Strain clamp state diagnosis method based on vibration signals

Abstract

A strain clamp state diagnosis method based on vibration signals comprises the steps that firstly, a vibration exciter is adopted to apply tiny vibration to a strain clamp, and an optical fiber sensor is adopted to collect vibration signals of air around the strain clamp; then, the collected vibration signals are decomposed according to various parameters, and noise signals in the vibration signals are filtered out; calculating the importance weight of the parameters, and selecting the parameters capable of reflecting the damage degree of the strain clamp as input parameters for state diagnosis of the strain clamp according to the importance weight; performing normalization processing on numerical values corresponding to the input parameters, and mapping data to the same dimensionless range; and finally, comparing the obtained data with data in a normal state, judging whether the strain clamp is damaged or not, and obtaining the damage degree of the strain clamp.

Description

Technical Field

[0001] This invention relates to the field of power system technology, and in particular to a method for diagnosing the condition of tension clamps based on vibration signals. Background Technology

[0002] Tension clamps, as crucial hardware for anchoring conductors in power distribution networks, directly impact the safety and stability of the distribution system. With increasing service life and the influence of external environmental factors, tension clamps are prone to fatigue fracture, leading to conductor breaks, widespread power outages, and serious threats to the normal operation of the power system. Therefore, regular condition monitoring of tension clamps is of paramount importance. However, in a vast power distribution network, the sheer number of tension clamps makes manual inspection not only time-consuming and labor-intensive but also ineffective in timely and accurate fault detection. Current methods for detecting equipment fault conditions using vibration signals largely rely on data-driven intelligent algorithms for vibration signal analysis and processing. However, these methods are complex in mechanism, have poor interpretability of the data processing, and require repeated training with large amounts of historical data to establish accurate data analysis and processing models, resulting in significant time and labor costs. Therefore, using vibration signal analysis and processing technology can significantly improve the inspection efficiency of tension clamps and reduce data analysis costs, providing strong support for the intelligent operation and maintenance of power distribution networks. Summary of the Invention

[0003] The purpose of this invention is to solve the technical problems of low inspection efficiency and high data analysis cost in the fault diagnosis of tension clamps in the existing technology, and thus proposes a tension clamp condition diagnosis method based on vibration signals.

[0004] The technical solution adopted in this invention is as follows: A method for diagnosing the condition of tension clamps based on vibration signals includes the following steps: Step 1: Apply minute vibrations to the tension clamp using a vibrator, and collect the time-domain vibration signal of the air surrounding the tension clamp using an optical fiber sensor; Step 2: Decompose the collected vibration signals according to multiple parameters and filter out noise signals in the vibration signals to retain effective vibration signal parameters; Step 3: For the effective vibration signal parameters, calculate their importance weight for the impact on air vibration, and then select the effective vibration signal parameters that can reflect the degree of damage to the tension clamp as the input parameters for tension clamp condition diagnosis based on the importance weight. Step 4: Normalize the values ​​corresponding to the input parameters to eliminate the influence of different units; Step 5: Calculate the deviation between each input parameter data and the normal state data to obtain the total deviation of the tension clamp state. Then, determine whether the tension clamp is damaged and the degree of damage by measuring the magnitude of the total deviation.

[0005] The parameters in step two include frequency, power spectral density, wavelength, amplitude, duration, phase, sound pressure level, and sound velocity.

[0006] In step two, the decomposition process specifically includes the following steps: Step 2-1) Decompose the detected time-domain vibration signal according to frequency type; Step 2-2) Convert the decomposed time-domain vibration signal into a frequency-domain vibration signal and filter out the high-frequency noise components.

[0007] In step 2-1), the detected time-domain vibration signal is decomposed using a series of complex exponential functions with different frequencies, amplitudes, and phases, according to frequency type; the length of the time-domain characteristic signal is denoted as... N , No. i The time-domain characteristic signals of each parameter are T [ i ],exist w i Vibration signals under frequency type are P [ w i The transformation formula is: .

