A vibration analysis and diagnosis method for hydraulic system faults

Abstract

The present application relates to the technical field of hydraulic system fault diagnosis, and specifically to a hydraulic system fault vibration analysis and diagnosis method, the method comprising: collecting the time series pressure sequence of the hydraulic system and the vibration amplitude sequence in each direction; constructing a singular matrix using the elements in each intrinsic mode function after modal decomposition of the vibration amplitude sequence; determining the membership degree based on the distance between any two singular vectors in the singular matrix; determining the decomposition state feature difference based on the membership degree and the pressure data difference corresponding to any singular vector and its difference vector; determining the multi-scale non-homomorphic coefficient of the intrinsic mode function based on the decomposition state feature difference under all different scale groups, which is used to screen the intrinsic mode function, and analyzing the vibration fault diagnosis result of the hydraulic system with the screened intrinsic mode function. The present application makes the fault diagnosis result of the hydraulic pump more accurate.

Description

Technical Field

[0001] The present application relates to the technical field of hydraulic system fault diagnosis, and in particular to a hydraulic system fault vibration analysis and diagnosis method. Background Art

[0002] The hydraulic pump is a power element and the core of the hydraulic system. It converts external mechanical energy into internal liquid pressure energy and then outputs it in a certain way. Because of its strong robustness, fast response speed, and stepless speed regulation, it is widely used in construction machinery, industrial machinery, transportation tools and other fields. Therefore, the fault diagnosis and prevention of hydraulic pumps is an important prerequisite for ensuring the smooth and safe production process. Once a hydraulic system fails, it may stop running and repair at the least, or endanger personal safety at the worst. At present, many experts and scholars have conducted research on the fault diagnosis method of hydraulic pumps, mainly for feature analysis of vibration signals, and then obtain diagnostic results after matching with existing fault features, but time-frequency analysis itself is relatively difficult.

[0003] Since the hydraulic pump needs to ensure a certain degree of sealing, different types of signals may be superimposed on each other in the vibration signal, and valuable information cannot be accurately identified. In addition, the period of the abnormal signal caused by the fault is irregular. The traditional method does not measure the regularity differences of the vibration signal from multiple scales, making the fault diagnosis effect of the hydraulic system inaccurate. Summary of the invention

[0004] In order to solve the above technical problems, the present application provides a hydraulic system fault vibration analysis and diagnosis method to solve the existing problems.

[0005] A hydraulic system fault vibration analysis and diagnosis method of the present application adopts the following technical solution:

[0006] An embodiment of the present application provides a hydraulic system fault vibration analysis and diagnosis method, the method comprising the following steps:

[0007] Collect the time series pressure sequence of the hydraulic system and the vibration amplitude sequence in all directions;

[0008] The vibration amplitude sequence is subjected to modal decomposition to obtain the intrinsic mode function of each decomposition layer; the three-dimensional vibration matrix is ​​determined based on the element values ​​at the corresponding positions of all decomposition layers; the three-dimensional vibration matrix is ​​grouped at different scales to obtain each batch sample matrix respectively;

[0009] Constructing a singular matrix based on the batch sample matrix; determining a difference vector of any singular vector based on the distance between any two singular vectors in the singular matrix; determining a degree of membership between any singular vector and its difference vector based on the distance;

[0010] Based on the difference in pressure data between any singular vector and its difference vector in the corresponding time period in the time series pressure sequence, the suppression coefficient ratio is determined; based on the membership degree and the suppression coefficient ratio, the difference in decomposition state characteristics of the intrinsic mode function is determined;

[0011] Determine the non-isomorphic jump coefficients of the intrinsic mode function at different step sizes based on the differences in decomposition state characteristics under all different scale groups; determine the multi-scale non-isomorphic coefficients of the intrinsic mode function based on the non-isomorphic jump coefficients under all step sizes;

[0012] The intrinsic mode functions are screened based on multi-scale non-homomorphic coefficients, and the screened intrinsic mode functions are used to analyze the vibration fault diagnosis results of the hydraulic system.

