Data cleaning and fault diagnosis method for gear box of main hoisting mechanism of quay crane

Abstract

The invention belongs to the technical field of fault diagnosis, and particularly relates to a data cleaning and fault diagnosis method for a main hoisting mechanism gearbox of a quay crane, which comprises the following steps of: S1, acquiring and preprocessing a signal; s2, the collected and preprocessed signals are screened; and S3, identifying poor gear meshing features. The data cleaning and fault diagnosis method for the main hoisting mechanism gearbox of the quay crane has the effects of realizing quantitative analysis and automatic diagnosis of gearbox faults and ensuring the accuracy of fault diagnosis.

Description

Technical Field

[0001] The invention belongs to the technical field of fault diagnosis, and in particular relates to a data cleaning and fault diagnosis method for a gear box of a main hoisting mechanism of a quay crane. Background Art

[0002] A shore container crane (abbreviated as a quay crane) is a large-scale equipment used for cargo loading and unloading at ports, and is generally installed on the shore of a port terminal. The power transmission system (gearbox) of the main lifting mechanism in the quay crane equipment is prone to bearing damage, gear wear and other faults. Therefore, the quay crane needs to have remote operation and maintenance capabilities such as remote fault diagnosis, fault prediction and predictive maintenance of the main lifting mechanism gearbox.

[0003] The existing technology mainly relies on manual inspection or installation of vibration sensors to diagnose the gearbox of the main lifting mechanism of the quay crane, and evaluates the equipment status through vibration signals. The former requires the staff to have rich experience and has high labor costs, and the diagnostic results are difficult to quantify and unify and are not accurate enough; the latter directly collects vibration signals without preprocessing and is suitable for monitoring gearboxes with stable speed. However, the gearbox speed of the main lifting mechanism of the quay crane is directly controlled by humans and changes speed frequently. There will be abnormal impacts in the collected vibration signals, and the spectrum obtained after Fourier transform will have chaotic spectrum lines. This phenomenon causes low accuracy of the diagnostic results.

[0004] The above-mentioned prior art solutions have the following defects: the diagnosis results of the gearbox of the main hoisting mechanism of the quay crane in the prior art are difficult to quantify and unify and are not accurate enough. Summary of the invention

[0005] The purpose of the present invention is to provide a data cleaning and fault diagnosis method for the gearbox of the main lifting mechanism of the quay crane, so as to solve the technical problem that the diagnosis results of the gearbox of the main lifting mechanism of the quay crane are difficult to quantify and unify and are not accurate enough, so as to achieve automatic fault diagnosis of the main lifting mechanism of the quay crane and ensure the accuracy of the diagnosis results.

[0006] In order to solve the above technical problems, the present invention provides a data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane, the method comprising the following steps:

[0007] S1. Signal acquisition and preprocessing. At least one vibration acceleration sensor is provided on the motor drive end and the gear box of the main lifting mechanism for signal acquisition. The vibration acceleration signal of the motor measuring point is x 0 , remove the vibration acceleration signal x 0 The DC bias component and low-frequency disturbance in the

[0008] S2, screening the collected and pre-processed signals;

[0009] Screening the collected and pre-processed signals specifically includes the following steps:

[0010] S2-1, remove shutdown and external interference data;

[0011] The signal x of the motor drive end measuring point length N 0 Divide into P 1 The signal y of group length n i (u), i = 1, 2...P 1 ; u = 1, 2...n, after the first data screening and the second data screening, a slice index B is formed;

[0012] S2-2, stable frequency identification, according to the slice index B from the signal y i (u) The corresponding slices are selected, and the above slices are subjected to adaptive tensor singular spectrum decomposition, and the envelope spectrum kurtosis value K is calculated respectively. g , and the correlation coefficient r g , perform the third data screening to obtain the one-dimensional array C for the next step;

[0013] S2-3, data reconstruction, obtain signal x 2 ;

[0014] S3, gear meshing fault feature recognition, by 2 The frequency domain signal F obtained by fast Fourier transform 2 ; and extract the frequency domain signal F 2 Gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The maximum value in the front and back intervals is used to quantitatively analyze the gear failure in the gearbox.

