Axle gearbox meshing fault identification method for engineering vehicles

Abstract

A method for identifying meshing faults of axle gearboxes of rail engineering vehicles is disclosed. In the method, a monitoring signal X of the axle gearbox of a rail engineering vehicle is collected, and a sparse dictionary Φ is constructed by using Q-switched wavelet transform. * The harmonic signal and the impulse signal in the monitoring signal X are sparsely represented, and the objective function of sparse decomposition is established and the objective function is solved by the Lagrangian contraction algorithm to decompose the monitoring signal X into the harmonic signal and the impulse signal. H The pulse index and the skewness index are extracted from the axle gearbox. When the pulse index is within the first range and the skewness index is within the second range, the axle gearbox is considered to be normal. Otherwise, a meshing fault occurs.

Description

Technical Field

[0001] The invention belongs to the technical field of axle gearbox meshing fault detection of rail engineering vehicles, and in particular relates to an axle gearbox meshing fault identification method of a rail engineering vehicle. Background Art

[0002] The axle gearbox is one of the important transmission components in rail engineering vehicles. Usually, the axle gearbox works in a high-speed and heavy-load environment. During the movement of rail engineering vehicles, the axle gearbox has to bear a lot of loads and impacts, and at the same time, it has to withstand the test of severe cold and heat. At present, the quality inspection of axle gearboxes is mostly carried out through the sound level of bench tests. This quality monitoring method is too general and cannot effectively identify axle gearbox faults.

[0003] The axle gearbox mainly relies on gear meshing to transmit motion and torque. Gear meshing is normal and can be directly reflected in the high-frequency vibration data. Due to the characteristics of gear meshing, normally meshed gears will also produce large vibrations. At the same time, these vibrations are also mixed with the vibrations of components such as the rotating shaft, support bearings, and axle box bearings. How to extract the signal components of the meshing gears from the composite vibration signal and extract indicators to characterize the gear state is the key to identifying meshing gear faults. To this end, the present invention proposes to use Q-switched wavelets to construct a sparse dictionary library, separate the gear vibrations in the high-frequency vibration signal, and extract indicators to identify axle gearbox meshing faults.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the invention and therefore it may contain information that does not form the prior art that is already known to a person of ordinary skill in the art. Summary of the invention

[0005] The purpose of the present invention is to provide a method for identifying meshing faults of axle gearboxes of rail engineering vehicles, which uses Q-switched wavelets to construct a sparse dictionary library, separate the gear vibrations in high-frequency vibration signals, and extract indicators to identify meshing faults of axle gearboxes. In order to achieve the above purpose, the present invention provides the following technical solutions:

[0006] A method for identifying an axle gearbox meshing fault of a rail engineering vehicle of the present invention comprises:

[0007] In the first step, the monitoring signal X of the axle gearbox of the railway engineering vehicle is collected, and the sparse dictionary Φ is constructed using two Q-switched wavelet transforms with different quality factors. * By sparsely representing the harmonic signal and impulse signal in the monitoring signal X, the low-pass filter L(ω) and the high-pass filter H(ω) of the Q-modulated wavelet transform are:

[0008]

[0009]

[0010] in |ω|≤π, which is the frequency response function, α and β are the scale factors of the low-pass filter L(ω) and the high-pass filter H(ω), respectively, 0<β≤1, 0<α≤1, |L(ω)| 2 +|H(ω)| 2 =1;

[0011] The second step is to establish the objective function of sparse decomposition and use the Lagrangian contraction algorithm to solve the objective function to decompose the monitoring signal X into harmonic signals and impulse signals, where: Corresponding to the harmonic signal x H and the impulse signal x L The sparse dictionary, W H and W L is the corresponding harmonic signal x H The sparse coefficient and impulse signal X L The sparse coefficient of , the objective function is:

[0012] The constraint condition is st

[0013] Among them, the parameter ε represents the noise level in the monitoring signal x, Respectively represent the ability to make ||w H ||1+||w L ||1 Take the minimum value of w H and w L .

