Automatic detection method and system for fault characteristics of railway vehicle gear transmission system

Abstract

The invention discloses a method for automatically detecting fault characteristics of a railway vehicle gear transmission system, comprising: iteratively extracting multiple time-frequency ridges from a time-frequency representation, identifying complete time-frequency ridges and time-frequency ridges with frequency jumps in the time-frequency ridges; decomposing the time-frequency ridges with frequency jumps into multiple continuous time-frequency ridge segments; respectively calculating the average frequency ratios of the complete time-frequency ridges, the time-frequency ridge segments and the fundamental frequency to obtain the calculation order; according to the gear box structure and kinematics knowledge, obtaining the theoretical order of the fault characteristics of the gear transmission system, and designing an upper triangular matrix; respectively comparing the calculation order with each row of the upper triangular matrix, selecting a row with the highest matching degree, thereby determining the physical meaning of the time-frequency ridge, and automatically judging whether the gear transmission system has a fault; the invention also discloses a system for automatically detecting fault characteristics of a railway vehicle gear transmission system; the invention realizes automatic monitoring of the service status of the gear transmission system.

Description

Technical Field

[0001] The invention relates to the technical field of fault detection of railway vehicle gear transmission systems, in particular to a method and system for automatically detecting fault characteristics of railway vehicle gear transmission systems. Background Art

[0002] The health status of the gear transmission system directly affects the operational safety of railway vehicles. Therefore, it is necessary to effectively detect the service status of the gear transmission system of railway vehicles. At present, the existing technology usually uses the time-frequency ridge extraction method to extract the time-frequency ridge from the time-frequency representation of the vibration signal, and then the user judges the physical meaning of the time-frequency ridge in combination with the speed information and performs offline fault diagnosis. However, the actual operation of railway vehicles does not allow sensors to be installed on the axle to directly measure the speed, making it difficult to obtain speed information. In addition, relying on experts for analysis is not only inefficient, but also unable to achieve real-time monitoring and diagnosis. Summary of the invention

[0003] In order to solve the problems existing in the prior art, the purpose of the present invention is to provide a method and system for automatically detecting fault characteristics of a railway vehicle gear transmission system. The present invention can automatically identify the physical meaning of time-frequency ridges without the need for speed information; the system does not require any human intervention in the entire process of extracting time-frequency ridges, parsing the physical meaning and outputting the health status of the gear transmission system, thereby realizing automatic monitoring of the service status of the gear transmission system.

[0004] To achieve the above object, the technical solution adopted by the present invention is: a method for automatically detecting fault characteristics of a railway vehicle gear transmission system, comprising the following steps:

[0005] Step 1, iteratively extract multiple time-frequency ridges from the time-frequency representation, and identify complete time-frequency ridges and time-frequency ridges with frequency jumps among the time-frequency ridges;

[0006] Step 2: Using statistical methods, decompose the time-frequency ridge with frequency jump into multiple continuous time-frequency ridge segments;

[0007] Step 3, select a ridge line from the complete time-frequency ridge line as the fundamental frequency, calculate the average frequency ratio of the complete time-frequency ridge line, the time-frequency ridge line segment and the fundamental frequency, and obtain the calculation order;

[0008] Step 4: According to the gearbox structure and kinematics knowledge, the theoretical order of the fault characteristics of the gear transmission system is obtained, and an upper triangular matrix is ​​designed to arrange all the theoretical orders in order from small to large;

[0009] Step 5: Compare the calculated order with each row of the upper triangular matrix, select the row with the highest matching degree, and thus determine the physical meaning of the time-frequency ridge line; based on the physical meaning of the time-frequency ridge line, automatically determine whether the gear transmission system has a fault.

[0010] As a further improvement of the present invention, the step 1 specifically comprises the following steps:

[0011] Step 1.1: Use the time-frequency ridge extraction algorithm to iteratively extract i time-frequency ridges from the time-frequency representation of the signal, denoted as f i (t n ), where n = 1, 2, 3, ..., N;

[0012] Step 1.2, from f i (t n ) selects a ridge line as the fundamental frequency, denoted as f b (t n ), calculate the point-to-point frequency ratio between each ridge and the fundamental frequency:

[0013] R i (t n )= f i (t n ) / f b (t n ) (1)

[0014] Step 1.3: Point-to-point frequency ratio R i (t n ) to sort, and then calculate the median E i :

