A vibration detection method and system for skived-tooth radiator

Abstract

The present invention relates to the technical field of the tightness monitoring of skived-tooth radiators, and in particular to a skived-tooth radiator vibration detection method and system. The present invention obtains all vibration components of the skived-tooth radiator vibration signal; further obtains the oscillation trend coefficient of each vibration component; obtains the fluctuation synergy between each vibration component and the corresponding adjacent vibration component; obtains the interference degree of each vibration component according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components; thereby screening out the interfering vibration components for filtering; and evaluating the vibration detection results of the skived-tooth radiator according to the filtering results. The present invention analyzes the interference degree of each vibration component based on the oscillation change trend of the decomposed vibration component and the fuzzy influence of the interference factors on the signal structure of the vibration component, thereby screening and filtering, reducing the evaluation of the interference on the vibration, and improving the accuracy of the tightness monitoring of the skived-tooth radiator.

Description

Technical Field

[0001] The invention relates to the technical field of skived-tooth radiator fastening monitoring, and in particular to a skived-tooth radiator vibration detection method and system. Background Art

[0002] A skived-tooth heat sink is a component used to dissipate heat from high-thermal-density components such as high-power devices. Usually, thermal paste is applied or a thermally conductive gasket is used as a spacer between the high-thermal-density component and the skived-tooth heat sink, and screws are used to fasten them. The heat generated by the high-thermal-density component is transferred to the skived-tooth heat sink through the thermally conductive gasket or thermal paste for rapid heat dissipation. Therefore, monitoring the fastening effectiveness of the skived-tooth heat sink is crucial to the heat conduction efficiency and heat dissipation performance.

[0003] When the fastening fails, there is usually a gap between the high heat density component and the skive-tooth heat sink. The skive-tooth heat sink will vibrate relatively irregularly along with the operating vibration of the equipment. The fastening effectiveness can be evaluated by monitoring the vibration of the skive-tooth heat sink. However, during the heat dissipation process, due to the effect of fluid dynamics, some eddy currents or irregular air flows may be generated, further causing the skive-tooth heat sink to be subjected to uneven fluid pressure, thereby causing tiny vibrations. These tiny vibrations will interfere with the vibration assessment, thereby reducing the accuracy of its fastening monitoring. Summary of the invention

[0004] In order to solve the technical problem of low accuracy of the prior art in monitoring the fastening of skived radiators, the present invention aims to provide a skived radiator vibration detection method and system, and the technical solution adopted is as follows:

[0005] The present invention provides a skived-tooth radiator vibration detection method, the method comprising:

[0006] Acquire a vibration signal of a skived-tooth heat sink, and acquire all vibration components of the vibration signal;

[0007] According to the fluctuation change and frequency characteristics of each vibration component, the oscillation trend coefficient of each vibration component is obtained; according to the difference in fluctuation between each vibration component and the adjacent vibration component, combined with the difference in the decomposition order of the corresponding vibration component, the fluctuation synergy between each vibration component and the corresponding adjacent vibration component is obtained; according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components, the interference degree of each vibration component is obtained;

[0008] According to the interference degree, the interference vibration component is selected from all the vibration components, and all the interference vibration components are filtered; and the vibration detection result of the skived radiator is evaluated according to the filtering result.

[0009] Furthermore, the method for obtaining the oscillation trend coefficient includes:

[0010] In each of the vibration components, the fluctuation parameter of each of the vibration components is obtained according to the amplitude difference between all adjacent extreme value points;

[0011] According to the phase change between all adjacent extreme value points in each of the vibration components, combined with the frequency of each of the vibration components, the frequency parameter of each of the vibration components is obtained;

[0012] The oscillation trend coefficient of each of the vibration components is obtained according to the fluctuation parameter and the frequency parameter, and both the fluctuation parameter and the frequency parameter are positively correlated with the oscillation trend coefficient.

[0013] Furthermore, the method for obtaining the fluctuation parameter includes:

[0014] Based on the amplitude differences between all adjacent extreme value points in the vibration component, information entropy is obtained, and a difference mean of the amplitude differences between all adjacent extreme value points is calculated;

[0015] A fluctuation parameter is obtained according to the information entropy and the difference mean, and both the information entropy and the difference mean are positively correlated with the fluctuation parameter.