[0014] To this end, the Lagrangian operator λ = [λ H ,λ L ] T , the constraint condition st Merge into the objective function to form the objective function of morphological component analysis:

[0015]

[0016] For w H and w L Derivative, take the extreme value, and get the harmonic signal x H and the impulse signal x L and its sparse coefficient w H 、w L ; Among them, λ H ,λ L Respectively represent w H and w L The Lagrangian operator.

[0017] Step 3: In the harmonic signal x H Extract the pulse index and skewness index from the , where the pulse index is:

[0018]

[0019] Skewness indicator

[0020] in, is the harmonic signal x H The average value of is the harmonic signal x H The variance of x H (n) is the harmonic signal, n is X H The serial number, N is x H The maximum number of sequence numbers.

[0021] When the pulse index is within the first range and the skewness index is within the second range, the axle gearbox is considered to be normal, otherwise, a meshing fault occurs.

[0022] In the axle gearbox meshing fault identification method for a rail engineering vehicle, the first range is [5, 8.5] and the second range is [-0.15, 0.1].

[0023] In the above technical scheme, the present invention provides a method for identifying the meshing fault of the axle gearbox of a rail engineering vehicle, which has the following beneficial effects: the present invention uses Q-switched wavelets to construct a sparse dictionary library, which can effectively extract the harmonic components in the high-frequency vibration signal, which can reflect the gear meshing state. Extracting statistical characteristic indicators from the harmonic signal can effectively and accurately determine the gear meshing fault in real time. The method is simple and easy to implement, and has engineering practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0025] FIG. 1( a ) and FIG. 1( b ) are schematic diagrams of signal sparse decomposition of an axle gearbox in a normal state and a fault state in the present invention, wherein FIG. 1( a ) is a schematic diagram of sparse decomposition of an axle gearbox in a normal state, and FIG. 1( b ) is a schematic diagram of sparse decomposition of an axle gearbox in a fault state;

[0026] Figure 2 It is a schematic diagram of identifying normal and fault indicators of the axle gearbox of the present invention;

[0027] Figure 3 It is a schematic diagram of identifying normal and fault indicators of the axle gearbox of the present invention. DETAILED DESCRIPTION

[0028] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0031] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0032] In order to enable those skilled in the art to better understand the technical solution of the present invention, Figure 1(a) to Figure 3 The present invention is further described in detail.

[0033] The method for identifying an axle gearbox meshing fault of a rail engineering vehicle comprises:

[0034] In the first step, the monitoring signal X of the axle gearbox of the railway engineering vehicle is collected, and the sparse dictionary Φ is constructed using two Q-switched wavelet transforms with different quality factors. * By sparsely representing the harmonic signal and impulse signal in the monitoring signal X, the low-pass filter L(ω) and the high-pass filter H(ω) of the Q-modulated wavelet transform are:

[0035]

[0036]

[0037] in |ω|≤π, which is the frequency response function, α and β are the scale factors of the low-pass filter L(ω) and the high-pass filter H(ω), respectively, 0<β≤1, 0<α≤1, |L(ω)| 2 +|H(ω)| 2 =1;

[0038] The second step is to establish the objective function of sparse decomposition and use the Lagrangian contraction algorithm to solve the objective function to decompose the monitoring signal X into harmonic signals and impulse signals, where: Corresponding to the harmonic signal X H and impulse signal X L The sparse dictionary, w H and w L is the corresponding harmonic signal x H The sparse coefficients and impulse signal x L The sparse coefficient of , the objective function is:

[0039]

[0040] Among them, the parameter ε represents the noise level in the monitoring signal X, Respectively represent the ability to make ||w H ||1+||w L ||1 Take the minimum value of w H and w L .

[0041] To this end, the Lagrangian operator λ = [λ H ,λ L ] T , the constraint condition st Merge into the objective function to form the objective function of morphological component analysis:

[0042]

[0043] For w H and w L Derivative, take the extreme value, and get the harmonic signal X H and the impulse signal x L and its sparse coefficient w H 、w L ; Among them, λ H ,λ L Respectively represent w H and w L The Lagrangian operator.

