A photovoltaic module steel frame detection method based on vibration data analysis

Abstract

The present invention relates to the technical field of data processing, and proposes a photovoltaic module steel frame detection method based on vibration data analysis, including: obtaining the original vibration data of the photovoltaic module steel frame; obtaining the spectrum of the original vibration data; obtaining the main frequency possibility of each frequency, and then clustering to obtain several initial main frequencies; performing EMD decomposition on the original vibration data to obtain several component signals; obtaining the frequency complexity of the original vibration data; obtaining the initial main frequency retention; obtaining the final main frequency range; correcting the original vibration data according to the final main frequency range to obtain corrected vibration data; based on the distribution of frequencies in the final main frequency range, and the time series distribution in the corrected vibration data, obtaining steel frame evaluation parameters. The present invention aims to solve the problem that when performing vibration analysis on the photovoltaic module steel frame, the frequency range of the vibration signal is wide, which makes it difficult to accurately obtain the abnormality of the frame.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a photovoltaic module steel frame detection method based on vibration data analysis. Background Art

[0002] Photovoltaic modules are the core components of solar power generation systems, responsible for converting solar energy into electrical energy. The stability and reliability of photovoltaic modules directly affect the performance and life of the entire photovoltaic power generation system; they are usually composed of glass, silicon wafers and steel frames. The steel frame is the supporting structure of the photovoltaic module, which has the function of supporting and protecting the glass and silicon wafers. Therefore, the quality inspection of the steel frame is crucial to the stability and reliability of photovoltaic modules. Vibration analysis of the steel frame can be performed by applying external force or vibration to the steel frame and monitoring and analyzing the vibration response to evaluate the strength and wind resistance of the steel frame.

[0003] In the prior art, when analyzing vibration data, a time-frequency conversion analysis method is usually used. Through Fourier transform, the collected time domain data is converted into the frequency domain space, and then the change of vibration data is analyzed according to the frequency change of different signals. The core idea of ​​the algorithm is that any periodic or non-periodic signal can be decomposed into a superposition of a series of sine and cosine functions, which have different frequencies, amplitudes and phases; however, when performing vibration tests on the steel frame of photovoltaic modules, the frequency of the vibration signal is very high and the frequency range is wide. If it is directly analyzed, the accuracy of the vibration data analysis will decrease due to the high complexity of the signal, which will affect the accuracy of the steel frame detection results. Summary of the invention

[0004] The present invention provides a photovoltaic module steel frame detection method based on vibration data analysis to solve the problem that when performing vibration analysis on the photovoltaic module steel frame, the vibration signal frequency range is wide, which makes it difficult to accurately obtain the frame abnormality. The technical solution adopted is as follows:

[0005] The present invention proposes a photovoltaic module steel frame detection method based on vibration data analysis, the method comprising the following steps:

[0006] Obtain the raw vibration data of the steel frame of the photovoltaic module;

[0007] A spectrum is obtained for the original vibration data, wherein each frequency in the spectrum corresponds to an amplitude; according to the amplitude of each frequency in the spectrum, the possibility of the main frequency of each frequency is obtained, and then a number of initial main frequencies are obtained by clustering;

[0008] Perform EMD decomposition on the original vibration data to obtain several component signals; obtain the frequency complexity of the original vibration data according to the amplitude changes in each component signal; obtain the initial main frequency retention degree by combining the frequency differences of adjacent component signals and the frequencies corresponding to each initial main frequency; adjust the initial main frequency based on the initial main frequency retention degree to obtain the final main frequency range;

[0009] The original vibration data is corrected according to the final main frequency range to obtain the corrected vibration data; based on the distribution of frequencies in the final main frequency range and the time series distribution in the corrected vibration data, the steel frame evaluation parameters are obtained.

[0010] Optionally, the method of obtaining the possibility of the main frequency of each frequency includes:

[0011] For any frequency, the ratio of the amplitude of the frequency to the cumulative value of the amplitudes of all frequencies is recorded as the main frequency possibility of the frequency.

