Downhole tool intelligent identification method and device based on curve abnormal point detection

Abstract

The invention relates to a downhole tool intelligent identification method and device based on curve abnormal point detection, and relates to the technical field of oil production engineering, and the method comprises the steps: obtaining a target well tool signal abnormal point difference value sequence; separating the data into a plurality of sections of data sets delta X, and extracting a tool length feature sequence Len and an abnormal point feature interval feature sequence MK in the data sets delta X; the delta X and the Len are updated according to the optimal separation combination of the mixed sequence determined by the MK, and an abnormal point feature spacing sequence is extracted from the delta X; determining optimal segmentation parameters of the water distributors according to the abnormal point feature spacing sequence, so as to determine the water distributors in delta X; and the length of the packer is determined according to other values except the corresponding water distributor in the Len, so that the packer and the short sleeve in the median of the delta X are determined. The problems that in the prior art, when underground tools are recognized, manual experience and operation need to be relied on, the accuracy of recognition results cannot be guaranteed, and meanwhile manual recognition is low in working efficiency and high in labor cost are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of oil production engineering, and in particular to a method and a device for intelligently identifying downhole tools based on curve abnormal point detection. Background Art

[0002] In water-driven oil fields, isotope injection profile logging has always been dominant and is the main means of dynamic monitoring of water-injected reservoirs. For this reason, a large number of isotope injection profile logging are carried out every year, which is the main source of tasks for production logging interpretation. At present, the intelligence level of isotope injection profile logging data interpretation software is very low, and a lot of work still requires manual intervention. The main links of the interpretation, such as curve depth calibration, tool annotation, curve superposition and water-absorbing layer division, still rely on the experience and judgment of the interpreters and manual operation. Not only is the work efficiency low and the labor cost high, but the quality of the interpretation cannot be guaranteed, and there are often ambiguous interpretation results. Therefore, it is necessary to carry out the automated interpretation of isotope injection profile logging. In this work, the automatic identification of tools is a crucial content and the main difficulty of this research. Summary of the invention

[0003] The present invention proposes a method and device for intelligent identification of downhole tools based on curve anomaly detection, so as to solve the problem that in the past, identification of downhole tools needed to rely on human experience and operation, and the accuracy of the identification results could not be guaranteed. At the same time, manual identification also had the problems of low work efficiency and high labor cost.

[0004] According to one aspect of the present invention, a method for intelligently identifying downhole tools based on curve abnormal point detection is provided, comprising: obtaining a target well tool signal abnormal point difference value sequence ΔY;

[0005] Taking the points in the difference sequence ΔY that are greater than or equal to the first predetermined value as segmentation points, ΔY is separated into multiple segments of data, and the collar segments contained in the multiple segments of data are excluded to obtain a multiple segment data set ΔX, and the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK are extracted from ΔX;

[0006] Determine the mixed sequence in ΔX according to Len, determine the optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine the updated Len according to the updated ΔX, and extract the abnormal point feature spacing sequence Num in the updated ΔX;

[0007] Determine the optimal segmentation parameter SP of the water distributor according to Num num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Numqt The corresponding updated values ​​in the Len sequence form the sequence Len qt ;

[0008] According to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

[0009] Preferably, before obtaining the target well tool signal abnormal point difference value sequence ΔY, the tool signal abnormal point difference value sequence ΔY is determined, and the method includes:

[0010] Obtain the magnetic positioning curve of the target well;

[0011] The time series anomaly detection algorithm is used to detect the abnormal points of the tool signal on the curve to form an abnormal point sequence Y;

[0012] Perform forward difference processing on the outlier sequence Y item by item to obtain the difference sequence ΔY.

[0013] Preferably, the method for excluding the collar segment contained in the multiple segments of data comprises:

[0014] If, among the multiple segments of data, there is a data segment whose total number of data included is less than the third predetermined value, and the maximum value in the segment of data is less than the fourth predetermined value, then the segment of data is a collar segment.

[0015] Preferably, the method for extracting the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK in ΔX comprises:

[0016] For each segment of data ΔX i Extract tool length L from i (i=1,2,3,...,n), forming a tool length feature sequence Len=(L1,L2,L3,...,L n ), where L i =∑(ΔX i );

[0017] For each segment of data ΔX i Extract the abnormal point feature distance MK i , forming an abnormal point feature spacing feature sequence MK = (MK1, MK2, MK3, ..., MK n ), where MK i = ΔX i A set of data whose values ​​are greater than or equal to a fifth predetermined value.

