Inspection data sharing method based on industrial cloud

Abstract

The invention relates to the technical field of data transmission sharing, in particular to an inspection data sharing method based on industrial cloud. Acquiring original ultrasonic data of the measuring point during detection; segmenting the original ultrasonic data, analyzing extreme points and signal segments to obtain corresponding mapping values, and then screening out the signal segments to determine target data; different decomposition results are obtained by iterating different time scales, component signals under adjacent iterations are analyzed, and constraint values are obtained in combination with data; the high-quality scale is determined according to the constraint value, and the component signal under the iteration number corresponding to the high-quality scale can be regarded as a signal without distortion; according to the method, the end point value can be predicted more accurately by using the high-quality scale, and finally, the end point value is obtained according to the end point predicted value, so that the original ultrasonic data is reduced, decomposed and reconstructed, the end point effect problem can be effectively solved, high-quality de-noised data is obtained, transmitted and shared, and the sharing value of the data is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data transmission sharing, and particularly relates to an inspection data sharing method based on an industrial industry cloud. BACKGROUND

[0002] With the continuous expansion of the Internet, computing, storage and various applications converge on the network, showing a growing trend, and industrial networks and the Internet are constantly merging to provide convenient operation for the industrial industry. Cloud computing and cloud services achieve elastic scheduling and scalability of resources through the Internet. However, with the widespread application of cloud computing and cloud services, the security of cloud data gradually appears in the process of remote sharing of related industrial product detection data in the cloud, especially the welding detection data has great sharing value.

[0003] Current industrial welding process detection mostly uses ultrasonic detection technology to detect the quality of the weld, and according to the ultrasonic scanning result to identify and detect defects to ensure that the welding quality meets the requirements. However, the welding detection ultrasonic signal will have a large noise interference, and the intrinsic time scale decomposition algorithm (Intrinsic Time-Scale Decomposition, ITD) is usually used for denoising, but the ITD decomposition algorithm will have an end effect problem, which will affect the denoising effect. The existing technology usually uses periodic extension method to solve the end effect problem, but it will cause the decomposition result to be distorted, thereby causing the denoising effect to be low, and ultimately affecting the sharing value of the data. SUMMARY

[0004] In order to solve the technical problem that the existing technology usually uses periodic extension method to solve the end effect problem, but it will cause the decomposition result to be distorted, thereby causing the denoising effect to be low, and ultimately affecting the sharing value of the data, the purpose of the present application is to provide an inspection data sharing method based on an industrial industry cloud, and the technical solution adopted is as follows: The present application provides an inspection data sharing method based on an industrial industry cloud, which comprises: Obtaining original ultrasonic data of each measuring point in the welding detection process based on a preset sampling frequency; According to the extreme value points in the original ultrasonic data, the signal segments are divided; according to the amplitude of the extreme value points, the length of the signal segments and the frequency of the length, the mapping value corresponding to the signal segments is obtained; and the target data is determined according to the mapping value; A preset upper limit of the iteration number is set, and the time scale corresponding to the target data at each iteration is obtained according to the iteration number; and the target data is decomposed according to the time scale in each iteration based on the ITD decomposition algorithm, and a preset number of component signals are obtained; obtaining a constraint value of each iteration according to the difference between the data corresponding to the time scale at each iteration and the difference between the component signals in adjacent iteration numbers; determining a quality scale according to the change of the constraint value in the iteration process; re-sampling the measuring points according to the quality scale to obtain prediction data of the measuring points, and obtaining an endpoint prediction value according to each prediction data; obtaining a final endpoint value according to the amplitude of the extreme point in the target data and the endpoint prediction value corresponding to all quality scales; restoring, decomposing and reconstructing the original ultrasonic data based on the ITD decomposition algorithm according to the final endpoint value to obtain denoising data for transmission and sharing.

[0005] Further, the signal segment is divided according to the extreme point in the original ultrasonic data, comprising: The data between the adjacent two extreme points is taken as a signal segment.

[0006] Further, the method for obtaining the mapping value comprises: The signal segments of the same length are taken as the same kind of signal segment; the frequency of the length of each kind of signal segment in all lengths is obtained; The average value of the left extreme point amplitude and the average value of the right extreme point amplitude of each kind of signal segment are obtained respectively; the difference between the average value of the left extreme point amplitude and the average value of the right extreme point amplitude and the length of the signal segment are taken as the average slope value corresponding to the signal segment; The average slope value corresponding to each kind of signal segment is normalized to be taken as an adjustment factor; the adjustment factor, the corresponding frequency of occurrence and a preset constant of each kind of signal segment are multiplied to be taken as the amplitude-frequency product corresponding to the signal segment; All kinds of signal segments are sorted in descending order based on the length of the signal segment to obtain a sorting sequence; any kind of signal segment in the sorting sequence is taken as an analyzed signal segment; the sum of the amplitude-frequency products corresponding to the analyzed signal segment and the subsequent signal segments in the sorting sequence is taken as an accumulated characteristic value; the accumulated characteristic value is rounded to obtain the mapping value of the analyzed signal segment.

[0007] Further, the signal segment is filtered out according to the mapping value to determine the target data, comprising: The mapping value corresponding to each kind of signal segment in the sorting sequence is compared with the mapping value corresponding to the adjacent next kind of signal segment in turn, if equal, the signal segment is deleted; initial data is obtained; The initial data is decentered and whitened to obtain the target data.

[0008] Further, the time scale corresponding to the target data at each iteration is obtained according to the number of iterations, comprising: The number of iterations is used as the merging number. Adjacent signal segments in the target data are merged according to the merging number. The length of the merged data segment is used as the time scale corresponding to the iteration of the target data.