[0013] Preferably, the method for collecting the time series pressure sequence and the vibration amplitude sequence in each direction includes:

[0014] A pressure sensor is installed at the oil outlet of the hydraulic pump, and the pressure data collected during the collection period are combined into a time-series pressure sequence;

[0015] Acceleration sensors are installed at several directions of the housing of the hydraulic pump, and the vibration data collected during the collection period are respectively combined into vibration amplitude sequences in each direction.

[0016] Preferably, the method for obtaining each batch sample matrix includes:

[0017] For the eigenmode functions of each decomposition layer, the elements of the eigenmode functions at the same position in all directions are arranged vertically to form three-dimensional vibration vectors at corresponding positions;

[0018] The three-dimensional vibration vectors at all positions in the eigenmode function are combined into a three-dimensional vibration matrix;

[0019] The three-dimensional vibration matrix is ​​scaled according to a preset scale, and each three-dimensional vibration vector and subsequent three-dimensional vibration vectors of a preset scale are combined into a batch sample matrix;

[0020] All batch sample matrices in the three-dimensional vibration matrix are combined into a three-dimensional vibration sub-matrix.

[0021] Preferably, the method for determining the difference vector of any singular vector comprises:

[0022] Perform singular value decomposition on the batch sample matrix, and form singular vectors from the decomposed singular values;

[0023] The singular vectors of all batch sample matrices in the three-dimensional vibration submatrix are respectively used as column vectors to form a singular matrix;

[0024] Calculate the sum of the absolute values ​​of the differences between all corresponding elements of any two singular vectors in the singular matrix as the distance between any two singular vectors;

[0025] For any singular vector, the singular vector with the largest distance between it and the singular vector is taken as the difference vector of any singular vector.

[0026] Preferably, the method for determining the similarity includes:

[0027] Calculate the result of raising the distance to a power of a preset number;

[0028] Calculate the inverse of the ratio of the result to a preset similarity tolerance factor as an exponent of an exponential function with a natural constant as a base;

[0029] The calculation result of the exponential function is used as the similarity.

[0030] Preferably, the method for determining the suppression coefficient ratio includes:

[0031] Calculate the mean value of the pressure data of any singular vector and its difference vector in the corresponding time period in the time series pressure sequence;

[0032] The normalized value of the absolute value of the difference between the two pressure data means is calculated as the suppression coefficient ratio.

[0033] Preferably, the method for determining the difference in decomposition state characteristics includes:

[0034] The average of the ratios of the suppression coefficients to the membership degrees between all singular vectors in the eigenmode function and their difference vectors is taken as the characteristic difference of the decomposition state of the eigenmode function.

[0035] Preferably, the method for determining the non-homomorphic jump coefficient includes:

[0036] Set the preset step size;

[0037] Calculate the absolute value of the difference between each scale and the decomposition state feature difference at the scale after the preset step length, and record it as the first absolute value of the difference;

[0038] The sum of the absolute values ​​of the first differences of all scales is taken as the non-isomorphic jump coefficient of the intrinsic mode function at a preset step size.

[0039] Preferably, the method for determining the multi-scale non-homomorphic coefficients includes:

[0040] The non-homomorphic jump coefficients of all step sizes are arranged into a sequence according to the step size from small to large, and the entropy value of the first-order difference sequence of the sequence is calculated as the multi-scale non-homomorphic coefficient.

[0041] Preferably, the analysis method of the vibration fault diagnosis result of the hydraulic system includes:

[0042] Discarding a preset number of intrinsic mode functions with the largest multi-scale non-homomorphic coefficients;

[0043] The remaining intrinsic mode functions after discarding the historically collected vibration amplitude sequences in all directions are used as the training set of the neural network, with two labels of "fault" and "non-fault", to obtain the fault diagnosis training model;

[0044] The vibration amplitude sequence to be diagnosed is input into the fault diagnosis training model and the label is output.