[0012] Optionally, the obtaining of several initial main frequencies includes the following specific methods:

[0013] All frequencies are clustered by density, and the distance metric uses the absolute value of the difference between the main frequency possibilities of the frequencies to obtain several clusters;

[0014] Calculate the mean of the main frequency possibilities of all frequencies in any cluster, and record it as the average main frequency possibility of the cluster; take all the frequencies in the cluster with the largest average main frequency possibility as the initial main frequency.

[0015] Optionally, the EMD decomposition of the original vibration data to obtain a plurality of component signals includes the following specific methods:

[0016] Perform EMD decomposition on the original vibration data to obtain several IMF components, which are recorded as several component signals.

[0017] Optionally, the frequency complexity of the original vibration data is obtained by:

[0018] The variance of the amplitude of each component signal is calculated respectively to obtain the amplitude variance of each component signal, and the product of the accumulated value of the amplitude variance of all component signals and the number of component signals is taken as the frequency complexity of the original vibration data.

[0019] Optionally, the obtaining of the initial main frequency retention degree includes the following specific methods:

[0020]

[0021] in, Indicates the maximum frequency among all initial main frequencies. Indicates the minimum frequency among all initial main frequencies. Indicates the frequency complexity of the original vibration data, represents the number of component signals, Indicates The frequency of the component signal, Indicates The frequency of the component signal, represents the absolute value function, Represents an exponential function with a natural constant as its base.

[0022] Optionally, the obtaining of the final main frequency range includes the following specific methods:

[0023] If the initial main frequency retention degree is greater than the retention threshold, all the initial main frequencies are used to form the final main frequency range;

[0024] If the initial main frequency retention degree is less than or equal to the retention threshold, the initial main frequency is adjusted; starting from the initial main frequency with the smallest frequency among all the initial main frequencies, if the absolute value of the difference between the frequencies of two adjacent initial main frequencies is greater than the removal frequency threshold, the previous initial main frequency is deleted and the latter initial main frequency continues to be judged; if the absolute value of the difference between the frequencies of two adjacent initial main frequencies is less than or equal to the removal frequency threshold, the initial main frequency is not deleted, and so on, until the initial main frequency with the second largest frequency among all the initial main frequencies is judged, and all the initial main frequencies that have not been deleted constitute the final main frequency range.

[0025] Optionally, the obtaining of the corrected vibration data includes the following specific methods:

[0026] According to the sine waves corresponding to all frequencies in the final main frequency range in the frequency domain space and the time domain space, the adjusted vibration data is obtained through inverse Fourier transform, and the adjusted vibration data is fitted by the least squares method, and the obtained fitting curve is recorded as the corrected vibration data.

[0027] Optionally, the obtaining of the steel frame evaluation parameters includes the following specific methods:

[0028] Calculate the variance of the amplitude of all data points in the corrected vibration data, which is recorded as the vibration variance of the corrected vibration data; obtain a number of data points with different amplitudes between the corrected vibration data and the adjusted vibration data, which are recorded as a number of corrected data points;

[0029] The steel frame evaluation parameters are obtained based on the final main frequency range, the vibration variance of the corrected vibration data and the time series distribution of the corrected data points.

[0030] Optionally, the steel frame evaluation parameters are specifically obtained by:

[0031]

[0032] in, represents the vibration variance of the corrected vibration data, Indicates the maximum frequency in the final main frequency range. Indicates the minimum frequency in the final main frequency range. represents the number of corrected data points, Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point; represents the absolute value function, Represents an exponential function with a natural constant as its base.

