Reservoir fluid identification method and system based on nuclear magnetic T2 spectrum morphological parameters

Abstract

The invention discloses a reservoir fluid identification method and system based on nuclear magnetic T2 spectrum morphological parameters, and the method comprises the following steps: calculating nuclear magnetic T2 spectrum morphological parameters based on nuclear magnetic logging data; constructing sample data according to the interpretation conclusion of the reservoir position and the T2 spectrum morphological parameters; training a pre-established fluid identification model by using the sample data; and inputting the target nuclear magnetic logging data into the trained fluid identification model to obtain a reservoir fluid identification result. And the fluid identification model is a Transform model. According to the method, an intelligent learning algorithm is adopted, rapid and effective reservoir fluid identification is realized based on nuclear magnetic T2 spectrum morphological parameters, the influence of human experience factors is reduced, and the fluid identification model can learn the form of the nuclear magnetic T2 spectrum corresponding to each explanation conclusion according to the label value; therefore, the reservoir fluid property can be quickly and effectively determined according to the morphological parameters of the new well T2 spectrum.

Description

Technical Field

[0001] The invention belongs to the technical field of petroleum well logging, and in particular relates to a reservoir fluid identification method and system based on nuclear magnetic T2 spectrum morphological parameters. Background Art

[0002] Reservoir fluid identification is an important part of logging interpretation and provides an important basis for reservoir evaluation. At present, the methods of using logging information to identify reservoir fluid properties mainly include resistivity and nuclear magnetic resonance logging methods. Among them, resistivity logging data can more intuitively identify reservoir fluid properties, but the measurement of resistivity is easily affected by factors such as rock skeleton. Nuclear magnetic resonance logging can directly measure hydrogen signals in the formation, and the measurement results are not affected by the rock skeleton, which has obvious advantages in determining fluid properties.

[0003] At present, the main methods for fluid identification using nuclear magnetic resonance logging data are differential spectrum method, shift spectrum method and nuclear magnetic water spectrum method. Among them, the differential spectrum method is to identify fluids based on the differential spectrum obtained by subtracting the long waiting time T2 spectrum from the short waiting time T2 spectrum, because all fluids in the formation are polarized under the long waiting time, and only water is polarized in the formation under the short waiting time, while oil and gas are not polarized, so the differential spectrum contains information about oil and gas. However, this method can only be used for qualitative identification of light hydrocarbons in the formation, because the determination of long and short waiting times is easily affected by human experience factors, resulting in water information in the differential spectrum, especially in large pore water layers. It is difficult for water to be fully polarized under short waiting times, and obvious differential spectrum signals will also appear at this time.

[0004] The spectrum shifting method uses the difference in diffusion coefficients to distinguish water from oil and gas or to distinguish heavy oil from water. This method collects two columns of echoes with the same waiting time but different echo intervals. The T2 spectrum with a long echo interval is shifted to the left and compressed due to the influence of diffusion. The diffusion coefficients of oil, gas and water in the formation are different, so oil, gas and water can be distinguished based on this. However, this method is more effective for gas and heavy oil. When the diffusion coefficients of light oil and water are close, it is difficult to distinguish between oil and water, especially the determination of the long and short echo intervals is easily affected by human experience, resulting in unclear spectrum shifting. In addition, the recognition results of the spectrum shifting method are greatly affected by the quality of the instrument measurement data, especially for eccentric nuclear magnetic logging instruments. Due to the small detection area, the signal-to-noise ratio of the measurement signal is lower, which makes the spectrum shifting method difficult to apply in practice.

[0005] The nuclear magnetic water spectrum method calculates the theoretical water spectrum under completely water-bearing formation conditions by measuring the parameters of the long waiting time short echo interval T2 spectrum and the long waiting time long echo interval component, and compares it with the long waiting time long echo interval T2 spectrum to determine the reservoir fluid properties. This method requires complex calculations and assumes that the long waiting time short echo interval T2 spectrum is all water signals, which is inconsistent with the actual situation. Therefore, the scope of application of this method is also limited, and the quality requirements of instrument measurement data are high.

[0006] In addition to the above methods, different oil fields have also summarized some empirical methods based on actual geological conditions, but they are only effective for individual areas and the versatility of the methods is poor. Therefore, further research is needed on the NMR logging fluid identification method to fully tap the application potential of NMR logging data and realize a fast and effective fluid identification method based on NMR logging data. Summary of the invention

[0007] The purpose of the present invention is to address the problems in the above-mentioned prior art and provide a reservoir fluid identification method and system based on nuclear magnetic resonance T2 spectrum morphological parameters, so as to achieve rapid and effective reservoir fluid identification based on nuclear magnetic resonance T2 spectrum morphological parameters, reduce the influence of human experience factors, and fully tap the application potential of nuclear magnetic resonance logging data.

[0008] In order to achieve the above object, the present invention has the following technical solutions:

[0009] A reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters comprises the following steps:

[0010] Based on the nuclear magnetic logging data, calculate the nuclear magnetic T2 spectrum morphological parameters;

[0011] Construct sample data based on the interpretation conclusions of reservoir location and T2 spectrum morphological parameters;

[0012] Use sample data to train the pre-established fluid recognition model;

[0013] The target NMR logging data is input into the trained fluid identification model to obtain the reservoir fluid identification results.

