A method for estimating hydrogen peak baseline in carbon-oxygen ratio spectrum

Abstract

The present invention discloses a method for estimating the baseline of the hydrogen peak in the carbon-oxygen ratio energy spectrum, which is used for processing the energy spectrum data collected by the carbon-oxygen ratio logging instrument, and mainly realizes the estimation of the baseline under the characteristic peak of the hydrogen element, laying a foundation for the subsequent processing and interpretation of the carbon-oxygen ratio data. Based on the asymmetric least squares principle, an objective function for baseline estimation is established. By selecting a pure baseline area near the peak area boundary to construct a peak area boundary matrix, additional constraints are introduced into the objective function. The objective function takes into account both the smoothness of the estimated baseline and the symmetry of the characteristic peak, and can effectively ensure the accuracy of the baseline estimation result. In terms of weight updating, it is proposed to use a hyperbolic tangent function combined with the residual mean and standard deviation to calculate the weight value in the iteration process, which effectively reduces the influence of noise on the result and improves the stability of the baseline estimation. Therefore, the baseline estimation result of the present invention is more reliable.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir evaluation after casing of oil and gas wells, and in particular to a method for estimating the baseline of a hydrogen peak in a carbon-oxygen ratio energy spectrum, which can effectively estimate the baseline under the characteristic peak of the hydrogen element in a non-elastic total spectrum and a capture spectrum in carbon-oxygen ratio logging, and lay a foundation for the processing and interpretation of subsequent logging data. Background Art

[0002] A large part of the information used in carbon-oxygen ratio logging comes from the inelastic scattered gamma rays produced by the inelastic scattering of fast neutrons with carbon (C), oxygen (O), silicon (Si), calcium (Ca) and other elements in the formation. Among them, the measurement of characteristic gamma rays of carbon and oxygen elements is the focus of carbon-oxygen ratio logging. Since the atomic nuclei of carbon and oxygen elements and the 14MeV fast neutrons emitted by the neutron source only undergo inelastic scattering, how to extract a pure inelastic spectrum from the actual measured inelastic total spectrum (including the inelastic spectrum of captured gamma components) is the basis for the interpretation of carbon-oxygen ratio logging. The key lies in how to calculate the hydrogen peak area in the total spectrum and the captured spectrum. Since hydrogen does not undergo inelastic nuclear reactions, the hydrogen peak area in the total spectrum represents the content of captured components. Since the carbon-oxygen ratio logging adopts pulsed neutron synchronous measurement technology for data acquisition, the ratio of the hydrogen peak area in the total inelastic spectrum to the hydrogen peak area in the capture spectrum can be calculated as the subtraction factor. The net inelastic spectrum can be extracted by subtracting the product of the subtraction factor and the capture spectrum from the total inelastic spectrum.

[0003] Since the energy of the characteristic gamma ray of hydrogen is low, only 2.23MeV, its characteristic peak will be located above a higher baseline due to the influence of factors such as the Compton scattering effect in the scintillation crystal of the logging instrument. The existence of the baseline will have an adverse effect on the subsequent extraction of the net inelastic spectrum. For this reason, it is necessary to design a data processing method to estimate the baseline under the characteristic peak of hydrogen in the carbon-oxygen ratio inelastic total spectrum and the capture spectrum, so as to realize the calculation of the deduction factor. The asymmetric least squares method is one of the current methods for estimating the baseline. This method believes that in instrument measurement, most of the measured spectrum signals are composed of a smooth and slowly changing baseline signal and a spectrum peak signal with characteristic information. Therefore, an effective baseline estimate can be obtained by combining the baseline smoothness and the fidelity of the spectrum peak constrained by asymmetric weights. This method uses a fixed weight update method to obtain a satisfactory baseline estimation effect in spectral data with a high signal-to-noise ratio. Some researchers have also proposed an adaptive iterative reweighted penalty least squares method, which uses an exponential function to update the weights, or adds a first-order derivative constraint on the basis of the asymmetric least squares method. Some researchers have also proposed an asymmetric reweighted penalized least squares method, which uses a logistic function combined with the mean and variance of the residual in the pure baseline region to update the weights, reducing the impact of noise on baseline estimation. All of the above methods use the principle of asymmetric least squares. For this principle, negative value areas are prone to appear when the spectrum to be processed is more complex. In addition, this type of method estimates the baseline by approximating the baseline characteristics and pays less attention to the peak characteristics of the spectrum, which can easily cause the characteristic peaks of the spectrum to be distorted after background baseline correction. Summary of the invention

