A method for selecting target modal components from vibration signal decomposition of rotating machinery

Abstract

A method for selecting a target modal component from the decomposition of a rotating machinery vibration signal, comprising: S1: collecting vibration signals on rotating machinery equipment; S2: decomposing the vibration signal to obtain multiple modal components; S3: calculating the square envelope spectrum of each modal component; S4: normalizing each square envelope spectrum; S5: calculating the energy concentration index of the fault characteristic frequency component of each modal component; S6: calculating the position accuracy index of the fault characteristic frequency of each modal component; S7: calculating the energy position index corresponding to each modal component; S8: finding the sequence number where the minimum value of the energy position index is located, and obtaining the target modal component. The present invention can accurately select the target modal component containing sufficient and complete fault characteristic information from the multiple modal components obtained by decomposing the rotating machinery vibration signal, so as to realize the fault diagnosis of the rotating machinery.

Description

Technical Field

[0001] The invention belongs to the technical field of mechanical fault diagnosis, and in particular relates to a method for selecting target modal components from the decomposition of a rotating mechanical vibration signal. Background Art

[0002] Rotating machinery and equipment usually contain rotating parts such as bearings and gears, which have a vital impact on the normal operation of rotating machinery. However, in practice, these rotating parts often have to operate under high speed, heavy load, variable speed or alternating load conditions, which are prone to various types of fault damage, affecting the use efficiency of rotating machinery and equipment. Therefore, it is necessary to adopt effective technical methods to monitor the state and diagnose faults of rotating machinery. Among them, the commonly used method is to collect the vibration signals of rotating machinery and equipment, process and analyze the vibration signals, so as to determine the health status of the rotating parts in the equipment.

[0003] An effective and commonly used method for processing and analyzing rotating machinery vibration signals is signal decomposition, which includes empirical mode decomposition (EMD), local mean decomposition (LMD), integrated empirical mode decomposition (EEMD), empirical wavelet transform (EWT), variational mode decomposition (VMD) and improved decomposition algorithms based on these decomposition algorithms. After decomposition, the rotating machinery vibration signal will obtain multiple modal components, some of which are false components, some are noise components, some are interference components, and some are components that are not completely decomposed and have modal aliasing problems. Therefore, how to accurately select the target modal components containing sufficient and complete fault feature information from these modal components, and then analyze the target modal components, extract fault features, and use them to identify rotating parts that may have fault damage in rotating machinery equipment is a major problem that needs to be solved in the application of these signal decomposition algorithms in the processing and analysis of rotating machinery vibration signals. Summary of the invention

[0004] The present invention aims to overcome the problem in the prior art that it is difficult to select a target modal component containing sufficient fault characteristic information, and proposes a method for selecting a target modal component from the decomposition of a rotating machinery vibration signal.

[0005] A method for selecting a target modal component from a rotating machinery vibration signal decomposition of the present invention comprises the following steps:

[0006] S1: On rotating machinery, s is the sampling frequency, and a vibration signal x(t) with a length of L is collected;

[0007] S2: Decompose the vibration signal x(t) using the signal decomposition algorithm to obtain M modal components u1(t), u2(t), ..., uM (t);

[0008] S3: Calculate each modal component u m The square envelope spectrum SES of (t)(m=1,2,…,M) m (f), where m represents the number of the modal component and f represents the frequency;

[0009] S4: For each square envelope spectrum SES m (f) Perform normalization processing to obtain the normalized square envelope spectrum NS m (f) The specific method is:

[0010]

[0011] Among them, max[SES m (f)] represents the square envelope spectrum SES m (f) the maximum amplitude;

[0012] S5: Calculate each normalized square envelope spectrum NS m (f) Energy concentration index EC(m) of the fault characteristic frequency component;

[0013] S6: Calculate each normalized square envelope spectrum NS m (f) Position accuracy index PA(m) of fault characteristic frequency;

[0014] S7: Calculate each modal component u m (t) corresponds to the energy position index EP(m), the formula is:

[0015]

[0016] Where p is the adjustment coefficient, and 0<p≤1, β(m) is the balance coefficient, and the calculation formula of β(m) is:

[0017]

[0018] S8: Using the relationship curve between the modal component number m and the energy position index EC(m), find the number m=T where the EC(m) has the minimum value, and obtain the target modal component u containing sufficient and complete fault feature information. T (t).

[0019] Preferably, in step S2, the signal decomposition algorithm includes empirical mode decomposition, local mean decomposition, integrated empirical mode decomposition, empirical wavelet transform and variational mode decomposition.