[0045] This application has at least the following beneficial effects:

[0046] The present application collects the time-series pressure sequence of the hydraulic system and the vibration amplitude sequence in all directions. Its beneficial effect is that the time-series pressure sequence is used to assist the vibration signal in time-frequency analysis, so as to eliminate the interference of pressure on the vibration signal. The present application uses modal decomposition of the vibration amplitude sequence to obtain the intrinsic mode function of each decomposition layer; determines the three-dimensional vibration matrix based on the element values ​​of the corresponding positions of all decomposition layers; groups the three-dimensional vibration matrix at different scales, and obtains each batch sample matrix respectively. Its beneficial effect is that the features can be extracted in a targeted manner, and deeper signal features can be discovered, and the local characteristics and frequency distribution of the signal can be better understood, so as to realize signal analysis, processing and feature extraction. By analyzing the changing characteristics of the vibration data near each moment, the three-dimensional vibration matrix is ​​grouped, so as to facilitate the analysis of local detail vibration data. The present application constructs a singular matrix based on a batch sample matrix; determines the difference vector of any singular vector based on the distance between any two singular vectors in the singular matrix; determines the membership between any singular vector and its difference vector based on the distance, and its beneficial effect is that the singular vector can reflect the characteristics of the batch sample matrix, which is a means of dimensionality reduction, and is convenient for the subsequent calculation of entropy values ​​at different scales; at the same time, by calculating the membership, the membership relationship between the two vectors is analyzed to accurately evaluate the possibility of failure of the hydraulic system. The present application determines the suppression coefficient ratio based on the pressure data difference between any singular vector and its difference vector in the corresponding time period in the time series pressure sequence; determines the difference in the decomposition state characteristics of the intrinsic mode function based on the membership and the suppression coefficient ratio, and its beneficial effect is that the suppression coefficient ratio is introduced to eliminate the deviation caused by the pressure change to a certain extent, and calculate the degree of influence of the entropy value at different scales in the intrinsic mode function, thereby timely weakening the influence on the original vibration intensity and increasing the accuracy of the analysis of the difference in the decomposition state characteristics of the intrinsic mode function. The present application determines the non-isomorphic jump coefficients of the intrinsic mode function at different step sizes based on the differences in decomposition state characteristics under all different scale groups; and determines the multi-scale non-isomorphic coefficients of the intrinsic mode function based on the non-isomorphic jump coefficients under all step sizes. The beneficial effect is that the differences in decomposition state characteristics under different scales are combined to analyze the differences in decomposition state characteristics of the intrinsic mode function as a whole at different scales, thereby screening out false components and excluding information that does not carry valuable information. The present application screens the intrinsic mode function based on multi-scale non-isomorphic coefficients, and uses the screened intrinsic mode function to analyze the vibration fault diagnosis results of the hydraulic system. The beneficial effect is that it extracts more adaptable features in the vibration signal, and the results of fault diagnosis of the hydraulic pump are more accurate compared to the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 A flow chart of a hydraulic system fault vibration analysis and diagnosis method provided in this application;

[0049] Figure 2 Construct a flow chart for the indicators of vibration fault diagnosis results. DETAILED DESCRIPTION

[0050] In order to further explain the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the following is a detailed description of the hydraulic system fault vibration analysis and diagnosis method proposed by the present application, its specific implementation method, structure, characteristics and effects, in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0051] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0052] The specific scheme of the hydraulic system fault vibration analysis and diagnosis method provided by the present application is described in detail below with reference to the accompanying drawings.

[0053] An embodiment of the present application provides a hydraulic system fault vibration analysis and diagnosis method.

[0054] Specifically, the following hydraulic system fault vibration analysis and diagnosis method is provided, please refer to Figure 1 , the method comprises the following steps:

[0055] Step S001, collecting the time-series pressure sequence and vibration amplitude sequence in each direction of the hydraulic system.

