Improved k-nearest neighbor algorithm for identifying fluid properties in gas reservoirs based on constraints of recorded and measured parameters

Abstract

The present invention provides a gas reservoir fluid property identification method based on an improved k-nearest neighbor algorithm that integrates logging parameter constraints. Step 1: Obtain gas logging and well logging data in a target depth interval; Step 2: Calculate the gas content of underground rock per unit volume; Step 3: Calculate the acoustic time difference porosity, neutron porosity and density porosity, and calculate the mining index; Step 4: Select the gas logging and well logging values ​​at the depth corresponding to the test production data, and calculate the gas content of underground rock per unit volume and the mining index; Step 5: Construct a fluid identification map, calculate the mining index and the gas content of rock per unit volume corresponding to the unknown well, and project them onto the fluid identification map to identify gas layers and water layers; Step 6: Input transitional fluid type samples in non-gas zones and non-water zones in the identification map into the improved k-nearest neighbor calculation module to complete the identification of transitional fluid properties; Repeat steps 2 to 6 to achieve the identification of different fluid properties of the target well and target layer.

Description

Technical Field

[0001] The invention belongs to the technical field of oil and gas field exploration and development, and relates to a gas reservoir fluid property identification method integrating recorded and measured parameter constraints with an improved k-nearest neighbor algorithm. Background Art

[0002] With the deepening of conventional oil and gas resource exploration and development and the accelerated process of energy structure diversification, deep and ultra-deep oil and gas reservoirs with huge resource potential have gradually become the main targets of oil and gas exploration and development at home and abroad, and have attracted widespread attention from the petroleum industry. Deep and ultra-deep reservoirs such as the Miocene Huangliu Formation in the southern section of the Ledong slope belt of the Yinggehai Basin in the South China Sea, the Lingshui Formation in the northern slope fault terrace belt of the Baodao Sag in the Qiongdongnan Basin, the Zhuhai Formation in the southern part of the Wenchang A Sag in the Pearl River Mouth Basin, the Shahejie Formation in the northern steep slope belt of the Yellow River Mouth Sag, and the Wenchang Formation in the Lufeng Sag of the Pearl River Mouth Basin have currently obtained relatively rich natural gas resources, demonstrating the huge exploration and development potential of deep and ultra-deep reservoirs in the sea area, and have gradually become the main exploration direction for many offshore oil fields and workers.

[0003] However, the geological conditions and reservoir relationships of low porosity and permeability reservoirs such as deep and ultra-deep layers are complex, the reservoir lithology and minerals are complex and changeable, the physical properties are severely heterogeneous, the reservoir space is complex, the accumulation process and stages are complex, and it is difficult to identify the fluid properties. Li Xia et al. (2013) used a combination of rock mechanical parameters obtained by array acoustic logging to establish a crossplot to achieve qualitative identification of gas layers in the Sulige gas field, and the quantitative evaluation of gas saturation was obtained by the dual-porosity saturation model (Li Xia, Shi Yujiang, Wang Ling, et al. Logging identification and evaluation technology of tight sandstone gas layers - a case study of the Sulige gas field. Natural Gas Geoscience, 2013, 24(1): 62-68); Fan Zhiqiang et al. (2015) used the P1 / 2 cumulative frequency method, crossplot method and multivariate discrimination method to identify the fluid properties of the Zizhou gas field reservoir (Fan Zhiqiang, Yang Guoping, Ding Xi, et al. Single well fluid identification method and application in the Shan 2 tight sandstone gas reservoir of Zizhou gas field. Natural Gas Geoscience, 2015, 26(06): 1113-1119); Deng Shaogui et al. (2016) obtained T based on nuclear magnetic resonance and acoustic data. 2gBased on the correlation between porosity and shear wave velocity, a nuclear magnetic resonance-acoustic wave joint intersection diagram method was established to identify fluid properties (Deng Shaogui, Niu Yunfeng, Zhao Yue, et al. Joint experiment of nuclear magnetic resonance-acoustic wave velocity in tight gas sandstone. Acta Petrolei Sinica, 2016, 37(06):768-776); Shi Yujiang et al. (2016) conducted wavelet analysis on the product of porosity and resistivity in the tight gas layer of Ordos, and used the analysis results to effectively identify gas and water layers (Shi Yujiang, Pan Baozhi, Jiang Bici, et al. Application of wavelet analysis in identification of tight sandstone gas layers. Geophysical Science, 2016, 41(12): 2127-2135); Yan Xingyu et al. (2019) aimed at the difficulty of evaluating tight sandstone gas reservoirs. They first used the plate method to extract lithologic sensitive parameters as input parameters for the training model, and used the XGBoost algorithm to identify the interpretation conclusions with an accuracy of 86.4% (Yan Xingyu, Gu Hanming, Xiao Yifei, et al. Application of XGBoost algorithm in well logging interpretation of tight sandstone gas reservoirs. Petroleum Geophysical Exploration, 2019, 54(02): 447-455+241). In terms of the integration of logging and well logging, Wang Xianggong et al. (2008) unified the logging and well logging data, established a comprehensive interpretation chart, and achieved certain evaluation results (Wang Xianggong, Wang Xuanran, Jiang Longsheng, et al. Research on quantitative evaluation of oil and water layers using logging and well logging data. Journal of Shandong University of Technology (Natural Science Edition), 2008, 22(06):23-25); Sheng Li (2011) used the MDT test LFA fluid analysis method to identify gas and water layers and achieved good results in Changshen 1 area (Sheng Li. Research on deep fluid identification and reserve evaluation in Changshen 1 area. Northeast Petroleum University , 2011); Gao Chuqiao et al. (2022) solved the multi-solution problem of identifying reservoir fluid properties with a single plate for a large number of deep / ultra-deep reservoirs. Based on the optimization of plates with good fluid property differentiation effect, they proposed a multi-plate comprehensive fluid property identification method that introduced the concept of plate identification reliability, fully utilized the optimized logging information, and improved the accuracy of reservoir fluid identification (Gao Chuqiao, Du Kun, Wang Bo, et al. Complex fluid property identification method based on multi-plate identification reliability analysis. Journal of Yangtze University (Natural Science Edition), 2022: 1-9).