[0014] Preferably, the nuclear magnetic resonance T2 spectrum morphological parameters include the maximum non-zero time of the T2 spectrum, the time of the last peak of the T2 spectrum, the ordinal number of the last peak of the T2 spectrum, the amplitude of the last peak of the T2 spectrum, the amplitude ratio of the last peak of the T2 spectrum, the time of the first peak of the T2 spectrum, the ordinal number of the first peak of the T2 spectrum, the amplitude of the first peak of the T2 spectrum, the amplitude ratio of the first peak of the T2 spectrum, the median time of the T2 spectrum, the right half-amplitude point offset of the last peak of the T2 spectrum, the left half-amplitude point offset of the last peak of the T2 spectrum, the time difference between the first and last peaks of the T2 spectrum, the ordinal difference between the first and last peaks of the T2 spectrum, the amplitude difference between the first and last peaks of the T2 spectrum, the amplitude ratio of the first and last peaks of the T2 spectrum, the average time of the T2 spectrum, the geometric mean of the T2 spectrum, the distribution coefficient of the T2 spectrum, the relative distribution coefficient of the T2 spectrum, the skewness of the T2 spectrum, the peak state of the T2 spectrum and the macropore coefficient of the T2 spectrum.

[0015] Furthermore, the maximum non-zero time of the T2 spectrum is the relaxation time corresponding to the last non-zero value on the T2 spectrum;

[0016] The last peak time of T2 spectrum is the relaxation time corresponding to the last peak on T2 spectrum;

[0017] The last peak number of the T2 spectrum is the number of the last peak in the T2 spectrum;

[0018] The amplitude of the final peak of the T2 spectrum is the porosity corresponding to the last peak on the T2 spectrum;

[0019] The proportion of the amplitude of the last peak of the T2 spectrum is the ratio of the porosity corresponding to the last peak on the T2 spectrum to the total porosity;

[0020] The first peak time of T2 spectrum is the relaxation time corresponding to the first peak on the T2 spectrum;

[0021] The first peak number of the T2 spectrum is the number of the first peak on the T2 spectrum in the T2 spectrum;

[0022] The amplitude of the final peak of the T2 spectrum is the porosity corresponding to the first peak on the T2 spectrum;

[0023] The first peak amplitude ratio of T2 spectrum is the ratio of the porosity corresponding to the first peak on T2 spectrum to the total porosity;

[0024] The median time of the T2 spectrum is the relaxation time corresponding to half the total porosity when the T2 spectrum is accumulated from the front to the back.

[0025] The offset of the right half-amplitude point of the last peak of the T2 spectrum is the difference between the corresponding sequence of the half-amplitude point on the right half of the last peak on the T2 spectrum and the sequence of the last peak of the T2 spectrum;

[0026] The offset of the left half amplitude point of the last peak of the T2 spectrum is the difference between the corresponding sequence number of the left half amplitude point of the last peak on the T2 spectrum and the sequence number of the last peak of the T2 spectrum;

[0027] The time difference between the first and last peaks of the T2 spectrum is the relaxation time difference between the last peak and the first peak on the T2 spectrum;

[0028] The ordinal difference between the first and last peaks of the T2 spectrum is the difference between the ordinal numbers of the last peak and the first peak on the T2 spectrum;

[0029] The amplitude difference between the first and last peaks of the T2 spectrum is the porosity difference between the last peak and the first peak on the T2 spectrum;

[0030] The amplitude ratio of the first and last peaks of the T2 spectrum is the porosity ratio corresponding to the last peak and the first peak on the T2 spectrum;

[0031] T2 spectrum average time is the average value of T2 relaxation time calculated by weighted average method;

[0032] The geometric mean of the T2 spectrum is the average value of the T2 relaxation time calculated using the geometric mean method;

[0033] The T2 spectrum distribution coefficient is a parameter that represents the concentration of the porosity component of the standard T2 spectrum, describing the degree of dispersion centered on the mean;

[0034] The relative distribution coefficient of T2 spectrum is the ratio of the distribution coefficient of T2 spectrum to the mean value of the pore component of T2 spectrum;

[0035] T2 spectrum skewness is a parameter that characterizes the symmetry of T2 spectrum distribution. If the T2 spectrum distribution is symmetrical, the value of T2 spectrum skewness is 0. A positive value indicates that the T2 spectrum distribution is more biased towards the macropore part, which is coarse skewness. A negative value indicates fine skewness.

[0036] T2 spectrum peak state is a parameter indicating the steepness of the T2 spectrum distribution curve. A value of T2 spectrum peak state greater than 1 indicates that the T2 spectrum is a high-peak curve, and a value less than 1 indicates a gentle peak or double-peak curve;

[0037] The macropore coefficient of T2 spectrum indicates the proportion of macropores in the upper part of T2 spectrum.