[0033] The beneficial effects of the present invention are as follows: the present invention obtains sine waves of several frequencies and several component signals by performing Fourier transform and EMD decomposition on the original vibration data respectively, quantifies the possibility of the main frequency through the amplitude of the sine wave and obtains the initial main frequency, quantifies the retention degree of the initial main frequency in combination with the fluctuation change of the component signal, and finally obtains the final main frequency range, thereby avoiding the problem that the wide frequency distribution range of the vibration signal makes it difficult to accurately analyze the changes in vibration data, thereby affecting the accuracy of steel frame quality detection. Among them, by performing Fourier transform on the original vibration data, the possibility of the main frequency is quantified according to the amplitude of each frequency, and then cluster analysis is used to extract several frequencies with a larger possibility of the main frequency and use them as the initial main frequency, so that the frequency with a larger amplitude in the spectrum is used as the initial main frequency, which preliminarily reduces the impact of the wider frequency distribution range in the original vibration data on the quality detection process, and reduces the interference of the vibration response of the frequency with a smaller amplitude on the quality detection; by performing EMD decomposition on the original vibration data, based on the amplitude change of the component signal and the initial main frequency, the frequency complexity is quantified, and then the initial main frequency retention is obtained, wherein the frequency complexity reflects the noise caused by abnormal vibration that may exist in the original vibration data, and the initial main frequency needs to be adjusted to avoid the interference caused by the noise, and then based on the initial main frequency retention, it is judged whether adjustment is needed, and the final main frequency range is obtained; based on the distribution of frequencies in the final main frequency range and the change of amplitude in the corrected vibration data, the detection of the steel frame of the photovoltaic module is completed. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0035] Figure 1 A schematic flow chart of a photovoltaic module steel frame detection method based on vibration data analysis provided by an embodiment of the present invention;

[0036] Figure 2 Schematic diagram of the experimental setup for vibration analysis of the steel frame of a photovoltaic module. DETAILED DESCRIPTION

[0037] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0038] See also Figure 1 , which shows a flow chart of a photovoltaic module steel frame detection method based on vibration data analysis provided by an embodiment of the present invention, the method comprising the following steps:

[0039] Step S001: obtaining original vibration data of the steel frame of the photovoltaic module.

[0040] The purpose of this embodiment is to perform quality inspection on the steel frame of the photovoltaic module through vibration analysis, so it is first necessary to collect the vibration signal of the steel frame of the photovoltaic module and use it as raw vibration data for subsequent processing and analysis.

[0041] Specifically, Figure 2 As shown, in this embodiment, the steel frame of the photovoltaic module is used as a test sample, which is placed on a vibration table through a test fixture, and the vibration of the vibration table is controlled by a signal conditioner, a vibration controller and a power amplifier. Finally, the vibration signal is monitored and acquired through a vibration acceleration sensor, and the obtained vibration signal is recorded as the original vibration data of the steel frame of the photovoltaic module.

[0042] Step S002, obtaining a frequency spectrum for the original vibration data; obtaining the possibility of the main frequency of each frequency according to the amplitude of each frequency in the frequency spectrum, and then clustering to obtain a number of initial main frequencies.

[0043] Preferably, in one embodiment of the present invention, the frequency spectrum is obtained from the original vibration data, and the specific method includes:

[0044] The original vibration data is subjected to Fourier transform and converted into the frequency domain space to obtain the spectrum of the original vibration data; in the obtained spectrum, the horizontal axis is the frequency and the vertical axis is the amplitude, and each frequency corresponds to an amplitude.

[0045] Preferably, in one embodiment of the present invention, the possibility of the main frequency of each frequency is obtained according to the amplitude of each frequency in the spectrum, and then a number of initial main frequencies are obtained by clustering, including the specific method of:

[0046] For any frequency, the ratio of the amplitude of the frequency to the accumulated value of the amplitudes of all frequencies is recorded as the main frequency possibility of the frequency; the main frequency possibility of each frequency is obtained according to the above method.

[0047] Furthermore, all frequencies are density clustered. In this embodiment, the density clustering adopts the DBSCAN clustering algorithm, and the distance measurement adopts the absolute value of the difference between the main frequency possibilities of the frequencies to obtain several clusters; the average of the main frequency possibilities of all frequencies in any cluster is calculated, which is recorded as the average main frequency possibility of the cluster; all frequencies in the cluster with the largest average main frequency possibility are taken as the initial main frequency.

[0048] At this point, by performing Fourier transform on the original vibration data, the possibility of the main frequency is quantified according to the amplitude of each frequency, and then cluster analysis is used to extract several frequencies with a higher possibility of main frequency and use them as the initial main frequency, so that the frequency with a larger amplitude in the spectrum is used as the initial main frequency, which preliminarily reduces the impact of the wider frequency distribution range in the original vibration data on the quality detection process, and reduces the interference of the vibration response of the frequency with a smaller amplitude on the quality detection.

