A method for correcting peak drift of carbon-oxygen ratio logging spectrum data

Abstract

The present invention discloses a method for correcting the peak drift of carbon-oxygen ratio logging energy spectrum data, which mainly realizes the positioning of characteristic peaks of elements in the carbon-oxygen ratio energy spectrum and the calculation of peak drift correction parameters. The present invention uses Mexican hat wavelet to perform continuous wavelet transform on energy spectrum data and converts the obtained wavelet coefficient matrix into a grayscale image. By segmenting the image, all regions containing peak positions are extracted, and the false peak features generated by noise are effectively eliminated. In order to determine the optimal segmentation threshold, the grayscale with the maximum frequency in the image grayscale histogram and the grayscale with the maximum inter-class variance in the segmentation result and all grayscale values ​​between the two are regarded as a fuzzy set. The corresponding membership is determined according to the inter-class variance of the image segmentation result with each grayscale as the threshold, and cluster analysis is used to determine the optimal grayscale threshold. Finally, all the searched element characteristic peaks are comprehensively utilized in combination with the corresponding standard peaks to realize the accurate calculation of the peak drift correction parameters.

Description

Technical Field

[0001] The present invention relates to a method for correcting the peak drift of energy spectrum data of carbon-oxygen ratio logging, which is mainly used to correct the energy scale change of energy spectrum caused by factors such as temperature and pressure in carbon-oxygen ratio logging, and lay the foundation for the subsequent extraction of important interpretation parameters such as carbon-oxygen ratio. Background Art

[0002] In order to determine the peak position and track address according to the gamma ray energy or vice versa, the gamma ray energy according to the peak position needs to be energy calibrated on the gamma detection system of the instrument. Energy calibration is to calibrate the relationship between energy and track address using gamma rays of known energy under the conditions of use determined by the detection system. With this relationship, the corresponding gamma ray energy can be determined by searching for the peak position of the characteristic peak. The typical energy calibration curve is approximately a straight line, but the straight line does not necessarily pass through the origin. During the measurement process of carbon-oxygen ratio logging, the unstable operation of the circuit system or the influence of external conditions such as downhole temperature and pressure will cause the corresponding relationship between the characteristic energy of gamma rays and the corresponding track address to change, that is, the spectrum peak drift phenomenon occurs. If the spectrum peak drift is not corrected in time and effectively after it occurs, the statistical error of the characteristic element counting rate will increase, affecting the accuracy of the formation information curves such as carbon-oxygen ratio and silicon-calcium ratio extracted from the energy spectrum, and even causing serious distortion. Therefore, for carbon-oxygen ratio logging, spectrum peak drift correction is an indispensable step before subsequent processing. The key to correcting the peak drift of gamma energy spectrum is to accurately find the peak position of the characteristic peak in the energy spectrum. In the process, we need to use a reliable peak-finding algorithm to locate the characteristic peak position of the element. Whether the peak position can be accurately identified directly affects the effect of energy spectrum drift correction.

[0003] The continuous wavelet transform method is a peak detection method that has attracted much attention at present. It has the advantages of flexibility, multi-resolution analysis capability and easy implementation. Some researchers directly use the Mexican hat wavelet to perform continuous wavelet transform on the spectral signal, and use the redundant information in the wavelet coefficients to identify the spectral characteristic peaks. Some researchers also proposed to make full use of the maximum, minimum and zero-crossing information in the wavelet coefficient matrix to improve the accuracy of peak detection. The existing continuous wavelet transform method realizes the detection of spectral characteristic peaks by searching for local maximum values ​​at various scales in the wavelet coefficient matrix. However, the noise signal in the spectrum will also produce a series of smaller local maximum points due to its randomness, forming false peak features, which will affect the final peak detection effect. Therefore, in practical applications, it is often necessary to cooperate with appropriate thresholds to eliminate false peaks. The choice of threshold is directly related to the effect of peak detection, which makes these algorithms very sensitive to the threshold and reduces the applicability of peak detection algorithms.