[0009] Furthermore, the method for calculating the constraint value includes: ;in, Indicates the first The constraint value at the next iteration This represents the total number of component signals. Indicates the first The first iteration One component signal Indicates the first The first iteration One component signal Indicates mean square error. This represents the number of pairwise combinations of all time scales in the target data at each iteration. Indicates the first During the nth iteration Mean square error between data segments corresponding to the time scale.

[0010] Furthermore, determining the quality metric based on the changes in constraint values ​​during the iteration process includes: Starting from the first iteration, and continuing until each iteration number, the constraint values ​​of all iterations are accumulated, and this is used as the constraint accumulation value for each iteration number. The constraint curve is obtained based on the cumulative constraint value of all iterations, where the horizontal axis of the constraint curve represents the number of iterations and the vertical axis represents the cumulative constraint value. The maximum inflection point of the constraint curve is obtained based on the elbow method. The number of iterations corresponding to the maximum inflection point is taken as the optimal number of iterations. All time scales from the first iteration to the optimal number of iterations are taken as high-quality scales.

[0011] Further, the step of resampling the measurement points according to the high-quality scale to obtain the predicted data of the measurement points, and obtaining the endpoint predicted value based on each of the predicted data, includes: The product of the inverse of the iteration number corresponding to each quality scale and the preset sampling frequency is used as the prediction sampling frequency; the measurement points are resampled according to the prediction sampling frequency to obtain a data sequence, and the data sequence is used as the prediction data corresponding to the quality scale; Based on the multinomial regression model and the prediction direction, the endpoint prediction value corresponding to each quality scale is obtained from the prediction data. When the prediction direction is the same as the time series direction, the obtained endpoint prediction value is the right endpoint prediction value. When the prediction direction is opposite to the time series direction, the obtained endpoint prediction value is the left endpoint prediction value.

[0012] Furthermore, the method for calculating the final endpoint value includes: ;in, Indicates the final endpoint value. This represents the total number of high-quality standards. Indicates the first Endpoint prediction values ​​corresponding to each high-quality scale This represents the mean of the magnitudes of all extreme points in the target data. It represents the standard deviation of the magnitudes of all extreme points in the target data.

[0013] Furthermore, the ITD-based decomposition algorithm restores, decomposes, and reconstructs the original ultrasound data according to the final endpoint value to obtain denoised data for transmission and sharing, including: Replace the endpoint values ​​in the original ultrasound data with the final endpoint values ​​to restore the original ultrasound data and obtain the data to be measured. The test data is decomposed based on the ITD decomposition algorithm, and a preset number of component signals are selected for reconstruction to obtain denoised data. The denoised data is compressed using the Huffman lossless compression algorithm to obtain compressed data; The compressed data is transmitted and shared.

[0014] The present invention has the following beneficial effects: The application mainly aims at the problem that signal data obtained in the weld detection is easily interfered by noise, the ITD decomposition algorithm has endpoint effect problem, the denoising effect is low, and finally the data sharing value is reduced; first, the original ultrasonic data of each measuring point in the weld detection process is obtained; since the ITD decomposition algorithm is a local decomposition algorithm, and the decomposition process is based on extreme points, therefore, the original ultrasonic data is segmented according to the extreme points, so as to analyze the data from the local; further, for the noise signal, the relatively gentle signal is an abnormal signal, therefore, the mapping value corresponding to the signal segment is obtained according to the amplitude of the extreme point, the signal segment length and the occurrence frequency, and then the target data is determined according to the mapping value; then, different time scales are obtained through different iteration numbers, and the ITD decomposition results under different time scales are obtained; the difference between the component signals under adjacent iteration numbers in the iteration process and the difference between the data corresponding to the time scales under each iteration number are evaluated on the decomposition result of each iteration, and the constraint value is obtained; therefore, the better time scale is determined according to the change of the constraint value in the iteration process, as the high-quality scale, and the component signal under the iteration number corresponding to the high-quality scale can be regarded as the signal without distortion; therefore, further, the measuring point is resampled by using the high-quality scale, and the prediction data is obtained, at this time, the prediction data can more accurately predict the endpoint value because the related information of the high-quality scale is combined, therefore, the endpoint prediction value is obtained based on the prediction data; finally, the adaptive acquisition of the final endpoint value is carried out according to the endpoint prediction value of the high-quality scale and the amplitude of the extreme point in the target data, the interference problem caused by the endpoint effect is effectively solved; and the original ultrasonic data is restored, decomposed and reconstructed based on the ITD decomposition algorithm and the final endpoint value, and high-quality denoising data is obtained; that is, the high-quality denoising data can be transmitted and shared, so as to ensure the sharing value of the data. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, and the advantages thereof, below, a brief introduction will be given to the drawings needed to be used in the embodiments or the prior art description, obviously, the drawings in the following description are only some embodiments of the application, and for those skilled in the art, other drawings can be obtained without creative labor based on these drawings.

[0016] Figure 1 The flow chart of the inspection data sharing method based on the industrial industry cloud provided by an embodiment of the application. DETAILED DESCRIPTION

[0017] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined inventive purpose, the specific implementation, structure, features and effects of the inspection data sharing method based on industrial cloud proposed according to the present application are described in detail below in combination with the drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.

[0019] Embodiment of the inspection data sharing method based on industrial cloud: The specific scheme of the inspection data sharing method based on industrial cloud provided by the present application is described below in combination with the drawings.

[0020] Please refer to Figure 1 which shows the method flowchart of the inspection data sharing method based on industrial cloud provided by one embodiment of the present application, which includes the following steps: Step S1: Obtain the original ultrasonic data of each measuring point in the weld detection process based on a preset sampling frequency.