[0018] Preferably, the method for determining the mixed sequence in ΔX according to Len comprises:

[0019] Select L in Len i ΔX corresponding to a value greater than 80 i is a mixed sequence;

[0020] The method of using MK to determine the optimal separation combination in each mixed sequence comprises:

[0021] MK corresponding to the mixed sequence i is the split point, dividing the mixed sequence into multiple subsequences;

[0022] Combine the subsequences with the goal of having 2 to 4 final combinations, sum the values ​​contained in each combination, and obtain a sum value sequence;

[0023] Calculate the coefficient of variation of each sum value sequence respectively, and select the combination corresponding to the minimum value of all coefficients of variation as the optimal separation combination.

[0024] Preferably, the method for obtaining the coefficient of variation of each sum value sequence comprises:

[0025] Use formula (1) to calculate the coefficient of variation of the sum value sequence;

[0026] CV = σ / μ (1);

[0027] In formula (1), CV is the coefficient of variation, σ is the standard deviation, μ is the average value of the sum,

[0028] in,

[0029] In formula (2), n is the number of sum values, x i is the sum value, μ is the average value of the sum value,

[0030] in,

[0031] In formula (3), x i is the sum value, (i=1,2,3......n), n is the number of sum values.

[0032] Preferably, the method for extracting the abnormal point characteristic spacing sequence Num in the updated ΔX comprises:

[0033] For each segment of data ΔX in the updated ΔX i Extract the number of outliers Num i , forming the characteristic sequence of the number of abnormal points Num = (Num1, Num2, Num3, ..., Num n ), where Numi = ΔX i The amount of data in .

[0034] Preferably, the optimal segmentation parameter SP of the water distributor is determined according to Num num methods, including:

[0035] In Num, separate Num a = {a∈Num|Q1≤a≤Q2} and Num b ={b∈Num|b≥Q3}, where Q1 is the lower quartile, Q2 is the median, and Q3 is the upper quartile;

[0036] According to Num a and Num b , using formula (4), determine the optimal partition parameter SP of the water distributor num ;

[0037]

[0038] Preferably, the qt The method for determining the packer length value FgLen includes:

[0039] According to Len qt , use formula (5) to determine the packer length value FgLen;

[0040]

[0041] Where: d i For Len qt The i-th value in (i=1,2,3...P), d j For Len qt The jth value in (j=1,2,3...P), P is Len qt The number of elements in .

[0042] According to one aspect of the present invention, there is provided a downhole tool intelligent identification device based on curve abnormal point detection, comprising: an acquisition unit, for acquiring a target well tool signal abnormal point difference value sequence ΔY;

[0043] A feature extraction unit is used to use points in the difference sequence ΔY that are greater than or equal to a first predetermined value as segmentation points, separate ΔY into multiple segments of data, exclude the collar segment contained in the multiple segments of data, obtain a multiple segment data set ΔX, and extract a tool length feature sequence Len and an abnormal point feature spacing feature sequence MK from ΔX;

[0044] A mixed sequence separation unit is used to determine the mixed sequence in ΔX according to Len, determine the optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine the updated Len according to the updated ΔX, and extract the abnormal point feature spacing sequence Num in the updated ΔX;

[0045] The water distributor identification unit is used to determine the optimal segmentation parameter SP of the water distributor according to Num. num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt ;

[0046] Packer and short sleeve identification unit, used according to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

[0047] The present invention has at least the following beneficial effects:

[0048] The present invention proposes a method and device for intelligently identifying downhole tools based on curve abnormal point detection. The method detects the downhole tool signals in the magnetic positioning curve as abnormal points, uses difference clustering, feature extraction and other processing methods on the position data of the abnormal points, and then extracts the signal characteristics of different tools, thereby finally realizing accurate identification of downhole tools. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The accompanying drawings herein are incorporated into and constitute a part of the specification. These drawings illustrate embodiments consistent with the present invention and, together with the specification, are used to explain the technical solutions of the present invention.