[0038] Preferably, when constructing sample data according to the interpretation conclusion of the reservoir position and the T2 spectrum morphological parameters, the nuclear magnetic resonance T2 spectrum morphological parameters are calculated for the T2 spectrum data of the three components A, B, and D in the nuclear magnetic resonance logging data respectively; component A represents a long waiting time and a short echo interval, component B represents a short waiting time and a short echo interval, and component D represents a long waiting time and a long echo interval. If component D is not measured, only components A and B are calculated; the porosity values ​​of the T2 spectrum intervals of the three components A, B, and D are calculated respectively; and the total porosity, effective porosity, mud bound water porosity and permeability calculated from the T2 spectrum of component A are taken; a one-dimensional curve corresponding to the depth distribution of each parameter is obtained as the attribute value; each reservoir is defined as a sample, each reservoir contains multiple depth points, and the depth of each reservoir is different. No matter how many depth points there are originally in the reservoir, all corresponding attribute values ​​in the current reservoir are interpolated to a length equal to the number of attribute values ​​to form sample data with the same dimension.

[0039] Preferably, the sample data determines the corresponding sample label according to the interpretation conclusion, and the interpretation conclusion includes a dry layer, a water layer, an oil layer, a poor oil layer, a gas layer and a poor gas layer, wherein the poor oil layer includes an oil-containing water layer and a water-containing oil layer, and the poor gas layer includes a gas-containing water layer and a water-containing gas layer.

[0040] Preferably, the sample data is selected in accordance with the following principles: the expanded diameter section data is not selected; the section data with tailing in the T2 spectrum is not selected; and the data length of the T2 spectrum is consistent, and data with inconsistent lengths do not appear in the same sample set.

[0041] Preferably, the fluid identification model is a Transformer model, and the step of training the pre-established fluid identification model using sample data includes:

[0042] Perform statistics on each attribute in the sample data, obtain the maximum and minimum values ​​of each attribute of the sample data, and save them in the sample data file;

[0043] Read all data from the sample data file, define all data with continuous depth and consistent labels in the sample data file of the same well as a sample, specify the label as the label value of the current single point data, and assume that the number of samples read in is N after all wells are read; interpolate each attribute value in each sample to a length equal to the number of attribute values, forming N sample data with the same dimensions;

[0044] Scatter and sort the N sample data read into the memory; randomly select a part of the samples as the training set data, and select another part of the samples as the validation set data;

[0045] Input the training set data and validation set data into the model for learning;

[0046] Save the learned model as an h5 model file.

[0047] Preferably, the step of inputting the target nuclear magnetic resonance logging data into the trained fluid identification model to obtain the reservoir fluid identification result comprises:

[0048] By performing standard T2 spectrum analysis and T2 spectrum morphological parameter calculation on the target nuclear magnetic logging data, the nuclear magnetic porosity and permeability curve and the interval porosity and morphological parameter curves of the three components A, B and D are generated;

[0049] Divide the reservoir based on the target NMR logging data;

[0050] Read data according to the reservoir position, and the data of each reservoir is a sample;

[0051] Read the sample data file to obtain the normalized maximum and minimum values ​​of each curve, and perform normalization processing on each curve;

[0052] Interpolate the attribute value of the current sample from the original length to a length equal to the number of attribute values ​​to form usable sample data with the same dimension;

[0053] Call the saved h5 model file and load the available sample data into the trained fluid identification model for calculation to obtain the reservoir fluid identification results.

[0054] A reservoir fluid identification system based on nuclear magnetic T2 spectrum morphological parameters, comprising:

[0055] The NMR T2 spectrum morphological parameter calculation module is used to calculate the NMR T2 spectrum morphological parameters based on the NMR logging data;

[0056] A sample data construction module is used to construct sample data according to the interpretation conclusion of the reservoir position and T2 spectrum morphological parameters;

[0057] A model training module is used to train a pre-established fluid recognition model using sample data;

[0058] The model output module is used to input the target nuclear magnetic logging data into the trained fluid identification model to obtain the reservoir fluid identification results.

[0059] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters is implemented.

[0060] Compared with the prior art, the present invention has at least the following beneficial effects:

[0061] The nuclear magnetic logging data is used to calculate the nuclear magnetic T2 spectrum morphological parameters, and then the sample data is constructed according to the interpretation conclusion of the reservoir position and the T2 spectrum morphological parameters, and the fluid identification model is trained. The fluid type judgment problem of the reservoir segment formed by the correlation of data at different depth points is solved by the pre-established fluid identification model. Each reservoir is defined as a sample, and each reservoir contains multiple depth points. Considering the relationship between different depth points, the new target nuclear magnetic logging data is directly input into the trained fluid identification model to obtain the reservoir fluid identification result. The present invention adopts an intelligent learning algorithm to reduce the influence of human experience factors. The fluid identification model can learn the morphology of the nuclear magnetic T2 spectrum corresponding to each interpretation conclusion according to the label value, thereby quickly and effectively determining the reservoir fluid properties according to the morphological parameters of the new well T2 spectrum. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0063] Figure 1 Flow chart of a reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to an embodiment of the present invention;

[0064] Figure 2 Example diagram of T2 spectrum morphological parameters according to an embodiment of the present invention;

[0065] Figure 3 An example diagram of the sample data size unification process according to an embodiment of the present invention;

[0066] Figure 4 An example of a T2 spectrum with serious diameter expansion effect in an embodiment of the present invention;

[0067] Figure 5 The present invention embodiment has a T2 spectrum example diagram with tailing;

[0068] Figure 6 A schematic diagram of the neural network structure of the fluid recognition model according to an embodiment of the present invention; DETAILED DESCRIPTION

[0069] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, ordinary technicians in this field can also obtain other embodiments without making creative work.