[0004] When the peak drift occurs, it means that the zero intercept and gain in the energy spectrum calibration relationship have changed. The so-called peak drift correction is to convert the energy spectrum after the drift to the energy spectrum that should be obtained under the theoretical calibration conditions. Some researchers have proposed to use the characteristic peaks of hydrogen and iron in the carbon-oxygen ratio spectrum as the benchmark to calculate the peak drift correction parameters. However, in actual processing, due to the differences between measuring instruments and the actual measurement environment, the measured spectral line characteristics will inevitably not remain unchanged. Searching for hydrogen peaks and iron peaks in the energy spectrum is currently a common method, but the use of a fixed correction mode cannot meet the needs of any situation. For example, some scintillation crystals have a higher detection efficiency and are more sensitive to the full energy peak of the element; some scintillation crystals are more sensitive to the escape peak of the element. For example, when the mineralization of the local formation is low, there is less chlorine, and the gamma rays produced by the capture reaction of chlorine nuclei are also less, which will not cover the characteristic peaks of other elements. Therefore, the characteristic peaks produced by the release of medium-energy gamma rays such as silicon and calcium will be more obvious. For the above types of situations, a more flexible method is needed to achieve peak drift correction. Summary of the invention

[0005] The present invention mainly overcomes the shortcomings of the existing energy spectrum peak search method, and realizes accurate identification and positioning of the element characteristic peaks in the carbon-oxygen ratio energy spectrum. At the same time, the actual data characteristics of the energy spectrum are integrated to provide a more flexible processing method for drift correction of carbon-oxygen ratio logging energy spectrum data. In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0006] When a wavelet with similar characteristics to the characteristic peak of an element is used as the mother wavelet for continuous wavelet transform of the energy spectrum, the wavelet coefficient value is large near the peak position and the wavelet coefficient will have a local maximum at the peak position. The local maxima at different scales are extracted, and the addresses where they are located can be used to characterize the peak position. However, the energy spectrum collected by carbon-oxygen ratio logging will inevitably contain noise signals due to radioactive statistical fluctuations or instrument stability. Due to the randomness of the noise, it will produce a series of smaller local maxima in the wavelet coefficient matrix. The biggest challenge faced by energy spectrum peak search is the need to eliminate such false peak features generated by noise signals. The maximum value that characterizes the characteristic peak often has a larger amplitude, while the maximum value generated by noise usually has a smaller amplitude due to its lower energy. Therefore, if the entire wavelet coefficient matrix is ​​converted into a grayscale image, the true peak features with larger amplitudes have higher grayscale values, while the false peak features of noise have lower grayscale values. By calculating a suitable image segmentation threshold, the potential peak area can be segmented out, thereby directly eliminating the false peak features in the wavelet coefficient matrix and improving the efficiency and accuracy of peak search. Based on the above principle, the present invention proposes a method for correcting the peak drift of carbon-oxygen ratio well logging energy spectrum data, which is characterized in that the method comprises the following steps:

[0007] S1: Continuous wavelet transform of energy spectrum. The Mexican hat wavelet is used as the mother wavelet to perform continuous wavelet transform on the carbon-oxygen ratio energy spectrum under the set scale to obtain a two-dimensional wavelet coefficient matrix.

[0008] S2: Convert the wavelet coefficient matrix into a grayscale image. Convert the two-dimensional wavelet coefficient matrix into an image with a grayscale range of 0 to 255. The present invention uses a nonlinear mapping method to convert the wavelet coefficient value into a grayscale value in the range of 0 to 255, as shown in formula (1).

[0009] This mapping method combines the mean and standard deviation of the wavelet coefficients, which can effectively avoid the problem of low resolution of the grayscale image after conversion.

[0010]

[0011] In the formula, c i ——Energy spectrum wavelet coefficients;

[0012] t i ——Energy spectrum wavelet coefficient c i The gray value of the map;

[0013] c max ——The maximum value among the energy spectrum wavelet coefficients;

[0014] c min ——The minimum value of the energy spectrum wavelet coefficients;

[0015] m——the average value of the energy spectrum wavelet coefficients;

[0016] σ——standard deviation of energy spectrum wavelet coefficients.

[0017] S3: Calculate the upper limit of the image segmentation threshold. Use different grayscale as the threshold to segment the wavelet coefficient image into foreground and background. Based on the grayscale histogram of the image, extract the grayscale value t that maximizes the inter-class variance between the foreground and background. max , where the between-class variance can be calculated using formula (2):

[0018]

[0019] In the formula, — the inter-class variance between foreground and background in the image;

[0020] μ T ——The average gray value of all pixels in the image;

[0021] μ0——average gray value of pixels in the foreground;

[0022] μ1——the average gray value of pixels in the background;

[0023] ω0——the proportion of foreground pixels;

[0024] ω1——The proportion of background pixels.

[0025] S4: The threshold segmentation effect is blurred. Since the element characteristic peak area accounts for a relatively small proportion of the entire spectrum, the gray value t with the largest frequency in the gray histogram min Should come from the background baseline area. Statistics t min to max All gray values ​​between the thresholds are used to calculate the inter-class variance of the segmentation results. The segmentation effects obtained by all optional gray values ​​are fuzzified and regarded as a fuzzy set S, whose domain is the gray value (threshold) used for image segmentation. This fuzzy set S can be expressed as formula (3) using Zadeh representation.