[0015] Furthermore, eliminating shutdown and external interference data specifically includes the following steps:

[0016] S2-1-1, the length of the motor measuring point is N, the signal x 0 Divide into P 1 The signal y of group length n i (u), i = 1, 2...P 1 , u=1,2...n, and calculate the root mean square RMS of each group of signals i and peak factor CF i ;

[0017] S2-1-2, perform the first data screening, retain the corresponding slice index to form a one-dimensional array A;

[0018] S2-1-3, perform a second data screening, that is, sort the one-dimensional array A from small to large, and divide the one-dimensional array A into w one-dimensional arrays based on continuity, retaining the array with the longest length to form the slice index B;

[0019] Furthermore, the root mean square RMS i and peak factor CF i The calculation formula is as follows:

[0020]

[0021]

[0022] Further, the first data screening, namely the root mean square RMS i and peak factor CF i and the power-on threshold T RMS and abnormal shock threshold T CF Compare and meet Rms i >T RMS And CF i <T CF The second condition is the slice index that meets the first data screening.

[0023] Furthermore, the second data screening is to sort the one-dimensional array A from small to large, and divide A into w one-dimensional arrays based on continuity, and retain the array with the longest length, which is the second data screening.

[0024] Furthermore, S2-2 specifically includes the following steps:

[0025] S2-2-1. From signal y according to slice index B i Filter out the corresponding slices in (u) and obtain P 2 The signal z of group length n j (u), j = 1, 2...P 2 ;

[0026] S2-2-2, for z j (u) Perform adaptive tensor singular spectrum decomposition to obtain the sum of g IMF components, Imf g，j Signal z j The g-th IMF component of (u);

[0027] S2-2-3. Calculate the envelope spectrum kurtosis value K of each IMF component g , and each IMF component is related to the original signal z j The correlation coefficient value r of (u) g ;

[0028] S2-2-4. Select the products that meet Kg × g -mean(K g × g )>0, and the frequency domain signal Z is obtained. j (u);

[0029] S2-2-5. Extract Z j (u) The frequency coordinate corresponding to the maximum amplitude in a specific interval, denoted by f j ;

[0030] S2-2-6, reverse the rotation frequency fr of the current motor and gearbox input shaft through the ideal double power frequency and rotation frequency relationship j ;

[0031] S2-2-7, establish a sliding window of length p, take the rotational frequency at the stable rotational speed stage, record the slice index corresponding to the rotational frequency, remove the repeated part, and obtain a one-dimensional array C;

[0032] S2-2-8, perform the third data screening, and the one-dimensional array C that meets the third data screening proceeds to the next step;

[0033] Furthermore, for z j (u) The specific steps of signal decomposition are as follows:

[0034] S2-2-2-1, z j (u) is constructed into a third-order tensor Z, the jth positive slice Z in the tensor Z ：，：，j It is expressed as:

[0035]

[0036] S2-2-2-2. Perform singular value decomposition on the tensor and obtain:

[0037]

[0038] Where L is the first-order dimension of the tensor S;

[0039] The rules for tensor multiplication are as follows:

[0040]

[0041] Among them, the variable k 1 , k 2 , ...k m+1-2o is a positive integer, p represents the order of the tensors involved in the tensor multiplication operation, <1, 1> means that the order of operations of tensor X and tensor Y during tensor multiplication is positive; m is the order of tensor X, l is the order of tensor Y; S 1is the dimension of the second order of tensor X, S o is the dimension of the (o + 1)-th order of tensor X;

[0042] S2-2-2-3. Let A g = U × 1，<1，1> S g，:， : × 2，<1，1> V T , and the first column of the j-th frontal slice of tensor A g is imf g，j , then we get:

[0043]

[0044] where Imf g，j is called the g-th IMF component of signal z j (u).