[0015] E i =Q 2 (sort(R i (t n ))) (2)

[0016] Where sort means sorting the array, Q 2 It means to find the median of an array;

[0017] Step 1.4: If f i (t n ) and fundamental frequency f b (t n ) are all complete time-frequency ridges, then f i (t n ) and f b (t n )The frequency ratios at different times t are close, that is, R i (t n ) are distributed in the median E inearby; hence the following definition:

[0018]

[0019] Formula (3) shows that if R i (t n ) has more than 95% of the elements in the same i If the relative error is less than 3.5%, the ridge line f i (t n ) and f b (t n ) are all complete time-frequency ridges, where t k ∈t n , if R i If they all fail to satisfy equation (3), then it is considered to be the fundamental frequency f b (t n ) frequency jump occurs, the base frequency f b (t n ) and then repeat steps 1.2 to 1.4;

[0020] Step 1.5: After all complete time-frequency ridges are identified, assume that there are p complete time-frequency ridges, and record the p complete time-frequency ridges as f p (t n ), the q frequency jump ridges are denoted as f q (t n ), where q=ip.

[0021] As a further improvement of the present invention, the step 2 is specifically as follows:

[0022] According to formula (1), f is calculated q (t n ) and f b (t n ) point-to-point average frequency ratio R q (t n ), and used statistical analysis methods to analyze the R q (t n ) to classify, thereby decomposing the ridge line with frequency jump into multiple segments; the decomposed ridge line segments are recorded as f q (t k ), where t k ∈t n .

[0023] As a further improvement of the present invention, the step 3 is specifically as follows:

[0024] Calculate all complete time-frequency ridges f p (t n ) and fundamental frequency f b (t n)’s average frequency ratio:

[0025] O p =mean(f p (t n ) / f b (t n )) (4)

[0026] Calculate the ridge segment f q (t k ) and f b (t k ), where f b (t k ) is the fundamental frequency f b (t n ) and t k The corresponding part:

[0027] O q =mean(f q (t k ) / f b (t k )) (5)

[0028] O p With O q Merge to get array O, the elements in array O represent: the ridge line f b (t n ) is used as a reference, and the calculation orders corresponding to other time-frequency ridges are

[0029] As a further improvement of the present invention, the step 4 is specifically as follows:

[0030] According to the structure of the gearbox, the characteristic frequencies are listed one by one, and then the order of each characteristic frequency is calculated and sorted in ascending order. Assuming that there are x characteristic frequencies in total, a single-row matrix A with x elements is obtained; find the transposed matrix A of the single-row matrix A T , and transpose the matrix A T Find the reciprocal of each element in:

[0031] B=1 / A T (6)

[0032] Multiply matrix A by matrix B to get matrix C:

[0033] C=A×B (7)

[0034] Let all elements below the diagonal of matrix C be 0, and an upper triangular matrix D is obtained. The upper triangular matrix D contains all theoretical orders of the characteristic frequencies of the gear transmission system.

[0035] The present invention also provides a railway vehicle gear transmission system fault characteristic automatic detection system, comprising:

[0036] Data acquisition module: used to obtain the original time-frequency distribution of the vibration signal of the gearbox of the railway vehicle;

[0037] Ridge extraction module: used to extract multiple time-frequency ridges from the original time-frequency distribution, and identify the complete time-frequency ridges and the time-frequency ridges with frequency jumps;

[0038] Ridge decomposition module: used to decompose the time-frequency ridge with frequency jump into multiple continuous time-frequency ridge segments;

[0039] Order calculation module: used to select one of the time-frequency ridges as the fundamental frequency, calculate the average frequency ratio of the complete time-frequency ridge, the time-frequency ridge segment and the fundamental frequency, and obtain the calculation order;

[0040] Matrix generation module: used to generate an upper triangular matrix containing the theoretical order of fault characteristics of the gear transmission system according to the gearbox structural parameters;

[0041] Feature recognition module: used to match the calculated order with the theoretical order in the upper triangular matrix to obtain the fault features corresponding to the time-frequency ridge line;

[0042] Fault detection module: used to output the health status of the gear transmission system based on the extracted fault characteristics.