[0026]

[0027] Where, t min ——The gray value with the maximum frequency in the histogram, the lower limit of the threshold;

[0028] t max ——The gray value used to obtain the maximum inter-class variance, the upper threshold limit;

[0029] t i ——Between t min and t max Grayscale values ​​between, alternative thresholds;

[0030] S(t min )——the membership degree of the lower limit of the threshold;

[0031] S(t max )——Membership degree of the upper threshold limit;

[0032] S(t i )——The membership degree corresponding to each alternative threshold.

[0033] S5: Determination of membership. The membership of each grayscale is determined based on the inter-class variance after image segmentation. Because the energy spectrum characteristic peaks have different morphological characteristics, when the scale of the continuous wavelet transform is determined, the grayscale values ​​in the obtained wavelet coefficient grayscale image are non-uniform. max When the threshold is segmented, the variance between classes is the largest, so the threshold has the most stringent requirements on grayscale. Only a part of the area with higher grayscale value can be extracted, and its membership is set to 1. min When used as the segmentation threshold, the grayscale requirement is the most relaxed, so the membership degree is set to 0. iWhen used as the threshold for image segmentation, the corresponding membership should be between 0 and 1. If the membership of a certain threshold segmentation effect is larger, it means that the difference between the foreground and the background after segmentation is greater, and the grayscale requirement is stricter. The inter-class variance can be used to calculate the membership of each grayscale, as shown in formula (4).

[0034]

[0035] In the formula, BCVar(t min )——Use t min is the inter-class variance after threshold segmentation;

[0036] BCVar(t i )——Use t i is the inter-class variance after threshold segmentation;

[0037] BCVar(t max )——Use t max is the inter-class variance after threshold segmentation.

[0038] S6: Optimal threshold calculation. Based on the membership of each gray threshold, cluster analysis is used to divide the fuzzy set S into two categories, and the optimal threshold is selected. max The minimum grayscale value in the same class is used as the final threshold for image segmentation processing. At this time, it is believed that the threshold can meet the basic requirements for grayscale differences in image segmentation results.

[0039] S7: Extract characteristic peaks. Search for the maximum values ​​at different scales in each segmented area, and count the addresses of these maximum values. The final characteristic peak is determined by the address with the largest proportion of maximum values ​​in these segmented areas.

[0040] S8: Calculation of peak drift correction parameters. Based on the identified characteristic peak positions of all elements and the energy scale of the carbon-oxygen ratio spectrum, a set of equations for solving the peak drift correction parameters is established, as shown in equation (5). The peak drift correction parameters are obtained by solving the least squares solution of the overdetermined set of equations.

[0041]

[0042] Where n is the number of characteristic peaks of the element searched.

[0043] x p1 ,x p2 …x pn ——respectively the theoretical peak positions of peak 1, peak 2…peak n;

[0044] Δx p1 ,Δx p2 …Δx pn ——respectively the drift of peak 1, peak 2…peak n;

[0045] z——the change of zero cut Z ΔZ;

[0046] g——Percentage of change in gain G.

[0047] The specific calculation method is: rewrite formula (5) into matrix form:

[0048]

[0049] Write equation (6) in matrix equation form:

[0050] AX=B (7)

[0051] According to formula (7), the peak drift correction parameter can be obtained as:

[0052] X=(A T A) -1 A T B (8)

[0053] S9: Spectral peak drift correction. According to the peak drift correction parameters z and g, the new channel address i' corresponding to the energy spectrum channel address i can be calculated, as shown in formula (9), thereby completing the spectral peak drift correction.

[0054] i′=i+Δi=z+(1+g)i (9)

[0055] Where, i′ is the new address corresponding to address i;

[0056] Δi——the change of address i.

[0057] The present invention designs a peak-finding method for element characteristic peaks of carbon-oxygen ratio energy spectrum based on fuzzy set theory and image segmentation theory. The advantage of the present invention is that it can effectively eliminate the false peak characteristics generated by noise and is insensitive to peak-finding parameters, and the peak-finding results are more accurate. At the same time, the present invention makes full use of all detected characteristic peak information in the calculation of peak drift correction parameters, and the correction results are more flexible and reliable. In addition, the peak-finding method proposed in the present invention can also be applied to peak-finding processing of other types of spectral data, and has a wide range of application value. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is the effect diagram of the spectral peak drift correction of the carbon-oxygen ratio logging near-probe measured spectral line;

[0059] Figure 2 This is the image segmentation effect of wavelet coefficients of the spectrum line measured by the near-probe of carbon-oxygen ratio logging;

[0060] Figure 3 This is the effect diagram of the peak drift correction of the spectrum line measured by the carbon-oxygen ratio logging remote probe;

[0061] Figure 4 This is the effect diagram of wavelet coefficient image segmentation of the spectrum line measured by the remote probe of carbon-oxygen ratio logging. DETAILED DESCRIPTION

[0062] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with application examples. It should be understood that the specific application examples described herein are only used to explain the present invention and are not used to limit the present invention.