[0016] Furthermore, the method for obtaining the frequency parameter includes:

[0017] Performing time-frequency conversion on each of the vibration components to obtain a frequency domain vibration component and an average frequency of the frequency domain vibration component; obtaining an average phase difference between all adjacent extreme value points in each of the vibration components;

[0018] A frequency parameter is obtained according to the average frequency and the average phase difference; the average frequency and the average phase difference are both positively correlated with the frequency parameter.

[0019] Furthermore, the method for obtaining the fluctuation synergy includes:

[0020] Taking a preset number of the vibration components that are closest in decomposition order to each of the vibration components as neighboring vibration components of each of the vibration components;

[0021] Between each of the vibration components and each of the adjacent vibration components, according to the difference in the number of extreme value points and the difference in amplitude between all signal points with the same sequence number, obtaining a fluctuation difference coefficient between each of the vibration components and each of the adjacent vibration components;

[0022] Using the negative correlation mapping result of the decomposition order difference between each vibration component and each adjacent vibration component as the reference weight of the fluctuation difference coefficient between the corresponding vibration components;

[0023] The fluctuation difference coefficient is weighted by using the reference weight, and the weighted result is negatively correlated to obtain the fluctuation synergy between each vibration component and the corresponding adjacent vibration component.

[0024] Furthermore, the method for obtaining the fluctuation difference coefficient includes:

[0025] The variance of the amplitude difference between all signal points with the same serial number is calculated; the product of the variance and the absolute value of the number difference of the extreme points is used as the fluctuation difference coefficient between each vibration component and the corresponding adjacent vibration component.

[0026] Furthermore, the method for obtaining the interference degree includes:

[0027] Obtaining the coefficient difference of the oscillation trend coefficient between each vibration component and the next vibration component; obtaining the accumulated value of all the coefficient differences between each vibration component and all the adjacent vibration components, and obtaining the accumulated value of the absolute values ​​of all the coefficient differences; taking the ratio of the accumulated value to the accumulated value of the absolute values ​​as a reference coefficient;

[0028] In each of the vibration components and all the adjacent vibration components, the fluctuation synergy between each of the vibration components and each of the adjacent vibration components is accumulated, the accumulated result is weighted by using the reference coefficient, and the weighted result is normalized to obtain the interference degree of each of the vibration components.

[0029] Furthermore, the method for obtaining the interference vibration component includes:

[0030] The vibration component whose interference degree is greater than a preset threshold is regarded as an interference vibration component.

[0031] Furthermore, the method for filtering the interfering vibration component includes:

[0032] The difference between a preset positive constant and the interference degree is used as a filter kernel adjustment weight corresponding to the vibration component, the filter kernel is adjusted, and the corresponding interference vibration component is filtered using the adjusted filter kernel.

[0033] The present invention also proposes a skived-tooth radiator vibration detection system, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the skived-tooth radiator vibration detection method are implemented.

[0034] The present invention has the following beneficial effects:

[0035] The present invention obtains all vibration components of the vibration signal of the shovel-tooth radiator; obtains the oscillation trend coefficient of each vibration component according to the fluctuation change and frequency characteristics of each vibration component; the oscillation trend coefficient characterizes the fluctuation change and oscillation trend of each vibration component, so as to evaluate whether it is interfered by noise in the subsequent combination with the oscillation trend change between vibration components of adjacent decomposition orders; obtains the fluctuation synergy between each vibration component and the corresponding adjacent vibration component according to the difference in fluctuation between each vibration component and the adjacent vibration component, combined with the difference in decomposition order of the corresponding vibration component; the fluctuation synergy accurately evaluates the fluctuation difference between each vibration component and the corresponding adjacent vibration component while reducing the possibility of overlapping influence, and the smaller the fluctuation difference, the more similar the vibration components are, indicating that the possibility of component signal structure fuzziness caused by interference factors is greater; according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components, the interference degree of each vibration component is obtained; according to the interference degree, the interference vibration component is screened out from all vibration components, and all the interference vibration components are filtered; and the vibration detection result of the shovel-tooth radiator is evaluated according to the filtering result. The present invention is based on the fluctuation and oscillation trend of the decomposed vibration components and combines the fuzzy influence of interference factors on the signal structure of the vibration components to analyze the interference degree of each vibration component, thereby screening and filtering it, reducing the impact of interference on vibration assessment and improving the accuracy of tightness monitoring of the skived tooth radiator. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention 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 invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0037] Figure 1 A flow chart of a skived-tooth radiator vibration detection method provided by one embodiment of the present invention;

[0038] Figure 2 A flow chart of a method for obtaining an oscillation trend coefficient provided by an embodiment of the present invention;

[0039] Figure 3 A flowchart of a method for obtaining fluctuation synergy provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0040] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of the specific implementation method, structure, features and effects of a shovel-tooth radiator vibration detection method and system proposed by the present invention 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.