[0045] Furthermore, the data reconstruction specifically includes the following steps:

[0046] S2-3-1. Use the one-dimensional array C to extract the corresponding data from the slice signal z j (u) of the motor drive end measurement point and complete the signal reconstruction in the order of the one-dimensional array C, and finally obtain the signal x 1 ;

[0047] S2-3-2. Obtain the envelope spectrum of x 1 through discrete Hilbert transform and fast Fourier transform and take the modulus to obtain the frequency-domain signal F 1 ;

[0048] S2-3-3. Repeat steps S2-2-3, S2-2-4, S2-2-5, and S2-2-6 to obtain the finally stable rotation frequency Fr of the signal;

[0049] S2-3-4. Slice the gearbox measurement point signal according to step S2-2-1, and extract the slice signal corresponding to the index in the one-dimensional array C to complete the signal reconstruction to obtain the signal x 2 ;

[0050] Furthermore, the calculation methods of the envelope spectrum kurtosis value K g of each IMF component, and the correlation coefficient value r j between each IMF component and the original signal z g (u) are as follows:

[0051]

[0052] where, in the formula, e g is the envelope signal obtained after the signal is demodulated by Hilbert, and σ e is eg Standard deviation;

[0053]

[0054] Among them, r is the correlation coefficient, Cov(imf g , z j ) is the covariance between the IMF component and the original signal, is the variance of the IMF component, is the variance of the original signal.

[0055] Further, select the g × g -mean(K g × g )>0, the envelope spectrum of the reconstructed signal is obtained by discrete Hilbert transform and fast Fourier transform, and the frequency domain signal Z is obtained. j (u)

[0056] Furthermore, extract Z j (u) in the specific interval [FL a -ε 1 , F.L. a +ε 1 ];

[0057] Among them, FL a is the preset ideal double power frequency, ε 1 is the power supply frequency error threshold.

[0058] Furthermore, the ideal double power frequency and rotation frequency relationship can be used to reversely calculate the rotation frequency fr of the current motor and gearbox input shaft. j The specific formula is as follows:

[0059]

[0060] in, f a is the rated speed of the motor, FL b is the rated power frequency of the motor.

[0061] Furthermore, the determination method of being in the stable speed stage is as follows:

[0062] fr J+q -mean(fr J , fr J+1 …fr J+p-1 )≤ε 2 , q=0, 1…p-1, J=1, 2…np,

[0063] Then determine fr J， fr J+1 ...frJ+p-1 is the stable speed stage, where ε 2 is the speed deviation threshold.

[0064] Furthermore, the third data screening is to determine whether the one-dimensional array C meets the following conditions:

[0065] length(C)>ε 3

[0066] Among them, ε 3 is the manually set threshold, ε 3 The setting is used to avoid the problem of insufficient frequency resolution of the combined data due to the one-dimensional array being too short.

[0067] Furthermore, the calculation formula of the gear characteristic frequency is specifically as follows:

[0068] f mesh =z1×Fr in =z2×Fr out

[0069]

[0070]

[0071] Among them, z1, z2 are input shaft gear and output shaft gear; Fr in is the input shaft rotation frequency, Fr out is the output shaft rotation frequency, Fr in It is equal to the final stable frequency Fr of the signal in step S2-3-3.

[0072] Furthermore, the gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The specific intervals are:

[0073] [f mesh -ε 4 , f mesh +ε 4 ]、

[0074] Among them, ε 4 , ε 5 , ε 6 is the error threshold.

[0075] Furthermore, the identification of the characteristics of poor gear meshing specifically includes the following steps:

[0076] S3-1. Calculate the gear characteristic frequency of the gearbox gear of the main hoisting mechanism of the quay crane, which includes the input shaft rotation frequency Fri n , output shaft rotation frequency Fr out , gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft

[0077] S3-2. Obtain signal x through fast Fourier transform 2 The frequency spectrum of the frequency domain signal F 2 ;

[0078] S3-3, extract F 2 At the gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The maximum value in the front and back intervals can be used to quantitatively analyze the gear failure in the gearbox.