[0021] Current industrial welding process detection mostly uses ultrasonic detection technology to detect the quality of the weld to ensure that the industrial welding quality meets the requirements; therefore, the ultrasonic weld detection data has great sharing value, which is an effective way for users to understand product quality, and is also a powerful reference for other producers and managers to improve production process, summarize technical problems and eliminate potential risks; therefore, the quality of ultrasonic weld detection data is an important factor to ensure its sharing value. However, the data may be disturbed by various noises during the collection process, such as environmental noise, weld surface stray noise, axial and radial scattering noise, and sensor noise, which will result in low quality of the detection data. Therefore, the main purpose of the present application is to improve the denoising precision of the data, thereby ensuring the sharing value of the data.

[0022] Firstly, the ultrasonic detection device with a transverse wave oblique probe is used to detect the weld based on a preset sampling frequency, and the ultrasonic signal data of each measuring point in the detection process is collected as the original ultrasonic data of each measuring point. The horizontal axis of the original ultrasonic data is time, and the vertical axis is sound pressure or sound intensity. It should be noted that the preset sampling frequency in the embodiment of the present application is set to 50 kHz, and the specific sampling frequency can be adjusted according to the implementation scene, which is not limited herein. At the same time, the specific sampling device and equipment implementer can also adjust according to the implementation scene, which is not limited herein.

[0023] Step S2: dividing the signal segment according to the extreme point in the original ultrasonic data; obtaining the mapping value corresponding to the signal segment according to the amplitude of the extreme point in the original ultrasonic data, the length of the signal segment and the frequency of the length; and determining the target data according to the mapping value.

[0024] Since the amplitude of the sound pressure of the original ultrasonic data obtained in the ultrasonic weld seam detection process will be distorted when it is interfered by noise, the weld seam structure information fed back will be wrong, in order to avoid the indiscriminate smoothing problem of the traditional filtering algorithm, the ITD decomposition algorithm is used to separate the signal components of the signal, and then the noise is removed.

[0025] The ITD decomposition algorithm is a signal processing algorithm for analyzing nonlinear and non-stationary time series data, and the decomposition process introduces the estimation of the local time scale, so that the signal decomposition process becomes a local decomposition, which is used to improve the mode aliasing problem that may occur when the traditional empirical mode decomposition algorithm processes some special signals, but the ITD still cannot avoid the end effect problem, and since the estimation of the intrinsic time scale is based on adjacent extreme points, the influence of the end effect will be amplified.

[0026] Based on the above analysis, the embodiment of the present application first divides the signal segment according to the extreme point in the original ultrasonic data, so as to analyze the local data.

[0027] Preferably, in an embodiment of the present application, the signal segment is divided according to the extreme point in the original ultrasonic data, comprising: In order to more accurately analyze the characteristics of the local data, the data between the two adjacent extreme points is taken as a signal segment. It should be noted that the method for obtaining the extreme point can use the derivative method, the peak detection method, etc., which is not limited and described here.

[0028] Since the ITD is a local decomposition algorithm, the relatively flat signal segment can be regarded as not having locality, and the signal segment with slow change is an abnormal signal for the noise signal. These signal segments will cause the predicted data to tend to be smooth when obtaining the predicted data, so that it is difficult to obtain accurate end point prediction value. Therefore, the mapping value corresponding to the signal segment can be obtained according to the amplitude of the extreme point in the original ultrasonic data, the length of the signal segment and the frequency of the length, so as to screen and process the signal segment according to the mapping value, and obtain the target data.

[0029] Preferably, in an embodiment of the present application, the method for obtaining the mapping value comprises: Firstly, the signal segments are classified, and the specific classification method is to regard the signal segments with the same length as the same kind of signal segment; and then the frequency of the length of each kind of signal segment in all lengths is obtained, and the calculation method of the frequency is the ratio of the frequency of the length of each kind of signal segment and the total number.

[0030] Then, the average amplitude of the left extreme point and the average amplitude of the right extreme point are obtained for each type of signal segment. The ratio of the difference between the average amplitude of the left extreme point and the average amplitude of the right extreme point to the length of the signal segment is used as the average slope value of the corresponding signal segment. Taking the method of obtaining the average amplitude of the left extreme point of each type of signal segment as an example, the amplitudes of the left extreme points of all signal segments in each segment are summed and the average value is obtained as the average amplitude of the left extreme point of that type of signal segment.

[0031] Then, the average slope value corresponding to each signal segment is normalized and used as the adjustment factor for the corresponding signal segment. Finally, all types of signal segments are sorted in descending order based on their length to obtain a sorted sequence. Any signal segment in this sorted sequence is taken as the signal segment to be analyzed. The sum of the adjustment factors corresponding to the signal segment to be analyzed and subsequent signal segments in the sorted sequence is taken as the accumulated feature value of the signal segment to be analyzed. The product of the accumulated feature value, the frequency of occurrence of the signal segment to be analyzed, and a preset constant is rounded down to obtain the mapping value of the signal segment to be analyzed. The formula model for the mapping value can be as follows: in, This represents the mapping value of the signal segment to be analyzed. This indicates the sort number of the signal segment to be analyzed in the sorting sequence. Indicates the number of signal segment types. Indicates the first position in the sorted sequence The average amplitude of the left extreme point of the signal segment. Indicates the first position in the sorted sequence The average amplitude of the right extreme point of the signal segment. Indicates the first The length of the signal segment, This represents a preset constant. Indicates the first The frequency of occurrence of this type of signal segment This represents the rounding function. This represents the normalization function.