[0008] In step 2-2), a mother wavelet function and the vibration signal are used to perform an inner product operation to extract the time-frequency features of the vibration signal. Then, the decomposed time-domain vibration signal is converted into a frequency-domain vibration signal through scaling and displacement transformation, and high-frequency noise components are filtered out. The transformed value is denoted as... W ( α , β ), α Indicates the scale of the transformation. β Indicates the magnitude of the displacement under transformation. Let be the complex conjugate of the mother wavelet function. t If the frequency domain vibration signal represents time, then... P [ w i ( t The transformation of )] is: ; The regulations stipulate that when W ( α , β ()∈[10kHz,∞) represents high frequency. This rule filters out noise components and retains effective vibration signal parameters.

[0009] The frequency types mentioned above include sine wave signals, cosine wave signals, and fundamental frequency signals, and all of them are time-related.

[0010] In step three, the importance weights are calculated using the analytic hierarchy process (AHP). Let the weight obtained using the AHP be the [number of weights]. k The importance weight of each parameter is: The first value obtained by using the entropy method k The importance weight of each parameter is: The first one obtained by using the grey relational method k The importance weight of each parameter is: Then the final number k The importance weight of each parameter for: .

[0011] In step four, record the selected first... m The input parameters contain q The input data is r m,1 , r m,2 ,……, r m,q The normalization process used is as follows: .

[0012] in, For the first m The first input parameter contained in o One input data, For input data at the 1st m The sorting of the input parameters For the first m The first input parameter contained in o The result after normalizing the input data.

[0013] Step five specifically includes the following steps: Step 5-1) Extract the data of the tension clamp under normal conditions according to the selected input parameters; Step 5-2) Record the first m The corresponding input parameters q The data for a tension clamp under normal conditions is as follows: r ’* m,1 , r ’* m,2 ,……, r ’* m,q Then the normalization process is as follows: ; Step 5-3) Calculate the first m The deviation between the input parameter data and the normal data D m for: ; in, For the first m The corresponding input parameter in the nth input parameter i Data for a tension clamp under normal conditions i For the data of the tension clamp under normal conditions, in the first... m Sort among the input parameters.

[0014] Step 5-4) Calculation a Average deviation of each input parameter D ,in a =3: ; Step 5-5) Based on the average deviation D Determine if the tension clamp has been damaged.

[0015] In step five, if D If ∈[0,0.2), then the tension clamp is determined to be undamaged; if D If ∈[0.2,0.4), then the tension clamp is determined to be slightly damaged; if D If the value is ∈ [0.4, 0.6), then the tension clamp is determined to be moderately damaged; if D a If the value is ∈ [0.6, 0.8), then the tension clamp is determined to be severely damaged; if D If the value is ∈[0.8,0.1), then the tension clamp is determined to have broken.

[0016] Compared with the prior art, the present invention has the following technical effects: 1) This invention integrates weight calculation models based on different mathematical principles to comprehensively calculate the importance weight of each vibration parameter, avoiding the subjectivity or data bias that may be caused by a single method; 2) By combining signal decomposition methods and a quantitative evaluation model based on the average deviation, this invention constructs the diagnostic process into an axiomatic quantitative procedure that can be explained by mathematical principles, which significantly improves the objectivity and interpretability of tension clamp status judgment. Attached Figure Description

[0017] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is an overall flowchart of the present invention; Figure 2 A schematic diagram illustrating the steps to determine whether a tension clamp is damaged and to obtain the extent of the damage. Detailed Implementation

[0018] like Figure 1 As shown, a method for diagnosing the condition of tension clamps based on vibration signals specifically includes the following steps: Step 1: Apply minute vibrations to the tension clamp using a vibrator, and collect the time-domain vibration signal of the air surrounding the tension clamp using an optical fiber sensor; Step 2: The collected vibration signals are decomposed and processed using waveform transformation according to parameters such as frequency, power spectral density, wavelength, amplitude, duration, phase, sound pressure level, and sound velocity to filter out noise signals in the vibration signals; First, the detected time-domain vibration signal is decomposed using a series of complex exponential functions with different frequencies, amplitudes, and phases, and then according to frequency type. The length of the time-domain characteristic signal is denoted as... N , No. i The time-domain characteristic signals of each parameter are T [ i ],exist w i Vibration signals under frequency type are P [ w i The transformation formula is: ; The decomposed time-domain vibration signal is then converted into a frequency-domain vibration signal, and high-frequency noise components are filtered out. The transformed value is denoted as... W ( α , β ), α Indicates the scale of the transformation. β Indicates the magnitude of the displacement under transformation. Let be the complex conjugate of the mother wavelet function. t If the frequency domain vibration signal represents time, then... P [ w i ( t The transformation of )] is: ; The regulations stipulate that when W ( α , β ()∈[10kHz,∞) represents high frequency, and noise components are filtered out by this rule.