[0004] The present invention mainly overcomes the shortcomings of existing baseline estimation technology, and aims to provide a method for accurately estimating the baseline under the characteristic peak of hydrogen element in carbon-oxygen ratio energy spectrum data, laying a foundation for the subsequent processing and interpretation of well logging data. At the same time, this method can also be used as a general technology for baseline estimation of other types of spectral data. In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0005] The present invention proposes a method for estimating a hydrogen peak baseline in a carbon-oxygen ratio energy spectrum, characterized in that the baseline estimation method comprises the following steps:

[0006] S1: Initialize the weight matrix W. This matrix represents the weight of a certain data in the baseline estimation. For an energy spectrum containing m data channels, the weight of each data channel can be initialized to 1 and expressed as an m×m diagonal matrix:

[0007]

[0008] S2: Initialize the second-order difference matrix D. For an energy spectrum containing m channels of data, the matrix can be expressed as an m-2×m matrix:

[0009]

[0010] S3: Select a pure baseline region on both sides of the hydrogen peak in the carbon-oxygen ratio spectrum that does not contain peak shape features, and determine the address of the data point therein.

[0011] S4: Initialize the peak boundary matrix E. Since the characteristic peak of the carbon-oxygen ratio energy spectrum is symmetrical, that is, the count rates at the left and right boundaries of the characteristic peak of the hydrogen element should be similar, additional constraints are set for the baseline estimation based on this symmetrical feature of the energy spectrum characteristic peak, thereby effectively ensuring the accuracy of the peak shape characteristics when performing baseline estimation. According to the channel address selected in S3, the peak boundary matrix E can be expressed as:

[0012]

[0013] For an energy spectrum containing m channels of data, the number of columns of the matrix is ​​equal to the number of channel addresses m of the energy spectrum, and the number of rows is equal to the number of selected peak regions n (n=1 for baseline correction under the hydrogen peak of the carbon-oxygen ratio spectrum). In the peak region boundary matrix E, the elements at the left boundary of the peak region are set to 1, the elements at the right boundary of the peak region are set to -1, and the elements at the remaining positions are set to 0. In actual processing, multiple channel addresses near the boundary can also be selected to improve the statistical accuracy. This constraint condition can effectively ensure the symmetry of the characteristic peak of the hydrogen element during the processing process and improve the effect of baseline estimation.

[0014] S5: Constructing the objective function. According to the data of each channel of the energy spectrum, the objective function of baseline estimation can be constructed through the second-order difference matrix D, the peak area boundary matrix E, and the weight matrix W. The objective function proposed in the present invention is expressed as:

[0015] F(z)=(xz) T W(xz)+λ 1 z T D T Dz+λ 2 (xz) T E T Ε(xz)

[0016] In the formula, λ 1 and λ 2 are the smoothing factor and symmetry factor, which respectively describe the smoothness of the estimated baseline and the symmetry of the energy spectrum characteristic peak. x is the energy spectrum data vector, and z is the vector composed of the estimated baseline data.

[0017] S6: Find the minimum value of the objective function. The derivative of the objective function can be used to obtain the representation of the estimated baseline:

[0018] z=(W+λ 1 D TD+λ 2 E T E) -1 (λ 2 E T Ex+Wx)

[0019] By selecting a suitable weight update method to iteratively update the weight matrix W, the loop is terminated and the estimated baseline is output when the difference between the two iteration results is less than a given error δ. The weight update method proposed in the present invention is:

[0020]

[0021] In the formula, For the set d - The mean of For the set d - The standard deviation of the set d is xz, d - To satisfy x i <z i The function tanh is a hyperbolic tangent function, which is used to update the weight value of each channel according to the above parameters. The specific weight calculation expression is:

[0022]

[0023] This weight update method can make the weight values ​​of each channel in the pure baseline area affected by noise close to each other, and for the channel containing the energy spectrum characteristic peak, its weight is close to 0. This effectively avoids the low estimation result of the pure baseline area due to the influence of noise.