[0032] In the formula model of the mapped value This represents the average slope value of the signal segment. The average slope value reflects the changes and trends of the data within that signal segment. The smaller the average slope value, the smoother the data within that signal segment. It is normalized and used as the adjustment factor corresponding to that signal segment. At this point, the smoother the data within a signal segment, the smaller the corresponding adjustment factor. The amplitude-frequency product is obtained by multiplying the adjustment factor, the corresponding frequency of occurrence, and the preset constant for each signal segment. ; then the signal segment to be analyzed and the amplitude-frequency product corresponding to the subsequent signal segment in the sorting sequence are accumulated to obtain an accumulated characteristic value, and the accumulated characteristic value is rounded to obtain a mapping value. It should be noted that in the embodiment of the present application, the preset constant is set to 50, and the specific value can be adjusted according to the implementation scene, which is not limited here.

[0033] Then the signal segment is screened out by the mapping value to determine the target data.

[0034] Preferably, in an embodiment of the present application, the signal segment is screened out according to the mapping value to determine the target data, comprising: The mapping value corresponding to each signal segment in the sorting sequence is compared with the mapping value corresponding to the adjacent next signal segment in turn, and if they are equal, the signal segment is deleted; thereby obtaining the initial data. It should be noted that the basis for deleting the signal segment is that according to the formula model of the mapping value, the accumulated part in the mapping value corresponding to a certain signal segment in the sorting sequence and the mapping value of the adjacent next signal segment should be smaller than the value corresponding to the adjacent next signal segment. However, when the mapping value corresponding to the next signal segment is equal to the mapping value of the current signal segment, it can be explained that the mapping value of the current signal segment occurs in the following two cases during the acquisition process: one is that the corresponding frequency is small, and the other is that the corresponding adjustment factor is small. The smaller the frequency is, the fewer the number of the signal segment is, and the smaller the corresponding adjustment factor is, the more gentle the data change in the signal segment is. Therefore, by comprehensively analyzing the two, the signal segment with fewer distribution quantities and more gentle data in the original ultrasonic data should be deleted.

[0035] Then the initial data is decentered and whitened to obtain the target data. It should be noted that decentering and whitening of the data are both well-known technical means to those skilled in the art, which will not be described here.

[0036] At this point, after deleting the signal segment with gentle data change and fewer distribution quantities in the original ultrasonic data, the prediction accuracy of the subsequent endpoint prediction value can be improved without causing signal distortion.

[0037] Step S3: preset an upper limit of the iteration number, obtain the time scale corresponding to the target data at each iteration according to the iteration number; based on the ITD decomposition algorithm, decompose the target data according to the time scale in each iteration to obtain a preset number of component signals.

[0038] The present application further combines adjacent signal segments in the target data as the time scale, and then sets an upper limit of the iteration number to obtain the decomposition result of the target data in each iteration process through iteration.

[0039] Preferably, the time scale corresponding to each iteration in the process of obtaining the target data in an embodiment of the present application comprises: The number of iterations is taken as the merging number, and adjacent signal segments in the target data are merged according to the merging number, and the length of the merged data segment is taken as the time scale corresponding to the iteration of the target data segment. The method for obtaining the time scale is further described by taking the last two iterations as an example. For example, in the first iteration, the merging number is 1, so the signal segments in the target data do not need to be merged, and thus the time scale at this time is the length of each signal segment in the target data. In the second iteration, the merging number is 2, so two adjacent signal segments in the target data need to be merged, and thus the time scale corresponding to this iteration is the length of each adjacent merged data segment. The method for obtaining the time scale corresponding to each subsequent iteration is the same. It should be noted that in the process of merging the signal segments, if the number of remaining signal segments is not enough for the merging number, the remaining signal segments can be directly merged.

[0040] After obtaining the time scale corresponding to each iteration, the target data can be decomposed according to the corresponding time scale in each iteration based on the ITD decomposition algorithm, and then a preset number of component signals can be obtained. It should be noted that in the embodiment of the present application, the number of component signals is set to 6, the upper limit of the preset number of iterations is set to 10, and the specific values can be adjusted by the implementer according to the implementation scenario, which is not limited herein. The ITD decomposition algorithm is a well-known technical means for those skilled in the art, which is not described herein.

[0041] Step S4: obtaining a constraint value of each iteration according to the difference between the data corresponding to the time scale at each iteration and the difference between the component signals in adjacent iteration numbers; determining a quality scale according to the change of the constraint value in the iteration process; re-sampling the measuring points according to the quality scale to obtain predicted data of the measuring points, and obtaining an endpoint prediction value according to each predicted data.

[0042] In step S3, different numbers of signal segments are merged to obtain the time scale, and then the decomposition result is obtained, so that the constraint value of each iteration can be obtained according to the difference between the data corresponding to all time scales at each iteration and the difference between the component signals in adjacent iteration numbers. The constraint value can reflect the influence of the time scale at each iteration on signal decomposition.

[0043] Preferably, the method for calculating the constraint value in an embodiment of the present application comprises: First, all the time scales in the target data at each iteration are paired two by two, and the combination number is denoted as Then, the mean square error between the data segments corresponding to the two time scales in each combination is analyzed to reflect the difference between the data corresponding to all time scales at each iteration; then the mean square error between the component signals corresponding to adjacent iteration times is obtained to reflect the difference between the component signals at adjacent iteration times; finally, the two are combined to obtain the constraint value.

