A multi-frequency array eddy current testing device and method thereof

Abstract

A multi-frequency array eddy current detection device and method, belonging to the technical field of eddy current detection, which detects and extracts outliers in the extracted phase difference based on the fluctuation intensity and fluctuation range of the phase difference, and calibrates the outliers, which can effectively prevent the defect that the phase difference detection is inaccurate due to the interference of external noise changes; according to the relationship between the eddy current detection value and the phase difference of the horizontal pipeline, according to the similarity of the eddy current detection conditions of the corresponding numerical clusters of each previous numerical point and the eddy current detection conditions of the numerical cluster where the outlier is located, and the action amplitude of each numerical point in the numerical cluster on the outlier, the previous numerical value similar to the outlier is detected, and the outlier is calibrated, which can improve the accuracy of the phase difference, and thus improve the calculation of the pipeline wall thickness based on the phase difference, until the accuracy of detecting local micro-crack defects of the pipeline is achieved.

Description

Technical Field

[0001] The present invention belongs to the technical field of eddy current testing, and particularly relates to a multi-frequency array eddy current testing device and method. Background Art

[0002] The multi-frequency array eddy current testing technology adopts a multi-channel testing method, which can detect multiple defects simultaneously, greatly improving the testing efficiency. At the same time, its testing results are highly accurate and can detect smaller defects.

[0003] In practical applications, the current multi-frequency array eddy current testing technology often uses the existing technical solution of the master's degree thesis titled "Far-field Unit of Multi-frequency Eddy Current Casing Damage Detector and Its Data Processing Design" by Lu Xiangyu and supervised by Shi Yibing, which was publicly disclosed by the University of Electronic Science and Technology in June 2016. It mainly consists of 12 groups of sensors connected to a single-chip microcomputer distributed circumferentially along the inner wall of the transverse pipeline. The 12 groups of sensors are used to pick up the weak far-field induction signals, digitize them and then transmit them to the single-chip microcomputer. The single-chip microcomputer is used to extract the phase difference based on the digitized weak far-field induction signals, and then perform the operation of the pipeline wall thickness based on the phase difference until the detection of local micro-crack defects in the pipeline is achieved.

[0004] However, during the transmission of the digitized weak far-field induction signals to the single-chip microcomputer, they are often affected by external noise interference. For example, as the intensity of the external magnetic field interference increases, some of the extracted phase differences become inaccurate, which is not conducive to performing the operation of the pipeline wall thickness based on the phase difference and thus affects the accuracy of detecting local micro-crack defects in the pipeline. Summary of the Invention

[0005] To solve the defects existing in the prior art, the present invention proposes a multi-frequency array eddy current testing device and method. According to the fluctuation intensity and fluctuation range of the phase difference, the outliers in the extracted phase difference are detected, and the outliers are calibrated, which can effectively prevent the defect that the phase difference detection is inaccurate due to the change and interference of external noise. According to the relationship between the eddy current detection value and the phase difference of the transverse pipeline, based on the similarity between the eddy current detection conditions of the corresponding numerical clusters of each previous numerical point and the eddy current detection conditions of the numerical cluster where the outlier is located, and the action amplitude of each numerical point in the numerical cluster on the outlier, the previous numerical value similar to the outlier is detected, and the outlier is calibrated, which can improve the accuracy of the phase difference, and thus improve the accuracy of performing the operation of the pipeline wall thickness based on the phase difference until the detection of local micro-crack defects in the pipeline is achieved.

[0006] The present invention adopts the following technical solutions.

[0007] A multi-frequency array eddy current testing method, comprising:

[0008] 12 groups of sensors pick up weak far-field induction signals, digitize them and transmit them to the single-chip microcomputer. The single-chip microcomputer extracts the phase difference based on the digitized weak far-field induction signals for processing, and then calculates the pipeline wall thickness based on the processed phase difference;

[0009] The method for the single-chip microcomputer to extract the phase difference based on the digitized weak far-field induction signals for processing includes:

[0010] Step1, obtain the phase differences of multiple detection points of the horizontal pipeline and the eddy current detection values corresponding to the phase differences at the same time point;

[0011] Step2, obtain the near-region values of the phase difference to form a near-region value cluster of the phase difference, analyze the value change of the near-region value cluster of any phase difference, obtain the outlier amplitude of the phase difference, select the outlier amplitude of the value to be analyzed, and obtain the outlier value;

[0012] Step3, according to the difference between the outlier value and the eddy current detection values of the previous period in the predefined time period until the actual value of the outlier value is obtained;

[0013] Step4, replace all the outlier values of the horizontal pipeline with the actual values to obtain the phase difference of the horizontal pipeline after replacement, and the phase difference after replacement is the processed phase difference.

[0014] Further, in Step1, obtain the phase differences of multiple detection points of the horizontal pipeline, define the phase differences in the latest predefined time period as the values to be analyzed, and define the non-values to be analyzed as the previous period values;

[0015] Obtain the eddy current detection values in the predefined time period before any phase difference The eddy current detection values include the digitized weak far-field induction signals corresponding to the phase difference and the pipeline wall thickness of the horizontal pipeline.