[0050] Figure 1 A flowchart showing a method for intelligently identifying downhole tools based on curve abnormal point detection according to an embodiment of the present invention is shown;

[0051] Figure 2 A diagram showing the separation result of a packer and a short sleeve mixing tool according to an embodiment of the present invention;

[0052] Figure 3 A diagram showing the separation result of a packer and a water distributor mixing tool according to an embodiment of the present invention;

[0053] Figure 4 A downhole tool identification curve graph of a magnetic positioning curve according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0054] Various exemplary embodiments, features and aspects of the present invention will be described in detail below with reference to the accompanying drawings. The same reference numerals in the accompanying drawings represent elements with the same or similar functions. Although various aspects of the embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless otherwise specified.

[0055] The word “exemplary” is used exclusively herein to mean “serving as an example, example, or illustration.” Any embodiment described herein as “exemplary” is not necessarily to be construed as preferred or advantageous over other embodiments.

[0056] The term "and / or" herein is only a description of the association relationship of the associated objects, indicating that there may be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the term "at least one" herein represents any combination of at least two of any one or more of a plurality of. For example, including at least one of A, B, and C can represent including any one or more elements selected from the set consisting of A, B, and C.

[0057] In addition, in order to better illustrate the present invention, numerous specific details are provided in the following specific embodiments. It should be understood by those skilled in the art that the present invention can be implemented without certain specific details. In some examples, methods, means, components and circuits well known to those skilled in the art are not described in detail in order to highlight the subject matter of the present invention.

[0058] Figure 1 A flowchart showing a method for intelligently identifying downhole tools based on curve abnormal point detection according to an embodiment of the present invention is shown; Figure 2 A diagram showing the separation result of a packer and a short sleeve mixing tool according to an embodiment of the present invention; Figure 3 A diagram showing the separation result of a packer and a water distributor mixing tool according to an embodiment of the present invention;

[0059] Figure 4 A magnetic positioning curve downhole tool identification curve diagram according to an embodiment of the present invention is shown. Figure 1-4As shown, a method for intelligently identifying downhole tools based on curve abnormal point detection includes: step S01: obtaining a target well tool signal abnormal point difference sequence ΔY; step S02: using points in the difference sequence ΔY that are greater than or equal to a first predetermined value as segmentation points, separating ΔY into multiple segments of data, excluding the collar segments contained in the multiple segments of data, obtaining a multiple segment data set ΔX, extracting a tool length feature sequence Len and an abnormal point feature spacing feature sequence MK from ΔX; step S03: determining a mixed sequence in ΔX according to Len, determining an optimal separation combination in each mixed sequence using MK, updating ΔX using the optimal separation combination, determining an updated Len according to the updated ΔX, and extracting an abnormal point feature spacing sequence Num from the updated ΔX; step S04: determining an optimal segmentation parameter SP of a water distributor according to Num num , in Num is greater than SP num The value in ΔX corresponding to the value of is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt ; Step S05: According to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

[0060] The method for intelligently identifying downhole tools based on curve abnormal point detection provided by an embodiment of the present invention specifically comprises the following steps:

[0061] Step S01: Obtain a target well tool signal abnormal point difference value sequence ΔY.

[0062] In the present invention, before obtaining the target well tool signal abnormal point difference value sequence ΔY, the tool signal abnormal point difference value sequence ΔY is determined, and the method includes: obtaining the target well magnetic positioning curve; using a time series abnormality detection algorithm to detect the tool signal abnormal points on the curve to form an abnormal point sequence Y; performing forward difference processing on the abnormal point sequence Y item by item to obtain a difference sequence ΔY.

[0063] In an embodiment of the present invention, before using the time series anomaly detection algorithm to detect abnormal points of tool signals on the curve, it is also necessary to eliminate some interference noise in the acquired magnetic positioning curve through wavelet filtering, which can improve the recognition environment to a certain extent; after removing the noise, the sampling interval of the logging curve is unified, and the minimum sampling interval of 0.025m used in logging is uniformly used. If it is not met, linear interpolation processing is required.

[0064] The time series anomaly detection algorithm SH-ESD can detect the tool signal on the magnetic positioning curve as an abnormal point, forming an abnormal point sequence represented by the index value (sampling point number), recorded as Y. The abnormal point position sequence Y is processed forward one by one, that is, starting from the second index value, each index value is subtracted from the previous index value adjacent to it, and the difference sequence ΔY can be obtained.

[0065] Step S02: Using the points in the difference sequence ΔY that are greater than or equal to the first predetermined value as segmentation points, ΔY is separated into multiple segments of data, and the coupling segments contained in the multiple segments of data are excluded to obtain a multiple segment data set ΔX, and the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK are extracted from ΔX.