[0004] The above intersection methods based on logging or mud logging or a combination of logging and well logging have good application effects in conventional reservoir fluid identification and can meet engineering needs; however, for low porosity and permeability reservoirs, the effect of using only logging data or a combination of logging and well logging is poor, and the identification accuracy cannot fully meet the existing oil and gas layer exploration needs. In particular, the resolution ability for transitional fluid properties (such as gas and water in the same layer) is limited. Therefore, how to effectively solve the problem of fluid property identification in complex gas reservoirs is a technical problem that needs to be solved urgently in this field. Summary of the Invention

[0005] In order to overcome the problem of poor fluid identification rate based on logging or mud logging or combined logging and logging in the existing technology, the present invention provides an improved k-nearest neighbor algorithm for identifying gas reservoir fluid properties that integrates logging and logging parameter constraints. This method has strong applicability and can effectively identify fluid properties such as gas / water layers, gas-water layers, and gas-water layers. In particular, it solves the problem of identifying transitional fluid properties in low-porosity and complex reservoirs, meeting the engineering needs of gas reservoir exploration and development.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solution: a method for identifying gas reservoir fluid properties using an improved k-nearest neighbor algorithm that integrates recorded and measured parameter constraints, characterized in that it comprises the following steps:

[0007] Step 1: Obtain gas logging and well logging data in the target depth interval;

[0008] Step 2: Calculate the gas content of underground rock per unit volume;

[0009] Step 3: Calculate the sonic transit time porosity, neutron porosity, and density porosity, and calculate the mining index;

[0010] Step 4: Select the gas logging and well logging values ​​at the depth corresponding to the test production data to calculate the gas content of the rock per unit volume and the excavation index;

[0011] Step 5: Construct a fluid identification map, calculate the excavation index and rock gas content per unit volume corresponding to the unknown well, and project them onto the fluid identification map to identify gas and water layers;

[0012] Step 6: Input the transitional fluid type samples that fall in the non-gas area and the non-water area in the identification plate into the improved k-nearest neighbor calculation module to complete the transitional fluid property identification;

[0013] Optionally, the method further includes step 7: repeating steps 2 to 6 to identify different fluid properties of the target well and target layer.