[0016] Further, in Step2, initially take any phase difference and the adjacent phase differences before it, and the adjacent phase differences after it to form a near-region value cluster of the phase difference, which is the predefined index one;

[0017] Then define the range of the near-region value cluster of the phase difference as the value change range of the near-region value cluster of the phase difference;

[0018] Subsequently, obtain the numerical fluctuation amplitude of the near-region value cluster of any phase difference according to the change of adjacent phase differences in the near-region value cluster of the phase difference;

[0019] Obtain the outlier amplitude of the phase difference by combining the numerical variation range of the near-region numerical cluster of the random phase difference and the numerical fluctuation amplitude of the near-region numerical cluster of the phase difference.

[0020] Further, in Step2, the method for obtaining the numerical fluctuation amplitude of the near-region numerical cluster of the random phase difference specifically includes:

[0021] Calculate the following equation:

[0022] ;

[0023] In the equation, represents the numerical fluctuation amplitude of the near-region numerical cluster of the th phase difference, represents the number of phase differences contained in the near-region numerical cluster of the phase difference, represents the th value of the th phase difference in the near-region numerical cluster of the th phase difference, represents the th value of the

[0024] Subsequently, obtain the outlier amplitude of the random phase difference based on the numerical variation range of the near-region numerical cluster of the phase difference and the numerical fluctuation amplitude of the near-region numerical cluster of the phase difference.

[0025] Further, in Step2, the method for obtaining the outlier amplitude of the random phase difference specifically includes:

[0026] Calculate the following equation:

[0027] ;

[0028] In the equation, represents the outlier amplitude of the th phase difference, represents the numerical fluctuation amplitude of the near-region numerical cluster of the th phase difference, represents the numerical variation range of the near-region numerical cluster of the th phase difference, represents performing standardization processing on using the Z-score method;

[0029] Subsequently, select all the numerical values to be analyzed whose outlier amplitude is higher than the predefined index two, and define them as outlier values.

[0030] Further, Step3 specifically includes:

[0031] Step3-1, obtain the initial similarity between the outlier and the previous values based on the difference in eddy current detection values between the outlier and the previous values in a predefined period;

[0032] Step3-2, obtain the time-point effect factor of any phase difference in the near-region value cluster of the outlier based on the time interval of the phase difference within the near-region value cluster of the outlier;

[0033] Step3-3, aggregate the initial similarity between the outlier and the previous values and the time-point effect factor of the phase difference in the near-region value cluster of the outlier to obtain the final similarity between the outlier and the previous values, and select all the similar previous values for the outlier;

[0034] Step3-4, combine the final similarity between the outlier and all the similar previous values, the values of the similar previous values, and the outlier amplitude of the similar previous values to obtain the actual value of the outlier.

[0035] Further, in Step3-1, initially, the eddy current detection values of any phase difference are obtained through Step1, and the eddy current detection values corresponding to each time point in a predefined period before any phase difference are obtained. The eddy current detection values include the digitized far-field weak induction signals corresponding to the detection sites where the phase difference is located in the predefined period before the phase difference and the pipe wall thickness of the transverse pipe;

[0036] Next, for any outlier and any previous value, perform a ∪ operation on the digitized far-field weak induction signals of the outlier and the previous value to obtain all the digitized far-field weak induction signals of the outlier and the previous value;

[0037] Finally, based on the difference in eddy current detection values between the outlier and the previous values in a predefined period, obtain the initial similarity between the outlier and the previous values.

[0038] Further, in Step3-1, the method for obtaining the initial similarity between any outlier and any previous value specifically includes:

[0039] Calculate the following equation:

[0040] ;

[0041] In the equation, represents the initial similarity between the th outlier and the th previous value, represents all the digitized far-field weak induction signals of the th outlier and the th previous value, represents the eddy current detection value of the th outlier at the The wall thickness of the horizontal pipe represents the wall thickness of the horizontal pipe in the eddy current detection values of the th past value; is the predefined index three, representing the execution of standardization processing on using the Z-score method.

[0042] Furthermore, in Step3-2, the method for obtaining the time-point effect factor with an arbitrary phase difference within the near-region numerical cluster of the outlier specifically includes:

[0043] Calculating the following equation:

[0044] ;

[0045] In the equation, represents the time-point effect factor of the th phase difference in the near-region numerical cluster of the th outlier, represents the time interval between the th outlier and the th phase difference in the near-region numerical cluster, is the predefined index three, representing the execution of standardization processing on using the Z-score method.

[0046] Furthermore, in Step3-3, the method for obtaining the final similarity between an arbitrary outlier and an arbitrary past value specifically includes:

[0047] Calculating the following equation:

[0048] ;

[0049] In the equation, represents the final similarity between the th outlier and the th past value; represents the number of phase differences included in the near-region numerical cluster of the phase difference; represents the th starting similarity of the corresponding phase difference in the near-region numerical cluster of the th outlier and the th past value; represents the th time-point effect factor of the corresponding phase difference in the near-region numerical cluster of the th outlier and the th past value;

[0050] For random outliers, select all previous values within the entire set of previous values that have a similarity to the outlier higher than a pre-defined index four, and define them as the similar previous values of the outlier.

[0051] Furthermore, in Step3 - 4, the method for obtaining the influence factor of the random similar previous value of a random outlier specifically includes:

[0052] Calculate the following equation:

[0053] ;

[0054] In the equation, represents the influence factor of the th similar previous value of the th outlier, represents the outlier amplitude of the th similar previous value of the th outlier, is the pre-defined index three.