[0066] In the embodiment of the present invention, the first predetermined value is 60; ΔY is segmented by setting a threshold of 60, that is, in the difference sequence ΔY, all values ​​greater than or equal to 60 are used as segmentation points (excluding segmentation points), and ΔY is separated into multiple segments of data. Since the value in ΔY is the value of the index value forward item by item, if the value in ΔY is greater than or equal to 60, it means that there are no abnormal points within the longer distance between the two subtracted index values, that is, no abnormal points are detected in the corresponding curve segment, and this part can be excluded. At the same time, by segmenting with 60, each separated data segment will contain the difference between multiple adjacent abnormal points, and the tool category can be identified from these separated data segments.

[0067] In the present invention, the method for excluding the collar segment contained in multiple segments of data includes: if, among the multiple segments of data, the total number of data contained in a data segment is less than a third predetermined value, and the maximum value in the segment of data is less than a fourth predetermined value, then the segment of data is a collar segment.

[0068] In the embodiment of the present invention, the third predetermined value is: 4; the fourth predetermined value is: 10.

[0069] If, among multiple segments of data, the total number of values ​​contained in a segment of data is less than 4, and the maximum value in the segment of data is less than 10, then the segment of data is considered to correspond to a collar; for example, if the multiple segments of data contain two segments of data (1,1) and (3,21,1,2,1,2,22,1,2,1), the number of data in the data segment (1,1) is 2, and the maximum value is 1, then the segment of data is a collar.

[0070] Exclude it from ΔY, and finally form multiple groups of data, recorded as ΔX = (ΔX1, ΔX2, ΔX3, ..., ΔX n ), where the value ΔX in ΔX i =(a1, a2, a3, ..., a m)(i=1,…,n) is a set of ordered difference data, a i Represents the ΔX remaining after removing the collar data from the difference sequence separated into multiple data segments i The abnormal points contained in are forward-differed one by one. At this time, ΔX i The difference sequences indicating the packer (fg), water distributor (ps) or short sleeve (dt), as well as the temporarily indistinguishable combination of the three in different ways, complete the preliminary clustering of the tools.

[0071] In the present invention, the method for extracting the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK in ΔX comprises: respectively extracting each segment of data ΔX in ΔX i Extract tool length L from i (i=1,2,3,...,n), forming a tool length feature sequence Len=(L1,L2,L3,...,L n ), where L i =∑(ΔX i ); for each segment of data ΔX i Extract the abnormal point feature distance MK i , forming an abnormal point feature spacing feature sequence MK = (MK1, MK2, MK3, ..., MK n ), where MK i = ΔX i A set of data whose values ​​are greater than or equal to a fifth predetermined value.

[0072] In the embodiment of the present invention, the tool length characteristic sequence Len is: i The a contained in is summed, that is, ΔX i The corresponding L i =a1+a2+a3+…+a m . Determine each ΔX i The corresponding length and L i .

[0073] The fifth predetermined value is 10; the abnormal point characteristic spacing MK is: for each ΔX i All data with a greater than or equal to 10 are extracted, and each ΔX i The set of all data greater than or equal to 10 extracted is denoted as MK i , each MK i Corresponding to multiple data greater than or equal to 10.

[0074] Step S03: Determine the mixed sequence in ΔX according to Len, determine the optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine the updated Len according to the updated ΔX, and extract the abnormal point feature spacing sequence Num in the updated ΔX.

[0075] In the present invention, the method for determining the mixed sequence in ΔX according to Len comprises: selecting L in Len i ΔX corresponding to a value greater than 80 i The method of using MK to determine the optimal separation combination in each mixed sequence includes: using the MK corresponding to the mixed sequence i The mixed sequence is divided into multiple subsequences with 2 to 4 subsequences as the split point; the subsequences are combined with the goal of finally combining 2 to 4 subsequences, and the values ​​contained in each combination are summed to obtain the sum value sequence; the coefficient of variation of each sum value sequence is calculated respectively, and the combination corresponding to the minimum value of all coefficients of variation is selected as the optimal separation combination.

[0076] In the embodiment of the present invention, due to L i represents the corresponding ΔX i The sum of the differences between all the abnormal points contained in and the previous item, that is, the data segment ΔX i The length of the tool. Select a length greater than 80, that is, L in Len. i ΔX corresponding to data with a length greater than 80 i is a mixed sequence; that is, ΔX i The total length of the tools contained in the data segment is greater than 80, indicating that there are multiple indistinguishable tools contained in the data segment.

