearings of a train running gear, which coordinates the vibration characteristics and temperature characteristics of multiple nodes, comprehensively analyzes multi-dimensional information, and improves the accuracy and reliability of fault diagnosis. Through this coordinated and integrated strategy, potential faults can be discovered earlier and early warnings can be issued in time to ensure the continuous and stable operation of rail trains, providing a strong guarantee for the safety of train operations.

[0109] like Figure 1 As shown, the train running gear axle box bearing fault monitoring method includes the following steps:

[0110] Step (1), installing vibration temperature sensors on multiple axle box bearings of the train running gear to collect temperature and vibration data in real time;

[0111] Step (2), extracting the features of the collected temperature and vibration data and performing collaborative processing to obtain corresponding collaborative feature values;

[0112] The collaborative eigenvalues ​​are fused at the feature level to make a preliminary diagnosis of the health status of the bearing;

[0113] Step (3), using the fusion diagnosis method based on DS evidence theory, the corresponding fault probability is calculated through the collaborative eigenvalues ​​to perform secondary diagnosis on the health status of the bearing;

[0114] Step (4), based on the results of the preliminary diagnosis and the secondary diagnosis, if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is also in a faulty state, an early warning message is issued and the corresponding axle box bearing is inspected and maintained in a timely manner; if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is in a normal state, no early warning message is issued.

[0115] In the detection of the axle box bearing status of the running gear of a rail train, the data of a single vibration sensor is often easily disturbed by environmental factors such as changes in road conditions, leading to misjudgment of abnormal system status. If only relying on the characteristics of a single sensor, such as RMS value, peak-to-peak value or kurtosis value, false alarms may occur due to environmental changes.

[0116] By coordinating the data from multiple sensors, features such as coordinated RMS value, coordinated peak-to-peak value, coordinated kurtosis value and coordinated root mean square frequency can be extracted, effectively reducing the impact of noise and errors, and more stably and reliably capturing the true vibration condition of the system, thereby improving the detection accuracy of abnormal conditions of axle box bearings and reducing the possibility of false alarms.

[0117] In a preferred embodiment of the present invention, the method for extracting the features of the collected vibration data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is:

[0118] Four corresponding synergistic features in the time and frequency domain are extracted from the vibration data, namely: synergistic RMS value, synergistic peak-to-peak value, synergistic kurtosis value and synergistic root mean square frequency;

[0119] Synergy RMS value X RMS for:

[0120]

[0121] Among them, RMSa represents the RMS value of the vibration of the axle box under test, RMS i represents the vibration RMS value of the i-th axle box, RMSa is not included in all RMSi, and N represents the total number of axle boxes;

[0122] Synergistic peak value X P for:

[0123]

[0124] Where Pa represents the peak-to-peak value of the vibration of the axle box under test, P i represents the peak-to-peak value of the vibration of the i-th axle box, and all Pi do not include Pa;

[0125] Co-kurtosis value X K for:

[0126]

[0127] Among them, Ka represents the vibration kurtosis of the axle box under test, K i represents the vibration kurtosis of the i-th axle box, all Ki excluding Ka;

[0128] Coordinated RMS frequency X RMSF for:

[0129]

[0130] Among them, RMSFa represents the root mean square frequency of the tested axle box, RMSFi represents the root mean square frequency of the i-th axle box, and all RMSFi do not include RMSFa.

[0131] In the temperature monitoring of the axle box bearings of the running gear of rail trains, the temperature value and temperature rise rate of a single sensor may be affected by the temperature changes in the vehicle's operating environment, making it difficult to accurately reflect the actual temperature state of the bearing. If only a single sensor is relied upon, changes in ambient temperature may lead to misjudgment.

[0132] By collaboratively processing the temperature data of multiple sensor nodes and comprehensively calculating the collaborative characteristics of temperature and temperature rise rate, the interference of environmental factors can be effectively reduced, the real changes in bearing temperature can be captured more accurately, and false alarms caused by environmental fluctuations can be avoided.

[0133] In a preferred embodiment of the present invention, the method for extracting the features of the collected temperature data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is:

[0134] Two corresponding collaborative features are extracted from the temperature data, namely: collaborative absolute temperature X T and the coordinated temperature rise rate X VT =:

[0135]

[0136] Where Ta represents the temperature of the axle box under test, T i Indicates the temperature of the i-th axle box, all Ti does not include Ta, N indicates the total number of axle boxes; VTa indicates the temperature rise rate of the axle box under test, VT i Represents the temperature rise rate of the i-th axle box, excluding VTa from all VTi.