[0043] The beneficial effects of the present invention are:

[0044] 1. The present invention proposes a method for automatically identifying and screening complete time-frequency ridges. The time-frequency representation of the measured gearbox vibration signal contains both complete time-frequency ridges and time-frequency ridges with frequency jumps. The existing methods rely on visual observation and manual selection of complete time-frequency ridges when extracting time-frequency ridges. The method proposed in the present invention provides a theoretical criterion for distinguishing complete ridges, and is more intelligent and automated, reducing the interference of human factors.

[0045] 2. The present invention proposes a method for obtaining fault characteristic information from time-frequency ridges with frequency jumps. Existing methods do not care about time-frequency ridges with frequency jumps, nor do they attempt to extract fault characteristic information therein, but only using complete time-frequency ridges often cannot complete the fault diagnosis task. In fact, time-frequency ridges with frequency jumps also contain a large amount of fault characteristic information. The method proposed in the present invention can separate the fault characteristic frequency from the time-frequency ridges with frequency jumps, providing richer and more reliable information for subsequent fault diagnosis.

[0046] 3. The present invention proposes a method for automatically identifying faults in railway vehicle gear transmission systems. According to the gearbox structural parameters, this method counts all potential characteristic frequencies in the vibration signal and calculates the theoretical orders corresponding to all characteristic frequencies, and then arranges the theoretical orders to generate an upper triangular matrix. The frequency ratio between the time-frequency ridges is calculated, the calculated order is matched with each row of the upper triangular matrix, and the row with the highest matching degree is selected. The fault type corresponding to each element of this row is the physical meaning of the time-frequency ridge. Compared with traditional methods, this algorithm does not require speed information, can automatically extract fault features and identify their physical meanings, and can automatically detect the health status of the gear transmission system without human intervention. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A flowchart of the method steps of an embodiment of the present invention;

[0048] Figure 2 is a flow chart of a method algorithm of an embodiment of the present invention;

[0049] Figure 3 Schematic diagram of the system structure of an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The embodiments of the present invention are described in detail below with reference to the accompanying drawings.

[0051] Example 1

[0052] The fault characteristic frequency and its multiples of the gear transmission system of railway vehicles exist in the form of time-frequency ridges in the time-frequency representation. After the existing methods extract the time-frequency ridges from the time-frequency representation, expert analysis and combination with prior knowledge (time-varying speed information) are required to determine the physical meaning of the time-frequency ridges, and then determine whether there is a fault in the gear transmission system. In fact, it is not allowed to install speed sensors on the axles of operating vehicles, making it difficult to obtain accurate speed information. At the same time, relying on expert analysis for fault diagnosis lacks real-time performance and cannot meet the actual needs of monitoring and diagnosis of railway vehicle gear transmission systems. To this end, the present invention proposes a method that does not require speed information, automatically identifies the physical meaning of time-frequency ridges, and determines the health status of the gear transmission system. The method includes: iteratively extracting multiple time-frequency ridges from the time-frequency representation, identifying the complete time-frequency ridges and the time-frequency ridges with frequency jumps in the time-frequency ridges according to the algorithm; using a statistical method, decomposing the time-frequency ridges with frequency jumps into multiple continuous time-frequency ridge segments; selecting a ridge from the complete time-frequency ridge as the fundamental frequency, respectively calculating the average frequency ratio of the complete time-frequency ridge, the time-frequency ridge segment and the fundamental frequency, and obtaining the calculation order; according to the gearbox structure and kinematics knowledge, obtaining the theoretical order of the fault characteristics of the gear transmission system, and designing an upper triangular matrix to arrange all theoretical orders in order from small to large. Compare the calculation order with each row of the upper triangular matrix, select the row with the highest matching degree (at this time, the calculation order coincides with the theoretical order), so as to determine the physical meaning of the time-frequency ridge. According to the physical meaning of the time-frequency ridge, automatically judge whether the gear transmission system has a fault.

[0053] The present embodiment is further described below. Figure 1 and Figure 2 As shown, a method for detecting faults in a railway vehicle gear transmission system comprises:

[0054] S1: Use the time-frequency ridge extraction algorithm to iteratively extract i time-frequency ridges from the time-frequency representation of the signal, denoted as f i (t n );

[0055] S2: From f i (t n ) selects a ridge line as the fundamental frequency, denoted as f b (t n ), calculate the point-to-point frequency ratio R between each ridge and the fundamental frequency i (t n );

[0056] S3: The point-to-point frequency ratio R i (t n ) to sort and calculate the median E i :

[0057] S4: Determine the time-frequency ridge line f according to formula (3) b (t n ) whether a frequency jump occurs. If the ridge line f b (t n ) when a frequency jump occurs, the ridge is deleted, and then steps S2 to S4 are repeated.