[0077] If the value of L3 is 85, then the corresponding ΔX3 is a mixed sequence. Then, using MK3 corresponding to ΔX3 as the segmentation point (including the segmentation point), all the values ​​a in ΔX3 are segmented into multiple subsequences. For example, the mixed sequence ΔX3 = (1, 2, 22, 1, 2, 1, 2, 34, 11). Since MK3 is the data greater than or equal to 10 in ΔX3, that is, 22, 34 and 11, 22, 34 and 11 are used as segmentation points to segment the mixed sequence ΔX3 into multiple subsequences. When segmenting, the corresponding MK3 in the mixed sequence is i The value can be divided into the previous subsequence, or become a subsequence alone, or be divided into the next subsequence. For example, the data in ΔX3 is divided into subsequence 1 = (1, 2), subsequence 2 = (22), subsequence 3 = (1, 2, 1, 2), subsequence 4 = (34, 11), or divided into subsequence 1 = (1, 2), subsequence 2 = (22, 1, 2, 1, 2), subsequence 3 = (34), subsequence 4 = (11).

[0078] Recombine the segmented subsequences into 2 to 4 groups and enumerate all the combinations. For example, if they are recombined into 3 groups, they can be ([1,2,22],[1,2,1,2],[34,11]) and ([1,2],[22,1,2,1,2],[34,11]) and so on.

[0079] Find the sum sequence of each combination, that is, sum all the values ​​in each combination. For example, the sum sequence corresponding to the group ([1,2,22], [1,2,1,2], [34,11]) is (25, 6, 45).

[0080] Calculate the coefficient of variation corresponding to each sum value sequence respectively, and select the combination corresponding to the minimum value of all coefficients of variation as the optimal separation combination.

[0081] In the present invention, the method for obtaining the coefficient of variation of each sum value sequence includes: using formula (1) to obtain the coefficient of variation of the sum value sequence;

[0082] CV = σ / μ (1);

[0083] In formula (1), CV is the coefficient of variation, σ is the standard deviation, μ is the average value of the sum,

[0084] in,

[0085] In formula (2), n is the number of sum values, x i is the sum value, μ is the average value of the sum value,

[0086] in,

[0087] In formula (3), x i is the sum value, (i=1,2,3......n), n is the number of sum values.

[0088] In the embodiment of the present invention, the sum value, the average sum value, and the number of sum values ​​in each sum value sequence are respectively substituted into formulas (1) to (3) to calculate the coefficient of variation corresponding to each sum value sequence.

[0089] If the combination corresponding to the sum sequence with the smallest coefficient of variation is combination 2, then the subsequence contained in combination 2 directly replaces the corresponding value in the original multiple sets of data ΔX, that is, ΔX3. Since combination 2 contains two subsequences, at this time, it is equivalent to increasing the number of values ​​in the entire ΔX by one, from the original ΔX3 to ΔX3 and ΔX4, corresponding to subsequence 2 and subsequence 3; the original ΔX4 becomes ΔX5... and so on.

[0090] Use the optimal separation combination corresponding to all mixed sequences to convert the corresponding ΔX iAfter the replacement is completed, the update of ΔX is completed, and the corresponding Len is recalculated for the updated ΔX, that is, each value (ΔX i ) corresponding to L i .

[0091] At the same time, the outlier feature spacing sequence Num is extracted from the updated ΔX.

[0092] In the present invention, the method for extracting the abnormal point characteristic spacing sequence Num in the updated ΔX comprises: respectively extracting each segment of data ΔX in the updated ΔX i Extract the number of outliers Num i , forming the characteristic sequence of the number of abnormal points Num = (Num1, Num2, Num3, ..., Num n ), where Num i = ΔX i The amount of data in .

[0093] In the embodiment of the present invention, if ΔX in the updated ΔX i =(a1, a2, a3, ..., a m ), then ΔX i Corresponding Num i =m.

[0094] Step S04: Determine the optimal segmentation parameter SP of the water distributor according to Num num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt .

[0095] In the present invention, the optimal segmentation parameter SP of the water distributor is determined according to Num. num The method comprises: in Num, separating Num a = {a∈Num|Q1≤a≤Q2} and Num b ={b∈Num|b≥Q3}, where Q1 is the lower quartile, Q2 is the median, and Q3 is the upper quartile; according to Num a and Num b , using formula (4), determine the optimal partition parameter SP of the water distributor num ;

[0096]

[0097] In the embodiment of the present invention, all values ​​in the sequence Num are sorted from small to large, wherein the lower quartile is the value corresponding to the 25% position in the sorted sequence Num, the median is the value corresponding to the 50% position in the sorted sequence Num, and the lower quartile is the value corresponding to the 75% position in the sorted sequence Num. Thus, all values ​​less than or equal to Q2 and greater than or equal to Q1 can be screened out in the sequence Num to form a set Num a , and all values ​​greater than or equal to Q3, forming the set Num b ; Substitute it into formula (4) to calculate the optimal segmentation parameter SP of the water distributor num .

[0098] Each data in Num is respectively compared with SP num For comparison, if Num i Greater than SP num , then the Num i The corresponding ΔX i is a water distributor; then the Num sequence is divided into two parts, namely the sequence Num corresponding to the water distributor ps , except for the values ​​corresponding to the water distributor, the other values ​​form the sequence Num qt Correspondingly, the updated Len sequence is also divided into two parts, namely ps The value of the corresponding value forms a sequence of Len ps , and the difference with Num ps The sequence Len formed by other values ​​outside the corresponding qt .

[0099] Therefore, it can be concluded that with the sequence Num ps The value of ΔX corresponding to the value of corresponds to the downhole sampling point position which is the water distributor, and the other values ​​of ΔX except the water distributor correspond to the packer and short sleeve.

[0100] Step S05: According to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

[0101] In the present invention, the qt The method for determining the packer length value FgLen comprises: qt , use formula (5) to determine the packer length value FgLen;

[0102]

[0103] Where: d i For Len qt The i-th value in (i=1,2,3...P), d j For Len qt The jth value in (j=1,2,3...P), P is Len qt The number of elements in .

[0104] In the embodiment of the present invention, the second predetermined value is 5. Formula (5) is the value of Len qt The values ​​in the sequence are used to find the difference sequence matrix, and then the sum of each row of the difference matrix is ​​calculated, and the one with the smallest sum is recorded as FgLen.

[0105] The sequence Len qt Substituting the value in into formula (5), we can get the most representative packer length value in the magnetic positioning curve of the target well, which is recorded as FgLen. qt Search for satisfying | FgLen-L i |≤5, determine the ΔX corresponding to the subscript of the searched value i For the packer.

[0106] Then, in the multi-segment data set ΔX, except for the water distributor determined in step S04 and the packer determined in step S05, the remaining values ​​correspond to short sleeves, thereby completing the identification of the downhole tool of the target well.

[0107] In the process of the present invention, in order to verify the reliability of the method, actual logging data of the isotope injection profile five-parameter combination logging were used to carry out verification experiments on hundreds of wells, achieving a 100% success rate. Figure 2 This is the result diagram of separation of packer and short sleeve hybrid tool for a target well; Figure 3 This is a diagram showing the separation results of the packer and water distributor hybrid tool for a target well; Figure 4 The results of automatic identification of downhole tools from magnetic positioning curves of three wells are given. There is a mixed tool of a packer and a short sleeve near 1021m in the first curve, and a mixed tool of a packer and a water distributor near 1101m in the second curve. Figures 2 to 4 It can be seen that the identification method of the present invention can accurately separate and identify mixed tools. The third curve is the automatic identification of tools in the presence of strong interference, and the tools are also accurately identified. From the application effect, the automatic identification method of downhole tools is stable and reliable, and can be promoted and applied in actual well logging interpretation. Its emergence has liberated heavy manual labor, improved interpretation efficiency and quality, and laid the foundation for realizing the intelligent interpretation of isotope injection profile logging.

[0108] It can be understood that the above-mentioned various method embodiments mentioned in the present invention can be combined with each other to form a combined embodiment without violating the principle logic. Due to space limitations, the present invention will not go into details.