[0014] More specifically, it includes the following steps:

[0015] Step 1: Perform gas logging and conventional logging in the target layer of the target well to obtain gas logging and logging data in the target depth interval;

[0016] Specifically, the gas logging data includes C1, C2, C3, iC4 and nC4 and iC5 and nC5 content curves; the logging data includes acoustic time difference Δt, compensated neutron Bulk density ρ b Well logging and natural gamma ray (GR) logging curves;

[0017] Among them, C1 represents methane (CH4), C2 represents ethane (C2H6), C3 represents propane (C3H8), iC4 represents isobutane (iC4H 10 ), nC4 represents n-butane (nC4H 10 ), iC5 represents isopentane (iC5H 12 ), nC5 represents n-butane (nC5H 12 );

[0018] Step 2: record the gas logging C1, C2, C3, iC4 and nC4 and iC5 and nC5 measurement values ​​and the drill bit diameter D, drilling fluid density ρ, displacement drilling Q, drilling time t, drilling fluid outlet temperature T out , geothermal gradient G t Temperature and pressure parameters, wellbore vertical depth H, and deviation compressibility coefficient Z of each component n Substitute the parameters into formula (1) to calculate the gas content of underground rock per unit volume C ug ;

[0019] C ug =12159∑(Z n C n )tQ(HG t +273.15) / [ρH(T out +273.15)πD2 (1)

[0020] Where D is the drill diameter, mm; T out is the drilling fluid outlet temperature, °C; H is the vertical depth of the wellbore, m; ρ is the drilling fluid density, g / cm 3 ; G t is the geothermal gradient, ℃ / m; Q is the displacement (including the sum of the circulation displacement and the booster pump displacement), L / min; t is the drilling time, min / m; C n (n=1,2,3,4,5,represent CH4、C2H6、C3H8、iC4H 10 、nC4H 10 、iC5H 12 、nC5H 12 ) represents the gas measured value of each hydrocarbon component, ppm; the deviation compressibility coefficient Z of each component n (n=1, 2, 3, 4, 5);

[0021] Step 3: The acoustic time difference Δt and the compensation neutron Bulk density ρ b The well logging and natural gamma logging values ​​are substituted into formula (3), formula (4) and formula (5) to calculate the acoustic time difference porosity Neutron porosity and density porosity Further substitute into formula (6) to calculate the mining index I ex ;

[0022] Step 4: Select the gas logging and typical logging values ​​at the depth corresponding to the test production data, and calculate the gas content C of the rock per unit volume in the underground according to steps 2 and 3. ug and mining index I ex ;

[0023] Step 5, build C ug- I ex Fluid Identification Plate, select I ex is the horizontal axis, C ug As the vertical coordinate, calculate the I corresponding to the unknown well ex and C ug , and projected to C ug- I ex In the fluid identification chart, if the data falls into the gas layer or water layer area, it will be identified as the gas layer or water layer accordingly;

[0024] Step 6, will fall on C ug- I ex The transitional fluid type samples of the non-gas and non-water areas in the image are input into the improved k-nearest neighbor calculation module to complete the identification of transitional fluid properties. Combined with step 5, the fluid properties of gas layers, water layers, gas-water layers, and gas-water layers can be identified.

[0025] Preferably, step 7 is also included: repeating steps 2 to 6 above to analyze the fluid properties at different depths, and effectively identify the fluid properties of the reservoir gas layer, water layer, gas-water layer, and gas-water layer in the target well and target layer.

[0026] Preferably, the specific calculation process of step 3 is as follows:

[0027] Step 301: Calculate the mud content V corresponding to a certain depth point on the natural gamma ray logging curve GR according to formula (2)-1. sh , where GCUR is the Hillch index, where I GR Calculated by formula (2)-2, GR max With GR min Take the natural gamma logging values ​​corresponding to the large set of mudstone layers and the large set of sandstone layers of the target well respectively;

[0028]

[0029] I GR =(GR-GR min ) / (GR max -GR min ) (2)-2

[0030] Step 302: The acoustic time difference Δt and the compensation neutron Bulk density ρ b The logging value and the mud content V at the depth point calculated in step 301 sh Substituting into formulas (3), (4), and (5), the apparent acoustic time difference porosity can be calculated: Neutron porosity and apparent porosity The obtained acoustic transit time porosity Neutron porosity and apparent porosity Substituting into formula (6), we can get the mining index I of the corresponding depth point: ex value.

[0031]

[0032]

[0033]

[0034]

[0035] Among them, Δt, ρ b , the units of GR are us / m, dimensionless, g / cm 3 , API; Δt ma , ρ ma are the rock logging values ​​of acoustic wave time difference, compensated neutron and volume density of rock skeleton, with the same units as above; Δt f , ρ f are the logging values ​​of the fluid acoustic wave time difference, compensated neutron, and volume density respectively; V sh Dimensionless.