[0049] Step S003, perform EMD decomposition on the original vibration data to obtain several component signals; obtain the frequency complexity of the original vibration data according to the amplitude changes in each component signal; obtain the initial main frequency retention degree by combining the frequency difference of adjacent component signals and the frequency corresponding to each initial main frequency; adjust the initial main frequency based on the initial main frequency retention degree to obtain the final main frequency range.

[0050] Preferably, in one embodiment of the present invention, the original vibration data is subjected to EMD decomposition to obtain a plurality of component signals, including the specific method of:

[0051] Perform EMD decomposition on the original vibration data to obtain several IMF components, which are recorded as several component signals.

[0052] It should be noted that when performing vibration analysis on the steel frame of a photovoltaic module, the frequency of the vibration signal will change, and there will be multiple frequencies corresponding to the original vibration data, resulting in a wider frequency distribution range. However, if there are quality problems with the steel frame, high-frequency noise will be generated, and high-frequency noise can reflect abnormal vibrations of the steel frame. Therefore, it is necessary to quantify the frequency complexity of the component signal, and reflect the complexity of the frequency change based on the change in the amplitude of the component signal, and then preliminarily reflect the possible abnormal vibration of the steel frame through the frequency complexity.

[0053] Preferably, in one embodiment of the present invention, the frequency complexity of the original vibration data is obtained according to the amplitude change in each component signal, and the specific method includes:

[0054] The variance of the amplitude of each component signal is calculated respectively to obtain the amplitude variance of each component signal, and the product of the accumulated value of the amplitude variance of all component signals and the number of component signals is taken as the frequency complexity of the original vibration data.

[0055] It should be noted that the greater the fluctuation of the amplitude change in the component signal, that is, the greater the amplitude variance, the more complex the frequency change contained in the original vibration data, and the more component signals there are, the more frequencies they contain and the greater the frequency complexity.

[0056] Preferably, in one embodiment of the present invention, the frequency difference of adjacent component signals and the frequency corresponding to each initial main frequency are combined to obtain the initial main frequency retention degree, and the specific method includes:

[0057] Get the maximum and minimum frequencies of all initial main frequencies, then the initial main frequency retention degree The calculation method is:

[0058]

[0059] in, Indicates the maximum frequency among all initial main frequencies. Indicates the minimum frequency among all initial main frequencies. Indicates the frequency complexity of the original vibration data, represents the number of component signals, Indicates The frequency of the component signal, Indicates The frequency of the component signal, represents the absolute value function, represents an exponential function with a natural constant as the base. In this embodiment, Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.

[0060] It should be noted that the frequency range of all initial main frequencies is obtained by the difference between the maximum frequency and the minimum frequency. The larger the frequency range and the wider the distribution range, the smaller the initial main frequency retention is, and the more adjustment is needed. At the same time, the frequency difference of adjacent component signals is obtained and the average is calculated, combined with the frequency complexity, to reflect the frequency change characteristics of the component signal. The larger the frequency change characteristics and the more concentrated the corresponding frequency part of the component signal are, the more it is necessary to adjust the range of the initial main frequency and the smaller the initial main frequency retention is.

[0061] Preferably, in one embodiment of the present invention, the initial main frequency is adjusted based on the initial main frequency retention degree to obtain the final main frequency range, and the specific method includes:

[0062] A retention threshold is preset. In this embodiment, the retention threshold is described as 0.7. If the initial main frequency retention degree is greater than the retention threshold, no adjustment is required, and all initial main frequencies constitute the final main frequency range; if the initial main frequency retention degree is less than or equal to the retention threshold, the initial main frequency needs to be adjusted to obtain the final main frequency range.

[0063] Furthermore, starting from the initial main frequency with the smallest frequency among all the initial main frequencies, a frequency removal threshold is preset. In this embodiment, the frequency removal threshold is described as 3. If the absolute value of the difference between the frequencies of two adjacent initial main frequencies is greater than the frequency removal threshold, the previous initial main frequency is deleted and the latter initial main frequency continues to be judged; if the absolute value of the difference between the frequencies of two adjacent initial main frequencies is less than or equal to the frequency removal threshold, the initial main frequency does not need to be deleted, and so on, until the initial main frequency with the second largest frequency among all the initial main frequencies is judged, and all the initial main frequencies that have not been deleted constitute the final main frequency range.