[0019] Step 3: For effective vibration signal parameters, calculate their importance weight for the impact on air vibration, and then select parameters that can reflect the degree of damage to the tension clamp as input parameters for tension clamp condition diagnosis based on the importance weight.

[0020] The importance weights of each parameter were calculated using the analytic hierarchy process (AHP), entropy method, and grey relational analysis. Let the importance weight obtained using the AHP be the first... k The importance weight of each parameter is: z kc The first value obtained by using the entropy method k The importance weight of each parameter is: z ks The first one obtained by using the grey relational method k The importance weight of each parameter is: z kh Then the final number k The importance weight of each parameter z k for: ; The parameters are ranked from highest to lowest importance, and the top three parameters are considered to reflect the degree of damage to the tension clamp, and are used as input parameters for the condition diagnosis of the tension clamp.

[0021] Step 4: Normalize the values ​​corresponding to the input parameters to eliminate the influence of different units. A normalization method is used. Let the selected first... m The input parameters contain q The input data is r m,1 , r m,2 ,……, r m,q ( m =1,2,3). Among them... m =1 represents the input parameter 1. m =2 represents the input parameter 2. m =3 represents the input parameter 3. The normalization process is as follows: ; in, For the first m The first input parameter contained in o One input data, For input data at the 1st m The sorting of the input parameters For the first m The first input parameter contained in o The result after normalizing the input data.

[0022] Step 5: Calculate the deviation between each input parameter data and the normal state data to obtain the total deviation of the tension clamp's state. Based on the magnitude of the total deviation, determine whether the tension clamp is damaged and the extent of the damage. Specific steps are attached. Figure 2 A schematic diagram illustrates the steps for determining whether a tension clamp is damaged and obtaining the extent of its damage: Step 5-1) Extract the data of the tension clamp under normal conditions according to the selected input parameters.

[0023] Step 5-2) Record the first m The corresponding input parameters q The data for a tension clamp under normal conditions is as follows: r * m,1 , r * m,2 ,……, r * m,q The normalization process is then as follows: ; Step 5-3) Calculate the first m The deviation between the input parameter data and the normal data D m for: ; in, For the first m The corresponding input parameter in the nth input parameter i Data for a tension clamp under normal conditions i For the data of the tension clamp under normal conditions, in the first... m Sort among the input parameters.

[0024] Step 5-4) Calculation a Average deviation of each input parameter D ,in a =3: ; Step 5: Based on the average deviation D Determine if the tension clamp is damaged, and assess the extent of the damage. If D If ∈[0,0.2), then the tension clamp is determined to be undamaged; if D If ∈[0.2,0.4), then the tension clamp is determined to be slightly damaged; if D If the value is ∈ [0.4, 0.6), then the tension clamp is determined to be moderately damaged; if D If the value is ∈ [0.6, 0.8), then the tension clamp is determined to be severely damaged; if DIf the value is ∈[0.8,0.1), then the tension clamp is determined to have broken.

[0025] Example: Step 1: Use an exciter to generate minute vibrations. Assume that the fiber optic sensor collects a signal of a certain length. N =8 time-domain signal T [ i The acquired original time-domain signal sequence is as follows: T [ i =[0.5,1.2,0.8,-0.3,-1.0,-0.7,0.2,0.9] i =0,1,2,3,4,5,6,7), sampling frequency F It is 50kHz.