[0070] See also Figure 1In order to fully explore the application potential of nuclear magnetic logging data and reduce the influence of human experience factors, the embodiment of the present invention realizes rapid and effective reservoir fluid identification based on nuclear magnetic T2 spectrum morphological parameters. A reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters is proposed, which mainly includes the following steps:

[0071] Based on the nuclear magnetic logging data, calculate the nuclear magnetic T2 spectrum morphological parameters;

[0072] Construct sample data based on the interpretation conclusions of reservoir location and T2 spectrum morphological parameters;

[0073] Use sample data to train the pre-established fluid recognition model;

[0074] The target NMR logging data is input into the trained fluid identification model to obtain the reservoir fluid identification results.

[0075] Furthermore, when calculating the nuclear magnetic T2 spectrum morphology parameters, the parameters that can quantitatively describe the T2 spectrum morphology are calculated based on the 200-point T2 spectrum in the nuclear magnetic logging data (data of MRT and iMRT nuclear magnetic logging instruments of China Petroleum Logging Company and data of MRIL-P and MRIL-XL nuclear magnetic logging instruments of Halliburton) or the 64-point T2 spectrum (data of Schlumberger logging instruments), including the maximum non-zero time of the T2 spectrum, the end peak time of the T2 spectrum, the end peak sequence of the T2 spectrum, the end peak amplitude of the T2 spectrum, and the end peak amplitude of the T2 spectrum. There are 23 parameters in total, including proportion, T2 spectrum first peak time, T2 spectrum first peak ordinal number, T2 spectrum first peak amplitude, T2 spectrum first peak amplitude proportion, T2 spectrum median time, T2 spectrum last peak right half amplitude point offset, T2 spectrum last peak left half amplitude point offset, T2 spectrum first and last peak time difference, T2 spectrum first and last peak ordinal difference, T2 spectrum first and last peak amplitude difference, T2 spectrum first and last peak amplitude ratio, T2 spectrum average time, T2 spectrum geometric mean, T2 spectrum distribution coefficient, T2 spectrum relative distribution coefficient, T2 spectrum skewness, T2 spectrum peak state, and T2 spectrum large pore coefficient.

[0076] The specific meaning of each parameter is as follows:

[0077] (1) Maximum non-zero time of T2 spectrum (T2TMAX): the relaxation time corresponding to the last non-zero value on the T2 spectrum;

[0078] (2) T2 spectrum last peak time (T2LPT): the relaxation time corresponding to the last peak on the T2 spectrum;

[0079] (3) T2 spectrum last peak number (T2LPP): the number of the last peak in the T2 spectrum;

[0080] (4) T2 spectrum last peak amplitude (T2LPV): the porosity corresponding to the last peak on the T2 spectrum;

[0081] (5) T2 spectrum last peak amplitude ratio (T2LPVR): the ratio of the porosity corresponding to the last peak on the T2 spectrum to the total porosity;

[0082] (6) T2 spectrum first peak time (T2BPRPT): the relaxation time corresponding to the first peak on the T2 spectrum;

[0083] (7) T2 spectrum first peak number (T2BPRPP): the number of the first peak in the T2 spectrum;

[0084] (8) T2 spectrum final peak amplitude (T2BPRPV): the porosity corresponding to the first peak on the T2 spectrum;

[0085] (9) T2 spectrum first peak amplitude ratio (T2BPRPVR): the ratio of the porosity corresponding to the first peak on the T2 spectrum to the total porosity;

[0086] (10) T2 spectrum median time (T2MVT): T2 spectra are accumulated from front to back, and the accumulated result is the relaxation time corresponding to half of the total porosity;

[0087] (11) T2 spectrum last peak right half amplitude point offset (T2LPROFF): the difference between the corresponding sequence of the right half amplitude point of the last peak on the T2 spectrum and the sequence of the last peak of the T2 spectrum;

[0088] (12) T2 spectrum last peak left half amplitude point offset (T2LPLOFF): the difference between the corresponding sequence of the left half amplitude point of the last peak on the T2 spectrum and the sequence of the last peak of the T2 spectrum;

[0089] (13) T2 spectrum first and last peak time difference (T2LFPTD): the relaxation time difference between the last peak and the first peak on the T2 spectrum;

[0090] (14) T2 spectrum first and last peak ordinal difference (T2LFPPD): the difference between the ordinal numbers of the last peak and the first peak on the T2 spectrum;

[0091] (15) T2 spectrum first and last peak amplitude difference (T2LFPVD): the porosity difference between the last peak and the first peak on the T2 spectrum;

[0092] (16) T2 spectrum first to last peak amplitude ratio (T2LFPVR): the porosity ratio corresponding to the last peak and the first peak on the T2 spectrum;

[0093] (17) T2 spectrum average time (T2AVG): The average value of T2 relaxation time calculated by weighted average method. The calculation expression is as follows:

[0094]

[0095] Where, T2,i is the time of T2 spectrum distribution, is the porosity component value of the i-th point, i=0,1,2…200.