[0058] S5: Through the above steps, all complete time-frequency ridges are identified. The p complete time-frequency ridges are denoted as f p (t n ), the q frequency jump ridges are denoted as f q (t n ) where q = ip.

[0059] S6: The frequency hopping ridge contains multiple segments of different signal components, and there may be noise. Calculate f according to formula (1): q (t n ) and f b (t n ) point-to-point average frequency ratio R q (t n ), and used statistical analysis methods to analyze the R q (t n ) to classify, thereby decomposing the ridge line with frequency jump into multiple segments. The decomposed ridge line segments are recorded as f q (t k ), where t k ∈t n .

[0060] S7: Calculate all complete time-frequency ridges f p (t n ) and fundamental frequency f b (t n ) has an average frequency of O p , calculate the ridge segment f q (t k ) and f b (t k ) has an average frequency of O q , O p With O q Merge to get array O, the elements in array O represent: the ridge line f b (t n ) is used as a reference, and the calculation orders corresponding to other time-frequency ridges are

[0061] S8: According to the gearbox structure and motion law, the input shaft frequency f r1 , bearing inner ring fault characteristic frequency f I1 , outer ring fault characteristic frequency f o1 , rolling element fault characteristic frequency fR1 , cage failure characteristic frequency f C1 , output shaft speed f r2 , bearing inner ring fault characteristic frequency f I2 , outer ring fault characteristic frequency f o2 , rolling element fault characteristic frequency f R2 and cage fault characteristic frequency f C2 , gear meshing frequency f m List the characteristic frequencies one by one (the input shaft bearing and output shaft bearing models of railway vehicle gearbox may be different), then calculate the order of each characteristic frequency and sort them in ascending order of order to obtain a single-row matrix A; find the transposed matrix A of the single-row matrix A T , and transpose the matrix A T Find the inverse of each element in to get the matrix B, find the matrix C = A × B, then set all elements below the diagonal of the matrix C to 0, and get an upper triangular matrix D;

[0062] S9: Match the array O with the upper triangular matrix D, and select the row with the highest numerical matching degree. At this time, the calculated order matches the theoretical order, and the row with the highest matching degree with O in the matrix D is selected, so as to automatically determine the physical meaning of the time-frequency ridge line, and then obtain the health status of the gear transmission system.

[0063] When f i (t n ) When there are both complete ridges and frequency-jumping ridges, the algorithm can automatically identify the complete ridges and frequency-jumping ridges, specifically including the following steps:

[0064] S2: From f i (t n ) selects a ridge line as the fundamental frequency, denoted as f b (t n ), calculate the point-to-point frequency ratio R between each ridge and the fundamental frequency i (t n );

[0065] R i (t n )= f i (t n ) / f b (t n ) (1)

[0066] S3: The point-to-point frequency ratio R i (t n ) are sorted, and then the median E is calculated i :

[0067] E i =Q 2 (sort(Ri (t n ))) (2)

[0068] Where sort means sorting the array, Q 2 It means finding the median of an array.

[0069] S4: If f i (t n ) and fundamental frequency f b (t n ) are all complete time-frequency ridges, then f i (t n ) and f b (t n ) The frequency ratios at different times t are relatively close, that is, R i (t n ) most of the elements are distributed in the median E i Therefore, the following definitions can be given:

[0070]

[0071] Formula (3) shows that if R i (t n ) has more than 95% of the elements in the same i If the relative error is less than 3.5%, the ridge line f i (t n ) and f b (t n ) are all complete time-frequency ridges, where t k ∈t n If R i If they all fail to satisfy equation (3), then it is considered to be the fundamental frequency f b (t n ) frequency jump occurs, the base frequency f b (t n ) and then repeat steps S2 to S4.

[0072] S5: Through the above steps, all complete time-frequency ridges are identified. Assume that there are p complete time-frequency ridges, and record the p complete time-frequency ridges as f p (t n ), the q frequency jump ridges are denoted as f q (t n ), where q=ip.