[0079] The beneficial effects of the present invention are:

[0080] 1. The signal reconstructed by the gearbox has a stable speed and high amplitude spectrum clarity. The corresponding frequency components can be extracted by combining the gearbox characteristic frequency calculation formula to realize quantitative analysis and automatic diagnosis of gearbox faults.

[0081] 2. The long-term data is carefully segmented through data slicing, and the signal noise reduction is completed by using adaptive tensor singular spectrum decomposition to extract the power frequency of each group of data slices. The slice index when the power frequency is stable in the slice data is determined by the sliding window method, and finally the group index is applied to the signal of the gearbox measurement point, and the signal reconstruction of the motor measurement point and the gearbox measurement point is completed at the same time to ensure the accuracy of fault diagnosis.

[0082] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0084] Figure 1 The present invention is a flow chart of a method for data cleaning and fault diagnosis of a gearbox of a main hoisting mechanism of a quay crane. Detailed implementation mode

[0085] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0086] Embodiment:

[0087] As Figure 1 shown, a data cleaning and fault diagnosis method for the gearbox of the main hoisting mechanism of a quay crane, the method comprising the following steps:

[0088] S1. Signal acquisition and preprocessing. A vibration acceleration sensor is installed at the motor drive end of the main hoisting mechanism, and several vibration acceleration sensors are installed on the gearbox according to the actual structure and customer needs. The acquisition method is trigger acquisition, that is, the motor measurement point will continuously acquire vibration acceleration data and calculate the root mean square value. If the calculated value exceeds the preset signal acquisition threshold, the sensors on the gearbox of the current mechanism will immediately acquire vibration acceleration signals x with a duration of T and a data point number of N 0 , and the DC bias component and low-frequency disturbance in the vibration acceleration signal x 0 are removed by using functions including but not limited to the DETREND function in MATLAB software;

[0089] S2. Screen the acquired and preprocessed signals;

[0090] The screening of the acquired and preprocessed signals specifically includes the following steps:

[0091] S2-1. Eliminate shutdown and external interference data;

[0092] The method of triggering the acquisition of a fixed-duration signal can only ensure that the front section of the signal is in the power-on state. There may be a situation where the device is manually shut down during the acquisition process, and the working conditions at the quay crane site are complex. The vibration impact caused by the start and stop of other devices around the measurement point will interfere with the signal acquisition. Therefore, it is necessary to eliminate the data containing these two situations. In addition, the corresponding device speed is not stable during the process of triggering the acquisition signal, and it is also necessary to screen out the vibration signals corresponding to the stable speed stage. Eliminating shutdown and external interference data specifically includes the following steps:

[0093] S2-1-1. Divide the signal x with a length of N at the motor measurement point 0 equally into P 1 groups of signals y with a length of n i (u), i = 1, 2... P 1, u=1,2…n, and calculate the RMS of each group of signals i and peak factor CF i ;

[0094] Among them, the root mean square RMS i and peak factor CF i The calculation formula is as follows:

[0095]

[0096]

[0097] S2-1-2, perform the first data screening to form a one-dimensional array A;

[0098] The first data screening is the root mean square RMS i and peak factor CF i and the power-on threshold T RMS and abnormal shock threshold T CF Compare and meet Rms i >T RMs And CF i <T CF The second condition is the slice index that meets the first data screening.

[0099] S2-1-3, perform a second data screening, that is, sort the one-dimensional array A from small to large, and divide the one-dimensional array A into w one-dimensional arrays based on continuity, retaining the array with the longest length to form the slice index B;

[0100] S2-2, stable frequency identification;

[0101] Stable frequency identification specifically includes the following steps:

[0102] S2-2-1. From signal y according to slice index B i Filter out the corresponding slices in (u) and obtain P 2 The signal z of group length n j (u), j = 1, 2...P 2 ;

[0103] S2-2-2, for z j (u) Decompose the signal to obtain the sum of g IMF components;

[0104] Specifically, for z j (u) The specific steps of signal decomposition are as follows:

[0105] S2-2-2-1, z j (u) is constructed into a third-order tensor Z, the jth positive slice Z in the tensor Z ：，：，jIt is expressed as:

[0106]

[0107] S2-2-2-2. Perform singular value decomposition on the tensor and obtain:

[0108]

[0109] Where L is the first-order dimension of the tensor S;

[0110] The rules for tensor multiplication are as follows:

[0111]

[0112] Among them, the variable k 1 , k 2 , ...k m+1-2o is a positive integer, p represents the order of the tensors involved in the tensor multiplication operation, <1, 1> means that the order of operations of tensor X and tensor Y during tensor multiplication is positive; m is the order of tensor X, l is the order of tensor Y; S 1 is the second-order dimension of the tensor X, S o is the dimension of the tensor X at the o+1th order;

[0113] S2-2-2-3, let A g =U× 1，<1，1> S g，:， :× 2，<1，1> V T , and the tensor A g The first column of the jth frontal slice is imf g，j , we get:

[0114]

[0115] Among them, Imf g，j Signal z j The g-th IMF component of (u).

[0116] S2-2-3. Calculate the envelope spectrum kurtosis value K of each IMF component g , and each IMF component is related to the original signal z j The correlation coefficient value r of (u) g ;

[0117] Envelope spectrum kurtosis value K g , and each IMF component is related to the original signal z j The correlation coefficient value r of (u) g The specific calculation method is as follows:

[0118]

[0119] Among them, in the formula, e g is the envelope signal obtained after Hilbert demodulation, σ e for e g Standard deviation;

[0120]

[0121] Among them, r is the correlation coefficient, Cov(imf g , z j ) is the covariance between the IMF component and the original signal, is the variance of the IMF component, is the variance of the original signal.

[0122] S2-2-4. Select the products that meet K g × g -mean(K g × g )>0, the envelope spectrum of the reconstructed signal is obtained by discrete Hilbert transform and fast Fourier transform, and the modulus is taken to perform noise reduction and reconstruction to obtain the frequency domain signal Z j (u);

[0123] S2-2-5. Extract Z j (u) In a specific interval [FL a -ε 1 , F.L. a +ε 1 The frequency coordinate corresponding to the maximum amplitude value within ] is denoted as f j ;

[0124] Among them, FL a is the preset ideal double power frequency, ε 1 is the power frequency error threshold.

[0125] S2-2-6, reverse the rotation frequency fr of the current motor and gearbox input shaft through the ideal double power frequency and rotation frequency relationship j ;

[0126] Specifically, the ideal double power frequency and rotation frequency relationship can be used to infer the rotation frequency fr of the current motor and gearbox input shaft. j The specific formula is as follows:

[0127]

[0128] Among them, f a is the rated speed of the motor, FL b is the rated power frequency of the motor.

[0129] S2-2-7, establish a sliding window with a length of p (1<p<P2), take the rotation frequency at the stable rotation speed stage, record the slice index corresponding to the rotation frequency, remove the repeated part, and obtain a one-dimensional array C;

[0130] Specifically, the method for determining whether the vehicle is in the stable speed stage is as follows:

[0131] fr J+q -mean(fr J , fr J+1 …fr J+p-1 )≤ε 2 , q=0, 1…p-1, J=1, 2…np,

[0132] Then determine fr J , fr J+1 ...fr J+p-1 is the stable speed stage, where ε 2 is the speed deviation threshold.

[0133] S2-2-8. Perform the third data screening. Specifically, the third data screening is to determine whether the one-dimensional array C meets the following conditions:

[0134] length(C)>ε 3

[0135] Among them, ε 3 is the manually set threshold, ε 3 The setting is used to avoid the problem that the one-dimensional array length is too short, resulting in insufficient frequency resolution of the combined data; the one-dimensional array C that meets the third data screening proceeds to the next step;

[0136] S2-3, data reconstruction;

[0137] The purpose of this step is to improve the spectral resolution of the signal and facilitate the subsequent characteristic frequency extraction. Data reconstruction specifically includes the following steps:

[0138] S2-3-1. Use the one-dimensional array C to slice the signal z from the motor drive end measurement point j (u) and complete the signal reconstruction in the order of the one-dimensional array C, and finally obtain the signal x 1 ;