[0064] At this point, by performing EMD decomposition on the original vibration data, the frequency complexity is quantified based on the amplitude changes of the component signals and the initial dominant frequency, and then the initial dominant frequency retention is obtained. The frequency complexity reflects the noise caused by abnormal vibrations that may exist in the original vibration data. In this case, the initial dominant frequency needs to be adjusted to avoid interference caused by noise, and then it is determined whether adjustment is needed based on the initial dominant frequency retention, and the final dominant frequency range is obtained.

[0065] Step S004, correct the original vibration data according to the final main frequency range to obtain corrected vibration data; based on the frequency distribution in the final main frequency range and the time series distribution in the corrected vibration data, obtain the steel frame evaluation parameters to complete the photovoltaic module steel frame detection.

[0066] Preferably, in one embodiment of the present invention, the original vibration data is corrected according to the final main frequency range to obtain the corrected vibration data, and the specific method includes:

[0067] According to the sine waves corresponding to all frequencies in the final main frequency range in the frequency domain space and the time domain space, the adjusted vibration data is obtained through inverse Fourier transform, and the adjusted vibration data is fitted by the least squares method. The obtained fitting curve is recorded as the corrected vibration data, wherein the least squares fitting is a well-known technology and will not be described in detail in this embodiment.

[0068] It should be noted that the adjusted vibration data is obtained by superimposing the sine waves corresponding to the frequencies in the final main frequency range (inverse Fourier transform). At this time, due to the removal of sine waves of some frequencies, the adjusted vibration data does fluctuate, so fitting is required to obtain the corrected vibration data. Subsequently, by analyzing the time points when there are differences between the corrected vibration data and the adjusted vibration data, as well as the fluctuation changes of the corrected vibration data, the quality assessment of the steel frame can be achieved.

[0069] Preferably, in one embodiment of the present invention, based on the distribution of frequencies in the final main frequency range and the time series distribution in the modified vibration data, the steel frame evaluation parameters are obtained, including the specific method of:

[0070] The variance of the amplitude of all data points in the corrected vibration data is calculated and recorded as the vibration variance of the corrected vibration data; a number of data points with different amplitudes in the corrected vibration data and the adjusted vibration data are obtained and recorded as a number of corrected data points, then the steel frame evaluation parameters The calculation method is:

[0071]

[0072] in, represents the vibration variance of the corrected vibration data, Indicates the maximum frequency in the final main frequency range. Indicates the minimum frequency in the final main frequency range. represents the number of corrected data points, Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point; represents the absolute value function, represents an exponential function with a natural constant as the base. In this embodiment, Model to present inverse proportional relationship and normalization processing, As the input of the model, the implementer can set the inverse proportional function and normalization function according to the actual situation.

[0073] It should be noted that the smaller the vibration variance of the corrected vibration data and the smaller the frequency range of the final main frequency range, the more concentrated the frequency distribution, and the fewer abnormal fluctuations in the corrected vibration data, the less abnormal vibration of the steel frame, and the larger the steel frame evaluation parameter; the closer the time difference between adjacent corrected data points, the closer the ratio is to 1, and the more regular the distribution of the corrected data points, the more likely the sine wave corresponding to the removed frequency is to be a part with a smaller amplitude but normal vibration, and the larger the steel frame evaluation parameter is.

[0074] Furthermore, a preset evaluation threshold is used. In this embodiment, the evaluation threshold is described as 0.56. If the steel frame evaluation parameter is greater than or equal to the evaluation threshold, the steel frame of the photovoltaic module is qualified; if it is less than the evaluation threshold, the steel frame of the photovoltaic module is unqualified.

[0075] At this point, this embodiment is completed.