[0026] Step 2: Convert the time-domain signal T [ i Transform to the frequency domain; take the frequency sequence as [ w Given [0, 6.25, 12.50, 18.75, 25.00, 31.25, 37.50, 43.75], we can obtain values ​​at different frequencies. w i The following components P [ w i ]=[1.600+j0.000, 3.692−j2.094, 0.900−j0.700, 1.006−j1.168, −0.600+j0.000, 1.006+j1.168, 0.900+j0.700, 3.692+j2.094].

[0027] Choosing the Morlet wavelet as the mother wavelet, then ( w 0=6); to cover the entire signal length, β Select 0, 1, 2, 3, 4, 5, 6, 7; Set α Critical Scale of Values α c =( ω 0⋅ F ) / (2 π ⋅ fc )≈4.77, therefore it is acceptable. α =6.0; the noise-removed signal obtained by solving is P new [ w i ]=[1.75+ j 0.08, 1.25− j 0.75, 0.85−j 0.25, 0.45− j 0.12, 0.60+ j 0.02, 0.45+ j 0.12, 0.85+ j 0.25, 1.25+ j The corresponding amplitudes are [1.75, 1.46, 0.88, 0.46, 0.60, 0.46, 0.88, 1.46]. Therefore, the maximum amplitude of the DC component occurs at... i =0, the maximum amplitude of the AC component occurs at... i =1 and i =7, therefore the main frequency range is set to 6.25kHz, which is taken as the frequency. The speed of sound is 343m / s, the wavelength is 0.0549m, and the amplitude is... A 1 is 1.05, and the duration is N / F =0.16ms, phase is -0.540rad; assuming reference sound pressure is A If 2=1, then the sound pressure level is 20log 10 ( A 1 / A 2)≈0.42dB; Using the total power as the power spectral density parameter, the calculated power spectral density is: Σ T i 2 / N≈ 1.48.

[0028] Step 3: To calculate the weights, multiple signal segments are used to construct a dataset containing multiple samples. This example uses 5 samples for calculation, and the data is shown in the table below:

[0029] The weights calculated using the analytic hierarchy process (AHP), entropy method, and grey relational analysis method are as follows: Importance weights z are calculated using the analytic hierarchy process. kc ( k =0,1,2,3,4,5,6,7), the calculation results are as follows:

[0030] Importance weights are calculated using the entropy method. z ks ( k =0,1,2,3,4,5,6,7), the calculation results are as follows:

[0031] The importance weights are calculated using the grey relational analysis method. zkh ( k =0,1,2,3,4,5,6,7), the calculation results are as follows:

[0032] Averaging the importance weights obtained from the above results, the final importance weight for the frequency parameter is... The final weights of the remaining parameters are shown in the table below:

[0033] Based on the final weight calculation results, the top three parameters are: power spectral density, frequency, and sound pressure level. Therefore, power spectral density, frequency, and sound pressure level are used as input parameters for diagnosing the condition of tension clamps.

[0034] Step 4: Normalize the values ​​corresponding to the input parameters, where... m =3, q If the value is 5, the calculation result is as follows:

[0035] The deviation between the input parameter data and the normal data is shown in the following figure:

[0036] Then the average deviation of the input parameter D = 1 / 3 * [0.72 + (-0.72) + 0.72] = 0.24; Step 5: Based on the judgment criteria, D If the value is ∈[0.2,0.4), then the tension clamp is determined to be slightly damaged.

[0037] In summary, this invention proposes a method for diagnosing the condition of tension clamps based on vibration signals. Experimental results show that this method can be applied to the condition judgment of tension clamps, verifying the effectiveness of the condition diagnosis method proposed in this invention.

Claims

1. A vibration signal-based diagnosis method for the state of a strain clamp, characterized by, The method comprises the following steps: Step 1: a small vibration is applied to the strain clamp by using a vibration exciter, and a time-domain vibration signal of air around the strain clamp is collected by using an optical fiber sensor; Step 2: the collected vibration signal is decomposed according to multiple parameters, noise signals in the vibration signal are filtered out, and effective vibration signal parameters are reserved; Step 3: for the effective vibration signal parameters, the importance weight of the influence of the effective vibration signal parameters on air vibration is calculated, so that the effective vibration signal parameters reflecting the damage degree of the strain clamp are selected as input parameters of the strain clamp state diagnosis according to the importance weight; Step 4: the values corresponding to the input parameters are normalized to eliminate the influence of different dimensions; Step 5: the deviation between each input parameter data and normal state data is calculated, so that the total deviation of the strain clamp state is obtained, and whether the strain clamp is damaged and the damage degree of the strain clamp are determined according to the size of the total deviation.