[0024] The objective function designed for the method for estimating the baseline of the hydrogen peak in the carbon-oxygen ratio energy spectrum provided by the present invention takes into account both the smoothness of the estimated baseline and the symmetry of the characteristic peak of the hydrogen element, and can effectively ensure the accuracy of the baseline estimation result under the hydrogen peak. In terms of the weight update method, it is proposed to use the hyperbolic tangent function combined with the residual mean and standard deviation to calculate the weight value in the iterative process, which effectively reduces the impact of noise on the estimation result and improves the stability of the baseline estimation. Therefore, the calculation result of the method proposed by the present invention is more reliable. The present invention can also be applied to the baseline estimation of other spectral data, and has a good processing effect on Raman spectroscopy, infrared spectroscopy, etc., and has a wide range of application value.

[0025] In terms of algorithm performance, the present invention has the following three advantages: First, it is not necessary to determine all peak areas in the energy spectrum when estimating the baseline. Generally speaking, better results can be obtained by selecting several typical peak areas in different spectral segments of the energy spectrum. Second, the constraints determined by the peak area boundary matrix E are also applicable to overlapping peaks. Because even for overlapping peaks, the channel address counts at the left and right boundaries of the entire peak area are also approximately equal. Third, the present invention does not need to accurately determine the boundaries of the peak area, and usually selects a pure baseline area near the boundary to set constraints to achieve better results. These three advantages also improve the applicability of the present invention in practical applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 This is an example of a polynomial baseline synthetic spectrum;

[0027] Figure 2 This is an example of an exponential baseline synthetic spectrum;

[0028] Figure 3 This is the result diagram of the estimation of the hydrogen peak baseline of the total spectrum of carbon-oxygen ratio;

[0029] Figure 4 This is the result diagram of the estimation of the hydrogen peak baseline of the carbon-oxygen ratio capture spectrum;

[0030] Figure 5 This is the hydrogen peak fitting result diagram of the total spectrum of carbon-oxygen ratio and the capture spectrum after baseline correction. DETAILED DESCRIPTION

[0031] 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.

[0032] Application Example 1:

[0033] In order to illustrate the baseline estimation effect of the present invention, two types of baselines and characteristic peaks are first used to form synthetic spectra for verification. The baseline types are exponential baseline and polynomial baseline respectively. The characteristic peaks in the energy spectrum are represented by Gaussian functions. The parameters of the synthetic spectrum are shown in Table 1. Four different energy spectrum characteristic peaks are designed in this paper, namely, a narrow single peak (Peak 1), a wide single peak (Peak 2), partially overlapping peaks (Peak 3, Peak 4) and overlapping peaks (Peak 5, Peak 6).

[0034] Table 1 Characteristic peak parameters of synthetic spectrum

[0035]

[0036] The expressions of the two types of baselines are:

[0037] b 1 =-0.0006i 2 +0.1i+130

[0038]

[0039] The synthetic spectrum containing two types of baselines is as follows Figure 1 and Figure 2 As shown, the vertical dotted line in the figure represents the boundary of the peak area, and the dotted line is the baseline. In order to prove the baseline estimation effect of the present invention, the baseline estimation effects of several commonly used baseline correction methods are compared with the present invention, mainly including three algorithms: asLS, arPLS, and SNIP. The parameter settings of the algorithms used and the root mean square errors (RMSE) of the two types of estimated baselines are shown in Table 2. It can be seen from the calculation results that the present invention has the smallest RMSE in the processing of both types of baselines. Compared with arPLS, the RMSE of the polynomial baseline is reduced by 69.04%; the RMSE of the processing result of the exponential baseline is close to 1 / 10 of that of the arPLS algorithm. This shows that the estimation result of the present invention is closest to the actual baseline, and also shows that the corrected characteristic peak has the best fidelity.