[0044] wherein, represents the constraint value at the i-th iteration, represents the total number of component signals, represents the i-th component signal at the i-th iteration, represents the i-th component signal at the i-th iteration, represents the i-th component signal at the i-th iteration, represents the mean square error, represents the number of combinations of the two-by-two pairing of all time scales in the target data at each iteration, represents the mean square error between the data segments corresponding to the j-th pair of time scales at the i-th iteration. In the calculation method of the constraint value, when the mean square error between the component signals corresponding to adjacent two iteration times is smaller, the average mean square error will be smaller, which indicates that the time scales corresponding to the i-th iteration are only changed from the time scales corresponding to the i-th iteration, i.e., the resolution is reduced from the combination of k signal segments to the combination of k-1 signal segments; however, since the average mean square error is smaller, it indicates that the final decomposition result does not have decomposition distortion; then represents the mean value of the mean square error between the data segments corresponding to the two-by-two pairing of all time scales at the i-th iteration, and the smaller the mean value, the better the repeatability of the data at the time scales corresponding to the iteration, and the higher the prediction accuracy; therefore, the smaller the final constraint value, the better the decomposition result at the time scales corresponding to the iteration. It should be noted that when , it indicates that at the first time of the current iteration, the constraint value obtained at this time is set to be equal to the constraint value obtained at the i-th iteration, in order to avoid errors when obtaining high-quality scales according to the maximum turning point.

[0045]

[0046] ​​​​​​​​​​​​​​Based on the above method, the constraint value of each iteration can be obtained, and then the better time scale, denoted as the optimal scale, can be determined according to the change of the constraint value in the iteration process.

[0047] Preferably, in an embodiment of the present application, the optimal scale is determined according to the change of the constraint value in the iteration process, comprising: From the first iteration, to each iteration in turn, the constraint values of all iterations are accumulated as the constraint cumulative value of each iteration, and then the constraint curve is obtained according to the constraint cumulative values of all iterations, wherein the horizontal axis of the constraint curve is the iteration number, and the vertical axis is the constraint cumulative value.

[0048] According to the analysis in the calculation method of the constraint value, the smaller the constraint value is, the better the decomposition result under the time scale is, so when the decomposition result under the time scale does not distort in the iteration process, the change rate of the constraint curve will be smaller, and until the decomposition result under the corresponding time scale distorts at a certain iteration, the change rate of the constraint curve will surge, so the maximum inflection point of the constraint curve is obtained based on the elbow method, the iteration number corresponding to the maximum inflection point is taken as the optimal iteration number, and then from the first iteration to the optimal iteration number, the corresponding all time scales are the optimal scales. It should be noted that the elbow method is a well-known technical means for those skilled in the art, and will not be described here.

[0049] After the optimal scale is determined, the data of the measuring point can be obtained again according to the optimal scale, and the endpoint prediction analysis can be performed.

[0050] Preferably, in an embodiment of the present application, the measuring point is resampled according to the optimal scale to obtain the prediction data of the measuring point, and the endpoint prediction value is obtained according to each prediction data, comprising: The product of the reciprocal of the iteration number corresponding to each optimal scale and the preset sampling frequency is taken as the prediction sampling frequency, the measuring point is resampled according to the prediction sampling frequency to obtain a data sequence, and the data sequence is taken as the prediction data corresponding to the optimal scale. Then, the endpoint prediction value corresponding to each optimal scale is obtained based on the polynomial regression model and the prediction direction, wherein when the prediction direction is the same as the time sequence direction, the obtained endpoint prediction value is the right side endpoint prediction value, and when the prediction direction is opposite to the time sequence direction, the obtained endpoint prediction value is the left side endpoint prediction value. It should be noted that in the embodiment of the present application, it is considered that the noise influence caused by environmental change does not change at different sampling times.

[0051] At this point, after the optimal scale is screened out, the endpoint prediction value obtained based on the optimal scale will also be more accurate, so the accuracy will be greatly improved in the subsequent analysis process.

[0052] Step S5: Obtain the final endpoint value based on the amplitude of the extreme points in the target data and the endpoint prediction values ​​corresponding to all high-quality scales; based on the ITD decomposition algorithm, restore, decompose, and reconstruct the original ultrasound data according to the final endpoint value to obtain denoised data for transmission and sharing.

[0053] Based on the steps above, the endpoint prediction values ​​corresponding to all high-quality scales have been obtained. Therefore, the endpoint prediction values ​​corresponding to different high-quality scales can be weighted and combined with the magnitude of the extreme points in the target data to obtain the final endpoint value.

[0054] Preferably, in one embodiment of the present invention, the method for calculating the final endpoint value includes: Obtain the mean and standard deviation of the amplitudes of all extreme points in the target data, and then obtain the final endpoint value based on the endpoint prediction value corresponding to each quality scale and the obtained mean and standard deviation.

[0055] in, Indicates the final endpoint value. This represents the total number of high-quality standards. Indicates the first Endpoint prediction values ​​corresponding to each high-quality scale This represents the mean of the magnitudes of all extreme points in the target data. It represents the standard deviation of the magnitudes of all extreme points in the target data.

[0056] In the formula for calculating the final endpoint value, since the main approach to solving the endpoint effect is to turn the current endpoint into an extreme point, or close to an extreme point, the mean of the amplitudes of all extreme points in the target data obtained in this embodiment of the invention is used to evaluate the prediction effect of the endpoint prediction value. The closer the value is to 1, the more accurate the endpoint prediction. The smaller the value, the higher the weight of the predicted value at that endpoint should be, therefore... Perform negative correlation mapping to correct logical relationships, and then... As endpoint predicted value The weights are multiplied together, and finally the products corresponding to all quality scales are summed and averaged to obtain the final endpoint value. It should be noted that in this embodiment of the invention, only the calculation method of the final endpoint value is given, and the final endpoint values ​​on the left and right sides can be obtained according to this method.

[0057] The final endpoint value combines the endpoint prediction values corresponding to all quality scales, and the endpoint prediction values are weighted and analyzed, so that the prediction accuracy of the final endpoint value can be further improved, that is, the endpoint effect can be better solved; therefore, the denoised data obtained according to the final endpoint value and the transmission and sharing can greatly guarantee the sharing value of the data.