[0055] Furthermore, in Step3 - 4, the method for obtaining the actual value of a random outlier specifically includes:

[0056] ;

[0057] In the equation, represents the actual value of the th outlier, represents the number of non-outlier similar previous values of the th outlier, represents the influence factor of the th similar previous value on the th outlier, represents the final similarity between the th outlier and the th similar previous value, represents the value of the th similar previous value of the th outlier, is the pre-defined index three.

[0058] A multi-frequency array eddy current detection device, comprising:

[0059] 12 groups of sensors connected to a single-chip microcomputer and distributed circumferentially along the inner wall of the transverse pipeline;

[0060] 12 groups of sensors are used to pick up far - field weak induction signals, digitize them and then transmit them to a single - chip microcomputer. The single - chip microcomputer is used to extract the phase difference based on the digitized far - field weak induction signals for processing, and then calculate the pipeline wall thickness based on the processed phase difference;

[0061] The modules running on the single - chip microcomputer include:

[0062] A detection module, which is used to obtain the phase differences of multiple detection sites of the horizontal pipeline and the eddy current detection values corresponding to the phase differences at the same time point;

[0063] An outlier module, which is used to obtain the near - zone values of the phase difference to form a near - zone value cluster of the phase difference, analyze the value change situation of the near - zone value cluster of any phase difference, obtain the outlier amplitude of the phase difference, select the outlier amplitude of the value to be analyzed, and obtain the outlier value;

[0064] A similarity module, which is used to obtain the initial similarity between the outlier value and the previous values according to the difference in eddy current detection values between the outlier value and the previous values in a predefined time period; obtain the time - point influence factor of any phase difference in the near - zone value cluster of the outlier value according to the time interval between the outlier value and the phase difference in the near - zone value cluster; aggregate the initial similarity between the outlier value and the previous values, the time - point influence factor of the phase difference in the near - zone value cluster of the outlier value, obtain the final similarity between the outlier value and the previous values, and select all the similar previous values of the outlier value; combine the final similarity between the outlier value and all the similar previous values, the values of the similar previous values and the outlier amplitude of the similar previous values, and obtain the actual value of the outlier value;

[0065] A replacement module, which is used to replace all the outlier values of the horizontal pipeline with the actual values to obtain the phase difference after replacement of the horizontal pipeline. The phase difference after replacement is the processed phase difference.

[0066] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0067] According to the fluctuation intensity and fluctuation range of the phase difference, the outlier values in the extracted phase difference are detected and the outlier values are calibrated, which can effectively prevent the defect of inaccurate phase difference detection caused by external noise changes; according to the relationship between the eddy current detection value and the phase difference of the horizontal pipeline, according to the similarity of the eddy current detection conditions of the value clusters corresponding to each previous value point and the eddy current detection conditions of the value cluster where the outlier value is located, and the influence amplitude of each value point in the value cluster on the outlier value, the previous values similar to the outlier value are detected and the outlier value is calibrated, which can improve the accuracy of the phase difference, and thus improve the calculation of the pipeline wall thickness based on the phase difference, until the accuracy of detecting local micro - crack defects of the pipeline is achieved. Description of the Drawings

[0068] Figure 1is the flowchart of the multi-frequency array eddy current detection method described in the present invention;

[0069] Figure 2 is a partial structural diagram of the multi-frequency array eddy current detection device described in the present invention. Detailed implementation manners

[0070] To make the objectives, technical solutions and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions of the present invention in conjunction with the accompanying drawings in the embodiments of the present invention. The embodiments described herein are only some of the embodiments of the present invention, rather than all of them. According to the spirit of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0071] As Figure 1 shown, a multi-frequency array eddy current detection method described in the present invention includes:

[0072] 12 groups of sensors pick up weak far-field induction signals, digitize them and transmit them to a single-chip microcomputer. The single-chip microcomputer extracts the phase difference from the digitized weak far-field induction signals for processing, and then calculates the pipeline wall thickness based on the processed phase difference until the detection of local small crack defects in the pipeline is completed;

[0073] The method by which the single-chip microcomputer extracts the phase difference from the digitized weak far-field induction signals for processing includes:

[0074] Step1, obtain the phase differences of multiple detection points on the horizontal pipeline and the eddy current detection values corresponding to the phase differences at the same time point;

[0075] The multi-frequency array eddy current detection of the horizontal pipeline involves the movement of 12 groups of sensors on the inner wall of the horizontal pipeline during the test cycle, so that the 12 groups of sensors perform detections at different detection points, that is, pick up weak far-field induction signals at different detection points, digitize them and transmit them to the single-chip microcomputer. Then, the single-chip microcomputer extracts the corresponding phase differences of different detection points from the digitized weak far-field induction signals at different detection points for processing, and then calculates the pipeline wall thickness of the horizontal pipeline at the corresponding different detection points based on the processed phase differences of different detection points.

[0076] In the preferred but non-limiting implementation manner of the present invention, in Step1, to obtain the relationship between the overall phase difference of the horizontal pipeline and the eddy current detection values based on the phase differences of different detection points, first, obtain the phase differences of multiple detection points on the horizontal pipeline and the eddy current detection values corresponding to the phase differences at the same time point. The method is as follows:

[0077] Obtain the phase differences of multiple detection points on the horizontal pipeline, and take the latest predefined time period The phase difference in it is defined as the value to be analyzed, and the value that is not to be analyzed is defined as the previous value;

[0078] Define the time period in advance before obtaining any random phase difference The eddy current detection value in it, the eddy current detection value includes the digitized far-field weak induction signal corresponding to the phase difference and the pipe wall thickness of the transverse pipe. The phase difference, the digitized far-field weak induction signal, and the pipe wall thickness of the transverse pipe are arranged in the order of the sequence in which they are formed.