[0073] The discontinuous time-frequency ridge with frequency jump can be decomposed into a plurality of continuous time-frequency ridge segments without frequency jump, which specifically includes the following steps:

[0074] S6: Calculate f according to formula (1) q (tn ) and f b (t n ) point-to-point average frequency ratio R q (t n ), using statistical analysis methods to analyze R q (t n ) to classify, thereby decomposing the mixed ridges into multiple fragments. The decomposed ridge fragments are denoted as f q (t k ), where t k ∈t n .

[0075] Automatically identify the physical meaning of the time-frequency ridges, including the following steps:

[0076] S7: S7.1 Calculate all complete time-frequency ridges f p (t n ) and fundamental frequency f b (t n )’s average frequency ratio:

[0077] O p =mean(f p (t n ) / f b (t n )) (4)

[0078] S7.2 Calculate the ridge segment f q (t k ) and f b (t k ), where f b (t k ) is the fundamental frequency f b (t n ) and t k The corresponding part:

[0079] O q =mean(f q (t k ) / f b (t k )) (5)

[0080] S7.3 will be O p With O q Merge to get array O, the elements in array O represent: the ridge line f b (t n ) is used as a reference, and the calculation orders corresponding to other time-frequency ridges are

[0081] S8: Establish a matrix D that reflects the ratio relationship between the fault characteristic frequencies of the gear transmission system:

[0082] S8.1 According to the structure of the gearbox, the input shaft rotation frequency and its multiples, the output shaft rotation frequency and its multiples, the bearing fault characteristic frequency and its multiples, the gear meshing frequency and its multiples, and other characteristic frequencies are listed one by one, and then the order of each characteristic frequency is calculated and sorted in ascending order of order (assuming that there are x characteristic frequencies in total), and a single-row matrix A with x elements is obtained; find the transposed matrix A of the single-row matrix A T , and transpose the matrix A T Find the reciprocal of each element in:

[0083] B=1 / A T (6)

[0084] S8.2 Multiply matrix A by matrix B to obtain matrix C:

[0085] C=A×B (7)

[0086] Let all elements below the diagonal of matrix C be 0, and an upper triangular matrix D is obtained. Matrix D contains all theoretical orders of characteristic frequencies of the gear transmission system.

[0087] S9: Match the array O with the upper triangular matrix D, and select the row with the highest numerical matching degree (the calculated order matches the theoretical order at this time). The fault type corresponding to each element of this row is the physical meaning of the time-frequency ridge line, thereby obtaining the health status of the gear transmission system.

[0088] The following table is an upper triangular matrix D including the ratio relationship of characteristic frequencies in the gear transmission system of a railway vehicle provided in an embodiment of the present application. The main transmission parts in the gear transmission system of a railway vehicle include input shaft bearings, pinions, gears and output shaft bearings. 1 Substituting the parameters such as the number of rolling elements z, the diameter of the rolling elements d and the contact angle α of the bearing into the calculation formula of the rolling bearing fault characteristic frequency can calculate the fault characteristic frequency of the rolling bearing, where f r1 Indicates the input shaft rotation frequency, f I1 、f o1 、f R1 and f C1 They represent the inner ring fault characteristic frequency, outer ring fault characteristic frequency, rolling element fault characteristic frequency and cage fault characteristic frequency of the input shaft bearing respectively; f r2 Indicates the output shaft speed, f I2 、f o2 、f R2 and f C2 They represent the inner ring fault characteristic frequency, outer ring fault characteristic frequency, rolling element fault characteristic frequency and cage fault characteristic frequency of the output shaft bearing respectively; according to the number of teeth Z of the pinion 1 and the number of teeth of the large gear Z 2The meshing frequency f of the meshing gear can be calculated m The first row in Table 1 shows the r2 When it is the fundamental frequency, the orders corresponding to other characteristic frequencies; the second line indicates the order of f C1 When it is the fundamental frequency, the orders corresponding to other characteristic frequencies; and so on.