2. The method of claim 1, wherein, The multiple parameters in step 2 include frequency, power spectral density, wavelength, amplitude, duration, phase, sound pressure level and sound velocity.

3. The method according to claim 1 or 2, characterized in that, In step 2, the decomposition processing specifically comprises the following steps: Step 2-1) the detected time-domain vibration signal is decomposed according to the frequency type; Step 2-2) the decomposed time-domain vibration signal is converted into a frequency-domain vibration signal, and high-frequency noise components are filtered out.

4. The method of claim 3, wherein, In step 2-1), the detected time-domain vibration signal is decomposed into a series of complex exponential functions with different frequencies, amplitudes and phases, and according to the frequency type; the length of the time-domain characteristic signal is denoted as N , the time-domain characteristic signal of the i th parameter is denoted as T [ i ], the vibration signal under the frequency type is denoted as w i , and the transformation formula is: P [ w i ] 。 5. The method of claim 3, wherein, In step 2-2), a mother wavelet function and the vibration signal are used to perform an inner product operation to extract the time-frequency characteristics of the vibration signal, then the decomposed time-domain vibration signal is converted into a frequency-domain vibration signal through a scale and displacement transformation, and high-frequency noise components are filtered out; the transformed value obtained after the transformation is denoted as W ( α , β ), α denotes the scale size of the transformation, β denotes the displacement size of the transformation, is a conjugate complex of the mother wavelet function, t denotes the time-frequency-domain vibration signal, then P [ w i ( t )] is transformed as follows: ； It is provided that when W ( α , β )∈[10kHz,∞) is high frequency, through the provision of noise components filter, keep effective vibration signal parameters.

6. The method of claim 4, wherein, The frequency type includes sine wave signal, cosine wave signal and fundamental frequency signal, and is related to time.

7. The method according to one of claims 1 to 6, characterized in that, In step three, the importance weight is calculated by using analytic hierarchy process, and the importance weight of the first k parameter obtained by using analytic hierarchy process is , the importance weight of the first k parameter obtained by using entropy method is , the importance weight of the first k parameter obtained by using grey correlation method is , and the importance weight of the first k parameter is : 。 8. The method of claim 1, wherein, In step four, the input data contained in the selected input parameters are m q r m,1 , r m,2 ,……, r m,q , and the normalization processing used is:​​ ； in, For the first m The first input parameter contained in o One input data, For input data at the 1st m The sorting of the input parameters For the first m The first input parameter contained in o The result after normalizing the input data.

9. The method of claim 1, wherein, In step 5, the following steps are specifically included: Step 5-1) according to the selected input parameters, data under the normal state of the strain clamp is extracted; Step 5-2) record the corresponding data of the input parameter in the normal state of the strain clamp m q r ’* m,1 , r ’* m,2 ,……, r ’* m,q Then the normalization processing is:​​ ； Step 5 - 3) Calculate the degree of deviation between the input data of the input parameter and the normal data m D m is:​ ； wherein, is the first m input parameter, i is the first i data of the strain clamp in the normal state, m is the ranking of the strain clamp data in the normal state in the first Step 5 - 4) Calculate a the average of the input parameter deviation D : ； Step 5 - 5) Average the deviation D Whether the strain clamp is damaged is determined.

10. The method of claim 9, wherein, In step five, if D ∈[0, 0.2), it is determined that the strain clamp is not damaged; if D ∈[0.2, 0.4), it is determined that the strain clamp is slightly damaged; if D ∈[0.4, 0.6), it is determined that the strain clamp is moderately damaged; if D a ∈[0.6, 0.8), it is determined that the strain clamp is severely damaged; if D ∈[0.8, 0.1), it is determined that the strain clamp has been broken.

Images