[0079] In a preferred but non-limiting embodiment of the present invention, in Step1, define the time period in advance and are five seconds and one second respectively.

[0080] Thus, the phase differences of multiple detection sites of the transverse pipe and the eddy current detection values corresponding to the phase differences at the same time point are obtained through the above method.

[0081] Step2, obtain the near-region values of the phase difference to form a cluster of near-region values of the phase difference, analyze the value change situation of the cluster of near-region values of any random phase difference, obtain the outlier amplitude of the phase difference, perform selection on the outlier amplitude of the value to be analyzed, and obtain the outlier value;

[0082] Because the change of external noise will cause the digitized far-field weak induction signal to be transmitted to the single-chip microcomputer and the extracted phase difference to fluctuate suddenly within a large range, resulting in a large difference in the extracted phase difference of a transverse pipe. Therefore, according to the action characterization of the external noise change on the phase difference, the outlier value in the extracted phase difference is determined.

[0083] During the detection of multiple detection sites of the transverse pipe, because it will be affected by external noise, the phase difference cannot reflect the actual phase difference of the transverse pipe. Therefore, it is necessary to identify the outlier value affected by the change of external noise in the extracted phase difference.

[0084] When the single-chip microcomputer extracts the phase difference according to the digitized far-field weak induction signal, when the value of the digitized far-field weak induction signal increases, the phase difference will increase. After the increasing stage, the carbon emissions will gradually spread to other places, and the detected phase difference will also gradually decrease. Therefore, the conventional phase difference fluctuates smoothly. When affected by external noise, the intensity of the digitized far-field weak induction signal transmitted to the single-chip microcomputer will be weakened. Therefore, after detecting that the phase difference increases, the extracted phase difference will suddenly decrease due to external noise interference. Therefore, it is possible to analyze whether the phase difference is an outlier (phase difference affected by noise interference) through the value change situation of the phase difference within a period of time.

[0085] In a preferred but non-limiting embodiment of the present invention, in Step 2, initially, a random phase difference, the phase differences adjacent to it in the front, and the phase differences adjacent to it in the back form a numerical cluster of near-region values of the phase difference, which is a predefined index one;

[0086] In a preferred but non-limiting embodiment of the present invention, in Step 2, the predefined index one is five.

[0087] Then, the full range within the numerical cluster of the random phase difference is defined as the numerical variation range of the numerical cluster of the phase difference;

[0088] The numerical variation range of the numerical cluster is the range where the values in the numerical cluster are located, that is, the difference between the highest value and the lowest value in the numerical cluster.

[0089] Subsequently, according to the variation of adjacent phase differences in the numerical cluster of the phase difference, the numerical fluctuation amplitude of the numerical cluster of the random phase difference is obtained;

[0090] The numerical fluctuation amplitude of the numerical cluster represents the variation of all adjacent values. The lower the variation of all adjacent values, the less significant the overall fluctuation of the numerical cluster at this time, and thus the lower the numerical fluctuation amplitude of the numerical cluster.

[0091] Combining the numerical variation range of the numerical cluster of the random phase difference and the numerical fluctuation amplitude of the numerical cluster of the phase difference, the outlier amplitude of the phase difference is obtained.

[0092] In a preferred but non-limiting embodiment of the present invention, in Step 2, the method for obtaining the numerical fluctuation amplitude of the numerical cluster of the random phase difference specifically includes:

[0093] Calculating the following equation:

[0094] ;

[0095] In the equation, represents the numerical fluctuation amplitude of the numerical cluster of the th phase difference, represents the number of phase differences included in the numerical cluster of the phase difference, represents the value of the th phase difference in the numerical cluster of the th phase difference, represents the value of the th phase difference in the numerical cluster of the th phase difference;

[0096] Subsequently, according to the numerical variation range of the near-region numerical cluster of the phase difference and the numerical fluctuation amplitude of the near-region numerical cluster of the phase difference, the outlier amplitude of any phase difference is obtained.

[0097] In a certain period, the higher the continuous variation of the phase difference, the higher the numerical fluctuation amplitude of the phase difference at this time, and the higher the outlier amplitude of the phase difference; the higher the range of the synchronous phase difference, the higher the numerical variation range of the phase difference at this time, and the higher the outlier amplitude of the phase difference.

[0098] In a preferred but non-limiting embodiment of the present invention, in Step2, the method for obtaining the outlier amplitude of any phase difference specifically includes:

[0099] Calculate the following equation:

[0100] ;

[0101] In the equation, represents the outlier amplitude of the th phase difference, represents the numerical fluctuation amplitude of the near-region numerical cluster of the th phase difference, represents the numerical variation range of the near-region numerical cluster of the th phase difference, represents the use of the Z-score method to perform standardization processing;

[0102] Subsequently, select all the numerical values to be analyzed with an outlier amplitude higher than the predefined index two, and define them as outliers.

[0103] In a preferred but non-limiting embodiment of the present invention, in Step2, the predefined index two is one-half.