[0096] (18) T2 spectrum geometric mean (T2LM): The mean value of T2 relaxation time calculated by the geometric mean method. The calculation expression is as follows:

[0097]

[0098] Where, T 2,i is the time of T2 spectrum distribution, is the porosity component value of the i-th point, i=0,1,2…200.

[0099] (19) T2 spectrum distribution coefficient (T2SP): a parameter of the concentration degree of the porosity component of the standard T2 spectrum, describing the degree of dispersion centered on the mean. The calculation expression is as follows:

[0100]

[0101] (20) T2 spectrum relative distribution coefficient (T2SPD): The ratio of the T2 spectrum distribution coefficient to the mean value of the T2 spectrum pore component. The calculation expression is as follows:

[0102]

[0103] (21) T2 spectral skewness (T2SKP): a parameter that characterizes the symmetry of T2 spectral distribution. If the T2 spectral distribution is symmetrical, its value is 0. If it is a positive value, it means that the T2 spectral distribution is more biased towards the macropore part, that is, coarse skewness, and a negative value indicates fine skewness. The calculation expression is as follows:

[0104]

[0105] (22) T2 spectrum peak state (T2KG): a parameter indicating the steepness of the T2 spectrum distribution curve, that is, the ratio of the distribution amplitude of the tail (thick tail and thin tail) of the distribution curve to the central amplitude. A value greater than 1 indicates that the T2 spectrum is a high peak curve, and a value less than 1 indicates a gentle peak or a double peak curve. The calculation expression is as follows:

[0106]

[0107] (23) T2 spectrum macropore ratio (T2BPR): represents the proportion of macropores in the upper part of the T2 spectrum; the calculation expression is as follows:

[0108]

[0109] In a possible implementation, regarding sample data generation, the T2 spectrum data of the three components A (long waiting time and short echo interval), B (short waiting time and short echo interval), and D (long waiting time and long echo interval) in the nuclear magnetic resonance logging data (if the D component is not measured, only the A and B components are calculated) are calculated as 23 T2 spectrum morphological parameters as above, for a total of 69 parameters; then the porosity values ​​of the T2 spectra of the three components A, B, and D in the 9 intervals after 8ms are taken, for a total of 27 parameters; and the total porosity (MSIGTA), effective porosity (MPHITA), mud bound water porosity (MCBWA) and permeability (PERM) calculated from the T2 spectrum of the A component are taken, for a total of 4 parameters. A total of 100 curves, all of which are one-dimensional curves distributed by depth, are used as 100 attribute values ​​participating in intelligent calculations, which are:

[0110]

[0111] According to the well logging data with determined interpretation conclusions, each reservoir is defined as one sample. Each reservoir contains multiple depth points. The depth of each reservoir is different. Therefore, in order to unify the sample size, no matter how many depth points there are in the reservoir, the above 100 attribute values ​​in the current reservoir are interpolated to a length of 100. The process of unifying the sample data size in the embodiment of the present invention is as follows: Figure 3 As shown, sample data with a dimension of 100*100 is formed.

[0112] The sample labels are determined according to the interpretation conclusions, where poor oil layers include oil-bearing water layers and water-bearing oil layers, and poor gas layers include gas-bearing water layers and water-bearing gas layers. This is done to prevent the problem of low recognition accuracy caused by data imbalance.

[0113] The one-to-one correspondence between sample labels and explanation conclusions is:

[0114] Explain the conclusion Tag value Dry layer 0 Water layer 1 Oil layer 2 Poor oil layer (including oil-water layer and water-oil layer) 3 Air layer 4 Poor gas layer (including gas-water layer and water-gas layer) 5

[0115] To facilitate storage and data conversion, the sample data is stored in a “.csv” table format and in depth order. After the data is read into memory, it is reprocessed into a single sample point.

[0116] The data of each well is saved in a separate file. The first column in the table is the depth, followed by the A group T2 spectrum porosity curve, interval porosity curve, T2 spectrum morphological parameter curve, Label, etc. The specific format is as follows:

[0117] Depth MSIGTA …… P1PRF …… P10DF T2MAX_A …… T2LPT_B …… T2BPR_D Label 2501.5 11.77 …… 1.03 …… 0 500.2 …… 321.5 …… 0.56 1 2501.6 11.49 …… 1.03 …… 0 460.8 …… 280.8 …… 0.78 1 2501.7 11.21 …… 1.11 …… 0 405.6 …… 214.5 …… 0.61 1 …… …… …… …… …… …… …… …… …… …… …… …… 2501.5 8.73 …… 0.08 …… 0 810.8 …… 221.5 …… 0.16 2 2501.6 9.19 …… 0.13 …… 0 760.7 …… 270.8 …… 0.88 2 2501.7 9.35 …… 0.51 …… 0 805.9 …… 314.5 …… 0.69 2 …… …… …… …… …… …… …… …… …… …… …… ……

[0118] The sample data of the embodiment of the present invention follows the following principles when selecting:

[0119] (1) The expansion section is not selected, such as Figure 4 As shown: if the reservoir section expands, the data of this section will not be selected as a sample;

[0120] (2) The well data with tailing in the T2 spectrum are not selected: If the tailing of the back end of the reservoir T2 spectrum is serious and the following points increase in sequence, such as Figure 5 If the data is not collected by the instrument, the quality may be low, so the data will not be selected as a sample.