[0044] Step 3: On the harmonic signal X HExtract the impulse index and skewness index from the , where the impulse index is:

[0045]

[0046] Skewness indicator

[0047] in, is the harmonic signal x H The average value of is the harmonic signal x H The variance of x H (n) is the harmonic signal, n is X H The serial number, N is X H The maximum number of sequence numbers.

[0048] When the pulse index is within the first range and the skewness index is within the second range, the axle gearbox is considered to be normal, otherwise, a meshing fault occurs.

[0049] In a preferred implementation of the method for identifying an axle gearbox meshing fault of a rail engineering vehicle, the first range is [5, 8.5] and the second range is [-0.15, 0.1].

[0050] In one embodiment, the method for identifying an axle gearbox meshing fault of a rail engineering vehicle comprises:

[0051] 1) Q-switched wavelet transform is selected to construct a sparse dictionary for sparsely representing the high-oscillation attributes and low-oscillation attributes in the signal.

[0052] 2) Establish the objective function of sparse decomposition of the signal, and use the Lagrangian contraction algorithm to solve the objective function of the sparse decomposition method based on the signal oscillation properties, so as to extract the gear vibration signal and decompose it into high-oscillation harmonic signals and low-oscillation impact signals.

[0053] 3) Extract the pulse index and skewness index from the harmonic signal. When the pulse index is within [5, 8.5] and the skewness index is within [-0.15, 0.1], the axle gearbox is considered to be normal, otherwise, a meshing fault occurs.

[0054] The specific method of step 1) is as follows:

[0055] Q-switched wavelet is a special wavelet transform method that can change the sampling frequency of the monitoring signal X. Its low-pass filter L(ω) and high-pass filter H(ω) are:

[0056]

[0057]

[0058] in |ω|≤π is the Daubechies frequency response with second-order vanishing moments. α and β are the scale factors of the low-pass filter and high-pass filter, respectively. The low-pass filter L(ω) and high-pass filter H(ω) of the Q-modulated wavelet transform satisfy the perfect reconstruction condition, that is, |L(ω)| 2 +|H(ω)| 2 =1. The scale parameters α and β determine the performance of the low-pass filter H0(ω) and the high-pass filter G0(ω). To ensure that the Q-modulated wavelet transform is not redundant, α and β must satisfy 0<β≤1, 0<α≤1. The low-pass filter L(ω) and the high-pass filter H(ω) are extracted from the monitoring signal x, which can sparsely characterize the high oscillation properties and low oscillation properties in the signal.

[0059] The specific method of step 2) is as follows:

[0060] Use the symbol Φ * The dictionary represents the low-pass filter L(ω) and the high-pass filter H(ω) constructed by the Q-modulated wavelet transform, that is, x = Φ * w. The signal X includes the harmonic signal x H and the impulse signal x L . X can be decomposed into: in Corresponding to the harmonic signal x H and the impulse signal x L The dictionary, w H and w L is the corresponding sparse coefficient. Therefore, the objective function is:

[0061]

[0062] The parameter ε represents the noise level in the signal X. To achieve the optimal solution to equation (3), the Lagrange multiplier method is used to give a suitable Lagrangian operator and transform the constrained optimization problem shown in equation (3) into the unconstrained optimization problem shown in the following equation:

[0063]

[0064] Where λ=[λ H ,λ L ] T is the Lagrangian operator, and λ is the signal x, and ε. Then the objective function of morphological component analysis is

[0065]

[0066] After the objective function is determined, the key to the sparse decomposition method based on the signal oscillation property lies in the construction of the dictionary and the solution of the objective function, which can obtain wH 、w L Finally, we get the harmonic signal X H and the impulse signal x L .

[0067] The specific method of step 3) is as follows:

[0068] Use the following formula to calculate the harmonic signal x H Extract the impulse indicator and skewness indicator from

[0069]

[0070]

[0071] in is x H The average value of is x H When the pulse index is within [5, 8.5] and the skewness index is within [-0.15, 0.1], the axle gearbox is considered normal, otherwise, a meshing fault occurs.