[0058] Preferably, in an embodiment of the present application, the original ultrasonic data is restored, decomposed and reconstructed according to the final endpoint value based on the ITD decomposition algorithm to obtain the denoised data for transmission and sharing, comprising: The final endpoint value is used to replace the endpoint value in the original ultrasonic data, that is, the original ultrasonic data is restored, and the restored original ultrasonic data is used as the to-be-tested data; then the to-be-tested data is decomposed based on the ITD decomposition algorithm and a preset number of component signals are selected for reconstruction to obtain the denoised data; finally, the denoised data is compressed based on the Huffman lossless compression algorithm to obtain the compressed data, and the compressed data is transmitted and shared. It should be noted that the preset number of component signals selected for reconstruction in the embodiment of the present application are the component signals discarded from the first two layers, because the component signals of the first two layers usually contain a large amount of details and noise, and have less influence on the overall characteristics of the signal, so discarding them can reduce the redundant information of the signal and extract more important component signals; the Huffman lossless compression algorithm is a technology known to those skilled in the art, and will not be described here.

[0059] In summary, the embodiment of the present application first obtains the original ultrasonic data of each measuring point during industrial welding detection based on a preset sampling frequency; since there is certain noise interference in the obtained original ultrasonic data, it is necessary to perform denoising processing on the original ultrasonic data, and the ITD decomposition algorithm is used to complete the denoising of the data in the embodiment of the present application. However, since the decomposition process is local decomposition, the end effect problem cannot be avoided, and since the estimation of the intrinsic time scale is based on adjacent extreme points, the influence of the end effect will be amplified. Therefore, the signal segment is divided according to the extreme point in the original ultrasonic data, so as to facilitate the analysis of the data from the local, and further, for the noise signal, the relatively gentle signal belongs to the abnormal signal, so the data in the signal segment is mapped according to the amplitude of the extreme point in the original ultrasonic data, the signal segment length and the occurrence frequency, to obtain the mapping value corresponding to each signal segment, so as to screen the signal segment according to the mapping value, to obtain the target data; then the different numbers of adjacent signal segments in the target data are combined to obtain the time scale, and then the target data is decomposed based on the ITD decomposition algorithm according to the time scale at each iteration to obtain a preset number of component signals; then the constraint value of each iteration is obtained according to the difference between the component signals under adjacent iteration numbers and the difference between the data under the corresponding time scale at each iteration, so as to determine the better time scale, which is referred to as the high-quality scale. Since there is a corresponding time scale at each iteration, the decomposition signal under the iteration number corresponding to the high-quality scale is regarded as a signal without distortion; the measuring point is resampled according to the high-quality scale to obtain the prediction data of the measuring point, and the prediction data at this time can more accurately predict the end point value since it combines the related information of the high-quality scale; finally, the end point prediction values corresponding to all high-quality scales are analyzed by weighting, and the amplitude of the extreme point in the target data is combined to obtain the final end point value; further, the final end point value is substituted for the end point value in the original ultrasonic data, so that the original ultrasonic data can be restored, and the end point value at this time can become an extreme point or be closer to the extreme point, so that the end effect can be eliminated. Therefore, after the data restored at this time is decomposed and reconstructed based on the ITD decomposition algorithm, a more accurate denoising effect can be obtained, and high-quality denoising data can be obtained; finally, the transmission and sharing of the denoising data can greatly guarantee the sharing value of the data.

[0060] An inspection data sharing method based on an industrial industry cloud The current industrial welding process detection mostly uses ultrasonic detection technology to detect the quality of the weld, and defect identification and detection are performed according to the ultrasonic scanning result to ensure that the welding quality meets the requirements. However, there is a large noise interference in the ultrasonic signal of the weld detection, and the intrinsic time scale decomposition algorithm (Intrinsic Time-Scale Decomposition, ITD) is usually used for denoising. However, the ITD decomposition algorithm has an endpoint effect problem, which affects the denoising effect. The existing technology usually uses the periodic extension method to solve the endpoint effect problem, but this will cause the decomposition result to be distorted, thereby causing the denoising effect to be low, and ultimately affecting the sharing value of the data. Therefore, the embodiment of the present application provides a test data sharing method based on an industrial industry cloud, comprising: Step S1: obtaining original ultrasonic data of each measurement point in the weld detection process based on a preset sampling frequency; Step S2: dividing a signal segment according to an extreme point in the original ultrasonic data; obtaining a mapping value corresponding to the signal segment according to the amplitude of the extreme point in the original ultrasonic data, the length of the signal segment, and the frequency of the length; and determining target data according to the mapping value; Step S3: presetting an upper limit of the number of iterations, obtaining a time scale corresponding to the target data at each iteration according to the number of iterations; and decomposing the target data according to the time scale at each iteration based on the ITD decomposition algorithm to obtain a preset number of component signals; Step S4: obtaining a constraint value of each iteration according to the difference between the data corresponding to the time scale at each iteration and the difference between the component signals in adjacent iteration numbers; determining a good scale according to the change of the constraint value in the iteration process; re-sampling the measurement points according to the good scale to obtain predicted data of the measurement points; and obtaining an endpoint prediction value according to each predicted data.

[0061] Among them, steps S1-S4 have been described in detail in the above embodiment of the test data sharing method based on the industrial industry cloud, and will not be repeated here.

[0062] Step S5: obtaining a final endpoint value according to the amplitude of the extreme point in the target data and the endpoint prediction value corresponding to all good scales; and restoring, decomposing, and reconstructing the original ultrasonic data according to the final endpoint value based on the ITD decomposition algorithm to obtain denoised data.