[0104] Thus, outliers are obtained through the above method.

[0105] Step3, according to the difference in eddy current detection values between the outlier value and the previous period values in a predefined period, until the actual value of the outlier value is obtained;

[0106] Under the same eddy current detection conditions, the phase differences should be similar. Therefore, according to the similarity of the eddy current detection conditions, find the previous period values similar to the outlier value, and calculate the actual phase difference at the corresponding time point of the outlier value based on the previous period values.

[0107] In a preferred but non-limiting embodiment of the present invention, Step3 specifically includes:

[0108] Step3-1, according to the difference in eddy current detection values between the outlier value and the previous period values in a predefined period, obtain the initial similarity between the outlier value and the previous period values;

[0109] The phase difference of the detection site is related to both the corresponding digitized far - field weak induction signal and the pipe wall thickness of the transverse pipe. Therefore, according to the similarity between the past values and the traffic flow condition of the outlier, the past values similar to the conventional phase difference that can reflect the outlier should be found.

[0110] In a preferred but non - restrictive embodiment of the present invention, in Step3 - 1, initially, the eddy current detection values with arbitrary phase differences are obtained through Step1, and the eddy current detection values corresponding to each time point in a predefined period before the arbitrary phase difference are obtained. The eddy current detection values include the corresponding digitized far - field weak induction signal at the detection site where the phase difference is located and the pipe wall thickness of the transverse pipe in the predefined period before the phase difference.

[0111] Then, for any outlier and any past value, a ∪ operation is performed on the digitized far - field weak induction signals of the outlier and the past value to obtain all the digitized far - field weak induction signals of the arbitrary outlier and the arbitrary past value.

[0112] For the same detection site, for any outlier and any past value, the closer the eddy current detection values are in the predefined period, the greater the similarity between the outlier and the past value at this time.

[0113] The similarity between the arbitrary outlier and the arbitrary past value in the predefined period of the eddy current detection values is characterized by: the similarity of the digitized far - field weak induction signal and the similarity of the pipe wall thickness of the transverse pipe. Therefore, for any outlier and any past value, the lower the difference in the pipe wall thickness of the transverse pipe, the greater the initial similarity between the outlier and the past value at this time.

[0114] Finally, based on the difference in the eddy current detection values of the outlier and the past value in the predefined period, the initial similarity between the outlier and the past value is obtained.

[0115] In a preferred but non - restrictive embodiment of the present invention, in Step3 - 1, the method for obtaining the initial similarity between an arbitrary outlier and an arbitrary past value specifically includes:

[0116] Calculate the following equation:

[0117] ;

[0118] In the equation, represents the initial similarity between the th outlier and the th past value, represents all the digitized far - field weak induction signals of the th outlier and the th past value, represents the The wall thickness of the th transverse pipe within the eddy current detection value of an outlier, represents the th wall thickness of the th transverse pipe among the eddy current detection values of previous periods, is the predefined index three, representing the implementation of standardization processing on

[0119] while represents that the closer the wall thicknesses of each transverse pipe within the near - zone numerical cluster of the th outlier and the th previous - period numerical value are to each other, the closer the eddy current detection values of the th outlier and the th previous - period numerical value are. Therefore, the starting similarity degree between the th outlier and the th previous - period numerical value is greater.

[0120] The value of the predefined index three can be one, used to make the divisor in the equation non - zero.

[0121] Step3 - 2: According to the time interval of the phase difference within the outlier and the near - zone numerical cluster, obtain the time - point effect factor of any phase difference in the near - zone numerical cluster of the outlier;

[0122] Since the influence of the phase difference at each time point (the time point is the formation time point of the corresponding phase difference) in the near - zone numerical cluster of the outlier on the outlier is different, the closer the time interval (the time length between time points) of the phase difference is, the higher the amplitude of the influence of the corresponding eddy current detection value on the outlier, and thus the higher the time - point effect factor of the phase difference.

[0123] In a preferred but non - restrictive embodiment of the present invention, in Step3 - 2, the method for obtaining the time - point effect factor of any phase difference in the near - zone numerical cluster of the outlier specifically includes:

[0124] Calculate the following equation:

[0125] ;

[0126] In the equation, represents the time - point effect factor of the th phase difference in the near - zone numerical cluster of the th outlier, represents the time interval between the th outlier and the th phase difference in the near - zone numerical cluster, is the predefined index three, Represent the use of the Z-score method for Perform standardization processing.

[0127] Step3-3: Aggregate the starting proximity between the outlier and the previous period values, and the time point effect factor of the phase difference in the near-region numerical cluster of the outlier, to obtain the final proximity between the outlier and the previous period values, and select all the previous period values with proximity to the obtained outlier;

[0128] For arbitrary outliers and arbitrary previous period values, the final proximity is reflected in: the starting proximity characterized by the proximity of eddy current detection values. The higher the starting proximity of each corresponding phase difference within the near-region numerical cluster of the arbitrary outlier and the arbitrary previous period values, the higher the final proximity at this time; it is also reflected in: the time point effect factor of each corresponding phase difference obtained through time interval. When the starting proximity of each corresponding phase difference is large, the time interval from the outlier time point is not large, so the representation amplitude of the starting proximity of the corresponding phase difference at the time point for the final proximity is higher.