[0040] Table 2 Baseline estimation RMSE of different methods

[0041]

[0042] Application Example 2:

[0043] The present invention is used to estimate the hydrogen peak baseline in the total spectrum and the captured spectrum collected at a certain depth of a measured well. The baseline estimation results of the total spectrum and the captured spectrum are as follows: Figure 3 and Figure 4 As shown. It can be seen that there are obvious peaks at the characteristic energy of hydrogen element in both the total spectrum and the capture spectrum, which indicates that the gamma rays released by the capture nuclear reaction exist in these two energy spectra. The baseline under the characteristic peak of hydrogen element in the total spectrum and the capture spectrum is estimated by the present invention, and the estimation results are shown as follows: Figure 3 and Figure 4 The baseline correction of the hydrogen peaks in the total spectrum and the captured spectrum was performed according to the estimated baseline. Figure 5 The results of the total spectrum of carbon-oxygen ratio and the captured spectrum after baseline correction are shown, and the peak curve of the hydrogen peak is fitted in the figure. Through calculation, it can be seen that the correlation coefficients of the fitting results of the hydrogen peak in the total spectrum and the captured spectrum can reach 0.8965 and 0.9652 respectively. This shows that the data after baseline correction can better reflect the peak shape characteristics of the hydrogen peak, which also proves the accuracy of the baseline estimation results of the present invention.

Claims

1. A method for estimating the hydrogen peak baseline in a carbon-oxygen ratio energy spectrum, characterized in that The following steps are involved: S1: Initialize the weight matrix W, which represents the weight of a certain data in the baseline estimation. For an energy spectrum containing m data channels, the weight of each data channel can be initialized to 1 and represented as an m×m diagonal matrix: S2: Initialize the second-order difference matrix D. For an energy spectrum containing m channels of data, the matrix can be expressed as an m-2×m matrix: S3: Select a pure baseline region without peak shape features on both sides of the hydrogen peak in the carbon-oxygen ratio spectrum, and determine the address of the data point therein; S4: Initialize the peak region boundary matrix E, and use the symmetry of the characteristic peak of the energy spectrum to set additional constraints for the objective function of baseline estimation. According to the channel address selected in S3, the peak region boundary matrix E can be expressed as: For a carbon-oxygen ratio spectrum containing m channels of data, the number of columns of the matrix is ​​equal to the number of channels m of the spectrum, and the number of rows is equal to the number of selected peak regions n. In the peak region boundary matrix E, the elements at the left boundary of the peak region are set to 1, the elements at the right boundary of the peak region are set to -1, and the elements at the remaining positions are set to 0. In actual processing, multiple channels near the boundary can also be selected to improve the statistical accuracy. S5: Construct an objective function. According to the data of each channel of the energy spectrum, the objective function of baseline estimation can be constructed through the second-order difference matrix D, the peak area boundary matrix E, and the weight matrix W. The objective function proposed by a method for estimating the baseline of the hydrogen peak in the carbon-oxygen ratio energy spectrum is expressed as: F(z)=(xz) T W(xz)+λ1z T D T Dz+λ2(xz) T E T Ε(xz) In the formula, λ1 and λ2 are the smoothing factor and symmetry factor, which respectively describe the smoothness of the estimated baseline and the symmetry of the energy spectrum characteristic peak, x is the energy spectrum data vector, and z is the vector composed of the estimated baseline data; S6: Solve for the minimum value of the objective function and derive the objective function to obtain the representation of the estimated baseline: z=(W+λ1D T D+λ2E T E) -1 (λ2E T Ex+Wx) The weight matrix W is iteratively updated by selecting a suitable weight update method until the difference between the two iteration results is less than the given error δ, then the loop is terminated and the estimated baseline is output. A weight update method proposed by a method for estimating the hydrogen peak baseline in a carbon-oxygen ratio energy spectrum is: In the formula, For the set d - The mean of For the set d - The standard deviation of the set d is xz, d - To satisfy x i <z i The function tanh is a hyperbolic tangent function, which is used to update the weight value of each channel according to the above parameters.

Images