[0137] In a preferred embodiment of the present invention, Figure 2 As shown in the figure, the collaborative eigenvalues ​​are fused at the feature level to preliminarily diagnose the health status of the bearing. The specific steps are as follows:

[0138] Perform principal component analysis (PCA) on the extracted collaborative features and standardize the obtained collaborative feature quantities:

[0139]

[0140] in, is the standardized feature quantity, ξ j is the characteristic quantity, E(ξ j ) is the mean value of the characteristic quantity, S(ξ j ) is the standard deviation of the characteristic quantity;

[0141] All standardized features are combined into a sample matrix X n×m ,for:

[0142]

[0143] Among them, each row represents an observation sample, and each column represents a collaborative feature;

[0144] Through the sample matrix X n×m The covariance matrix M is calculated as:

[0145] M=cov(X n×m );

[0146] The eigenvalues ​​λ of the covariance matrix M i The values ​​are arranged in descending order, and their corresponding eigenvectors form the feature matrix K:

[0147]

[0148] Calculate the contribution rate G of the characteristic quantity i :

[0149]

[0150] Take the first j features whose cumulative contribution rate reaches 85% as the principal element P j ,for:

[0151]

[0152] The pivot element P j Input the trained support vector machine SVM to preliminarily determine whether the bearing is faulty.

[0153] In a preferred embodiment of the present invention, Figure 3 As shown in the figure, a fusion diagnosis method based on DS evidence theory is used to calculate the corresponding fault probability through collaborative eigenvalues ​​to perform secondary diagnosis of the health status of the bearing. The specific method is as follows:

[0154] Determine the identification framework H for bearing fault diagnosis. Since it is only necessary to determine whether a fault occurs, the identification framework is:

[0155] H={h1,h2}

[0156] Among them, h1 indicates a fault, and h2 indicates no fault;

[0157] It is necessary to calculate the basic probability function of fault m(h1) and the basic probability function of no fault m(h2). At the same time, under the DS evidence theory, the basic probability function of uncertain fault is represented by m(H).

[0158] Although a single sensor has high sensitivity to fault monitoring, it is also susceptible to external interference factors, such as changes in road conditions and ambient temperature, which may lead to errors in fault probability calculation, thereby increasing the risk of false alarms. Therefore, through the coordinated processing of multi-sensor signals, the influence of noise and external interference can be effectively suppressed, thereby improving the accuracy of diagnosis and the reliability of the system.

[0159] Based on the collaborative eigenvalues ​​of temperature and vibration data, the fault probability value, no-fault probability value and fault uncertainty probability value corresponding to each collaborative eigenvalue are calculated respectively;

[0160] Normalizing the fault probability value, no-fault probability value and fault uncertainty probability value of each collaborative eigenvalue to obtain the basic probability function of each collaborative eigenvalue;

[0161] Based on the basic probability functions of each collaborative eigenvalue, decision fusion is performed to obtain the final basic probability function of bearing fault; wherein the basic probability function includes fault, no fault and fault uncertainty, and whether there is a fault is determined according to the probability size.

[0162] In a preferred embodiment of the present invention, based on the collaborative eigenvalues ​​of temperature and vibration data, the steps of respectively calculating the fault probability value, the non-fault probability value and the fault uncertainty probability value corresponding to each collaborative eigenvalue are as follows:

[0163] By cooperating RMS value X RMS Calculate the probability of failure m RMS , the probability of no failure n RMS , Failure uncertainty probability p RMS :

[0164]

[0165] Among them, X RMS represents the collaborative RMS value, α, β, σ represent the correction coefficients of each formula;

[0166] By the co-kurtosis value X K Calculate the probability of failure m K , the probability of no failure n K , Failure uncertainty probability p K :

[0167]

[0168] Among them, X K represents the co-kurtosis value;

[0169] By coordinating the absolute temperature X T Calculate the probability of failure m T , the probability of no failure n T, Failure uncertainty probability p T :

[0170]

[0171]

[0172] Among them, X T represents the coordinated absolute temperature value;

[0173] Through the coordinated temperature rise rate X VT = Calculate the probability of failure m VT , the probability of no failure n VT , Failure uncertainty probability p VT :

[0174]

[0175] Among them, X VT represents the coordinated temperature rise rate.