[0063] Application examples:

[0064] In order to illustrate the application effect of the present invention, the captured energy spectrum measured by carbon-oxygen ratio logging of a certain example well is used for verification. In actual processing, the energy scale used by the instrument is known, so the theoretical peak position channel address of each element characteristic peak can be determined accordingly. The present invention is used to perform energy spectrum peak search to identify all potential formation element characteristic peak positions in the carbon-oxygen ratio energy spectrum, and then the identified element characteristic peak positions are compared with the standard peak positions of each element characteristic peak. If the channel address difference between the identified characteristic peak position and the standard peak position of a certain element characteristic peak satisfies ±n channels (n is the set channel address error), the identified peak position is matched with the characteristic element and used for drift correction of the energy spectrum.

[0065] According to the energy scale relationship of the energy spectrum of this example, the characteristic peak of hydrogen is fixed at the 64th channel (2.23MeV), and the characteristic peak of iron is fixed at the 218th channel (7.64MeV). The carbon-oxygen ratio logging tool used in this example well has two gamma detectors with long and short source distances, which respectively collect energy spectrum data under long and short source distances. The capture energy spectrum collected under the short source distance is shown in Figure 2. Figure 1 As shown, the results of peak searching using the present invention are also Figure 1 It is marked in . It can be seen that the chlorine photoelectric peak (Cl, 1.95MeV), hydrogen photoelectric peak (H, 2.23MeV), silicon photoelectric peak (Si, 3.54MeV), silicon first escape peak (Si1, 3.03MeV), calcium photoelectric peak (Ca, 6.42MeV), calcium first escape peak (Ca1, 5.91MeV), calcium second escape peak (Ca2, 5.40MeV) and iron photoelectric peak (Fe, 7.64MeV) were searched in the spectrum measured by the short source distance probe. The result after segmenting the wavelet coefficient image using the image segmentation threshold determined by the present invention is as follows. Figure 2As shown, it can be seen that the wavelet coefficient is significantly larger at the peak position. The image segmentation threshold determined by the present invention can effectively identify the area where the element characteristic peak position in the energy spectrum is located, and then the accurate characteristic peak position can be extracted by judging the address of the local maximum value at different scales in each area. In addition, the present invention can effectively eliminate the false peak characteristics generated by noise and avoid misjudgment of the peak position. The drift amount of each peak position of the short source distance capture spectrum calculated based on the peak search result combined with the energy scale is shown in Table 1.

[0066] Table 1 Drift of characteristic peaks in short source distance capture spectrum

[0067]

[0068] According to the drift amount and the theoretical peak position of the searched characteristic peak, the equation group is established according to equation (5), and the least square method is used to solve the equation group. The peak drift correction parameters z = -2.93, g = 0.04 are calculated. The peak drift correction parameters are used to perform drift correction on the short source distance capture spectrum. The correction results are shown in Figure 1 It can be seen that the drift of each channel of the energy spectrum measured is different. Compared with the low energy segment, the peak drift is more serious. The comprehensive use of the characteristic peak information of all detected elements can make the correction result more accurate.

[0069] The present invention is also used to perform energy spectrum peak search and drift correction on the captured spectrum line measured by the long source distance probe, and the results are as follows: Figure 3 As shown. It can be seen that the hydrogen photoelectric peak (H, 2.23MeV), silicon photoelectric peak (Si, 3.54MeV), silicon first escape peak (Si1, 3.03MeV), calcium first escape peak (Ca1, 5.91MeV) and iron photoelectric peak (Fe, 7.64MeV) were searched in the spectrum measured by the long source distance probe. Similarly, the wavelet coefficient image segmentation result after image segmentation threshold processing determined by the present invention is as follows Figure 4 The peak drift values ​​calculated based on the peak search results and energy scale are shown in Table 2. The peak drift correction parameters calculated based on this are z = -0.18, g = -0.0017. The results of the peak drift correction for the long source distance capture spectrum based on the peak drift correction parameters are shown in Figure 3 As shown by the dotted line in the figure, it can be seen that the spectrum peak drift of the long source distance is lighter than that of the short source distance, the drift of each channel is relatively small, and the corrected spectrum line is basically consistent with the original spectrum line.