[0036] In a specific embodiment, the test production data in step 4 refers to the gas test and MDT sampling data, which are used to directly determine the fluid properties, such as gas layer, water layer, and gas-water layer. The gas logging and typical logging values ​​at the corresponding depth of the deterministic fluid properties are selected and substituted into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point exWhen the test production data is insufficient, the gas logging and typical logging values ​​corresponding to the depth point where the logging interpretation conclusion is consistent with the well logging interpretation conclusion can be selected, and the values ​​can be substituted into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point ex value;

[0037] In a preferred embodiment, the specific method of step 5 is as follows:

[0038] The C ug- I ex The fluid identification chart is to select the I calculated in step 4 ex As the horizontal axis, C ug As the vertical axis, the gas content C of the rock per unit volume at the corresponding depth with the determined fluid properties in step 4 is ug and mining index I ex Values ​​plotted at C ug- I ex In the intersection diagram of the two coordinates, we can get C ug- I ex The fluid identification chart is recorded; the corresponding depth gas logging and typical logging values ​​of the unknown fluid properties in the unknown well target layer are substituted into the C calculated by formula (2) and formula (6) respectively. ug and I ex , and projected to the standard C of known fluid properties obtained in step 4 ug- I ex In the chart, the gas and water layers in the target layer of unknown wells (i.e. wells in the same area or region where gas or water layers are to be identified) can be identified, and the identification boundaries of water and gas layers can be obtained. When the reservoir is a gas reservoir, the porosity index I ex Greater than 1, water layer I ex Less than 1, the boundaries of the air layer and the water layer are "C ug >1、I ex >1" and "C ug <0.4”;

[0039] In another preferred embodiment, the improved k-nearest neighbor calculation module in step 6 includes three main processes:

[0040] I- Balancing of Transitional Fluid Type Sample Points: The improved k-nearest neighbor classification described in step 6 usually adopts the majority voting classification principle. Since different transitional fluid samples often have an imbalanced number problem, that is, the number of transitional fluid sample points of different types is very different, the k-nearest neighbor Gaussian fuzzy formula classification effect is prone to overfitting or underfitting. Therefore, randomly select C ug- Iex For any sample x in the transitional fluid sample points of the non-gas zone and non-water zone in the fluid identification chart, calculate the distance to the nearest type of sample w (such as gas-water coexistence layer) and another type of sample v (such as gas-bearing water layer) according to formula (7). Calculate the difference between the distances of x to w and x to v respectively. When it satisfies formula (8), retain the sample x; otherwise, eliminate the sample. Complete the calculation of all transitional fluid sample points in turn through loops to obtain the set of balanced transitional fluid sample points;

[0041]

[0042]

[0043]

[0044] where x i , x j respectively represent all transitional fluid sample points of the gas content rate C of underground unit volume rock calculated according to formula (1) ug and the excavation index I calculated according to formula (6) ex . The distance between the two is represented by L p , p is the parameter of the distance formula, x i and x j are both vectors in the n-dimensional feature space;

[0045] ||x - w|| - ||x - v|| > rδ n (8)

[0046] δ n = min{||x i - x j ||: l(x i ) ≠ l(x j )}

[0047] where δ n represents the minimum distance between two samples in w and v. r takes a value of 0 < r < 1 for threshold adjustment of formula (8). l(x i ) represents the category of x i , l(x j ) represents the category of x j , that is, they respectively represent different transitional fluid properties;

[0048] II-Calculate the distance between a sample point of unknown fluid properties at a certain depth and a sample point of known fluid properties after equalization: After the above equalization process, calculate the distance between a sample point of unknown fluid properties at a certain drilling depth and all sample points of known transitional fluid properties according to formula (7). If there are N sample points of all known fluid properties, then N distances can be calculated and sorted from small to large.

[0049] III-Determine the fluid properties of the unknown sample point: Take the N sorted samples in step II The integer part of is used as the k value of the improved k-nearest neighbor method, and the number of data points in each category is counted. The new sample points of unknown fluid properties are assigned to the category with the largest number, that is, the fluid properties of the unknown sample points are classified as the fluid properties to which the sample points with the largest number of statistics belong.