[0121] (3) The data instrument types selected as samples must be consistent or similar, ensuring that all T2 spectra are 200-point data (China Petroleum Logging Company MRT, iMRT nuclear magnetic logging instrument data and Halliburton MRIL-P, MRIL-XL nuclear magnetic logging instrument data) or 64-point data (Schlumberger logging instrument data). 200-point data and 64-point data T2 spectra cannot appear in the same sample set;

[0122] Furthermore, in the process of establishing the fluid identification model, the problem of the present invention is to determine the fluid type of the reservoir segment formed by the correlation of data at different depth points. It is necessary to consider the relationship between different depth points. Therefore, in order to improve the calculation accuracy, the embodiment of the present invention adopts the Transformer model with super sequence modeling ability and global information perception ability for modeling. The neural network structure of the specific fluid identification model is as follows: Figure 6 shown.

[0123] In addition, since the data is interpolated during the sample construction process, in order to increase the calculation speed, convolution and pooling layers are added before the data is input into the transformer model to reduce information redundancy and improve the calculation speed.

[0124] After the model is built, follow the steps below to learn it:

[0125] (1) Data normalization: To ensure that the training data and application data use the same normalization parameters, firstly, the 100 attributes in the sample data are counted, their maximum and minimum values ​​are obtained, and saved in the file normalize.txt;

[0126] (2) Read all data from the sample data file, define all data with continuous depth and consistent labels in the same well file as a sample, specify the label as the label value of the current single point data, and assume that the number of samples read in is N after all wells are read;

[0127] (3) Each attribute value in each sample is interpolated to a length of 100, forming N sample data of size 100*100;

[0128] (4) Scatter and sort the N sample data read into the memory;

[0129] (5) Randomly select 80% of the samples as the training set data TrainDataSample, and select 20% of the samples as the validation set data ValiDataSample;

[0130] (6) Input the training set data and the validation set data into the model for learning;

[0131] (7) Save the learned model as an h5 model file.

[0132] Finally, after building the model using the above method, the model is applied to the new well data. The specific calculation method is as follows:

[0133] (1) In LEAD4.0, standard T2 spectrum analysis and T2 spectrum morphological parameter calculation were performed on the new well data through the "standard T2 spectrum analysis" and "T2 spectrum morphological parameter calculation" modules to generate nuclear magnetic porosity and permeability curves and interval porosity and morphological parameter curves of the three components A, B, and D;

[0134] (2) Divide the reservoir of the new well data based on conventional data and nuclear magnetic logging data;

[0135] (3) Read data according to the reservoir position. The data of each reservoir is a sample. If the current reservoir has M depth points, the size of the current sample is M*100.

[0136] (4) Read the file normalize.txt to obtain the maximum and minimum values ​​of each normalized curve, and perform normalization on each curve;

[0137] (5) Interpolate the 100 attribute values ​​of the current sample from a length of M to a length of 100, forming a 100*100 usable sample;

[0138] (6) Call the saved h5 model and load the read data into the model for calculation to obtain the calculation results of all samples;

[0139] (7) Write the obtained results and the corresponding depth of the reservoir into the "Interpretation Conclusion" table in LEAD4.0.

[0140] The following is a complete description of the technical solution and process of the present invention in conjunction with the actual well data processing process of a block:

[0141] 1) Select the nuclear magnetic logging data and conventional logging data of 40 wells in a block, ensure that all 40 wells contain interpretation conclusions, of which 35 wells are used as training wells and 5 wells are used as test wells, and load all the data into LEAD4.0 software;

[0142] 2) The data of 40 wells were processed using the "standard T2 spectrum analysis" and "T2 spectrum morphological parameter calculation" modules developed based on the LEAD4.0 platform to obtain the porosity and permeability parameters, the interval porosity of the three components of ABD, and the T2 spectrum morphological parameters;

[0143] 3) Extract sample data containing 100 attribute values ​​of the reservoir location based on the interpretation conclusion data of the 40 wells. The sample data of each well is saved as an independent .csv file according to the format, and a total of 40 sample files are generated;

[0144] 4) Develop intelligent fluid identification algorithm based on Python script and Tensorflow framework in LEAD4.0 to achieve:

[0145] (1) Read the data of 35 sample files of TrainWell, and read a total of 2780 sample data. Rearrange these samples into a 100*100 two-dimensional array, and then scatter and sort these sample data;

[0146] (2) Randomly select 80% of the samples as the training set data TrainDataSample, a total of 2224 samples, and select 20% of the samples as the validation set data ValiDataSample, a total of 556 samples;

[0147] (3) The training set data and the validation set data were input into the model for learning. The learning accuracy reached 92.53% and the validation accuracy reached 90.48%. The learned model was saved as an h5 model file named "FluidIdentify_T2Shape.h5".

[0148] (4) The “FluidIdentify_T2Shape.h5” file was called to process the five wells in TestWell, and the test accuracy was 83.58%.

[0149] Another embodiment of the present invention further provides a reservoir fluid identification system based on nuclear magnetic T2 spectrum morphological parameters, comprising:

[0150] The NMR T2 spectrum morphological parameter calculation module is used to calculate the NMR T2 spectrum morphological parameters based on the NMR logging data;

[0151] A sample data construction module is used to construct sample data according to the interpretation conclusion of the reservoir position and T2 spectrum morphological parameters;

[0152] A model training module is used to train a pre-established fluid recognition model using sample data;

[0153] The model output module is used to input the target nuclear magnetic logging data into the trained fluid identification model to obtain the reservoir fluid identification results.