[0056] In one embodiment of the present application, taking an axial piston pump as an example, a vibration signal is collected on the housing of the axial piston pump as a vibration signal of the entire pump.

[0057] Furthermore, in this embodiment, the outer shell of the axial piston pump area is regarded as a cylinder. In the operating posture, the center of the circular surface is used as the origin to make an x-axis horizontal to the right and a y-axis vertically upward with the ground, and then a z-axis perpendicular to the circular surface is made. The three axes are mutually perpendicular, and then an acceleration sensor is installed on each axis to detect the vibration signal. The vibration signals within the acquisition period are respectively organized into a vibration amplitude sequence in time order, and the vibration amplitude sequence of the x-axis, y-axis, and z-axis can be obtained. At the same time, a pressure sensor is installed at the oil outlet of the axial piston pump to obtain the time-series pressure sequence within the acquisition period.

[0058] It should be noted that, in this embodiment, the sampling frequency K of the acceleration sensor and the pressure sensor is set to 2 kHz, and the collection period is set to 10 s. The implementer can set it according to the actual situation.

[0059] At this point, the time series pressure sequence of the hydraulic system and the vibration amplitude sequence in each direction can be obtained, and the time series pressure sequence is used to assist the vibration signal in time-frequency analysis. In other embodiments of the present application, other hydraulic pumps can also be used to collect and analyze the hydraulic system fault vibration data.

[0060] Step S002, using modal decomposition of the vibration amplitude sequence to obtain the intrinsic mode function of each decomposition layer; determining the three-dimensional vibration matrix based on the element values ​​of the corresponding positions of all decomposition layers; grouping the three-dimensional vibration matrix at different scales to obtain each batch sample matrix respectively.

[0061] In one embodiment of the present application, since the vibration signals of the three axes are non-stationary signals, directly inputting them into the neural network cannot perform targeted feature extraction and it is difficult to discover deeper signal features. Therefore, it is necessary to perform variational mode decomposition (VMD) on them separately.

[0062] It should be noted that variational mode decomposition (VMD) can decompose the vibration amplitude sequence into multiple intrinsic mode functions (IMFs). By decomposing the signal into IMFs, the local characteristics and frequency distribution of the signal can be better understood, thereby realizing signal analysis, processing and feature extraction.

[0063] In this embodiment, the number of VMD decomposition layers is set to 9, that is, each vibration amplitude sequence obtains 9 intrinsic mode functions IMFs, which are In other embodiments of the present application, the number of VMD decomposition levels may be set according to actual conditions.

[0064] Preferably, in the decomposition process of VMD, since the number of decomposition levels and the termination conditions are preset, some information without actual semantics, ie, false components, may be generated.

[0065] Based on this, this embodiment calculates the entropy values ​​of the data in different intrinsic mode functions of IMFs at different scales to screen out the false components in IMFs, thereby performing more accurate vibration analysis and diagnosis of hydraulic system faults.

[0066] Furthermore, for the intrinsic mode function In this embodiment, the x-axis is for ; y-axis for ; z-axis for , L is the sequence length, and each element value represents the vibration amplitude.

[0067] in three directions Take one element of the corresponding position in order and arrange them vertically. Taking the i-th position as an example, we get the three-dimensional vibration vector , T represents matrix transposition, and then the three-dimensional vibration vectors of all positions are combined into a three-dimensional vibration matrix .

[0068] Similarly, the three-dimensional vibration vector and three-dimensional vibration matrix of each decomposition layer can be obtained.

[0069] In one embodiment of the present application, since details are easily overlooked when analyzing overall vibration data, this embodiment groups the three-dimensional vibration matrix by analyzing the changing characteristics of the vibration data near each moment, thereby facilitating the analysis of local detail vibration data.