[0089]

[0090] Table 1

[0091] For example Figure 1 As shown, the vibration signal and gear structure parameters are input into the computer device, and then Figure 2 In the calculation module shown, after the time-frequency representation of the signal is obtained by the time-frequency analysis method, multiple time-frequency ridges are iteratively extracted from the time-frequency representation by the ridge extraction algorithm, and the point-to-point frequency ratio R between the time-frequency ridges is calculated. i , and calculate the median E of the frequency ratio i Identify and select complete ridges, and calculate the average frequency ratio of the complete ridges to the fundamental frequency. p Statistical analysis is used to classify the frequency-hopping ridges, separate the ridge segments, and calculate the average frequency ratio of the ridge segments to the fundamental frequency. q , merge O p and O q Get array O. Calculate the ratio between characteristic frequencies according to the gearbox transmission parts parameters, and arrange the ratio relationship to generate an upper triangular matrix D, as shown in the table above. Match array O with each row of the upper triangular matrix. The fault type corresponding to the row with the highest matching degree is the physical meaning of the time-frequency ridge line. Output the health status of the gear transmission system based on the result.

[0092] Example 2

[0093] For example Figure 1 and Figure 2 As shown, a method for detecting faults in a gear transmission system is provided. The specific steps are as follows:

[0094] S1: Use the time-frequency ridge extraction algorithm to iteratively extract i time-frequency ridges from the time-frequency representation of the signal, denoted as f i (t n ), where n = 1, 2, 3, ..., N;

[0095] S2: From f i (t n ) selects a ridge line as the fundamental frequency, denoted as f b (t n ), calculate the point-to-point frequency ratio between each ridge and the fundamental frequency:

[0096] R i (tn )= f i (t n ) / f b (t n ) (1)

[0097] S3: point-to-point frequency ratio R i (t n ) to sort, and then calculate the median E i :

[0098] E i =Q 2 (sort(R i (t n ))) (2)

[0099] Where sort means sorting the array, Q 2 It means finding the median of an array.

[0100] S4: If f i (t n ) and fundamental frequency f b (t n ) are all complete time-frequency ridges, then f i (t n ) and f b (t n ) The frequency ratios at different times t are relatively close, that is, R i (t n ) are distributed in the median E i Therefore, the following definitions can be given:

[0101]

[0102] Formula (3) shows that if R i (t n ) has more than 95% of the elements in the same i If the relative error is less than 3.5%, the ridge line f i (t n ) and f b (t n ) are all complete time-frequency ridges, where t k ∈t n , if R i If they all fail to satisfy equation (3), then it is considered to be the fundamental frequency f b (t n ) frequency jump occurs, the base frequency f b (t n ) and then repeat steps S2 to S4.

[0103] S5: Through the above steps, all complete time-frequency ridges are identified. Assume that there are p complete time-frequency ridges, and record the p complete time-frequency ridges as f p (t n ), the q frequency jump ridges are denoted as f q (t n ), where q=ip.

[0104] S6: The frequency hopping ridge contains multiple segments of different signal components, and there may be noise. Calculate f according to formula (1): q (t n ) and f b (t n ) point-to-point average frequency ratio R q (t n ), and used statistical analysis methods to analyze the R q (t n ) to classify, thereby decomposing the ridge line with frequency jump into multiple segments. The decomposed ridge line segments are recorded as f q (t k ), where t k ∈t n .

[0105] S7: Calculate all complete time-frequency ridges f p (t n ) and fundamental frequency f b (t n )’s average frequency ratio:

[0106] O p =mean(f p (t n ) / f b (t n )) (4)

[0107] Calculate the ridge segment f q (t k ) and f b (t k ), where f b (t k ) is the fundamental frequency f b (t n ) and t k The corresponding part:

[0108] O q =mean(f q (t k ) / f b (t k )) (5)

[0109] O p With Oq Merge to get array O, the elements in array O represent: the ridge line f b (t n ) is used as a reference, and the calculation orders corresponding to other time-frequency ridges are

[0110] S8: According to the gearbox structure parameters, the input shaft frequency f r1 , bearing inner ring fault characteristic frequency f I1 , outer ring fault characteristic frequency f o1 , rolling element fault characteristic frequency f R1 , cage failure characteristic frequency f C1 , output shaft speed f r2 , bearing inner ring fault characteristic frequency f I2 , outer ring fault characteristic frequency f o2 , rolling element fault characteristic frequency f R2 and cage fault characteristic frequency f C2 , gear meshing frequency f m The eigenfrequencies are listed one by one, and then the order of each eigenfrequency is calculated and sorted in ascending order of order to obtain a single-row matrix A;

[0111] Find the transposed matrix A of a single row matrix A T , and transpose the matrix A T Find the reciprocal of each element in:

[0112] B=1 / A T (6)

[0113] Multiply matrix A by matrix B to get matrix C:

[0114] C=A×B (7)

[0115] Let all elements below the diagonal of matrix C be 0, and we get an upper triangular matrix D, as shown in Table 1. The first row in Table 1 represents the matrix f r2 When it is the fundamental frequency, the orders corresponding to other characteristic frequencies; the second line indicates the order of f C1 When it is the fundamental frequency, the order corresponding to other characteristic frequencies is the same. And so on.