[0072] Refer to Figure 1(a) and Figure 1(b) for the sparse decomposition of the signal of the axle gearbox in normal and faulty states; in Figure 1(a) and Figure 1(b), the upper figure is the harmonic signal after sparse decomposition, the horizontal axis is time t / s, and the vertical axis is amplitude / g; the lower figure is the impact signal after sparse decomposition, the horizontal axis is time t / s, and the vertical axis is amplitude / g. It can be seen from Figure 1(a) that when the axle gearbox is normal, there is still a certain impact component in the harmonic signal, and the corresponding impact is relatively regular. Compared with Figure 1(b), when the gearbox fails, the impact component in the harmonic signal is more obvious, but the impact signal becomes very irregular, and the value fluctuates. For this reason, the harmonic signal extraction index is selected to identify the axle gearbox fault.

[0073] refer to Figure 2 The normal and fault indicators of the axle gearbox shown in the figure are pulse indicators on the horizontal axis and skewness indicators on the vertical axis. "*" indicates the monitoring data indicators of the axle gearbox with normal meshing, and "□" indicates the monitoring data indicators of the axle gearbox with serious faults. When the pulse indicator is within [5, 8.5] and the skewness indicator is within [-0.15, 0.1], the axle gearbox is normal; the axle gearbox data outside this range corresponds to meshing faults.

[0074] refer to Figure 3The indicator identification of minor faults shown in the figure, the horizontal axis is the pulse indicator, and the vertical axis is the skewness indicator. "*" indicates the monitoring data indicator of the axle gearbox with normal meshing, and "o" indicates the monitoring data indicator of the axle gearbox with minor faults. "□" indicates the monitoring data indicator of the axle gearbox with serious faults. The minor fault data still conforms to the rule that "when the pulse indicator is within [5, 8.5] and the skewness indicator is within [-0.15, 0.1], the axle gearbox is normal".

[0075] Methods include,

[0076] 1) Q-switched wavelet transform is selected to construct a sparse dictionary for sparsely representing the high-oscillation attributes and low-oscillation attributes in the signal.

[0077] 2) Establish the objective function of sparse decomposition of the signal, and use the Lagrangian contraction algorithm to solve the objective function of the sparse decomposition method based on the signal oscillation properties, so as to extract the gear vibration signal and decompose it into high-oscillation harmonic signals and low-oscillation impact signals.

[0078] 3) Extract the pulse index and skewness index from the harmonic signal. When the pulse index is within [5, 8.5] and the skewness index is within [-0.15, 0.1], the axle gearbox is considered to be normal, otherwise, a meshing fault occurs.

[0079] The specific method of step 1) is as follows:

[0080] Q-switched wavelet is a special wavelet transform method that can change the sampling frequency of the monitoring signal X. Its low-pass filter L(ω) and high-pass filter H(ω) are:

[0081]

[0082]

[0083] in |ω|≤π, is the frequency response function, α and β are the scale factors of the low-pass filter and high-pass filter respectively. The low-pass filter L(ω) and high-pass filter H(ω) of the Q-modulated wavelet transform meet the perfect reconstruction condition, that is, |L(ω)| 2 +|H(ω)| 2 =1. The scale parameters α and β determine the performance of the low-pass filter H0(ω) and the high-pass filter G0(ω). To ensure that the Q-modulated wavelet transform is not redundant, α and β must satisfy 0<β≤1, 0<α≤1. The low-pass filter L(ω) and the high-pass filter H(ω) are extracted from the monitoring signal x, which can sparsely characterize the high oscillation properties and low oscillation properties in the signal.