[0070] Further, each three-dimensional vibration matrix is ​​batch grouped. In this embodiment, the scale is set , for each three-dimensional vibration vector, take The three-dimensional vibration vectors are grouped into a group, and the group division method is overlapping selection until a certain The right side is insufficient vectors.

[0071] Based on this, we can get the three-dimensional vibrator matrix , batch sample matrix For size The matrix of .

[0072] Step S003, constructing a singular matrix based on the batch sample matrix; determining a difference vector of any singular vector based on the distance between any two singular vectors in the singular matrix; and determining a degree of membership between any singular vector and its difference vector based on the distance.

[0073] Preferably, in one embodiment of the present application, SVD singular value decomposition is performed on each batch sample matrix to obtain a scale The singular values ​​of each batch sample matrix under the condition are arranged vertically to obtain singular vectors, and the i-th batch sample matrix is For example, the i-th batch sample matrix The singular vectors of , the singular vectors of all batch sample matrices in the three-dimensional oscillator submatrix are taken as column vectors to form a singular matrix Q.

[0074] It should be noted that the singular vector can reflect the characteristics of the batch sample matrix and is a means of dimensionality reduction, which facilitates the subsequent calculation of entropy values ​​at different scales. SVD singular value decomposition is a well-known technology and will not be described in detail.

[0075] Further, calculate each singular vector and the remaining The distance between the singular vectors is calculated by taking the i-th singular vector as an example. The maximum value of the sum of the absolute values ​​of the differences between the corresponding elements of the two singular vectors is calculated, which is recorded as ,Will As the distance between the i-th singular vector and the j-th singular vector, the j-th singular vector is the vector corresponding to the maximum absolute value of the difference between the elements corresponding to the i-th singular vector.

[0076] Preferably, this embodiment uses the j-th singular vector as the difference vector of the ith singular vector.

[0077] Similarly, each singular vector in the singular matrix can obtain its corresponding difference vector.

[0078] Preferably, in one embodiment of the present application, in order to identify the similarity between the two singular vectors with large differences, the present embodiment calculates the membership degree between the two singular vectors:

[0079]

[0080] In the formula, Indicated in scale The membership degree between the next i-th singular vector and the j-th singular vector, is an exponential function with the natural constant e as base, Indicated in scale The distance between the next i-th singular vector and the j-th singular vector, r represents the similarity tolerance factor.

[0081] It should be noted that the similarity tolerance factor is a preset value. , implementers can set it according to actual conditions. is the standard deviation of the vibration amplitude sequence corresponding to the singular vector.

[0082] It should be noted that, when the distance between the i-th singular vector and the j-th singular vector is larger, the distance is magnified by L power, so that the similarity between the i-th singular vector and the j-th singular vector is smaller, that is, the affiliation between them is more distant.

[0083] Furthermore, the membership degree is controlled within the range of 0 to 1 through an exponential function. The larger the membership degree, the more similar the two singular vectors are, and the more likely the batch sample matrices containing the two singular vectors are to belong to the same vibration signal, that is, the smaller the possibility of a hydraulic system failure; the smaller the membership degree, the more dissimilar the two singular vectors are, and the more likely the batch sample matrices containing the two singular vectors are to belong to different vibration signals, that is, the greater the possibility of a hydraulic system failure.

[0084] Step S004, determining the suppression coefficient ratio based on the pressure data difference between any singular vector and its difference vector in the corresponding time period in the time series pressure sequence; determining the decomposition state characteristic difference of the intrinsic mode function based on the membership degree and the suppression coefficient ratio.

[0085] In one embodiment of the present application, since the faulty hydraulic component periodically impacts other components during operation, the vibration signal is more orderly and has smaller entropy values ​​at different scales compared to the normal hydraulic pump signal.

[0086] Furthermore, when the pressure at the outlet of the hydraulic oil is greater, the vibration of the hydraulic pump is more suppressed, and the entropy value of the vibration signal at different scales will be relatively reduced. Therefore, this embodiment introduces the suppression coefficient ratio to eliminate the deviation caused by the pressure change to a certain extent.