[0139] S2-3-2. Obtain x through discrete Hilbert transform and fast Fourier transform 1 The envelope spectrum of the signal is modulo and the frequency domain signal F is obtained. 1 ;

[0140] S2-3-3, repeat steps S2-2-3, S2-2-4, S2-2-5 and S2-2-6 to obtain the final stable frequency Fr of the signal;

[0141] S2-3-4, slice the gearbox measurement point signal according to step S2-2-1, and extract the slice signal corresponding to the index in the one-dimensional array C to complete signal reconstruction, and obtain the signal x 2 ;

[0142] S3, identification of poor gear meshing characteristics;

[0143] The gearbox of the main hoisting mechanism of the quay crane is mostly a three-stage gear. Taking the first-stage gear as an example, the identification of the poor gear meshing characteristics specifically includes the following steps:

[0144] S3-1. Calculate the gear characteristic frequency of the gearbox gear of the main hoisting mechanism of the quay crane. The gear characteristic frequency includes the input shaft rotation frequency Fr in , output shaft rotation frequency Fr out , gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft Specifically, the calculation formula of the gear characteristic frequency is as follows:

[0145] f mesh =z1×Fr in =z2×Fr out

[0146]

[0147]

[0148] Among them, z1, z2 are input shaft gear and output shaft gear; Fr in is the input shaft rotation frequency, Fr out is the output shaft rotation frequency, Fr in It is equal to the final stable frequency Fr of the signal in step S2-3-3.

[0149] S3-2. Obtain signal x through fast Fourier transform 2 The frequency spectrum of the frequency domain signal F 2 ;

[0150] S3-3, extract F 2 At the gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The interval between front and back mesh -ε 4 , f mesh +ε4 ]、 The maximum value within can be used to quantitatively analyze the gear failure in the gearbox; where ε 4 , ε 5 , ε 6 is the error threshold.

[0151] In summary, the present invention proposes a method for automatically identifying the speed stabilization stage during the operation of the variable speed mechanism of the quay crane by using the vibration acceleration signal of the motor measuring point, and helping the gearbox measuring point to complete signal screening and realize frequency extraction; the long-term data is finely divided by data slicing, and the signal noise reduction is completed by using adaptive tensor singular spectrum decomposition, and the power supply frequency of each group of data slices is extracted; the slice index when the power supply frequency is stable in the slice data is determined by using a sliding window method, and finally the group of indexes is applied to the signal of the gearbox measuring point, and the signal reconstruction of the motor measuring point and the gearbox measuring point is completed at the same time; the signal reconstructed by the gearbox has a stable speed and a high clarity of the amplitude spectrum, and the corresponding frequency components are extracted in combination with the gearbox characteristic frequency calculation formula, so that the quantitative analysis and automatic diagnosis of gearbox faults can be realized.

[0152] The various devices selected in this application are all universal standard parts or parts known to those skilled in the art, and their structures and principles can be known to those skilled in the art through technical manuals or conventional experimental methods.

[0153] In the description of the embodiments of the present invention, unless otherwise clearly specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0154] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the 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 limiting the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0155] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A data cleaning and fault diagnosis method for the gearbox of the main hoisting mechanism of the quay crane. It is characterized in that The method comprises the following steps: S1. Signal acquisition and preprocessing. The motor drive end and gear box of the main lifting mechanism are respectively equipped with at least one vibration acceleration sensor for acquisition. The vibration acceleration signal of the motor measuring point is x 0 , remove the vibration acceleration signal x 0 The DC bias component and low-frequency disturbance in the S2, screening the collected and pre-processed signals; Screening the collected and pre-processed signals specifically includes the following steps: S2-1, remove shutdown and external interference data; The signal x with a measurement point length of N at the motor drive end 0 is equally divided into P 1 groups of signals y with a length of n i (u), where i = 1, 2... P 1 ; u = 1, 2... n, and they go through the first data screening and the second data screening successively to form the slice index B; S2-2, stable frequency identification, according to the slice index B from the signal y i (u) The corresponding slices are selected, and the above slices are subjected to adaptive tensor singular spectrum decomposition, and the envelope spectrum kurtosis value K is calculated respectively. g , and the correlation coefficient r g , perform the third data screening to obtain the one-dimensional array C for the next step; S2-3, data reconstruction, obtain signal x 2 ; S3, gear meshing fault feature recognition, by 2 The frequency domain signal F obtained by fast Fourier transform 2 ; and extract the frequency domain signal F 2 Gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The maximum value in the front and back intervals is used to quantitatively analyze the gear failure in the gearbox.

2. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 1, It is characterized in that Eliminating shutdown and external interference data specifically includes the following steps: S2-1-1, the length of the motor measuring point is N, the signal x 0 Divide into P 1 Slice signal y with group length n i (u), i=1,2…P 1 , u=1,2…n, and calculate the RMS of each group of signals i and peak factor CF i ; S2-1-2, perform the first data screening, retain the corresponding slice index to form a one-dimensional array A; S2-1-3. Perform the second data screening, that is, sort the one-dimensional array A from small to large, and divide the one-dimensional array A into w one-dimensional arrays based on continuity, retaining the array with the longest length to form the slice index B.

3. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 2, It is characterized in that Root Mean Square RMS i and peak factor CF i The calculation formula is as follows: 。 4. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 3, It is characterized in that The first data screening is the root mean square RMS i and peak factor CF i and the power-on threshold T RMS and abnormal shock threshold T CF Compare and meet Rms i >T RMS And CF i <T CF The second condition is the slice index that meets the first data screening.

5. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 4, It is characterized in that The second data screening is to sort the one-dimensional array A from small to large, and divide A into w one-dimensional arrays based on continuity. The array with the longest length is retained, which is the second data screening.

6. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 1, It is characterized in that S2-2 specifically includes the following steps: S2-2-1. From signal y according to slice index B i Filter out the corresponding slices in (u) and obtain P 2 The signal z of group length n j (u), j = 1, 2…P 2 ; S2-2-2, for z j (u) Perform adaptive tensor singular spectrum decomposition to obtain g IMF components, Imf g,j Signal z j The g-th IMF component of (u); S2-2-3. Calculate the envelope spectrum kurtosis value K of each IMF component g , and each IMF component is related to the original signal z j The correlation coefficient value r of (u) g ; S2-2-4. Select the products that meet K g × g -mean(K g × g )>0 to perform noise reduction and reconstruction to obtain the frequency domain signal Z j (u); S2-2-5. Extract Z j (u) The frequency coordinate corresponding to the maximum amplitude in the previous and next intervals, denoted as f j ; S2-2-6, reverse the rotation frequency fr of the current motor and gearbox input shaft through the ideal double power frequency and rotation frequency relationship j ; S2-2-7, establish a sliding window of length p, take the rotational frequency at the stable rotational speed stage, record the slice index corresponding to the rotational frequency, remove the repeated part, and obtain a one-dimensional array C; S2-2-8. Perform the third data screening, and the one-dimensional array C that meets the third data screening will proceed to the next step.

7. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that Right j (u) The specific steps of signal decomposition are as follows: S2-2-2-1, z j (u) is constructed into a third-order tensor Z, the jth positive slice Z in the tensor Z :,:,j It is expressed as: S2-2-2-2. Perform singular value decomposition on the tensor and obtain: Where L is the first-order dimension of the tensor S; The rules for tensor multiplication are as follows: Among them, the variable k 1 ,k 2 ,…k m+l-2o is a positive integer, p represents the order of the tensors involved in the tensor multiplication operation, <1,1> means that the order of operations of tensor X and tensor Y during tensor multiplication is positive; m is the order of tensor X, l is the order of tensor Y; S 1 is the second-order dimension of the tensor X, S o is the dimension of the tensor X at the o+1th order; S2-2-2-3, let A g =U× 1,<1,1> S g,:,: × 2,<1,1> V T , and the tensor A g The first column of the jth frontal slice is imf g,j , we get: Among them, Imf g,j Signal z j The g-th IMF component of (u).

8. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 7, It is characterized in that Data reconstruction specifically includes the following steps: S2-3-1. Use the one-dimensional array C to slice the signal z from the motor drive end measurement point j (u) and complete the signal reconstruction in the order of the one-dimensional array C, and finally obtain the signal x 1 ; S2-3-2. Obtain x through discrete Hilbert transform and fast Fourier transform 1 The envelope spectrum of the signal is modulo and the frequency domain signal F is obtained. 1 ; S2-3-3, repeat steps S2-2-3, S2-2-4, S2-2-5 and S2-2-6 to obtain the final stable frequency Fr of the signal; S2-3-4, slice the gearbox measurement point signal according to step S2-2-1, and extract the slice signal corresponding to the index in the one-dimensional array C to complete signal reconstruction, and obtain the signal x 2 .

9. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that The envelope spectrum kurtosis value K of each IMF component g , and each IMF component is related to the original signal z j The correlation coefficient value r of (u) g The specific calculation method is as follows: Among them, in the formula, e g is the envelope signal obtained after Hilbert demodulation, σ e for e g Standard deviation; Among them, r is the correlation coefficient, Cov(imf g ,z j ) is the covariance between the IMF component and the original signal, is the variance of the IMF component, is the variance of the original signal.

10. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that Select the one that satisfies K g × g -mean(K g × g )>0, the envelope spectrum of the reconstructed signal is obtained by discrete Hilbert transform and fast Fourier transform, and the frequency domain signal Z is obtained. j (u).

11. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 10, It is characterized in that Extract Z j (u) in the specific interval [FL a -ε 1 ,FL a +ε 1 ]; Among them, FL a is the preset ideal double power frequency, ε 1 is the power supply frequency error threshold.

12. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that The ideal double power frequency and rotation frequency relationship is used to infer the rotation frequency fr of the current motor and gearbox input shaft j The specific formula is as follows: Among them, f a is the rated speed of the motor, FL b is the rated power frequency of the motor.

13. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that The determination method of being in the stable speed stage is as follows: fr J+q -mean(fr J ,fr J+1 …fr J+p-1 )≤ε 2 ,q=0,1…p-1,J=1,2…n-p, Then determine fr J ,fr J+1 …fr J+p-1 is the stable speed stage, where ε 2 is the speed deviation threshold.

14. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 6, It is characterized in that The third data screening is to determine whether the one-dimensional array C meets the following conditions: length(C)>ε 3 Among them, ε 3 is the manually set threshold, ε 3 The setting is used to avoid the problem of insufficient frequency resolution of the combined data due to the one-dimensional array being too short.

15. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 1, It is characterized in that The calculation formula of the gear characteristic frequency is specifically as follows: f mesh =z1×Fr in =z2×Fr out Among them, z1, z2 are input shaft gear and output shaft gear; Fr in is the input shaft rotation frequency, Fr out is the output shaft rotation frequency, Fr in It is equal to the final stable frequency Fr of the signal in step S2-3-3.

16. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 15, It is characterized in that Gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The specific intervals are: [f mesh -e 4 ,f mesh +e 4 ]、 Among them, ε 4 , ε 5 , ε 6 is the error threshold.

17. A data cleaning and fault diagnosis method for a gearbox of a main hoisting mechanism of a quay crane as claimed in claim 1, It is characterized in that The identification of gear meshing defects specifically includes the following steps: S3-1. Calculate the gear characteristic frequency of the gearbox gear of the main hoisting mechanism of the quay crane. The gear characteristic frequency includes the input shaft rotation frequency Fr in , output shaft rotation frequency Fr out , gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft S3-2. Obtain signal x through fast Fourier transform 2 The frequency spectrum of the frequency domain signal F 2 ; S3-3, extract F 2 At the gear meshing frequency f mesh , Input shaft meshing frequency sideband Meshing frequency sideband with output shaft The maximum value in the front and back intervals can be used to quantitatively analyze the gear failure in the gearbox.

Images