[0041] 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 invention belongs.

[0042] The specific scheme of the vibration detection method and system of a skived-tooth radiator provided by the present invention is described in detail below with reference to the accompanying drawings.

[0043] See also Figure 1 , which shows a flow chart of a skived-tooth radiator vibration detection method provided by an embodiment of the present invention, specifically comprising:

[0044] Step S1, obtaining a vibration signal of a skived-tooth heat sink, and obtaining all vibration components of the vibration signal.

[0045] The implementation scenarios targeted by the embodiments of the present invention generally refer to scenarios in which a mainboard of an air conditioner outdoor unit, a fan mainboard, etc., will generate mechanical vibrations during operation of the equipment, which may lead to failure of the fastening of the skived-tooth heat sink.

[0046] In one embodiment of the present invention, in order to monitor the tightness of the skived-tooth heat sink, the vibration signal of the skived-tooth heat sink when the high heat density element is operating is first obtained, specifically by installing a vibration sensor on the outermost heat sink or supporting structure of the skived-tooth heat sink to collect the vibration signal, wherein the collection frequency of the vibration sensor is once per second; in order to facilitate subsequent analysis and filter out vibration interference caused by air flow, all vibration components of the vibration signal are further obtained, and specifically, the inherent time scale decomposition algorithm (Intrinsic Time Decomposition, ITD) is used to decompose the vibration signal.

[0047] It should be noted that the collected vibration signal can be a vibration signal within a preset time period during operation, such as ten minutes, half an hour or even the vibration signal during the entire operation process; multiple vibration sensors can also be installed for analysis to comprehensively evaluate the degree of tightening; the sampling frequency can also be set by oneself; in other embodiments of the present invention, the implementer can also adopt other signal decomposition algorithms such as EMD, VMD, etc., which are well-known technologies with the ITD algorithm and will not be repeated here.

[0048] Step S2, according to the fluctuation change and frequency characteristics of each vibration component, obtain the oscillation trend coefficient of each vibration component; according to the difference in fluctuation between each vibration component and the adjacent vibration component, combined with the difference in decomposition order of the corresponding vibration components, obtain the fluctuation synergy between each vibration component and the corresponding adjacent vibration component; according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components, obtain the interference degree of each vibration component.

[0049] Ideally, the ITD decomposition algorithm will decompose the vibration signal into multiple components of different scales, each component corresponding to an oscillation mode at a different frequency or time scale; the algorithm will start decomposition from a shorter time scale and gradually extract the high-frequency components in the vibration signal. The high-frequency components correspond to vibration components that usually have the characteristics of violent fluctuations; as the decomposition process deepens, the decomposed vibration components are usually low-frequency components such as slowly changing trends or baseline signals in the signal, that is, the vibration components decomposed first often contain high-frequency components, which should have more obvious oscillation changes, and the vibration components decomposed later are usually low-frequency components, and the degree of fluctuation should be relatively reduced; but since the vibration signal is mixed with the vibration of the shovel-tooth radiator caused by the air flow caused by normal heat dissipation, the air flow vibration caused by heat dissipation will also appear more random due to different heat dissipation states of the equipment. The distribution of these random interferences will cause the fluctuation trend of the vibration component to become complex and difficult to quantify.

[0050] Therefore, the embodiment of the present invention obtains the oscillation trend coefficient of each vibration component according to the fluctuation change and frequency characteristics of each vibration component; the oscillation trend coefficient characterizes the fluctuation change and oscillation trend of each vibration component, so as to subsequently combine the oscillation trend changes between the vibration components in adjacent decomposition orders to evaluate whether they are interfered by noise.

[0051] Preferably, in one embodiment of the present invention, the method for obtaining the oscillation trend coefficient includes:

[0052] See also Figure 2 , which shows a flow chart of a method for obtaining an oscillation trend coefficient provided by an embodiment of the present invention, specifically comprising:

[0053] Step S201, in each vibration component, obtaining the fluctuation parameter of each vibration component according to the amplitude difference between all adjacent extreme value points.