[0129] In a preferred but non-limiting embodiment of the present invention, in Step3-3, the method for obtaining the final proximity between an arbitrary outlier and an arbitrary previous period value specifically includes:

[0130] Calculate the following equation:

[0131] ;

[0132] In the equation, Represents the final proximity between the th outlier and the th previous period value; Represents the number of phase differences contained in the near-region numerical cluster of the phase difference; Represents the th outlier and the th previous period value in the near-region numerical cluster, and the th corresponding phase difference starting proximity; Represents the th outlier and the th previous period value in the near-region numerical cluster, and the th corresponding phase difference time point effect factor.

[0133] And Represents that in the near-region numerical cluster of the th outlier and the th previous period value, the th corresponding phase difference time point effect factor is regarded as the importance, and the starting proximities of all corresponding phase differences in the near-region numerical cluster of the th outlier and the th previous period value are averaged after attaching the importance to form the The last similarity between an outlier and the th previous value.

[0134] For a random outlier, select all previous values within the entire set of previous values whose last similarity to the outlier is higher than a predefined criterion four, and define them as the similar previous values of the outlier.

[0135] Step3-4, Combine the last similarity between the outlier and all similar previous values, the values of the similar previous values, and the outlier amplitude of the similar previous values to obtain the actual value of the outlier.

[0136] If the previous values are subject to significant external noise interference, then the previous values will also be inaccurate. Therefore, according to Step2, calculate the outlier amplitude of each similar previous value of a random outlier. The higher the outlier amplitude of the similar previous value, the lower the amplitude of its effect during the calculation of the actual value of the outlier, that is, the lower the effect factor.

[0137] In a preferred but non-limiting embodiment of the present invention, in Step3-4, the method for obtaining the effect factor of any similar previous value of a random outlier specifically includes:

[0138] Calculate the following equation:

[0139] ;

[0140] In the equation, represents the effect factor of the th similar previous value of the th outlier, represents the outlier amplitude of the th similar previous value of the th outlier, is a predefined criterion three.

[0141] Among all the similar previous values of a random outlier, the outlier should be removed at the beginning because the outlier similar previous values will interfere with the actual value of the outlier.

[0142] Select the non-outlier similar previous values within all the similar previous values according to Step2.

[0143] The actual value of the outlier is obtained by using the values of the adjacent past periods that are close. Thus, for any adjacent past period value, the higher the similarity amplitude with the outlier, the greater the amplitude by which the adjacent past period value can reflect the actual value of the outlier, and the higher the formation ratio of the actual value of the outlier. Additionally, the adjacent past period values should also reflect the phase difference situation in this regard. That is, the lower the outlier amplitude of the adjacent past period value, the greater the effect of the adjacent past period value on the actual value of the outlier, and the higher the formation ratio of the actual value of the outlier. Therefore, the actual phase difference of the outlier can be calculated based on the values of all non-outlier adjacent past period values, the similarity amplitude of the outlier, and the influence factor of the adjacent past period value.

[0144] In a preferred but non-limiting embodiment of the present invention, in Step 3-4, the method for obtaining the actual value of any outlier specifically includes:

[0145] ;

[0146] In the equation, represents the actual value of the th outlier, represents the number of non-outlier adjacent past period values of the th outlier, represents the influence factor of the th adjacent past period value on the th outlier, represents the final similarity degree between the th outlier and the th adjacent past period value, represents the value of the th adjacent past period value of the th outlier, is a predefined index three.

[0147] And represents the average value obtained by attaching importance weights to the values of all adjacent past period values of the th outlier using the influence factor and similarity amplitude as importance weights, and then calculating the mean of all non-outlier adjacent past period values, forming the actual value of the th outlier. The horizontal pipe is the casing.

[0148] Thus, the actual value of any outlier is obtained through the above method.

[0149] Step 4, replace all outliers of the horizontal pipe with the actual values to obtain the phase difference after replacement of the horizontal pipe, and the phase difference after replacement is the processed phase difference.

[0150] Through the above method, the multi-frequency array eddy current detection method is achieved.

[0151] As Figure 2 shown, a multi-frequency array eddy current detection device shown in the present invention includes:

[0152] 12 groups of sensors connected to a single-chip microcomputer and distributed circumferentially along the inner wall of the transverse pipeline;

[0153] The 12 groups of sensors are used to pick up weak far-field induction signals, digitize them and then transmit them to the single-chip microcomputer. The single-chip microcomputer is used to extract the phase difference according to the digitized weak far-field induction signals for processing, and then calculate the pipeline wall thickness according to the processed phase difference until the detection of local micro-crack defects in the pipeline is completed;

[0154] The modules running on the single-chip microcomputer include:

[0155] A detection module, which is used to obtain the phase differences of multiple detection points of the transverse pipeline and the eddy current detection values corresponding to the phase differences at the same time point;

[0156] An outlier module, which is used to obtain the near-region values of the phase differences to form a near-region value cluster of the phase differences, analyze the value change situation of the near-region value cluster of any phase difference, obtain the outlier amplitude of the phase difference, select the outlier amplitude of the value to be analyzed, and obtain the outlier value;

[0157] A similarity module, which is used to obtain the initial similarity between the outlier value and the previous value according to the difference between the outlier value and the eddy current detection value of the previous value in a predefined period; obtain the time-point effect factor of any phase difference in the near-region value cluster of the outlier value according to the time interval between the outlier value and the phase differences in the near-region value cluster; aggregate the initial similarity between the outlier value and the previous value and the time-point effect factor of the phase differences in the near-region value cluster of the outlier value to obtain the final similarity between the outlier value and the previous value, and select all the similar previous values of the outlier value; combine the final similarity between the outlier value and all the similar previous values, the values of the similar previous values and the outlier amplitudes of the similar previous values to obtain the actual value of the outlier value;

[0158] A replacement module, which is used to replace all the outlier values of the transverse pipeline with the actual values to obtain the phase differences after replacement of the transverse pipeline, and the phase differences after replacement are the processed phase differences.