[0116] S9: Match the array O with the upper triangular matrix D, and select the row with the highest numerical matching degree (the calculated order matches the theoretical order at this time). The fault type corresponding to each element of this row is the physical meaning of the time-frequency ridge line. The algorithm automatically outputs the fault characteristic frequency corresponding to the time-frequency ridge line to obtain the health status of the gear transmission system.

[0117] Example 3

[0118] like Figure 3As shown, a schematic diagram of a system based on a gear transmission system fault detection method, the system includes:

[0119] Data acquisition module: obtains the original time-frequency distribution of the vibration signal of the railway vehicle gearbox;

[0120] Ridge extraction module: extract multiple time-frequency ridges from the original time-frequency distribution, and identify the complete time-frequency ridges and the time-frequency ridges with frequency jumps;

[0121] Ridge decomposition module: decomposes the time-frequency ridge with frequency jump into multiple continuous time-frequency ridge segments;

[0122] Order calculation module: select one of the time-frequency ridges as the fundamental frequency, calculate the average frequency ratio of the complete time-frequency ridge, the time-frequency ridge segment and the fundamental frequency, and obtain the calculation order;

[0123] Matrix generation module: Generates an upper triangular matrix containing the theoretical order of fault characteristics of the gear transmission system according to the gearbox structural parameters;

[0124] Feature recognition module: matches the calculated order with the theoretical order in the upper triangular matrix to obtain the fault feature corresponding to the time-frequency ridge line;

[0125] Fault detection module: outputs the health status of the gear transmission system based on the extracted fault characteristics.

[0126] The above-mentioned embodiments only express the specific implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for automatically detecting fault characteristics of a railway vehicle gear transmission system, characterized in that: The following steps are involved: Step 1, iteratively extract multiple time-frequency ridges from the time-frequency representation, and identify complete time-frequency ridges and time-frequency ridges with frequency jumps among the time-frequency ridges; Step 2: Using statistical methods, decompose the time-frequency ridge with frequency jump into multiple continuous time-frequency ridge segments; Step 3, select a ridge line from the complete time-frequency ridge line as the fundamental frequency, calculate the average frequency ratio of the complete time-frequency ridge line, the time-frequency ridge line segment and the fundamental frequency, and obtain the calculation order; Step 4: According to the gearbox structure and kinematics knowledge, the theoretical order of the fault characteristics of the gear transmission system is obtained. All theoretical orders are arranged in order from small to large to obtain a single-row matrix A. The transposed matrix A of the single-row matrix A is calculated. T , and for A T Calculate the inverse of each element in to get matrix B; multiply matrix A by matrix B to get matrix C, and set all elements below the diagonal of matrix C to 0, to get an upper triangular matrix of all theoretical orders containing the characteristic frequencies of the gear transmission system; Step 5: Compare the calculated order with each row of the upper triangular matrix, select the row with the highest matching degree, and thus determine the physical meaning of the time-frequency ridge line; based on the physical meaning of the time-frequency ridge line, automatically determine whether the gear transmission system has a fault.