[0054] Considering that the amplitude difference between adjacent extreme points can describe the fluctuation characteristics of the vibration component to a certain extent, the larger the amplitude difference, the more violent the fluctuation. Therefore, the amplitude difference between all adjacent extreme points can be comprehensively considered to evaluate the fluctuation of the vibration component.

[0055] In a preferred embodiment of the present invention, considering that the fluctuation oscillation trend in the vibration component should be relatively complex and irregular, the amplitude difference change between adjacent extreme points should also be relatively complex, and the information entropy can reflect the complexity and confusion of information to a certain extent; at the same time, the mean of the amplitude difference between the extreme points can reflect the approximate fluctuation amplitude of the vibration component as a whole, and indirectly reflect the degree of fluctuation intensity; therefore, the fluctuation parameters are obtained by combining the two; the method for obtaining the fluctuation parameters includes:

[0056] Based on the amplitude difference between all adjacent extreme points in the vibration component, the information entropy is obtained, and the difference mean of the amplitude difference between all adjacent extreme points is calculated; the fluctuation parameter is obtained according to the information entropy and the difference mean, and the information entropy and the difference mean are positively correlated with the fluctuation parameter.

[0057] As an example, the absolute value of the amplitude difference between each extreme point and the next adjacent extreme point is calculated to evaluate the amplitude difference, where the last extreme point can be ignored if it has no next adjacent extreme point; further calculate the information entropy, multiply the information entropy with the mean of the absolute value of the amplitude difference to obtain the fluctuation parameter; in other examples, implementers can also use basic mathematical operations such as addition or weighted summation or related mapping methods to combine the two to obtain the fluctuation parameter.

[0058] It should be noted that the acquisition of extreme points and information entropy is a well-known technology and will not be elaborated here.

[0059] In other embodiments of the present invention, the implementer may also directly calculate the variance of the amplitude difference between adjacent extreme points and other parameters used to characterize the degree of discreteness instead of information entropy to obtain the fluctuation parameters, which will not be elaborated here.

[0060] Step S202 , obtaining the frequency parameter of each vibration component according to the phase change between all adjacent extreme value points in each vibration component and the frequency of each vibration component.

[0061] Considering that the phase change between adjacent extreme points reflects the frequency to a certain extent, the greater the phase change, the more likely it is a high-frequency component; at the same time, the higher the frequency of the vibration component, the more violent the fluctuation characteristics, so the larger the frequency parameter of the vibration component, the greater the fluctuation oscillation trend should be.

[0062] In a preferred embodiment of the present invention, considering that time-frequency conversion can obtain the frequency domain signal of the vibration component, it is convenient to analyze the frequency component of the vibration component; the average phase difference comprehensively evaluates the frequency of the vibration component by the phase changes between all adjacent extreme points; both reflect the frequency characteristics of the vibration component; based on this, the method for obtaining the frequency parameter includes:

[0063] Perform time-frequency conversion on each vibration component to obtain the frequency domain vibration component and the average frequency of the frequency domain vibration component; obtain the average phase difference between all adjacent extreme value points in each vibration component;

[0064] The frequency parameter is obtained according to the average frequency and the average phase difference; the average frequency and the average phase difference are both positively correlated with the frequency parameter.

[0065] As an example, the vibration component is specifically obtained through Fourier transform to obtain the frequency domain vibration component, each frequency in the frequency domain vibration component is multiplied by the corresponding amplitude to obtain the weighted frequency, and all weighted frequencies are averaged to obtain the average frequency of each vibration component; then the absolute value of the phase difference between adjacent extreme points is obtained, and the average phase difference is further averaged to obtain the average phase difference; finally, the average frequency and the average phase difference are multiplied to obtain the frequency parameter; in other examples, the implementer may also use basic mathematical operations such as addition or weighted summation or related mapping methods to combine the two to obtain the frequency parameter.

[0066] In other embodiments of the present invention, implementers may also use the mode instead of the average, which will not be described in detail here.

[0067] Step S203, obtaining the oscillation trend coefficient of each vibration component according to the fluctuation parameter and the frequency parameter, wherein the fluctuation parameter and the frequency parameter are both positively correlated with the oscillation trend coefficient.