[0063] Based on the above steps, the endpoint prediction value corresponding to all good scales has been obtained, so the endpoint prediction values corresponding to different good scales can be weighted, and the final endpoint value can be obtained in combination with the amplitude of the extreme point in the target data.

[0064] Preferably, the calculation method of the final endpoint value in one embodiment of the present application comprises: Obtain the mean and standard deviation of the amplitudes of all extreme points in the target data, and then obtain the final endpoint value based on the endpoint prediction value corresponding to each quality scale and the obtained mean and standard deviation.

[0065] in, Indicates the final endpoint value. This represents the total number of high-quality standards. Indicates the first Endpoint prediction values ​​corresponding to each high-quality scale This represents the mean of the magnitudes of all extreme points in the target data. It represents the standard deviation of the magnitudes of all extreme points in the target data.

[0066] In the formula for calculating the final endpoint value, since the main approach to solving the endpoint effect is to turn the current endpoint into an extreme point, or close to an extreme point, the mean of the amplitudes of all extreme points in the target data obtained in this embodiment of the invention is used to evaluate the prediction effect of the endpoint prediction value. The closer the value is to 1, the more accurate the endpoint prediction. The smaller the value, the higher the weight of the predicted value at that endpoint should be, therefore... Perform negative correlation mapping to correct logical relationships, and then... As endpoint predicted value The weights are multiplied together, and finally the products corresponding to all quality scales are summed and averaged to obtain the final endpoint value. It should be noted that in this embodiment of the invention, only the calculation method of the final endpoint value is given, and the final endpoint values ​​on the left and right sides can be obtained according to this method.

[0067] The final endpoint value combines the endpoint predictions corresponding to all high-quality scales and performs weighted analysis on the endpoint predictions. Therefore, the prediction accuracy of the final endpoint value can be further improved, which means that the endpoint effect can be better resolved. Thus, the denoised data obtained from the final endpoint value is high-quality denoised data.

[0068] Preferably, in one embodiment of the present invention, the original ultrasound data is restored, decomposed, and reconstructed based on the final endpoint value using the ITD decomposition algorithm to obtain denoised data, including: The final endpoint value is substituted for the endpoint value in the original ultrasonic data, which is regarded as restoring the original ultrasonic data, and the restored original ultrasonic data is taken as the to-be-tested data; then the to-be-tested data is decomposed based on the ITD decomposition algorithm, and a preset number of component signals are selected for reconstruction to obtain denoised data. It should be noted that the preset number of component signals selected for reconstruction in the embodiment of the application are the component signals of the first two layers, and the reason is that the component signals of the first two layers usually contain a large amount of details and noise, and have less influence on the overall characteristics of the signal, so discarding them can reduce the redundant information of the signal and extract more important component signals.

[0069] At this point, high-quality denoised data is obtained.

[0070] The beneficial effects brought by the embodiment include: The application mainly aims at the problems that the signal data obtained in the welding seam detection is easily disturbed by noise, and the ITD decomposition algorithm has an endpoint effect problem, resulting in low denoising effect. First, the original ultrasonic data of each measuring point in the welding seam detection process is obtained. Since the ITD decomposition algorithm is a local decomposition algorithm, and for noise signals, the relatively flat signal is an abnormal signal, the original ultrasonic data is segmented, and the mapping value corresponding to the signal segment length is obtained according to the amplitude of the extreme point, the signal segment length and the occurrence frequency, and then the signal segment is screened out according to the mapping value to determine the target data. At this time, the target data can effectively improve the subsequent further, different time scales are obtained through different iteration numbers, and the ITD decomposition results under different time scales are obtained. Then, the difference between the component signals under adjacent iteration numbers in the iteration process and the difference between the data corresponding to the time scale under each iteration number are evaluated for the decomposition result of each iteration to obtain a constraint value. Thus, a better time scale is determined as a high-quality scale according to the constraint value. Further, the measuring point is resampled using the high-quality scale to obtain predicted data, and an endpoint prediction value is obtained based on the predicted data. Finally, the adaptive acquisition of the final endpoint value is performed according to the endpoint prediction value of the high-quality scale and the amplitude of the extreme point in the target data, effectively solving the interference caused by the endpoint effect. The original ultrasonic data is restored, decomposed and reconstructed based on the ITD decomposition algorithm and the final endpoint value, and high-quality denoised data is obtained.

[0071] It should be noted that the above-mentioned embodiments of the application are in the order of description only, and do not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0072] The various embodiments described in this specification are presented by way of example, and each embodiment is not inherently more important than any other embodiment.

Claims

1. A method for sharing inspection data based on an industrial industry cloud, characterized by, The method comprises: obtaining original ultrasonic data of each measuring point in a weld detection process based on a preset sampling frequency; dividing a signal segment according to an extreme point in the original ultrasonic data; obtaining a mapping value corresponding to the signal segment according to an amplitude of the extreme point in the original ultrasonic data, a length of the signal segment and a frequency of the length; determining target data according to the mapping value by screening the signal segment; obtaining a time scale corresponding to the target data at each iteration according to an iteration number, decomposing the target data according to the time scale in each iteration based on an ITD decomposition algorithm, and obtaining a preset number of component signals; obtaining a constraint value of each iteration according to a difference between data corresponding to the time scale at each iteration and a difference between component signals in adjacent iteration numbers; determining a high-quality scale according to a change of the constraint value in the iteration process; re-sampling the measuring points according to the high-quality scale to obtain predicted data of the measuring points, and obtaining an endpoint prediction value according to each of the predicted data; obtaining a final endpoint value according to the amplitude of the extreme point in the target data and the endpoint prediction value corresponding to all high-quality scales; restoring, decomposing and reconstructing the original ultrasonic data according to the final endpoint value based on the ITD decomposition algorithm to obtain denoising data for transmission and sharing.