[0176] In a preferred embodiment of the present invention, the method of normalizing the fault probability value, the non-fault probability value and the fault uncertainty probability value of each collaborative characteristic value to obtain the basic probability function of each collaborative characteristic value is as follows:

[0177] By cooperating RMS value X RMS Calculate the basic probability function m of fault 11 (h1), basic probability function of no fault m 11 (h2), basic probability function of failure uncertainty m 11 (H):

[0178]

[0179] By the co-kurtosis value X K Calculate the basic probability function m of fault 12 (h1), basic probability function of no fault m 12 (h2), basic probability function of failure uncertainty m 12 (H):

[0180]

[0181] By coordinating the absolute temperature X T Calculate the basic probability function m of fault 21 (h1), basic probability function of no fault m 21 (h2), basic probability function of failure uncertainty m 21 (H):

[0182]

[0183] Through the coordinated temperature rise rate X VT = Calculate the basic probability function m of fault 22 (h1), basic probability function of no fault m 22 (h2), basic probability function of failure uncertainty m 22 (H):

[0184]

[0185] In a preferred embodiment of the present invention, decision fusion is performed based on the basic probability function of each collaborative feature value, and the specific steps are as follows:

[0186] The synthesis formula of DS evidence theory is:

[0187]

[0188] Among them, under the definition of DS evidence theory, C is the fused target set, m(C) is the probability function of the target set, A and B represent two sets of different evidence supporting C, m1 and m2 are two basic probability functions under the same recognition framework, and K is the trust conflict degree that describes the conflict between the two, which is:

[0189] The synthesis formula can be written as:

[0190]

[0191] The basic probability function m for the cooperative RMS value 11 and the basic probability function m of the co-kurtosis 12 The basic probability function m1 of vibration is obtained by fusion, that is:

[0192]

[0193] The basic probability function m for the coordinated absolute temperature 21 and the basic probability function m of the coordinated temperature rise rate 22 The basic probability function m2 of temperature is obtained by fusion:

[0194]

[0195] The basic probability function m1 of vibration and the basic probability function m2 of temperature are combined to obtain the final basic probability function m of bearing failure:

[0196]

[0197] The present invention also provides a train running gear axle box bearing fault monitoring system, comprising a data acquisition module and a processing module, wherein the data acquisition module is used to collect temperature and vibration data in real time, and the output end of the data acquisition module is electrically connected to the input end of the processing module. The processing module executes the method of the present invention to monitor the train running gear axle box bearing fault.

[0198] In a preferred embodiment of the present invention, the data acquisition module includes a vibration temperature sensor (i.e., a vibration temperature sensing module) and a pre-processor, such as Figure 4 As shown, the vibration temperature sensor can be detachably installed on the axle box of the running gear of the rail train. The vibration temperature sensor is used to obtain the temperature information and vibration information of the axle box bearing of the running gear during the operation of the rail train.

[0199] The preprocessor is arranged on the running frame at the bottom of each carriage of the train. The preprocessor is electrically connected to the output end of the vibration and temperature sensor to convert the collected vibration and temperature information into digital signals and transmit them to the processing module (i.e., the on-board host) through the mesh network. The on-board host is responsible for storing the received data and processing the data to realize the bearing fault diagnosis of the rail train running gear.

[0200] The data acquisition module has a simple structure and uses a wireless mesh network to transmit data, reducing wiring requirements and simplifying installation and maintenance.

[0201] The present invention overcomes the shortcoming of a single sensor being susceptible to external interference by coordinating the vibration and temperature characteristic data of multiple acquisition nodes, and realizes more accurate bearing fault diagnosis. It adopts a diagnostic strategy combining feature-level fusion and decision-level fusion to avoid misdiagnosis and greatly improve the reliability and accuracy of diagnosis.

[0202] At the same time, the present invention uses wireless mesh network to transmit data, reduces wiring requirements, and simplifies installation and maintenance. The present invention can not only effectively reduce noise interference, but also monitor the bearing status in real time, provide early warning, and ensure safe operation and efficient maintenance of the train.

[0203] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0204] Although the embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the claims and their equivalents.