[0159] The beneficial effects of the present invention are that, compared with the prior art, the technical effects of the present invention include:

[0160] According to the fluctuation intensity and fluctuation range of the phase difference, outliers in the extracted phase difference are detected, and the outliers are calibrated, which can effectively prevent the defect of inaccurate phase difference detection caused by the interference of external noise changes; according to the relationship between the eddy current detection value and the phase difference of the horizontal pipeline, and according to the similarity between the eddy current detection conditions of the numerical clusters corresponding to each previous numerical point and the eddy current detection conditions of the numerical cluster where the outlier is located, and the action amplitude of each numerical point in the numerical cluster on the outlier, previous numerical values similar to the outlier are detected, and the outlier is calibrated, which can improve the accuracy of the phase difference, and thus improve the calculation of the pipeline wall thickness based on the phase difference, until the accuracy of detecting local micro-crack defects in the pipeline is achieved.

[0161] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that: modifications or equivalent replacements can still be made to the specific implementation manners of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention shall be covered by the protection scope of the claims of the present invention.

Claims

1. A multi-frequency array eddy current detection method, characterized in that: include: 12 groups of sensors pick up far-field weak induction signals and transmit them to the single-chip microcomputer after digitization. The single-chip microcomputer extracts the phase difference according to the digitized far-field weak induction signals and performs processing, and then calculates the pipe wall thickness according to the processed phase difference; The method of performing processing by extracting the phase difference based on the digitized far-field weak induction signal by the single chip microcomputer includes: Step 1, obtain the phase difference of multiple detection positions of the transverse pipeline and the eddy current detection value corresponding to the phase difference at the same time point; Step 2, obtain the near-zone value of the phase difference to form a near-zone value cluster of the phase difference, analyze the value change of the near-zone value cluster of the random phase difference, obtain the outlier amplitude of the phase difference, select the outlier amplitude of the value to be analyzed, and obtain the outlier value; Step 3, according to the difference of eddy current detection values ​​between the outlier and the previous values ​​in a predefined time period, obtain the initial similarity between the outlier and the previous values; according to the time distance of the phase difference within the outlier and the near-zone value cluster, obtain the time point effect factor of the random phase difference in the near-zone value cluster of the outlier; concentrate the initial similarity between the outlier and the previous values, the time point effect factor of the phase difference in the near-zone value cluster of the outlier, obtain the final similarity between the outlier and the previous values, and select all the past values ​​close to the outlier; combine the final similarity between the outlier and all the past values ​​close to each other, the value of the past values ​​close to each other and the outlier amplitude of the past values ​​close to each other to obtain the actual value of the outlier; Step 4, replace all outliers of the lateral pipeline with actual values, and obtain the phase difference of the lateral pipeline after replacement. The phase difference after replacement is the phase difference after treatment; In Step 2, start by placing the random phase difference and the phase difference in the previous adjacent Phase difference, phase difference is next to phase differences, forming a cluster of near-region values ​​of phase differences, It is a pre-defined indicator one; Then, the full range within the near-zone value cluster of the random phase difference is defined as the value variation range of the near-zone value cluster of the phase difference; Then, the value fluctuation amplitude of the near-zone value cluster of the random phase difference is obtained according to the change of the adjacent phase difference in the near-zone value cluster of the phase difference; The outlier amplitude of the phase difference is obtained by combining the numerical variation range of the nearby numerical clusters of the random phase difference and the numerical fluctuation amplitude of the nearby numerical clusters of the phase difference.

2. The multi-frequency array eddy current detection method according to claim 1, characterized in that: In Step 1, the phase difference of multiple detection points of the horizontal pipeline is obtained, and the latest pre-defined period is The phase difference in is defined as the value to be resolved, and the value that must be resolved is defined as the previous value; Define the time period before obtaining the random phase difference The eddy current detection value in the eddy current detection value includes the digitized far-field weak induction signal corresponding to the phase difference and the pipe wall thickness of the transverse pipe.

3. The multi-frequency array eddy current detection method according to claim 2, characterized in that: In Step 2, the method for obtaining the numerical fluctuation amplitude of the near-zone numerical cluster of arbitrary phase difference specifically includes: Operate the following equation: ； In the equation, Representative The numerical fluctuation amplitude of the near-zone numerical cluster with phase difference, The number of phase differences contained in the near-zone value cluster representing the phase difference, Representative The phase difference of the near-zone value cluster The phase difference value, Representative The phase difference of the near-zone value cluster The value of the phase difference; Then, according to the value variation range of the nearby value cluster of the phase difference and the value fluctuation amplitude of the nearby value cluster of the phase difference, the outlier amplitude of the random phase difference is obtained.