2. The industrial industry cloud based inspection data sharing method of claim 1, wherein, The method of dividing a signal segment according to an extreme point in the original ultrasonic data comprises: regarding data between two adjacent extreme points as a signal segment. 3.The industrial industry cloud-based inspection data sharing method of claim 2, wherein, The method of obtaining the mapping value comprises: regarding signal segments of the same length as the same kind of signal segment; obtaining a frequency of the length of each kind of signal segment in all lengths; respectively obtaining an average value of the amplitude of the left extreme point and an average value of the amplitude of the right extreme point of each kind of signal segment; regarding a difference between the average value of the amplitude of the left extreme point and the average value of the amplitude of the right extreme point and the length of the signal segment as an average slope value corresponding to the signal segment; normalizing the average slope value corresponding to each kind of signal segment to obtain an adjustment factor; multiplying the adjustment factor, the corresponding frequency of occurrence and a preset constant of each kind of signal segment to obtain an amplitude-frequency product corresponding to the signal segment; sequentially sorting all kinds of signal segments in descending order based on the length of the signal segment to obtain a sorting sequence; regarding any kind of signal segment in the sorting sequence as an analyzed signal segment; regarding a sum value of the amplitude-frequency products corresponding to the analyzed signal segment and subsequent signal segments in the sorting sequence as an accumulated characteristic value; rounding off the accumulated characteristic value to obtain a mapping value of the analyzed signal segment.

4. The industrial industry cloud based inspection data sharing method of claim 3, wherein, The method of determining target data by screening the signal segment according to the mapping value comprises: comparing the mapping value corresponding to each kind of signal segment in the sorting sequence with the mapping value corresponding to the adjacent next kind of signal segment in sequence, and if they are equal, deleting the signal segment; obtaining initial data; performing decentralization and whitening processing on the initial data to obtain target data.

5. The industrial industry cloud based inspection data sharing method as claimed in claim 1, wherein, The method of obtaining a time scale corresponding to the target data at each iteration according to an iteration number comprises: The number of iterations of each time is taken as a merging number, adjacent signal segments in the target data are merged according to the merging number, and the length of the merged data segment is taken as a time scale corresponding to iteration of the target data.

6. The industrial industry cloud based inspection data sharing method of claim 1, wherein, The method for calculating the constraint value comprises: ; wherein denotes the constraint value at the i-th iteration, denotes the constraint value at the i-th iteration, denotes the total number of component signals, denotes the i-th component signal at the i-th iteration, denotes the i-th component signal at the i-th iteration, denotes the i-th component signal at the i-th iteration, denotes the i-th component signal at the i-th iteration, denotes the i-th component signal at the i-th iteration, denotes the i-th component signal at the i-th iteration, denotes the mean square error, denotes the number of combinations of all time scales pairwise paired in the target data at each iteration, denotes the mean square error between the data segments corresponding to the i-th pair of time scales at the i-th iteration, denotes the mean square error between the data segments corresponding to the i-th pair of time scales at the i-th iteration, denotes the mean square error between the data segments corresponding to the i-th pair of time scales at the i-th iteration.

7. The industrial industry cloud based inspection data sharing method of claim 1, wherein, The method for determining the high-quality scale according to the change of the constraint value in the iteration process comprises: From the first iteration, the constraint values of all iterations are accumulated as constraint cumulative values of each iteration number. A constraint curve is obtained according to the constraint cumulative values of all iteration numbers, wherein the horizontal axis of the constraint curve is the iteration number, and the vertical axis is the constraint cumulative value. The maximum inflection point of the constraint curve is obtained based on the elbow method, the iteration number corresponding to the maximum inflection point is taken as an optimal iteration number, and all time scales corresponding to the first iteration to the optimal iteration number are taken as high-quality scales.

8. The industrial industry cloud based inspection data sharing method of claim 1, wherein, The method for resampling the measuring points according to the high-quality scales to obtain predicted data of the measuring points, and obtaining endpoint prediction values according to each predicted data comprises: The product of the reciprocal of the iteration number corresponding to each high-quality scale and the preset sampling frequency is taken as a predicted sampling frequency, the measuring points are resampled according to the predicted sampling frequency to obtain a data sequence, and the data sequence is taken as predicted data corresponding to the high-quality scale. The endpoint prediction values corresponding to each high-quality scale are obtained based on a polynomial regression model and a prediction direction, wherein when the prediction direction is the same as the time sequence direction, the obtained endpoint prediction value is a right-side endpoint prediction value, and when the prediction direction is opposite to the time sequence direction, the obtained endpoint prediction value is a left-side endpoint prediction value. 9.The industrial industry cloud-based inspection data sharing method of claim 1, wherein, The method for calculating the final endpoint value comprises: ; wherein, represents the final endpoint value, represents the total number of quality scales, represents the endpoint prediction value corresponding to the th quality scale, represents the mean of the amplitudes of all extreme points in the target data, represents the standard deviation of the amplitudes of all extreme points in the target data.

10. The industrial industry cloud based inspection data sharing method as claimed in claim 1 wherein, The method for restoring, decomposing, and reconstructing the original ultrasonic data according to the final endpoint value based on the ITD decomposition algorithm to obtain denoising data for transmission and sharing comprises: The final endpoint value is used to replace the endpoint value in the original ultrasonic data to restore the original ultrasonic data and obtain to-be-measured data. The to-be-measured data is decomposed based on the ITD decomposition algorithm, and a preset number of component signals are reconstructed to obtain denoising data. The denoising data is compressed based on the Huffman lossless compression algorithm to obtain compressed data. The compressed data is transmitted and shared.

Images