[0070] Table 2 Drift of characteristic peaks in long source distance capture spectrum

[0071]

Claims

1. A method for correcting peak drift of carbon-oxygen ratio logging spectrum data, characterized in that The method comprises the following steps: S1: Using the Mexican Hat wavelet as the mother wavelet, the carbon-oxygen ratio energy spectrum is subjected to continuous wavelet transform under the set scale to obtain a two-dimensional wavelet coefficient matrix; S2: A nonlinear mapping method is used to convert the wavelet coefficient values ​​in the two-dimensional wavelet coefficient matrix into grayscale values ​​ranging from 0 to 255. The specific mapping method is: In the formula, c i is the energy spectrum wavelet coefficient, t i is the energy spectrum wavelet coefficient c i The gray value of the map, c max is the maximum value of the energy spectrum wavelet coefficients, c min is the minimum value of the energy spectrum wavelet coefficients, m is the average value of the energy spectrum wavelet coefficients, and σ is the standard deviation of the energy spectrum wavelet coefficients; S3: Use different grayscale as threshold to divide the wavelet coefficient image into foreground and background. Based on the grayscale histogram of the image, extract the grayscale value t that maximizes the inter-class variance between the foreground and background. max , where the between-class variance is calculated as: In the formula, is the inter-class variance between foreground and background in the image, μ T is the average gray value of all pixels in the image, μ0 is the average gray value of the pixels in the foreground, μ1 is the average gray value of the pixels in the background, ω0 is the proportion of foreground pixels, and ω1 is the proportion of background pixels; S4: Count the gray value t with the maximum frequency in the gray histogram min , the gray value t that maximizes the inter-class variance between foreground and background max And all the gray values ​​between the two and calculate the inter-class variance of the segmentation results when they are used as thresholds. The segmentation effects obtained by all optional gray values ​​are fuzzy and regarded as a fuzzy set S, whose domain is the gray value used for image segmentation. This fuzzy set S can be expressed as: Where, t min is the gray value with the maximum frequency in the histogram, t max To obtain the gray value used to obtain the maximum inter-class variance, t i Between t min and t max The gray value between min ) is t min The threshold membership, S(t max ) is t max The threshold membership, S(t i ) is t min and t max The degree of membership corresponding to each grayscale threshold between; S5: The membership degree corresponding to each grayscale is determined according to the inter-class variance after image segmentation. max When the threshold is segmented, the variance between classes is the largest, so the threshold has the most stringent requirements on grayscale. Only a part of the area with higher grayscale value can be extracted, and its membership is set to 1. min When used as the segmentation threshold, the grayscale requirement is the most relaxed, so the membership degree is set to 0. i When used as the threshold for image segmentation, the corresponding membership should be between 0 and 1. The inter-class variance can be used to calculate the membership of each grayscale: In the formula, BCVar(t min ) is the use of t min is the inter-class variance after threshold segmentation, BCVar(t i ) is the use of t i is the inter-class variance after threshold segmentation, BCVar(t max ) Using t max is the inter-class variance after threshold segmentation; S6: Based on the membership of each gray threshold, cluster analysis is used to divide the fuzzy set S into two categories, and the max The minimum gray value in the same class is used as the final threshold for image segmentation. At this time, it is considered that the threshold can meet the basic requirements for gray difference in image segmentation results. S7: Search for maximum values ​​at different scales in each segmented area, and count the locations of these maximum values. The final characteristic peak position is determined by the location of the maximum value in the segmented area. S8: Based on the identified characteristic peak positions of all elements and the energy scale of the carbon-oxygen ratio spectrum, a set of equations for solving the peak drift correction parameters is established: Where n is the number of characteristic peaks of the element searched, x p1 , x p2 ,…,x pn is the theoretical peak position of peak 1, peak 2, ..., peak n, Δx p1 , Δx p2 , …, Δx pn is the drift of peak 1, peak 2, ..., peak n, z is the change of zero cutoff Z, g is the percentage of gain G change, and the peak drift correction parameters are obtained by solving the least square solution of the overdetermined equations; S9: The new channel address i′ corresponding to the channel address i can be calculated according to the peak drift correction parameters z and g: i′=i+Δi=z+(1+g)i (6) Where i′ is the new channel address corresponding to channel address i, and Δi is the change in channel address i. The peak drift correction of carbon-oxygen ratio energy spectrum data is achieved based on this relationship.

Images