[0068] As an example, the fluctuation parameter and the frequency parameter are multiplied and combined to obtain the oscillation trend coefficient of the corresponding vibration component; in other examples, the implementer may also use basic mathematical operations such as addition or weighted summation or related mapping methods to combine the two to obtain the oscillation trend coefficient.

[0069] At this point, the oscillation tendency coefficient of each vibration component is obtained.

[0070] Considering that in the absence of other external interference such as air flow, the corresponding vibration components at different time scales should have relatively obvious differences; however, the random distribution of interference may blur the structural differences between the vibration components, making the oscillation characteristics of the vibration components more similar, thereby reducing the accuracy and credibility of the decomposition; and considering that the algorithm may not be able to completely and accurately separate all overlapping frequency components during the decomposition process of the vibration signal, the vibration components obtained by adjacent decompositions may have certain overlapping similar components, which may lead to lower accuracy in the fluctuation similarity analysis between different vibration components.

[0071] Therefore, the embodiment of the present invention obtains the fluctuation synergy between each vibration component and the corresponding adjacent vibration component based on the difference in fluctuation conditions between each vibration component and the adjacent vibration component, combined with the difference in decomposition order of the corresponding vibration components; the fluctuation synergy is combined with the fluctuation difference between the adjacent vibration components to accurately evaluate whether the fluctuation oscillation trends between the vibration components are coordinated while reducing the possibility of overlapping effects; the more coordinated and similar the synergy is, the greater the possibility that the component signal structure is blurred due to interference factors.

[0072] Preferably, in one embodiment of the present invention, the method for obtaining the fluctuation synergy includes:

[0073] See also Figure 3 , which shows a flow chart of a method for obtaining fluctuation synergy provided by an embodiment of the present invention, specifically comprising:

[0074] Step S301 : taking a preset number of vibration components that are closest to the decomposition order of each vibration component as adjacent vibration components of each vibration component.

[0075] First, determine the adjacent vibration components to facilitate subsequent analysis; as an example, set the preset number to 4, that is, take each vibration component as the object to be analyzed, and use the first two and last two adjacent vibration components decomposed as the adjacent vibration components; that is, take each vibration component as the center to construct a neighborhood analysis window, which contains 1 central vibration component and 4 adjacent vibration components decomposed before and after.

[0076] It should be noted that there may be some vibration components that do not have adjacent decomposed vibration components in the front or back, then the combination of adjacent vibration components can be 0 followed by 4, 1 followed by 3, 3 followed by 1, or 4 followed by 0; in other examples, the implementer can also set the preset number by himself, which is not elaborated here.

[0077] Step S302 , obtaining the fluctuation difference coefficient between each vibration component and each adjacent vibration component according to the difference in the number of extreme value points and the difference in amplitude between all signal points with the same sequence number.

[0078] Considering that between two vibration components, the fluctuation difference is usually reflected in the difference in fluctuation amplitude and waveform, and the number of extreme points usually reflects the fluctuation to a certain extent, and the amplitude difference between the signal points corresponding to the same time sequence number further reflects the difference in fluctuation amplitude; based on this, the two can be combined to obtain the fluctuation difference coefficient; the fluctuation difference coefficient preliminarily reflects the difference in fluctuation trends between the vibration components. The larger the fluctuation difference coefficient, the lower the possibility that the vibration component is disturbed.

[0079] In a preferred embodiment of the present invention, a method for obtaining a fluctuation difference coefficient includes: calculating the variance of the amplitude difference between all signal points with the same serial number; and taking the product of the variance and the absolute value of the number difference of extreme points as the fluctuation difference coefficient between each vibration component and the corresponding adjacent vibration component.

[0080] In other embodiments of the present invention, the implementer may also use basic mathematical operations such as addition or weighted summation or related mapping methods to combine the two to obtain the fluctuation difference coefficient; or may use other parameters such as standard deviation to characterize the degree of dispersion instead of variance.

[0081] Step S303: using the negative correlation mapping result of the decomposition order difference between each vibration component and each adjacent vibration component as a reference weight of the fluctuation difference coefficient between the corresponding vibration components.

[0082] The closer the decomposition order is, the greater the possibility of overlapping similarities is, and the lower the confidence level of the corresponding volatility difference coefficient is; therefore, the difference in decomposition order is negatively correlated and mapped as a reference weight.