[0084] The specific method of step 2) is as follows:

[0085] Use the symbol Φ * The dictionary represents the low-pass filter L(ω) and the high-pass filter H(ω) constructed by the Q-modulated wavelet transform, that is, x = Φ * w. The signal X includes the harmonic signal x H and the impulse signal x L . X can be decomposed into: in Corresponding to the harmonic signal X H and the impulse signal x L The dictionary, w H and w L is the corresponding sparse coefficient. Therefore, the objective function is:

[0086]

[0087] The parameter ε represents the noise level in the signal X. To achieve the optimal solution to equation (3), the Lagrange multiplier method is used to give a suitable Lagrangian operator and transform the constrained optimization problem shown in equation (3) into the unconstrained optimization problem shown in the following equation:

[0088]

[0089] Where λ=[λ H ,λ L ] T is the Lagrangian operator, and λ is the signal X, and ε. Then the objective function of morphological component analysis is

[0090]

[0091] After the objective function is determined, the key to the sparse decomposition method based on the signal oscillation property lies in the construction of the dictionary and the solution of the objective function, which can obtain w H 、w L Finally, we get the harmonic signal x H and the impulse signal x L .

[0092] The specific method of step 3) is as follows:

[0093] Use the following formula to calculate the harmonic signal x H Extract the impulse indicator and skewness indicator from

[0094]

[0095]

[0096] in is x H The average value of is x H When the pulse index is within [5, 8.5] and the skewness index is within [-0.15, 0.1], the axle gearbox is considered normal, otherwise, a meshing fault occurs.

[0097] Example:

[0098] This embodiment provides a specific implementation process of the present invention in an axle gearbox test and verifies the effectiveness of the invention.

[0099] The axle gearbox of a rail engineering vehicle was tested on a test bench with an input speed of 1412r / min and an output speed of 758r / min. High-frequency vibration data of 10.4K per second was collected, and two axle gearboxes of the same model were selected for testing, one with normal meshing and the other with meshing failure. The sparse decomposition results of the high-frequency vibration signal using a sparse dictionary constructed using Q-switched wavelet are shown in Figure 1(a) and Figure 1(b). 28 sets of data were collected for the normal axle gearbox and 26 sets of data were collected for the faulty gearbox. Statistical features were extracted, such as Figure 2 When the pulse index of the harmonic signal after sparse decomposition is within [5, 8.5] and the skewness index is within [-0.15, 0.1], the axle gearbox is meshing normally. If it is beyond this range, the meshing gearbox fault occurs. Then a group of axle gearbox tests with slight faults are added, such as Figure 3 As shown, this is in line with the rule.

[0100] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for identifying an axle gearbox meshing fault of a rail engineering vehicle, characterized in that: It includes the following steps: The first step is to collect the monitoring signal of the axle gearbox of the rail engineering vehicle , using Q-switched wavelet transform to construct sparse dictionary Characterizing the monitoring signal by sparse Harmonic signals and impulse signals in the Q-modulated wavelet transform low-pass filter and high pass filter for: ， ， in , , which is the frequency response function, and Low pass filter and high pass filter The scale factor, , , ; The second step is to establish the objective function of sparse decomposition and use the Lagrangian shrinkage algorithm to solve the objective function to convert the monitoring signal It will be decomposed into harmonic signal and impulse signal, where , , Corresponding to harmonic signals and shock signal A sparse dictionary, and is the corresponding harmonic signal The sparse coefficient and impact signal of The sparse coefficient of , the objective function is: ， The constraint condition is st , Among them, the parameters Characterization monitoring signal The noise level in Respectively, they can allow Take the minimum value and ; To this end, the Lagrangian operator is introduced , the constraint condition st Merge into the objective function to form the objective function of morphological component analysis: ， Respectively and Derivative, take the extreme value, and get the harmonic signal and shock signal and its sparse coefficient , ;in, Respectively and The Lagrangian operator of Step 3: On the harmonic signal Extract the impulse index and skewness index from the , where the impulse index is: ， Skewness indicator , in, It is a harmonic signal The average value of It is a harmonic signal The variance of is the harmonic signal, n is The serial number, N is The maximum number of sequence numbers; When the pulse index is within the first range and the skewness index is within the second range, the axle gearbox is considered to be normal, otherwise, a meshing fault occurs.

2. The method for identifying an axle gearbox meshing fault of a rail engineering vehicle according to claim 1, characterized in that: The first range is [5, 8.5] and the second range is [-0.15, 0.1].

Images