[0087] In the time series pressure sequence, calculate the average pressure value in the time period corresponding to the i-th singular vector and the j-th singular vector , The difference is normalized to obtain the suppression coefficient ratio between the i-th singular vector and the j-th singular vector. :

[0088]

[0089] in, represents the suppression coefficient ratio between the i-th singular vector and the j-th singular vector, , They represent the average pressure values ​​in the time periods corresponding to the i-th singular vector and the j-th singular vector, respectively. , They respectively represent the maximum pressure data and the minimum pressure data in the time series pressure sequence.

[0090] It should be noted that the greater the difference between the normalized average pressure values ​​in the time periods corresponding to the i-th singular vector and the j-th singular vector, that is, the greater the suppression coefficient ratio, the greater the influence of the pressure data at this time on the vibration data, and the more it is necessary to eliminate this influence to a certain extent.

[0091] In one embodiment of the present application, the decomposition state characteristic difference of the eigenmode function is constructed by combining the membership degree and suppression coefficient ratio between all singular vectors in the eigenmode function and their difference vectors. .in, Indicated in scale The decomposition state characteristics of the eigenmode functions under .

[0092] Preferably, the method for constructing the decomposition state characteristic difference is: taking the average of the ratio of the suppression coefficient to the membership degree between all singular vectors in the intrinsic mode function and their difference vectors as the decomposition state characteristic difference of the intrinsic mode function.

[0093] In other embodiments of the present application, the suppression coefficient ratio between all singular vectors in the intrinsic mode function and their difference vectors may also have other negative correlations with the membership degree, such as difference, negative exponential relationship, etc., which can be set by the implementer according to actual conditions.

[0094] It should be noted that the suppression coefficient ratio between each singular vector and its difference vector is divided by the membership degree, and the suppression coefficient ratio is used to calculate the weighted average of the inverse of the membership degree, so that when the suppression coefficient ratio is larger, it means that the pressure has a greater impact on the calculation of the entropy value at different scales in the intrinsic mode function, which leads to the weakening of the original vibration intensity, and it is necessary to increase the characteristic difference of the decomposition state of the intrinsic mode function; when the membership degree is smaller, the larger its inverse is, which means that the maximum difference between the singular vectors in the intrinsic mode function is also larger, that is, the more significant the characteristics of the incidental vibration amplitude sequence in the intrinsic mode function are, and the larger the multi-dimensional entropy value is.

[0095] Step S005, determining the non-isomorphic jump coefficients of the intrinsic mode function at different step sizes based on the differences in decomposition state characteristics under all different scale groups; and determining the multi-scale non-isomorphic coefficients of the intrinsic mode function based on the non-isomorphic jump coefficients under all step sizes.

[0096] In one embodiment of the present application, the difference in the decomposition state characteristics at different scales can reflect the information complexity or uncertainty of the data at different spatial scales.

[0097] Based on this, this embodiment will scale Growth ,in, Without exceeding the value range of the scale, steps S003 and S004 are repeated to obtain the decomposition state feature difference at each scale within the scale range.

[0098] Preferably, this embodiment combines the differences in decomposition state characteristics at different scales to further construct non-isomorphic jump coefficients of the intrinsic mode function at different step sizes:

[0099]

[0100] In the formula, represents the non-isomorphic jump coefficient of the intrinsic mode function under step size d, , Respectively expressed in scale , The decomposition state characteristic difference of the intrinsic mode function under the present embodiment is set to a positive integer from 1 to 5, and the specific value can be set according to the actual situation.

[0101] It should be noted that by calculating the difference in decomposition state characteristics between each scale and its scale under different step-size jumps, the difference in decomposition state characteristics of the overall intrinsic mode function at different scales is analyzed. The larger the value, the greater the non-isomorphism between multiple scales under this step-size, which means that the characteristics in the intrinsic mode function under this step-size jump are more chaotic, lack the regularity of normal vibration or failure, are more of a false component, and carry less valuable information.