[0083] As an example, since the difference in decomposition order cannot be 0, the absolute value of the difference in the decomposition order numbers is inversely calculated and the logic is adjusted to obtain the reference weight; in other examples, implementers may also use other basic mathematical operations or related mapping methods to adjust the logic, which will not be elaborated here.

[0084] Step S304, weighting the fluctuation difference coefficient using the reference weight, performing negative correlation mapping on the weighted result, and obtaining the fluctuation synergy between each vibration component and the corresponding adjacent vibration component.

[0085] As an example, the reference weight is multiplied by the fluctuation difference coefficient to obtain a weighted result; the weighted result reduces the possibility of overlapping effects while accurately evaluating the fluctuation difference between each vibration component and the corresponding adjacent vibration component. The smaller the fluctuation difference, the more similar the vibration components are, and the greater the fluctuation synergy between the corresponding vibration components. In this case, the negative correlation mapping adjustment logic is required, specifically using the weighted result as a natural constant. Exponential function with base In ; In other examples, implementers may also adopt other negative correlation mapping methods, which are not described here.

[0086] In another embodiment of the present invention, the implementer may also evaluate the fluctuation difference coefficient between each vibration component and each adjacent vibration component by comparing the DTW distance between them; specifically, the DTW distance is negatively correlated and the DTW distance is used as a natural constant. Exponential function with base In , and obtain the volatility difference coefficient, which is then combined with the reference weight to obtain volatility synergy, which will not be elaborated here.

[0087] Considering that as the decomposition proceeds, the oscillation trend is relatively gentler when the decomposition order is later, that is, as the decomposition order gradually increases, the oscillation trend coefficient should gradually decrease; the gradual decrease in the oscillation trend coefficient between adjacent vibration components reflects the effectiveness of signal decomposition to a certain extent, and the higher the confidence for analyzing the fluctuation synergy in order to evaluate the presence of interference. Therefore, the embodiment of the present invention obtains the interference degree of each vibration component according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components. The interference degree is based on the fluctuation oscillation change trend of the decomposed vibration component, combined with the fuzzy influence of the interference factor on the signal structure of the vibration component, and analyzes the interference effect of each vibration component.

[0088] Preferably, in one embodiment of the present invention, considering that the oscillation trend coefficient of the vibration component should be gradually reduced, the coefficient difference of the oscillation trend coefficients between all adjacent vibration components can be integrated to evaluate its change, and then obtain a reference coefficient that can reflect the decomposition effect; considering that if the vibration component has a higher fluctuation synergy with all adjacent vibration components, the possibility of the vibration component being interfered is greater; based on this, the method for obtaining the interference degree includes:

[0089] Obtain the coefficient difference of the oscillation trend coefficient between each vibration component and the next vibration component; obtain the cumulative value of all coefficient differences between each vibration component and all adjacent vibration components, and obtain the absolute value cumulative value of all coefficient differences; use the ratio of the cumulative value to the absolute value cumulative value as a reference coefficient;

[0090] In each vibration component and all adjacent vibration components, the fluctuation synergy between each vibration component and each adjacent vibration component is accumulated, the accumulated result is weighted by using a reference coefficient, and the weighted result is normalized to obtain the interference degree of each vibration component.

[0091] As an example: The interference level is calculated as: ;in, For the The degree of interference of each vibration component; is the standard normalization function; For the first The sequence number of the vibration component in the neighborhood analysis window centered on the vibration component, include ; For the first The total number of vibration components within the neighborhood analysis window centered on the vibration component; For the first The vibration component is the center of the neighborhood analysis window. Oscillation tendency coefficient of each vibration component; For the first The vibration component is the center of the neighborhood analysis window. Oscillation tendency coefficient of each vibration component; For the first The sequence number of the vibration component in the neighborhood analysis window centered on the vibration component, Not included ; For the first In the neighborhood analysis window centered on the vibration component, The vibration component and The fluctuation synergy between the vibration components; is the reference coefficient.

[0092] In the above formula, the closer the reference coefficient is to 1, the greater the possibility that all coefficient differences are positive, that is, the change in the oscillation trend coefficient of adjacent vibration components is getting smaller and smaller, and the higher the confidence in analyzing the fluctuation synergy; the larger the cumulative result of the fluctuation synergy is, it means that the vibration component has similar fluctuation characteristics with all adjacent vibration components in the adjacent analysis window, and the greater the possibility that the component signal structure is blurred due to interference; the specific maximum and minimum value normalization method is used, and other normalization methods can also be used, which will not be repeated here.