[0020] Further preferably, in step S3, the modal component u mThe square envelope spectrum SES of (t)(m=1,2,…,M) m (f) The calculation includes the following steps:

[0021] S3-1: Calculate u m The square envelope signal SE of (t) m (t), the formula is:

[0022]

[0023] Where j is the imaginary unit, represents Hilbert transform, |·| represents the modulus of a complex number;

[0024] S3-2: Calculate SE m The squared envelope spectrum SES of (t) m (f), the formula is:

[0025]

[0026] in, represents Fourier transform.

[0027] Further preferably, in step S5, each normalized square envelope spectrum NS is calculated m The energy concentration index EC(m) of the fault characteristic frequency component in (f)(m=1,2,…,M) includes the following steps:

[0028] S4-1: Normalized square envelope spectrum NS m The amplitudes of (f) are rearranged from large to small to obtain the rearranged amplitude sequence SNS m (n)(n=1,2,…,N), where N represents the number of frequency points, N=L / 2;

[0029] S4-2: Calculating SNS m The energy concentration index EC(m) of the fault characteristic frequency is obtained by taking the average value of the difference between the first amplitude and the subsequent K amplitudes in (n), and the formula is:

[0030]

[0031] Further preferably, in step S6, each normalized square envelope spectrum NS is calculated m The position accuracy index PA(m) of the fault characteristic frequency in (f)(m=1,2,…,M) includes the following steps:

[0032] S5-1: Find the normalized square envelope spectrum NS m The frequency value f corresponding to the maximum amplitude in (f) ma (m);

[0033] S5-2: Calculate the position accuracy index PA (m), the formula is:

[0034]

[0035] Where C is the number of rotating parts in the rotating machinery, f i (i=1,2,…,C) is the theoretical fault characteristic frequency of each rotating component.

[0036] The positive effect achieved by the present invention is that: the present invention selects the target modal component by minimizing the proposed energy position index, which, on the one hand, maximizes the energy of the fault characteristic frequency component in the target modal component, ensuring the sufficiency and completeness of the fault characteristic information in the target modal component; on the other hand, it makes the overlap between the fault characteristic frequency in the target modal component and the theoretical fault characteristic frequency of a rotating component in the rotating machinery the highest, so that the rotating component with fault damage can be accurately judged based on this, ensuring the validity and usefulness of the target modal component; therefore, the present invention can accurately select the target modal component containing sufficient and complete fault characteristic information from the multiple modal components obtained by decomposing the mechanical vibration signal, for realizing the fault diagnosis of the rotating machinery. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a flow chart for the implementation of the present invention;

[0038] Figure 2 A time domain waveform diagram of a rolling bearing vibration signal in an embodiment of the present invention;

[0039] Figure 3 Graph 1 is a square envelope spectrum of a rolling bearing vibration signal in an embodiment of the present invention.

[0040] Figure 4 is a relationship curve between the sequence number m of the modal component and the energy position index EC(m) in an embodiment of the present invention.

[0041] Figure 5 A time domain waveform diagram of a selected target modal component in an embodiment of the present invention;

[0042] Figure 6 : is the square envelope spectrum of the selected target modal component in the embodiment of the present invention. DETAILED DESCRIPTION

[0043] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0044] Reference Figure 1 , a method for selecting a target modal component from a rotating machinery vibration signal decomposition, comprising the following steps:

[0045] S1: On rotating machinery, s is the sampling frequency, and a vibration signal x(t) with a length of L is collected;

[0046] S2: Decompose the vibration signal x(t) using the signal decomposition algorithm to obtain M modal components u1(t), u2(t), ..., u M (t);

[0047] S3: Calculate the modal component u m The square envelope spectrum SES of (t)(m=1,2,…,M) m (f), where m represents the number of the modal component and f represents the frequency;

[0048] S4: For each square envelope spectrum SES m (f) Perform normalization processing to obtain the normalized square envelope spectrum NS m (f) The specific method is:

[0049]

[0050] Among them, max[SES m (f)] represents the square envelope spectrum SES m (f) the maximum amplitude;

[0051] S5: Calculate each normalized square envelope spectrum NS m (f) Energy concentration index EC(m) of the fault characteristic frequency component;

[0052] S6: Calculate each normalized square envelope spectrum NS m (f) Position accuracy index PA(m) of fault characteristic frequency;

[0053] S7: Calculate each modal component u m (t) corresponds to the energy position index EP(m), the formula is:

[0054]

[0055] Where p is the adjustment coefficient, and 0<p≤1, β(m) is the balance coefficient, and the calculation formula of β(m) is:

[0056]

[0057] S8: Using the relationship curve between the modal component number m and the energy position index EC(m), find the number m=T where the EC(m) has the minimum value, and obtain the target modal component u containing sufficient and complete fault feature information. T (t).

[0058] In step S2, the signal decomposition algorithms include empirical mode decomposition, local mean decomposition, integrated empirical mode decomposition, empirical wavelet transform, variational mode decomposition, and improved decomposition algorithms based on these decomposition algorithms.