[0050] The present invention also provides a system for identifying gas reservoir fluid properties according to the method, characterized in that it comprises the following steps:

[0051] Data acquisition module: used to obtain gas logging and well logging data in the target depth interval;

[0052] Module for calculating gas content of underground rock per unit volume;

[0053] Mining index calculation module: used to calculate the sonic transit time porosity, neutron porosity and density porosity, and to calculate the mining index;

[0054] Module for calculating underground rock gas content per unit volume and excavation index: used to select gas logging and well logging values ​​corresponding to the depth of test production data to calculate the underground rock gas content per unit volume and excavation index;

[0055] Fluid identification chart construction module: used to calculate the excavation index and rock gas content per unit volume corresponding to the unknown well, and project them into the fluid identification chart to identify gas and water layers;

[0056] Transitional fluid property identification module: used to input transitional fluid type samples falling in the non-gas area and non-water area of ​​the identification plate into the improved k-nearest neighbor calculation module to complete the transitional fluid property identification;

[0057] Optionally, a calculation module is also circulated: for repeating the mining index calculation module, the underground unit volume rock gas content and mining index calculation module, the fluid identification plate construction module, and the transitional fluid property identification module, so as to finally realize the identification of different fluid properties of the target well and the target layer.

[0058] Optionally, a display module is also included, for example, displaying intermediate results and final results through a screen or a mobile terminal.

[0059] Therefore, the present invention also provides a device containing the system, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program, and the computer program code implements the system.

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

[0061] (1) The present invention provides an improved k-nearest neighbor algorithm for identifying gas reservoir fluid properties that integrates logging and measurement parameter constraints. It adopts a layer-by-layer fluid identification approach, uses the joint identification results of logging and measurement data as input, and realizes rapid transitional gas reservoir fluid identification through an improved k-nearest neighbor calculation module. It not only realizes the identification of gas layers and water layers with large differences in fluid properties, but also solves the problem of identifying transitional fluid properties with small differences in fluid properties, truly realizing the fluid property identification of complex formations such as gas layers, water layers, gas-water layers, and gas-water layers;

[0062] (2) The improved k-nearest neighbor algorithm for identifying gas reservoir fluid properties that integrates logging parameter constraints provided by the present invention utilizes logging composite parameters (see formula (1) and formula (6)), which can quickly and conveniently realize the effective identification of gas layers and water layers to be measured in the same area or region. At the same time, the computational time cost of the improved k-nearest neighbor classification is reduced, and the computational efficiency and recognition accuracy are greatly improved. Different from gas layers and water layers, transitional fluid types often have aliasing phenomena. For example, the difference in logging response between the gas-water layer and the gas-water layer or the water-gas layer is not significant. In C ug- I ex The images are often concentrated in local areas with poor boundary distinction. By combining well logging constraints with an improved k-nearest neighbor method, the recognition accuracy and computational efficiency are improved.

[0063] (3) The present invention provides an improved k-nearest neighbor algorithm for identifying gas reservoir fluid properties that integrates recorded and measured parameter constraints. This algorithm solves the problem that the traditional k-nearest neighbor algorithm cannot adapt to the imbalance of sample points through certain rules (see formula (8)). It is more applicable to the engineering problems involved in the present invention and significantly improves the accuracy of fluid property identification, especially the identification of transitional fluid properties. It plays an important role in exploration and development and realizes the problem of complex fluid property identification driven by both physics and data. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is the workflow diagram of the gas reservoir fluid property identification method using the improved k-nearest neighbor algorithm that integrates recorded and measured parameter constraints.

[0065] Figure 2 C ug- I ex Measure and record fluid identification diagram.

[0066] Figure 3 This is the improved k-nearest neighbor calculation flowchart. DETAILED DESCRIPTION

[0067] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. 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, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0068] The present invention belongs to the field of oil and gas field exploration and development technology, and provides a method for identifying gas reservoir fluid properties by using an improved k-nearest neighbor algorithm that integrates recorded and measured parameter constraints. Figure 1 As shown, the present invention aims to solve the problem of identifying fluid properties in complex gas reservoirs, especially transitional fluids. To this end, the present invention is described in detail below using a well area in the Baodao area of ​​the Qiongdongnan Basin as an example. This study area is primarily composed of natural gas reservoirs.