Claims

1. A method for monitoring axle box bearing faults of a train running gear, characterized in that: The steps include: Install vibration temperature sensors on multiple axle box bearings of the train running gear to collect temperature and vibration data in real time; Extract the features of the collected temperature and vibration data and perform collaborative processing to obtain the corresponding collaborative feature values; The collaborative eigenvalues ​​are fused at the feature level to make a preliminary diagnosis of the health status of the bearing; Using the fusion diagnosis method based on DS evidence theory, the corresponding fault probability is calculated through the collaborative eigenvalue to perform secondary diagnosis on the health status of the bearing; According to the results of the preliminary diagnosis and the secondary diagnosis, if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is also in a faulty state, an early warning message is issued; if the preliminary diagnosis shows that the bearing is faulty and the secondary diagnosis shows that the bearing is in a normal state, no early warning message is issued.

2. The train running gear axle box bearing fault monitoring method according to claim 1, characterized in that: The method of extracting the features of the collected vibration data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is as follows: Four corresponding synergistic features in the time and frequency domain are extracted from the vibration data, namely: synergistic RMS value, synergistic peak-to-peak value, synergistic kurtosis value and synergistic root mean square frequency; Synergy RMS value X RMS for: Among them, RMSa represents the RMS value of the vibration of the axle box under test, RMS i represents the vibration RMS value of the i-th axle box, RMSa is not included in all RMSi, and N represents the total number of axle boxes; Synergy Peak-to-Peak Value X P for: Where Pa represents the peak-to-peak value of the vibration of the axle box under test, P i represents the peak-to-peak value of the vibration of the i-th axle box, and all Pi do not include Pa; Co-kurtosis value X K for: Among them, Ka represents the vibration kurtosis of the axle box under test, K i represents the vibration kurtosis of the i-th axle box, all Ki excluding Ka; Coordinated RMS frequency X RMSF for: Among them, RMSFa represents the root mean square frequency of the tested axle box, RMSFi represents the root mean square frequency of the i-th axle box, and all RMSFi do not include RMSFa.

3. The train running gear axle box bearing fault monitoring method according to claim 1, characterized in that: The method of extracting the features of the collected temperature data and performing collaborative processing to obtain the corresponding collaborative feature values ​​is as follows: Two corresponding collaborative features are extracted from the temperature data, namely: collaborative absolute temperature X T and the coordinated temperature rise rate X VT =: Where Ta represents the temperature of the axle box under test, T i Indicates the temperature of the i-th axle box, all Ti does not include Ta, N indicates the total number of axle boxes; VTa indicates the temperature rise rate of the axle box under test, VT i Represents the temperature rise rate of the i-th axle box, excluding VTa from all VTi.

4. The train running gear axle box bearing fault monitoring method according to claim 1, characterized in that: The collaborative eigenvalues ​​are fused at the feature level to preliminarily diagnose the health status of the bearing. The specific steps are as follows: Perform principal component analysis (PCA) on the extracted collaborative features and standardize the obtained collaborative feature quantities: in, is the standardized feature quantity, ξ j is the characteristic quantity, E(ξ j ) is the mean value of the characteristic quantity, S(ξ j ) is the standard deviation of the characteristic quantity; All standardized features are combined into a sample matrix X n×m ,for: Among them, each row represents an observation sample, and each column represents a collaborative feature; Through the sample matrix X n×m The covariance matrix M is calculated as: M=cov(X n×m ); The eigenvalues ​​λ of the covariance matrix M i The values ​​are arranged in descending order, and their corresponding eigenvectors form the feature matrix K: Calculate the contribution rate G of the characteristic quantity i : Take the first j features whose cumulative contribution rate reaches 85% as the principal element P j ,for: The pivot element P j Input the trained support vector machine SVM to preliminarily determine whether the bearing is faulty.

5. The train running gear axle box bearing fault monitoring method according to claim 1, characterized in that: Using the fusion diagnosis method based on DS evidence theory, the corresponding fault probability is calculated through the collaborative eigenvalue, and the health status of the bearing is secondary diagnosed. The specific method is as follows: The identification framework H for bearing fault diagnosis is determined as: H={h1,h2} Among them, h1 indicates a fault, and h2 indicates no fault; Based on the collaborative eigenvalues ​​of temperature and vibration data, the fault probability value, no-fault probability value and fault uncertainty probability value corresponding to each collaborative eigenvalue are calculated respectively; Normalizing the fault probability value, no-fault probability value and fault uncertainty probability value of each collaborative eigenvalue to obtain the basic probability function of each collaborative eigenvalue; Based on the basic probability functions of each collaborative eigenvalue, decision fusion is performed to obtain the final basic probability function of bearing fault; wherein the basic probability function includes fault, no fault and fault uncertainty, and whether there is a fault is determined according to the probability size.