[0059] In step S3, the modal component u m The square envelope spectrum SES of (t)(m=1,2,…,M) m (f) The calculation includes the following steps:

[0060] S3-1: Calculate u m The square envelope signal SE of (t) m (t), the formula is:

[0061]

[0062] Where j is the imaginary unit, represents Hilbert transform, |·| represents the modulus of a complex number;

[0063] S3-2: Calculate SE m The squared envelope spectrum SES of (t) m (f), the formula is:

[0064]

[0065] in, represents Fourier transform.

[0066] In step S5, each normalized square envelope spectrum NS is calculated. m The energy concentration index EC(m) of the fault characteristic frequency component in (f)(m=1,2,…,M) includes the following steps:

[0067] S4-1: Normalized square envelope spectrum NS m The amplitudes of (f) are rearranged from large to small to obtain the rearranged amplitude sequence SNS m (n)(n=1,2,…,N), where N represents the number of frequency points, N=L / 2;

[0068] S4-2: Calculating SNS m The energy concentration index EC(m) of the fault characteristic frequency is obtained by taking the average value of the difference between the first amplitude and the subsequent K amplitudes in (n), and the formula is:

[0069]

[0070] In step S6, each normalized square envelope spectrum NS is calculated. mThe position accuracy index PA(m) of the fault characteristic frequency in (f)(m=1,2,…,M) includes the following steps:

[0071] S5-1: Find the normalized square envelope spectrum NS m The frequency value f corresponding to the maximum amplitude in (f) ma (m);

[0072] S5-2: Calculate the position accuracy index PA (m), the formula is:

[0073]

[0074] Where C is the number of rotating parts in the rotating machinery, f i (i=1,2,…,C) is the theoretical fault characteristic frequency of each rotating component.

[0075] The above invention is applied to the vibration signal processing of rolling bearings. The bearing data comes from the Bearing Data Center of Western Reserve University. The data used is the vibration data of the outer ring of the rolling bearing. The specific fault is a single point damage with a diameter of 0.021 inches and a depth of 0.011 inches at the 6 o'clock position of the outer ring. The load applied by the loading motor is 3HP. The output speed of the drive motor measured by the encoder is 1721r / min, and the corresponding rotation frequency is f r =1721 / 60=28.6833Hz, so the theoretical fault characteristic frequency of the inner ring of the rolling bearing is calculated to be f ir =5.4152·f r =155.3260Hz, the theoretical fault characteristic frequency of the outer ring is f or =3.5848·f r =102.8240Hz, the theoretical fault characteristic frequency of the rolling element is f ba =4.7135·f r =135.1989 Hz. The sampling frequency of the rolling bearing vibration data is 12000 Hz. The steps of selecting the target mode using the present invention are as follows.

[0076] S1: Select a vibration signal x(t) with a length of L=4000 from the data set corresponding to the rolling bearing. Its time domain waveform is as follows: Figure 2 As shown, its square envelope spectrum is as follows Figure 3 As shown, it can be seen that there are many large-value interference components, and the characteristic frequency component of the bearing outer ring fault indicated by f=102Hz is not prominent, so it is difficult to judge whether the rolling bearing has a fault damage;

[0077] S2: The vibration signal x(t) is decomposed using the variational modal decomposition algorithm, where the modal decomposition layer number K is set to 13 and the balance parameter α is set to 6650, so 13 modal components u1(t), u2(t), …, u 13 (t);

[0078] S3: Calculate each modal component u m Square envelope spectrum SES of (t)(m=1,2,…,13) m (f), where m represents the number of the modal component and f represents the frequency;

[0079] S4: For each square envelope spectrum SES m (f) Perform normalization processing to obtain the normalized square envelope spectrum NS m (f) The specific method is:

[0080]

[0081] Among them, max[SES m (f)] represents the square envelope spectrum SES m (f) the maximum amplitude;

[0082] S5: Calculate each normalized square envelope spectrum NS m The energy concentration index EC(m) of the fault characteristic frequency component in (f) is as follows:

[0083] S5-1: Normalized square envelope spectrum NS m The amplitudes of (f) are rearranged from large to small to obtain the rearranged amplitude sequence SNS m (n)(n=1,2,…,2000);

[0084] S5-2: Calculating SNS m The energy concentration index EC(m) of the fault characteristic frequency is obtained by taking the average value of the difference between the first amplitude and the following 20 amplitudes in (n), and the formula is:

[0085]

[0086] S6: Calculate each normalized square envelope spectrum NS m The position accuracy index PA(m) of the fault characteristic frequency in (f) is as follows:

[0087] S6-1: Find the normalized square envelope spectrum NS m The frequency value f corresponding to the maximum amplitude in (f) ma (m);

[0088] S6-2: Calculate the position accuracy index PA(m), considering the three rotating parts of the rolling bearing, the inner ring, the outer ring and the rolling element, then the calculation formula of PA(m) is:

[0089] PA(m)=(|f ma (m)-f ir |·|f ma (m)-f or |·|f ma (m)-f ba |) 1 / 3

[0090] Among them, f ir 、f or and f ba They are respectively the theoretical fault characteristic frequencies of the inner ring, outer ring and rolling element of the rolling bearing calculated above.