[0093] Step S3, screening out interference vibration components from all vibration components according to the interference degree, and filtering all interference vibration components; and evaluating the vibration detection result of the skived-tooth radiator according to the filtering result.

[0094] After obtaining the interference degree of each vibration component, the interfering vibration component can be further screened out; it can be further filtered to remove the influence of interfering factors such as air flow on the vibration signal.

[0095] Preferably, in one embodiment of the present invention, the method for obtaining the interference vibration component includes:

[0096] The vibration component with an interference degree greater than a preset threshold is regarded as an interference vibration component. The preset threshold is 0.7, which can also be set by the implementer.

[0097] In a preferred embodiment of the present invention, the implementer may use the difference between a preset positive constant and the interference degree as a filter kernel adjustment weight of the corresponding vibration component, adjust the filter kernel, and use the adjusted filter kernel to filter the corresponding interference vibration component.

[0098] As an example: set the preset normal number to 1.5, the interference degree is 0.7-1, and the filter kernel adjustment weight is 0.5-0.8; determine the filter kernel size during Gaussian filtering based on the standard deviation, which is a prior art known to those skilled in the art and will not be elaborated here; use the filter kernel weight to appropriately reduce the filter kernel. The smaller the filter kernel, the more detailed the smoothing effect. Otherwise, more features may be smoothed out; then use the adjusted filter kernel to filter the corresponding interference vibration component, so as to smooth out the interference influence to the greatest extent while ensuring the original oscillation fluctuation characteristics; in other examples, the implementer may also set the preset normal number to other values ​​between 1.5-1.7. When the preset normal number is 1.7, the filter kernel adjustment weight is 0.9-0.1. The adjustment range is relatively small, but it further improves the interference filtering effect. The implementer can choose by themselves.

[0099] The embodiment of the present invention filters all the interfering vibration components and reconstructs them to obtain the filtered vibration signal. Furthermore, within a certain historical time period, the historical noise-reduced vibration signals are manually observed to see whether the skived-tooth radiator is tightened, so as to be manually labeled with a tightened label or an untightened label, and a training sample data is constructed to train a neural network model. The trained neural network model is used to evaluate the label corresponding to the current noise-reduced vibration signal, so as to evaluate whether the skived-tooth radiator has failed to be tightened, and thus corresponding tightening or component re-soldering measures are taken to avoid abnormal effects such as equipment shutdown caused by desoldering.

[0100] It should be noted that the neural network model is a convolutional neural network model. The training application of the convolutional neural network model is a prior art well known to those skilled in the art and will not be described in detail here.

[0101] In summary, the embodiment of the present invention obtains all vibration components of the vibration signal of the shovel-tooth radiator; first obtains the oscillation trend coefficient of each vibration component; further obtains the fluctuation synergy between each vibration component and the corresponding adjacent vibration component; obtains the interference degree of each vibration component according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components; thereby screening out the interfering vibration component from all vibration components for filtering; and evaluates the vibration detection result of the shovel-tooth radiator according to the filtering result. The present invention is based on the fluctuation oscillation change trend of the decomposed vibration component, combined with the fuzzy influence of the interference factor on the signal structure of the vibration component, and analyzes the interference degree of each vibration component, thereby screening and filtering it, reducing the interference evaluation on the vibration, and improving the accuracy of the tightening monitoring of the shovel-tooth radiator.

[0102] The present invention also proposes a skived-tooth radiator vibration detection system, the system comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of a skived-tooth radiator vibration detection method when executing the computer program.

[0103] It should be noted that the sequence of the above embodiments of the present invention is for description only and does not represent the advantages and disadvantages of the embodiments. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0104] 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.