4. The multi-frequency array eddy current detection method according to claim 3, characterized in that: In Step 2, the method for obtaining the outlier amplitude of the random phase difference specifically includes: Operate the following equation: ； In the equation, Representative The outlier amplitude of the phase difference, Representative The numerical fluctuation amplitude of the near-zone numerical cluster with phase difference, Representative The range of value variation of the near-zone value cluster of phase differences, Represents the use of Z-score method to Implement standardized disposal; Then, the values ​​to be analyzed whose outlier amplitudes are higher than the pre-defined index 2 among all the values ​​to be analyzed are selected and defined as outliers.

5. The multi-frequency array eddy current detection method according to claim 4, characterized in that: In Step 3-1, initially, the eddy current detection value of the random phase difference is obtained through Step 1, and the corresponding eddy current detection value at each time point in the predefined period before the random phase difference is obtained, and the eddy current detection value includes the corresponding digitized far-field weak induction signal of the detection position where the phase difference is located in the predefined period before the phase difference and the pipe wall thickness of the transverse pipe; Next, a ∪ operation is performed on the random outliers and random past values, and on the digitized far-field weak sensing signals of the outliers and the past values, to obtain all digitized far-field weak sensing signals of the random outliers and the random past values; Finally, the initial similarity between the outlier value and the past value is obtained based on the difference in eddy current detection values ​​in a pre-defined time period between the outlier value and the past value.

6. The multi-frequency array eddy current detection method according to claim 5, characterized in that: In Step 3-1, the method for obtaining the initial similarity between the random outlier value and the random past value specifically includes: Operate the following equation: ； In the equation, Representative outliers and The initial similarity of the previous values, Representative outliers and The far-field weak induction signal after all the previous values ​​are digitized, Representative The outlier eddy current detection value is The wall thickness of the transverse pipe, Representative The eddy current detection value of the previous values The wall thickness of the transverse pipe, is the predefined indicator three, Represents the use of Z-score method to Implement standardized disposal; In Step 3-2, the method for obtaining the time point effect factor of the random phase difference in the near-zone value cluster of the outlier specifically includes: Operate the following equation: ； In the equation, Representative The outlier value is the nearest value in the cluster. The time factor of the phase difference is Representative outliers and nearby values ​​in the cluster The time interval between phase differences, is the predefined indicator three, Represents the use of Z-score method to Perform standardized disposal.

7. The multi-frequency array eddy current detection method according to claim 6, characterized in that: In Step 3-3, the method for obtaining the final similarity between the random outlier value and the random past value specifically includes: Operate the following equation: ； In the equation, Representative outliers and The last similarity of the previous values; The number of phase differences contained in the near-zone value cluster representing the phase difference; Representative outliers and In the cluster of values ​​near the previous values, The initial proximity of the corresponding phase differences; Representative outliers and In the cluster of values ​​near the previous values, The time point effect factor of the corresponding phase difference; For random outliers, select the past values ​​whose last similarity with the outlier value is higher than the pre-defined index 4 among all past values ​​and define them as the outlier's similar past values; In Step 3-4, the method for obtaining the influencing factor of the random close past values ​​of the random outlier value specifically includes: Operate the following equation: ； In the equation, Representative The outlier The factor of similar past values, Representative The outlier The outlier amplitude of similar past values, It is the pre-defined indicator three; In Step 3-4, the method for obtaining the actual value of the random outlier includes: ； In the equation, Representative The actual value of the outlier, Representative The number of non-outlier similar past values ​​of an outlier value, Representative Similar past values ​​for The outlier factor, Representative outliers and The last similarity of similar past values, Representative The outlier values ​​that are close to previous values, It is the pre-defined indicator three.

8. A multi-frequency array eddy current testing device, characterized in that: include: 12 groups of sensors connected to the single chip microcomputer distributed along the circumference of the inner wall of the transverse pipe; 12 sets of sensors are used to pick up far-field weak induction signals and transmit them to the single-chip microcomputer after digitization. The single-chip microcomputer is used to extract the phase difference according to the digitized far-field weak induction signals and perform processing, and then calculate the pipe wall thickness according to the processed phase difference; The modules running on the microcontroller include: A detection module, which is used to obtain the phase difference of multiple detection positions of the transverse pipeline and the eddy current detection value corresponding to the phase difference at the same time point; An outlier module is used to obtain the near-zone values ​​of the phase difference to form a near-zone value cluster of the phase difference, analyze the value change of the near-zone value cluster of the random phase difference, obtain the outlier amplitude of the phase difference, select the outlier amplitude of the value to be analyzed, and obtain the outlier value; A proximity module is used to obtain the initial proximity of the outlier and the previous values ​​according to the difference in eddy current detection values ​​between the outlier and the previous values ​​in a predefined time period; obtain the time point effect factor of the random phase difference in the near-zone value cluster of the outlier according to the time distance of the phase difference in the outlier and the near-zone value cluster; concentrate the initial proximity of the outlier and the previous values, the time point effect factor of the phase difference in the near-zone value cluster of the outlier, obtain the final proximity of the outlier and the previous values, and select all the past values ​​close to the outlier; combine the final proximity of the outlier and all the past values ​​close to each other, the value of the past values ​​close to each other and the outlier amplitude of the past values ​​close to each other to obtain the actual value of the outlier; The replacement module is used to replace all outliers of the lateral pipeline with actual values, and obtain the phase difference of the lateral pipeline after replacement. The phase difference after replacement is the phase difference after treatment.

Images