[0091] S7: Calculate each modal component u m (t) corresponds to the energy position index EP(m), the formula is:

[0092]

[0093] Where p = 1, β(m) is the balance coefficient, and the calculation formula of β(m) is:

[0094]

[0095] S8: Using the relationship curve between the modal component number m and the energy position index EC(m), such as Figure 4 As shown, the sequence number m=5 where the minimum value of EP(m) is located, the target modal component containing sufficient and complete fault characteristic information is the fifth modal component u5(t), and its time domain waveform is as follows Figure 5 As shown, the square envelope spectrum is Figure 6 As shown, it can be seen that the amplitude of the interference component has been suppressed to a large extent, and the extracted characteristic frequency f=102Hz of the rolling bearing outer ring fault is the most significant, indicating that the outer ring of the rolling bearing has indeed suffered from fault damage, which is consistent with the actual situation.

[0096] The contents described in the embodiments of this specification are merely an enumeration of the implementation forms of the inventive concept. The protection scope of the present invention should not be regarded as limited to the specific forms described in the embodiments. The protection scope of the present invention also extends to equivalent technical means that can be conceived by those skilled in the art based on the inventive concept.

Claims

1. A method for selecting a target modal component from a rotating machinery vibration signal decomposition, comprising the following steps: S1: On rotating machinery, s is the sampling frequency, and a vibration signal x(t) with a length of L is collected; S2: Decompose the vibration signal x(t) using the signal decomposition algorithm to obtain M modal components u1(t), u2(t), ..., u M (t); S3: Calculate each modal component u m The squared envelope spectrum SES of (t) m (f), where m represents the number of the modal component, m = 1, 2, ..., M, and f represents the frequency; S4: For each square envelope spectrum SES m (f) Perform normalization processing to obtain the normalized square envelope spectrum NS m (f) The specific method is: Among them, max[SES m (f)] represents the square envelope spectrum SES m (f) the maximum amplitude; S5: Calculate each normalized square envelope spectrum NS m (f) Energy concentration index EC(m) of the fault characteristic frequency component; S6: Calculate the normalized square envelope spectrum NS m (f) Position accuracy index PA(m) of fault characteristic frequency; S7: Calculate each modal component u m (t) corresponds to the energy position index EP(m), the formula is: Where p is the adjustment coefficient, and 0<p≤1, β(m) is the balance coefficient, and the calculation formula of β(m) is: S8: Using the relationship curve between the modal component number m and the energy position index EC(m), find the number m=T where the minimum value of EC(m) is located, and obtain the target modal component u containing sufficient and complete fault feature information. T (t); In step S3, the modal component u m The square envelope spectrum SES of (t) m (f) The calculation includes the following steps: S3-1: Calculate u m The square envelope signal SE of (t) m (t), the formula is: Where j is the imaginary unit, represents Hilbert transform, |·| represents the modulus of a complex number; S3-2: Calculate SE m The squared envelope spectrum SES of (t) m (f), the formula is: in, represents Fourier transform; In step S5, each normalized square envelope spectrum NS is calculated. m The energy concentration index EC(m) of the fault characteristic frequency component in (f) includes the following steps: S4-1: Normalized square envelope spectrum NS m The amplitudes of (f) are rearranged from large to small to obtain the rearranged amplitude sequence SNS m (n), n = 1, 2, ..., N, where N represents the number of frequency points, N = L / 2; S4-2: Calculating SNS m The energy concentration index EC(m) of the fault characteristic frequency is obtained by taking the average value of the difference between the first amplitude and the subsequent K amplitudes in (n), and the formula is: In step S6, each normalized square envelope spectrum NS is calculated. m The position accuracy index PA(m) of the fault characteristic frequency in (f) includes the following steps: S5-1: Find the normalized square envelope spectrum NS m The frequency value f corresponding to the maximum amplitude in (f) ma (m); S5-2: Calculate the position accuracy index PA (m), the formula is: Where C is the number of rotating parts in the rotating machinery, f i is the theoretical fault characteristic frequency of each rotating component, i=1,2,…,C.

2. A method for selecting target modal components from the decomposition of a rotating machinery vibration signal according to claim 1, characterized in that: In step S2, the signal decomposition algorithm includes empirical mode decomposition, local mean decomposition, integrated empirical mode decomposition, empirical wavelet transform and variational mode decomposition.

Images