[0109] The execution subject of the method for intelligently identifying downhole tools based on curve abnormal point detection may be a device for intelligently identifying downhole tools based on curve abnormal point detection. For example, the method for intelligently identifying downhole tools based on curve abnormal point detection may be executed by a terminal device or a server or other processing device, wherein the terminal device may be a user equipment (UE), a mobile device, a user terminal, a terminal, a cellular phone, a cordless phone, a personal digital assistant (PDA), a handheld device, a computing device, a vehicle-mounted device, a wearable device, etc. In some possible implementations, the method for intelligently identifying downhole tools based on curve abnormal point detection may be implemented by a processor calling computer-readable instructions stored in a memory. "

[0110] Those skilled in the art will appreciate that, in the above method of specific implementation, the order in which the steps are written does not imply a strict execution order and does not constitute any limitation on the implementation process. The specific execution order of the steps should be determined by their functions and possible internal logic.

[0111] The present invention also provides a downhole tool intelligent identification device based on curve abnormal point detection, comprising: an acquisition unit, used to acquire a target well tool signal abnormal point difference sequence ΔY; a feature extraction unit, used to use points in the difference sequence ΔY that are greater than or equal to a first predetermined value as segmentation points, separate ΔY into multiple segments of data, exclude the collar segment contained in the multiple segments of data, obtain a multiple segment data set ΔX, extract a tool length feature sequence Len and an abnormal point feature spacing feature sequence MK from ΔX; a mixed sequence separation unit, used to determine a mixed sequence in ΔX according to Len, determine an optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine an updated Len according to the updated ΔX, and extract an abnormal point feature spacing sequence Num from the updated ΔX; a water distributor identification unit, used to determine an optimal segmentation parameter SP of the water distributor according to Num num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt ; Packer and short sleeve identification unit, used according to Len qt Determine the packer length value FgLen, in Lenqt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

[0112] In some embodiments, the functions or modules and units included in the device provided by the embodiment of the present invention can be used to execute the method described in the above method embodiment. Its specific implementation can refer to the description of the above method embodiment. For the sake of brevity, it will not be repeated here.

[0113] The present invention converts the process of identifying downhole tool signals into the feature of analyzing the difference sequence of the position of the abnormal points through curve abnormal point detection. It does not require complex curve shape recognition. After simple data processing and separation operations, it can realize cluster recognition of downhole tool signals in one-dimensional data dimension. From the application effect, the method adopted by the present invention conforms to the characteristics of magnetic positioning logging curves, and is simple and reliable.

[0114] The method of the present invention is used to automatically identify the type and position of downhole tools from the magnetic positioning curve, which is suitable for application in the automatic interpretation of isotope injection profile logging, and is also suitable for promotion and application in other production logging projects that need to provide the position of downhole tools. Compared with the traditional manual marking of downhole tools, the entire tool identification process can be completed intelligently by computer programs without manual intervention, and the work efficiency and interpretation quality are greatly improved.

[0115] The invention has a reasonable design, reliable principle and convenient method, and solves the problems of complicated operation process, large workload and easy negligence and error in manual marking of downhole tools in injection profile logging interpretation. The tool identification method and device of the invention can realize batch processing of curve tool marking, greatly improve the data interpretation quality and work efficiency, and effectively reduce labor costs.

[0116] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for intelligently identifying downhole tools based on curve abnormal point detection, characterized in that: include: Obtain the abnormal point difference value sequence ΔY of the target well tool signal; Taking the points in the difference sequence ΔY that are greater than or equal to the first predetermined value as segmentation points, ΔY is separated into multiple segments of data, and the collar segments contained in the multiple segments of data are excluded to obtain a multiple segment data set ΔX, and the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK are extracted from ΔX; Determine the mixed sequence in ΔX according to Len, determine the optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine the updated Len according to the updated ΔX, and extract the abnormal point feature spacing sequence Num in the updated ΔX; Determine the optimal segmentation parameter SP of the water distributor according to Num num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt ; According to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

2. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 1, characterized in that: Before obtaining the target well tool signal abnormal point difference value sequence ΔY, the tool signal abnormal point difference value sequence ΔY is determined, and the method includes: Obtain the magnetic positioning curve of the target well; The time series anomaly detection algorithm is used to detect the abnormal points of the tool signal on the curve to form an abnormal point sequence Y; Perform forward difference processing on the outlier sequence Y item by item to obtain the difference sequence ΔY.

3. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 1, characterized in that: The method for excluding the collar segment included in the multiple segments of data comprises: If, among the multiple segments of data, there is a data segment whose total number of data included is less than the third predetermined value, and the maximum value in the segment of data is less than the fourth predetermined value, then the segment of data is a collar segment.

4. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 1, characterized in that: The method for extracting the tool length feature sequence Len and the abnormal point feature spacing feature sequence MK in ΔX includes: For each segment of data ΔX i Extract tool length L from i (i=1,2,3,...,n), forming a tool length feature sequence Len=(L1,L2,L3,...,L n ), where L i =∑(ΔX i ); For each segment of data ΔX i Extract the abnormal point feature distance MK i , forming an abnormal point feature spacing feature sequence MK = (MK1, MK2, MK3, ..., MK n ), where MK i = ΔX i A set of data whose values ​​are greater than or equal to a fifth predetermined value.

5. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 4, characterized in that: The method for determining the mixed sequence in ΔX according to Len comprises: Select L in Len i ΔX corresponding to a value greater than 80 i is a mixed sequence; The method of using MK to determine the optimal separation combination in each mixed sequence comprises: MK corresponding to the mixed sequence i is the split point, dividing the mixed sequence into multiple subsequences; Combine the subsequences with the goal of having 2 to 4 final combinations, sum the values ​​contained in each combination, and obtain a sum value sequence; Calculate the coefficient of variation of each sum value sequence respectively, and select the combination corresponding to the minimum value of all coefficients of variation as the optimal separation combination.

6. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 5, characterized in that: The method for obtaining the coefficient of variation of each sum value sequence includes: Use formula (1) to calculate the coefficient of variation of the sum value sequence; CV = σ / μ(1); In formula (1), CV is the coefficient of variation, σ is the standard deviation, μ is the average value of the sum, in, In formula (2), n is the number of sum values, x i is the sum value, μ is the average value of the sum value, in, In formula (3), x i is the sum value, (i=1,2,3......n), n is the number of sum values.

7. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 4, characterized in that: The method for extracting the abnormal point characteristic spacing sequence Num in the updated ΔX includes: For each segment of data ΔX in the updated ΔX i Extract the number of outliers Num i , forming the characteristic sequence of the number of abnormal points Num = (Num1, Num2, Num3, ..., Num n ), where Num i = ΔX i The amount of data in .

8. The method for intelligently identifying downhole tools based on curve abnormal point detection according to claim 1, characterized in that: Determine the optimal segmentation parameter SP of the water distributor according to Num num methods, including: In Num, separate Num a = {a∈Num|Q1≤a≤Q2} and Num b ={b∈Num|b≥Q3}, where Q1 is the lower quartile, Q2 is the median, and Q3 is the upper quartile; According to Num a and Num b , using formula (4), determine the optimal partition parameter SP of the water distributor num ; 9. The method for intelligently identifying downhole tools based on curve abnormal point detection according to any one of claims 1 to 8, characterized in that: According to Len qt The method for determining the packer length value FgLen includes: According to Len qt , use formula (5) to determine the packer length value FgLen; Where: d i For Len qt The i-th value in (i=1,2,3...P), d j For Len qt The jth value in (j=1,2,3...P), P is Len qt The number of elements in .

10. A downhole tool intelligent identification device based on curve abnormal point detection, characterized in that: include: An acquisition unit, used for acquiring a target well tool signal abnormal point difference value sequence ΔY; A feature extraction unit is used to use points in the difference sequence ΔY that are greater than or equal to a first predetermined value as segmentation points, separate ΔY into multiple segments of data, exclude the collar segment contained in the multiple segments of data, obtain a multiple segment data set ΔX, and extract a tool length feature sequence Len and an abnormal point feature spacing feature sequence MK from ΔX; A mixed sequence separation unit is used to determine the mixed sequence in ΔX according to Len, determine the optimal separation combination in each mixed sequence using MK, update ΔX using the optimal separation combination, determine the updated Len according to the updated ΔX, and extract the abnormal point feature spacing sequence Num in the updated ΔX; The water distributor identification unit is used to determine the optimal segmentation parameter SP of the water distributor according to Num. num , in Num is greater than SP num The value of the corresponding updated ΔX is the water distributor, then the other values ​​in Num other than the value corresponding to the water distributor form a sequence Num qt , and Num qt The corresponding updated values ​​in the Len sequence form the sequence Len qt ; Packer and short sleeve identification unit, used according to Len qt Determine the packer length value FgLen, in Len qt The value in ΔX corresponding to the value that satisfies the absolute value of the difference with FgLen is less than or equal to the second predetermined value is the packer, and the other values ​​in ΔX except the values ​​corresponding to the water distributor and the packer are short sleeves, thereby completing the downhole tool identification.

Images