2. The method for automatically detecting fault characteristics of a railway vehicle gear transmission system according to claim 1, characterized in that: The step 1 specifically comprises the following steps: Step 1.1: Use the time-frequency ridge extraction algorithm to iteratively extract i time-frequency ridges from the time-frequency representation of the signal, denoted as f i (t n ), where n = 1, 2, 3, ..., N; Step 1.2, from f i (t n ) selects a ridge line as the fundamental frequency, denoted as f b (t n ), calculate the point-to-point frequency ratio between each ridge and the fundamental frequency: R i (t n )= f i (t n ) / f b (t n ) (1) Step 1.3: Point-to-point frequency ratio R i (t n ) to sort, and then calculate the median E i : E i =Q2(sort(R i (t n ))) (2) Among them, sort means sorting the array, and Q2 means finding the median of the array; Step 1.4: If f i (t n ) and fundamental frequency f b (t n ) are all complete time-frequency ridges, then f i (t n ) and f b (t n )The frequency ratios at different times t are close, that is, R i (t n ) are distributed in the median E i nearby; hence the following definition: Formula (3) shows that if R i (t n ) has more than 95% of the elements in the same i If the relative error is less than 3.5%, the ridge line f i (t n ) and f b (t n ) are all complete time-frequency ridges, where t k ∈t n , if R i If they all fail to satisfy equation (3), then it is considered to be the fundamental frequency f b (t n ) frequency jump occurs, the base frequency f b (t n ) and then repeat steps 1.2 to 1.4; Step 1.5: After all complete time-frequency ridges are identified, assume that there are p complete time-frequency ridges, and record the p complete time-frequency ridges as f p (t n ), the q frequency jump ridges are denoted as f q (t n ), where q=ip.

3. The method for automatically detecting fault characteristics of a railway vehicle gear transmission system according to claim 2, characterized in that: The step 2 is specifically as follows: According to formula (1), f is calculated q (t n ) and f b (t n ) point-to-point average frequency ratio R q (t n ), and used statistical analysis methods to analyze the R q (t n ) to classify, thereby decomposing the ridge line with frequency jump into multiple segments; the decomposed ridge line segments are recorded as f q (t k ), where t k ∈t n .

4. The method for automatically detecting fault characteristics of a railway vehicle gear transmission system according to claim 3, characterized in that: The step 3 is as follows: Calculate all complete time-frequency ridges f p (t n ) and fundamental frequency f b (t n )’s average frequency ratio: O p =mean(f p (t n ) / f b (t n )) (4) Calculate the ridge segment f q (t k ) and f b (t k ), where f b (t k ) is the fundamental frequency f b (t n ) and t k The corresponding part: O q =mean(f q (t k ) / f b (t k )) (5) O p With O q Merge to get array O, the elements in array O represent: the ridge line f b (t n ) is used as a reference, and the calculation orders corresponding to other time-frequency ridges are 5. The method for automatically detecting fault characteristics of a railway vehicle gear transmission system according to claim 4, characterized in that: The step 4 is specifically as follows: According to the structure of the gearbox, the characteristic frequencies are listed one by one, and then the order of each characteristic frequency is calculated and sorted in ascending order. Assuming that there are x characteristic frequencies in total, a single-row matrix A with x elements is obtained; find the transposed matrix A of the single-row matrix A T , and transpose the matrix A T Find the reciprocal of each element in: B=1 / A T (6) Multiply matrix A by matrix B to get matrix C: C=A×B (7) Let all elements below the diagonal of matrix C be 0, and an upper triangular matrix D is obtained. The upper triangular matrix D contains all theoretical orders of the characteristic frequencies of the gear transmission system.

6. A railway vehicle gear transmission system fault feature automatic detection system, characterized in that: include: Data acquisition module: used to obtain the original time-frequency distribution of the vibration signal of the gearbox of the railway vehicle; Ridge extraction module: used to extract multiple time-frequency ridges from the original time-frequency distribution, and identify the complete time-frequency ridges and the time-frequency ridges with frequency jumps; Ridge decomposition module: used to decompose the time-frequency ridge with frequency jump into multiple continuous time-frequency ridge segments; Order calculation module: used to select a time-frequency ridge line from the complete time-frequency ridge line as the fundamental frequency, calculate the average frequency ratio of all complete time-frequency ridge lines, time-frequency ridge line segments and the fundamental frequency, and obtain the calculation order; Matrix generation module: used to list the characteristic frequencies one by one according to the gearbox structural parameters, then calculate the order of each characteristic frequency and sort them in ascending order to obtain a single-row matrix A, and calculate the transposed matrix A of the single-row matrix A T , and for A T Calculate the inverse of each element in to get matrix B; multiply matrix A by matrix B to get matrix C, and set all elements below the diagonal of matrix C to 0, to get an upper triangular matrix of all theoretical orders containing the characteristic frequencies of the gear transmission system; Feature recognition module: used to match the calculated order with the theoretical order in the upper triangular matrix to obtain the fault characteristics corresponding to the time-frequency ridge line; Fault detection module: used to output the health status of the gear transmission system based on the extracted fault characteristics.

Images