Claims

1. A vibration detection method for a skived-tooth radiator, characterized in that: The method comprises: Acquire a vibration signal of a skived-tooth heat sink, and acquire all vibration components of the vibration signal; According to the fluctuation change and frequency characteristics of each vibration component, the oscillation trend coefficient of each vibration component is obtained; according to the difference in fluctuation between each vibration component and the adjacent vibration component, combined with the difference in the decomposition order of the corresponding vibration component, the fluctuation synergy between each vibration component and the corresponding adjacent vibration component is obtained; according to the fluctuation synergy between each vibration component and all adjacent vibration components, and the change of the oscillation trend coefficient between adjacent vibration components, the interference degree of each vibration component is obtained; According to the interference degree, the interference vibration component is selected from all the vibration components, and all the interference vibration components are filtered; and the vibration detection result of the skived radiator is evaluated according to the filtering result; The method for obtaining the oscillation trend coefficient includes: In each of the vibration components, the fluctuation parameter of each of the vibration components is obtained according to the amplitude difference between all adjacent extreme value points; According to the phase change between all adjacent extreme value points in each of the vibration components, combined with the frequency of each of the vibration components, the frequency parameter of each of the vibration components is obtained; Obtaining an oscillation trend coefficient of each of the vibration components according to the fluctuation parameter and the frequency parameter, wherein both the fluctuation parameter and the frequency parameter are positively correlated with the oscillation trend coefficient; The method for obtaining the fluctuation synergy includes: Taking a preset number of the vibration components that are closest in decomposition order to each of the vibration components as neighboring vibration components of each of the vibration components; Between each of the vibration components and each of the adjacent vibration components, according to the difference in the number of extreme value points and the difference in amplitude between all signal points with the same sequence number, obtaining a fluctuation difference coefficient between each of the vibration components and each of the adjacent vibration components; Using the negative correlation mapping result of the decomposition order difference between each vibration component and each adjacent vibration component as the reference weight of the fluctuation difference coefficient between the corresponding vibration components; The fluctuation difference coefficient is weighted by using the reference weight, and the weighted result is negatively correlated to obtain the fluctuation synergy between each vibration component and the corresponding adjacent vibration component; The method for obtaining the interference degree includes: Obtaining the coefficient difference of the oscillation trend coefficient between each vibration component and the next vibration component; obtaining the accumulated value of all the coefficient differences between each vibration component and all the adjacent vibration components, and obtaining the accumulated value of the absolute values ​​of all the coefficient differences; taking the ratio of the accumulated value to the accumulated value of the absolute values ​​as a reference coefficient; In each of the vibration components and all the adjacent vibration components, the fluctuation synergy between each of the vibration components and each of the adjacent vibration components is accumulated, the accumulated result is weighted by using the reference coefficient, and the weighted result is normalized to obtain the interference degree of each of the vibration components.

2. According to the method for detecting vibration of a skived-tooth radiator as described in claim 1, it is characterized in that: The method for obtaining the fluctuation parameter includes: Based on the amplitude differences between all adjacent extreme value points in the vibration component, information entropy is obtained, and a difference mean of the amplitude differences between all adjacent extreme value points is calculated; A fluctuation parameter is obtained according to the information entropy and the difference mean, and both the information entropy and the difference mean are positively correlated with the fluctuation parameter.

3. According to the vibration detection method of skived-tooth radiator described in claim 1, it is characterized in that: The method for obtaining the frequency parameter includes: Performing time-frequency conversion on each of the vibration components to obtain a frequency domain vibration component and an average frequency of the frequency domain vibration component; obtaining an average phase difference between all adjacent extreme value points in each of the vibration components; A frequency parameter is obtained according to the average frequency and the average phase difference; the average frequency and the average phase difference are both positively correlated with the frequency parameter.

4. A skived-tooth radiator vibration detection method according to claim 1, characterized in that: The method for obtaining the fluctuation difference coefficient includes: The variance of the amplitude difference between all signal points with the same serial number is calculated; the product of the variance and the absolute value of the number difference of the extreme points is used as the fluctuation difference coefficient between each vibration component and the corresponding adjacent vibration component.

5. A skived-tooth radiator vibration detection method according to claim 1, characterized in that: The method for obtaining the interference vibration component includes: The vibration component whose interference degree is greater than a preset threshold is regarded as an interference vibration component.

6. A skived-tooth radiator vibration detection method according to claim 1, characterized in that: The method for filtering the interfering vibration component comprises: The difference between a preset positive constant and the interference degree is used as a filter kernel adjustment weight corresponding to the vibration component, the filter kernel is adjusted, and the corresponding interference vibration component is filtered using the adjusted filter kernel.

7. A skived-tooth radiator vibration detection system, characterized in that: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of a skived-tooth radiator vibration detection method as described in any one of claims 1 to 6 are implemented.

Images