[0154] Another embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters is implemented.

[0155] Exemplarily, the instructions stored in the memory may be divided into one or more modules / units, which are stored in a computer-readable storage medium and executed by the processor to complete the reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters of the present invention. The one or more modules / units may be a series of computer-readable instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the server.

[0156] The electronic device may be a computing device such as a smart phone, a notebook, a PDA, and a cloud server. The electronic device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the electronic device may also include more or fewer components, or a combination of certain components, or different components, for example, the electronic device may also include an input / output device, a network access device, a bus, etc.

[0157] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0158] The memory may be an internal storage unit of the server, such as a hard disk or memory of the server. The memory may also be an external storage device of the server, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the server. Furthermore, the memory may include both an internal storage unit of the server and an external storage device. The memory is used to store the computer-readable instructions and other programs and data required by the server. The memory may also be used to temporarily store data that has been output or is to be output.

[0159] It should be noted that the information interaction, execution process and other contents between the above-mentioned module units are based on the same concept as the method embodiment. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0160] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0161] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may at least include: any entity or device that can carry the computer program code to the camera device / terminal device, recording medium, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a disk or an optical disk.

[0162] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0163] The embodiments described above are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, a person skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. Such modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters, characterized in that: The following steps are involved: Based on the nuclear magnetic logging data, calculate the nuclear magnetic T2 spectrum morphological parameters; Construct sample data based on the interpretation conclusions of reservoir location and T2 spectrum morphological parameters; Use sample data to train the pre-established fluid recognition model; The target NMR logging data is input into the trained fluid identification model to obtain the reservoir fluid identification results.

2. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 1 is characterized in that: The morphological parameters of nuclear magnetic resonance T2 spectrum include the maximum non-zero time of T2 spectrum, the time of the last peak of T2 spectrum, the ordinal number of the last peak of T2 spectrum, the amplitude of the last peak of T2 spectrum, the amplitude ratio of the last peak of T2 spectrum, the time of the first peak of T2 spectrum, the ordinal number of the first peak of T2 spectrum, the amplitude of the first peak of T2 spectrum, the amplitude ratio of the first peak of T2 spectrum, the median time of T2 spectrum, the right half amplitude point offset of the last peak of T2 spectrum, the left half amplitude point offset of the last peak of T2 spectrum, the time difference between the first and last peaks of T2 spectrum, the ordinal difference between the first and last peaks of T2 spectrum, the amplitude difference between the first and last peaks of T2 spectrum, the amplitude ratio of the first and last peaks of T2 spectrum, the average time of T2 spectrum, the geometric mean of T2 spectrum, the distribution coefficient of T2 spectrum, the relative distribution coefficient of T2 spectrum, the skewness of T2 spectrum, the peak state of T2 spectrum and the macropore coefficient of T2 spectrum.

3. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 2 is characterized in that: The maximum non-zero time of the T2 spectrum is the relaxation time corresponding to the last non-zero value on the T2 spectrum; The last peak time of T2 spectrum is the relaxation time corresponding to the last peak on T2 spectrum; The last peak number of the T2 spectrum is the number of the last peak in the T2 spectrum; The amplitude of the final peak of the T2 spectrum is the porosity corresponding to the last peak on the T2 spectrum; The proportion of the amplitude of the last peak of the T2 spectrum is the ratio of the porosity corresponding to the last peak on the T2 spectrum to the total porosity; The first peak time of T2 spectrum is the relaxation time corresponding to the first peak on the T2 spectrum; The first peak number of the T2 spectrum is the number of the first peak on the T2 spectrum in the T2 spectrum; The amplitude of the final peak of the T2 spectrum is the porosity corresponding to the first peak on the T2 spectrum; The first peak amplitude ratio of T2 spectrum is the ratio of the porosity corresponding to the first peak on T2 spectrum to the total porosity; The median time of the T2 spectrum is the relaxation time corresponding to half the total porosity when the T2 spectrum is accumulated from the front to the back. The offset of the right half-amplitude point of the last peak of the T2 spectrum is the difference between the corresponding sequence of the half-amplitude point on the right half of the last peak on the T2 spectrum and the sequence of the last peak of the T2 spectrum; The offset of the left half amplitude point of the last peak of the T2 spectrum is the difference between the corresponding sequence number of the left half amplitude point of the last peak on the T2 spectrum and the sequence number of the last peak of the T2 spectrum; The time difference between the first and last peaks of the T2 spectrum is the relaxation time difference between the last peak and the first peak on the T2 spectrum; The ordinal difference between the first and last peaks of the T2 spectrum is the difference between the ordinal numbers of the last peak and the first peak on the T2 spectrum; The amplitude difference between the first and last peaks of the T2 spectrum is the porosity difference between the last peak and the first peak on the T2 spectrum; The amplitude ratio of the first and last peaks of the T2 spectrum is the porosity ratio corresponding to the last peak and the first peak on the T2 spectrum; T2 spectrum average time is the average value of T2 relaxation time calculated by weighted average method; The geometric mean of the T2 spectrum is the average value of the T2 relaxation time calculated using the geometric mean method; The T2 spectrum distribution coefficient is a parameter that represents the concentration of the porosity component of the standard T2 spectrum, describing the degree of dispersion centered on the mean; The relative distribution coefficient of T2 spectrum is the ratio of the distribution coefficient of T2 spectrum to the mean value of the pore component of T2 spectrum; T2 spectrum skewness is a parameter that characterizes the symmetry of T2 spectrum distribution. If the T2 spectrum distribution is symmetrical, the value of T2 spectrum skewness is 0. A positive value indicates that the T2 spectrum distribution is more biased towards the macropore part, which is coarse skewness. A negative value indicates fine skewness. T2 spectrum peak state is a parameter indicating the steepness of the T2 spectrum distribution curve. A value of T2 spectrum peak state greater than 1 indicates that the T2 spectrum is a high-peak curve, and a value less than 1 indicates a gentle peak or double-peak curve; The macropore coefficient of T2 spectrum indicates the proportion of macropores in the upper part of T2 spectrum.

4. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 1 is characterized in that: When constructing sample data according to the interpretation conclusion of the reservoir position and the T2 spectrum morphological parameters, the nuclear magnetic resonance T2 spectrum morphological parameters are calculated for the T2 spectrum data of the three components A, B, and D in the nuclear magnetic resonance logging data respectively; the A component represents a long waiting time and a short echo interval, the B component represents a short waiting time and a short echo interval, and the D component represents a long waiting time and a long echo interval. If the D component is not measured, only the A and B components are calculated; the porosity values ​​of the T2 spectrum intervals of the three components A, B, and D are calculated respectively; and the total porosity, effective porosity, mud bound water porosity and permeability calculated from the T2 spectrum of the A component are taken; a one-dimensional curve corresponding to the depth distribution of each parameter is obtained as the attribute value; each reservoir is defined as a sample, each reservoir contains multiple depth points, and the depth of each reservoir is different. No matter how many depth points there are originally in the reservoir, all corresponding attribute values ​​in the current reservoir are interpolated to a length equal to the number of attribute values ​​to form sample data with the same dimension.

5. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 1 is characterized in that: The sample data determines the corresponding sample label according to the interpretation conclusion, and the interpretation conclusion includes dry layer, water layer, oil layer, poor oil layer, gas layer and poor gas layer. The poor oil layer includes oil-containing water layer and water-containing oil layer, and the poor gas layer includes gas-containing water layer and water-containing gas layer.

6. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 1, characterized in that: The following principles are followed when selecting sample data: the data of the expanded well section are not selected; the data of the well section with tailing in the T2 spectrum are not selected; and the data length of the T2 spectrum is consistent, and data with inconsistent lengths do not appear in the same sample set.

7. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 1, characterized in that: The fluid recognition model is a Transformer model, and the steps of training the pre-established fluid recognition model using sample data include: Perform statistics on each attribute in the sample data, obtain the maximum and minimum values ​​of each attribute of the sample data, and save them in the sample data file; Read all data from the sample data file, define all data with continuous depth and consistent labels in the sample data file of the same well as a sample, specify the label as the label value of the current single point data, and assume that the number of samples read in is N after all wells are read; interpolate each attribute value in each sample to a length equal to the number of attribute values, forming N sample data with the same dimensions; Scatter and sort the N sample data read into the memory; randomly select a part of the samples as the training set data, and select another part of the samples as the validation set data; Input the training set data and validation set data into the model for learning; Save the learned model as an h5 model file.

8. The reservoir fluid identification method based on nuclear magnetic T2 spectrum morphological parameters according to claim 7 is characterized in that: The step of inputting the target nuclear magnetic logging data into the trained fluid identification model to obtain the reservoir fluid identification result comprises: By performing standard T2 spectrum analysis and T2 spectrum morphological parameter calculation on the target nuclear magnetic logging data, the nuclear magnetic porosity and permeability curve and the interval porosity and morphological parameter curves of the three components A, B and D are generated; Divide the reservoir based on the target NMR logging data; Read data according to the reservoir position, and the data of each reservoir is a sample; Read the sample data file to obtain the normalized maximum and minimum values ​​of each curve, and perform normalization processing on each curve; Interpolate the attribute value of the current sample from the original length to a length equal to the number of attribute values ​​to form usable sample data with the same dimension; Call the saved h5 model file and load the available sample data into the trained fluid identification model for calculation to obtain the reservoir fluid identification results.

9. A reservoir fluid identification system based on nuclear magnetic T2 spectrum morphological parameters, characterized in that: include: The NMR T2 spectrum morphological parameter calculation module is used to calculate the NMR T2 spectrum morphological parameters based on the NMR logging data; A sample data construction module is used to construct sample data according to the interpretation conclusion of the reservoir position and T2 spectrum morphological parameters; A model training module is used to train a pre-established fluid recognition model using sample data; The model output module is used to input the target nuclear magnetic logging data into the trained fluid identification model to obtain the reservoir fluid identification results.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method for identifying reservoir fluid based on nuclear magnetic T2 spectrum morphological parameters as described in any one of claims 1 to 8 is implemented.

Images