[0102] Furthermore, in this embodiment, the non-homomorphic jump coefficients under all step sizes are arranged into a sequence according to the step size from small to large, and the entropy value of the first-order difference sequence of the sequence is calculated as the multi-scale non-homomorphic coefficient of the intrinsic mode function.

[0103] Furthermore, the above operation is applied to all intrinsic mode functions to obtain corresponding multi-scale non-homomorphic coefficients.

[0104] Step S006, screening the intrinsic mode function based on the multi-scale non-homomorphic coefficients, and analyzing the vibration fault diagnosis result of the hydraulic system with the screened intrinsic mode function.

[0105] Preferably, in this embodiment, the five intrinsic mode functions with the largest multi-scale non-isomorphic coefficients are discarded, and the remaining four intrinsic mode functions are used as vibration amplitude sequences to be further analyzed. In other embodiments of the present application, the number of the discarded intrinsic mode functions can be set according to actual conditions.

[0106] The intrinsic mode functions remaining after discarding the historically collected vibration amplitude sequences in all directions are used as the training set of the neural network, and they are manually labeled with two labels, "fault" and "non-fault", to obtain the fault diagnosis training model. Among them, the optimization function uses SGD and the loss function uses cross entropy.

[0107] In one embodiment of the present application, the neural network adopts a CNN neural network, and the implementer can choose other suitable neural networks for training according to actual conditions.

[0108] Furthermore, the vibration amplitude sequence to be diagnosed is input into the fault diagnosis training model and a label is output.

[0109] At this point, accurate vibration fault diagnosis results of the hydraulic system can be obtained. The indicator construction flow chart of the vibration fault diagnosis results is as follows: Figure 2 shown.

[0110] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0111] It should be noted that, unless otherwise specified and limited, terms such as "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a circuit structure, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such article or device. In the absence of further restrictions, an element defined by the sentence "including one..." does not exclude the existence of other identical elements in the article or device including the element. In addition, the term "and\or" used herein includes any and all combinations of one or more related listed items.

[0112] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention invented herein. The present application is intended to cover any variations, uses or adaptations of the present application, which follow the general principles of the present application and include common knowledge or customary technical means in the art not invented by the present application.

[0113] It should be understood that the present application is not limited to the exact construction that has been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof.

Claims

1. A method for vibration analysis and diagnosis of hydraulic system faults. It is characterized in that The method comprises the following steps: Collect the time-series pressure sequence of the hydraulic system and the vibration amplitude sequence in all directions; The vibration amplitude sequence is subjected to modal decomposition to obtain the intrinsic mode function of each decomposition layer; the three-dimensional vibration matrix is ​​determined based on the element values ​​at the corresponding positions of all decomposition layers; the three-dimensional vibration matrix is ​​grouped at different scales to obtain each batch sample matrix respectively; Constructing a singular matrix based on the batch sample matrix; determining a difference vector of any singular vector based on the distance between any two singular vectors in the singular matrix; determining a degree of membership between any singular vector and its difference vector based on the distance; Based on the difference in pressure data between any singular vector and its difference vector in the corresponding time period in the time series pressure sequence, the suppression coefficient ratio is determined; based on the membership degree and the suppression coefficient ratio, the difference in decomposition state characteristics of the intrinsic mode function is determined; Determine the non-isomorphic jump coefficients of the intrinsic mode function at different step sizes based on the differences in decomposition state characteristics under all different scale groups; determine the multi-scale non-isomorphic coefficients of the intrinsic mode function based on the non-isomorphic jump coefficients under all step sizes; Based on the multi-scale non-isomorphic coefficients, the intrinsic mode functions are screened and the vibration fault diagnosis results of the hydraulic system are analyzed by the screened intrinsic mode functions. The expression of the suppression coefficient ratio is: in, represents the suppression coefficient ratio between the i-th singular vector and the j-th singular vector, , They represent the average pressure values ​​in the time periods corresponding to the i-th singular vector and the j-th singular vector, respectively. , They respectively represent the maximum pressure data and the minimum pressure data in the time series pressure sequence.

2. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for collecting the time series pressure sequence and the vibration amplitude sequence in each direction includes: A pressure sensor is installed at the oil outlet of the hydraulic pump, and the pressure data collected during the collection period are combined into a time-series pressure sequence; Acceleration sensors are installed at several directions of the housing of the hydraulic pump, and the vibration data collected during the collection period are respectively combined into vibration amplitude sequences in each direction.

3. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for obtaining each batch sample matrix includes: For the eigenmode functions of each decomposition layer, the elements of the eigenmode functions at the same position in all directions are arranged vertically to form three-dimensional vibration vectors at corresponding positions; The three-dimensional vibration vectors at all positions in the eigenmode function are combined into a three-dimensional vibration matrix; The three-dimensional vibration matrix is ​​scaled according to a preset scale, and each three-dimensional vibration vector and subsequent three-dimensional vibration vectors of a preset scale are combined into a batch sample matrix; All batch sample matrices in the three-dimensional vibration matrix are combined into a three-dimensional vibration sub-matrix.

4. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 3, It is characterized in that The method for determining the difference vector of any singular vector comprises: Perform singular value decomposition on the batch sample matrix, and form singular vectors from the decomposed singular values; The singular vectors of all batch sample matrices in the three-dimensional vibration submatrix are respectively used as column vectors to form a singular matrix; Calculate the sum of the absolute values ​​of the differences between all corresponding elements of any two singular vectors in the singular matrix as the distance between any two singular vectors; For any singular vector, the singular vector with the largest distance between it and the singular vector is taken as the difference vector of any singular vector.

5. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 4, It is characterized in that The method for determining the similarity includes: Calculate the result of raising the distance to a power of a preset number; Calculate the inverse of the ratio of the result to a preset similarity tolerance factor as an exponent of an exponential function with a natural constant as a base; The calculation result of the exponential function is used as the similarity.

6. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for determining the suppression coefficient ratio includes: Calculate the mean value of the pressure data of any singular vector and its difference vector in the corresponding time period in the time series pressure sequence; The normalized value of the absolute value of the difference between the two pressure data means is calculated as the suppression coefficient ratio.

7. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for determining the difference in decomposition state characteristics comprises: The average of the ratios of the suppression coefficients to the membership degrees between all singular vectors in the eigenmode function and their difference vectors is taken as the characteristic difference of the decomposition state of the eigenmode function.

8. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for determining the non-homomorphic jump coefficient comprises: Set the preset step size; Calculate the absolute value of the difference between each scale and the decomposition state feature difference at the scale after the preset step length, and record it as the first absolute value of the difference; The sum of the absolute values ​​of the first differences of all scales is taken as the non-isomorphic jump coefficient of the intrinsic mode function at a preset step size.

9. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The method for determining the multi-scale non-homomorphic coefficients includes: The non-homomorphic jump coefficients of all step sizes are arranged into a sequence according to the step size from small to large, and the entropy value of the first-order difference sequence of the sequence is calculated as the multi-scale non-homomorphic coefficient.

10. A hydraulic system fault vibration analysis and diagnosis method as claimed in claim 1, It is characterized in that The analysis method of the vibration fault diagnosis result of the hydraulic system includes: Discarding a preset number of intrinsic mode functions with the largest multi-scale non-homomorphic coefficients; The remaining intrinsic mode functions after discarding the historically collected vibration amplitude sequences in all directions are used as the training set of the neural network, with two labels of "fault" and "non-fault", to obtain the fault diagnosis training model; The vibration amplitude sequence to be diagnosed is input into the fault diagnosis training model and the label is output.

Images