[0076] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A photovoltaic module steel frame detection method based on vibration data analysis, characterized in that: The method comprises the following steps: Obtain the raw vibration data of the steel frame of the photovoltaic module; A spectrum is obtained for the original vibration data, wherein each frequency in the spectrum corresponds to an amplitude; according to the amplitude of each frequency in the spectrum, the possibility of the main frequency of each frequency is obtained, and then a number of initial main frequencies are obtained by clustering; Perform EMD decomposition on the original vibration data to obtain several component signals; obtain the frequency complexity of the original vibration data according to the amplitude changes in each component signal; obtain the initial main frequency retention degree by combining the frequency differences of adjacent component signals and the frequencies corresponding to each initial main frequency; adjust the initial main frequency based on the initial main frequency retention degree to obtain the final main frequency range; The original vibration data is corrected according to the final main frequency range to obtain the corrected vibration data; based on the distribution of frequencies in the final main frequency range and the time series distribution in the corrected vibration data, the steel frame evaluation parameters are obtained; The specific method of obtaining the steel frame evaluation parameters includes: Calculate the variance of the amplitude of all data points in the corrected vibration data, which is recorded as the vibration variance of the corrected vibration data; obtain a number of data points with different amplitudes between the corrected vibration data and the adjusted vibration data, which are recorded as a number of corrected data points; According to the final main frequency range, the vibration variance of the corrected vibration data and the time series distribution of the corrected data points, the steel frame evaluation parameters are obtained; The specific method for obtaining the steel frame evaluation parameters is as follows: in, represents the vibration variance of the corrected vibration data, Indicates the maximum frequency in the final main frequency range. Indicates the minimum frequency in the final main frequency range. represents the number of corrected data points, Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point is Indicates The time corresponding to the corrected data point; represents the absolute value function, represents an exponential function with a natural constant as base; The frequency complexity of the original vibration data is specifically obtained by: The variance of the amplitude of each component signal is calculated respectively to obtain the amplitude variance of each component signal, and the product of the accumulated value of the amplitude variance of all component signals and the number of component signals is taken as the frequency complexity of the original vibration data; The specific method of obtaining the initial main frequency retention degree includes: in, Indicates the maximum frequency among all initial main frequencies. Indicates the minimum frequency among all initial main frequencies. Indicates the frequency complexity of the original vibration data, represents the number of component signals, Indicates The frequency of the component signal, Indicates The frequency of the component signal; The specific method of obtaining the final main frequency range includes: If the initial main frequency retention degree is greater than the retention threshold, all the initial main frequencies are used to form the final main frequency range; If the initial main frequency retention degree is less than or equal to the retention threshold, the initial main frequency is adjusted; starting from the initial main frequency with the smallest frequency among all the initial main frequencies, if the absolute value of the difference between the frequencies of two adjacent initial main frequencies is greater than the removal frequency threshold, the previous initial main frequency is deleted and the latter initial main frequency continues to be judged; if the absolute value of the difference between the frequencies of two adjacent initial main frequencies is less than or equal to the removal frequency threshold, the initial main frequency is not deleted, and so on, until the initial main frequency with the second largest frequency among all the initial main frequencies is judged, and all the initial main frequencies that have not been deleted constitute the final main frequency range.

2. A photovoltaic module steel frame detection method based on vibration data analysis according to claim 1, characterized in that: The method of obtaining the possibility of the main frequency of each frequency includes: For any frequency, the ratio of the amplitude of the frequency to the cumulative value of the amplitudes of all frequencies is recorded as the main frequency possibility of the frequency.

3. A photovoltaic module steel frame detection method based on vibration data analysis according to claim 1, characterized in that: The specific method of obtaining a plurality of initial main frequencies includes: Density clustering is performed on all frequencies, and the distance metric uses the absolute value of the difference between the main frequency possibilities of the frequencies to obtain several clusters; Calculate the mean of the dominant frequency likelihood of all frequencies in any cluster, and record it as the average dominant frequency likelihood of the cluster; All frequencies in the cluster with the highest average dominant frequency probability are taken as the initial dominant frequencies.

4. A photovoltaic module steel frame detection method based on vibration data analysis according to claim 1, characterized in that: The specific method of performing EMD decomposition on the original vibration data to obtain a plurality of component signals includes: Perform EMD decomposition on the original vibration data to obtain several IMF components, which are recorded as several component signals.

5. The photovoltaic module steel frame detection method based on vibration data analysis according to claim 1, characterized in that: The specific method of obtaining the corrected vibration data includes: According to the sine waves corresponding to all frequencies in the final main frequency range in the frequency domain space and the time domain space, the adjusted vibration data is obtained through inverse Fourier transform, and the adjusted vibration data is fitted by the least squares method, and the obtained fitting curve is recorded as the corrected vibration data.

Images