6. The train running gear axle box bearing fault monitoring method according to claim 5, characterized in that: Based on the collaborative eigenvalues ​​of temperature and vibration data, the steps for calculating the fault probability value, no-fault probability value and fault uncertainty probability value corresponding to each collaborative eigenvalue are as follows: By cooperating RMS value X RMS Calculate the probability of failure m RMS , the probability of no failure n RMS , Failure uncertainty probability p RMS : Among them, X RMS represents the collaborative RMS value, α, β, σ represent the correction coefficients of each formula; By the co-kurtosis value X K Calculate the probability of failure m K , the probability of no failure n K , Failure uncertainty probability p K : Among them, X K represents the co-kurtosis value; By coordinating the absolute temperature X T Calculate the probability of failure m T , the probability of no failure n T , Failure uncertainty probability p T : Among them, X T represents the coordinated absolute temperature value; Through the coordinated temperature rise rate X VT = Calculate the probability of failure m VT , the probability of no failure n VT , Failure uncertainty probability p VT : Among them, X VT represents the coordinated temperature rise rate.

7. The train running gear axle box bearing fault monitoring method according to claim 6, characterized in that: The method of normalizing the fault probability value, no-fault probability value and fault uncertainty probability value of each collaborative eigenvalue to obtain the basic probability function of each collaborative eigenvalue is as follows: By cooperating RMS value X RMS Calculate the basic probability function m of fault 11 (h1), basic probability function of no fault m 11 (h2), basic probability function of failure uncertainty m 11 (H): By the co-kurtosis value X K Calculate the basic probability function m of fault 12 (h1), basic probability function of no fault m 12 (h2), basic probability function of failure uncertainty m 12 (H): By coordinating the absolute temperature X T Calculate the basic probability function m of fault 21 (h1), basic probability function of no fault m 21 (h2), basic probability function of failure uncertainty m 21 (H): Through the coordinated temperature rise rate X VT = Calculate the basic probability function m of fault 22 (h1), basic probability function of no fault m 22 (h2), basic probability function of failure uncertainty m 22 (H):

8. The train running gear axle box bearing fault monitoring method according to claim 5, characterized in that: Based on the basic probability function of each collaborative eigenvalue, decision fusion is performed. The specific steps are as follows: The synthesis formula of DS evidence theory is: Among them, under the definition of DS evidence theory, C is the fused target set, m(C) is the probability function of the target set, A and B represent two sets of different evidence supporting C, m1 and m2 are two basic probability functions under the same recognition framework, and K is the trust conflict degree that describes the conflict between the two, which is: The synthesis formula can be written as: The basic probability function m for the cooperative RMS value 11 and the basic probability function m of the co-kurtosis 12 The basic probability function m1 of vibration is obtained by fusion, that is: The basic probability function m for the coordinated absolute temperature 21 and the basic probability function m of the coordinated temperature rise rate 22 The basic probability function m2 of temperature is obtained by fusion: The basic probability function m1 of vibration and the basic probability function m2 of temperature are combined to obtain the final basic probability function m of bearing failure:

9. A train running gear axle box bearing fault monitoring system, characterized in that: It includes a data acquisition module and a processing module, wherein the data acquisition module is used to collect temperature and vibration data in real time, and the output end of the data acquisition module is connected to the input end of the processing module; The processing module executes the method described in any one of claims 1-8 to monitor the axle box bearing fault of the train running gear.

10. The train running gear axle box bearing fault monitoring system according to claim 9, characterized in that: The data acquisition module includes a vibration and temperature sensor and a preprocessor; The vibration temperature sensor can be detachably mounted on the axle box of the running gear of the rail train, and is used to obtain the temperature information and vibration information of the axle box bearing of the running gear during the operation of the rail train; The preprocessor is connected to the output end of the vibration and temperature sensor, and is used to convert the collected vibration and temperature information into a digital signal, and transmit it to the processing module through the mesh network.

Images