[0069] Step 1: Perform gas logging and conventional logging in the target layer of the target well to obtain gas logging and logging data in the target depth interval;

[0070] The gas logging data include C1, C2, C3, iC4 and nC4, iC5 and nC5 obtained by gas logging in the Lingshui interval of the Baodao area of ​​the Qiongdongnan Basin, which represent CH4, C2H6, C3H8, iC4H, etc. 10 、nC4H 10 、iC5H 12 、nC5H 12 The logging data include acoustic time difference Δt, compensated neutron Bulk density ρb logging and natural gamma ray GR logging curves;

[0071] Step 2: Substitute the gas logging, drilling, temperature and pressure parameters into formula (1) to calculate the gas content per unit volume of underground rock C ug ;

[0072] The gas content of underground rock per unit volume C ug The gas logging C1, C2, C3, iC4 and nC4 and iC5 and nC5 measured at a certain depth of the target layer of the target well and the drill bit diameter D, drilling fluid density ρ, displacement drilling Q, drilling time t parameter, drilling fluid outlet temperature T out , geothermal gradient G t Temperature and pressure parameters, wellbore vertical depth H, and deviation compressibility coefficient Z of each component nSubstituting the parameters (n=1, 2, 3, 4, 5) into formula (1), the gas content C of the rock per unit volume at the depth can be calculated. ug , as shown in Table 1;

[0073] C ug =12159Σ(Z n C n )tQ(HG t +273.15) / [ρH(T out +273.15)πD 2 ] (1)

[0074] Where D is the drill diameter, mm; T out is the drilling fluid outlet temperature, °C; H is the vertical depth of the wellbore, m; ρ is the drilling fluid density, g / cm 3 ; G t is the geothermal gradient, ℃ / m; Q is the displacement (including the sum of the circulation displacement and the booster pump displacement), L / min; t is the drilling time, min / m; C n (n=1, 2, 3, 4, 5, respectively represent CH4, C2H6, C3H8, iC4H 10 、nC4H 10 、iC5H 12 、nC5H 12 , the same below) represents the gas measured value of each hydrocarbon component, ppm; the deviation compression coefficient Z of each component n (n = 1, 2, 3, 4, 5, including isomeric hydrocarbons), the Z of this embodiment is obtained from the standard generalized compressibility factor chart n The value is 1.1 and is dimensionless.

[0075]

[0076]

[0077]

[0078] Step 3: The acoustic time difference Δt and the compensation neutron Bulk density ρ b The sonic transit time porosity is calculated from the well logging and natural gamma ray logging curves. Neutron porosity and density porosity And further calculate the mining index I ex ;

[0079] Step 301: Calculate the mud content V corresponding to a certain depth point on the natural gamma ray logging curve GR according to formula (2)-1. sh(dimensionless), where GCUR is the Hillch index, I GR Calculated by formula (2)-2, GR max With GR min Take the natural gamma logging values ​​corresponding to the large set of mudstone layers and the large set of sandstone layers of the target well respectively;

[0080]

[0081] I GR =(GR-GR min ) / (GR max -GR min ) (2)-2

[0082] Step 302: The acoustic time difference Δt and the compensation neutron Bulk density ρ b The logging value and the mud content V at the depth point calculated in step 301 sh Substituting into formulas (3), (4), and (5), the apparent acoustic time difference porosity can be calculated: Neutron porosity and apparent porosity Then calculate the mining index I according to formula (6): ex (dimensionless);

[0083]

[0084]

[0085]

[0086]

[0087] Among them, Δt, ρ b , the units of GR are us / m, dimensionless, g / cm 3 , API; Δt ma , ρ ma are the rock logging values ​​of acoustic wave time difference, compensated neutron and volume density of rock skeleton, with the same units as above; Δt f , ρ f They are the logging values ​​of the fluid's acoustic wave time difference, compensated neutron, and volume density, and the units are the same as above.

[0088] Table 2

[0089]

[0090] Table 3

[0091]

[0092] Step 4: Select the gas logging and well logging representative values ​​at the depth corresponding to the test production data, and calculate the underground unit volume rock gas content C according to steps 2 and 3 respectively. ug and mining index I ex ;

[0093] The test production data refers to the gas test and MDT sampling data. Based on the gas test and MDT sampling data, the deterministic fluid properties, such as gas layer, water layer, gas-water layer, etc., can be directly obtained. The gas logging and typical logging values ​​at the corresponding depth of the deterministic fluid properties are selected and substituted into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point ex When the test production data is insufficient, the gas logging and typical logging values ​​corresponding to the depth point where the logging interpretation conclusion is consistent with the well logging interpretation conclusion can be selected, and the values ​​can be substituted into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point ex values ​​(Table 2);

[0094] Step 5, build C ug- I ex Fluid Identification Plate, select I ex is the horizontal axis, C ug As the vertical coordinate, calculate the C corresponding to the unknown well ug and I ex , and projected to C ug- I ex In the fluid identification chart, the data falling into the gas layer and water layer areas are the gas layer and water layer respectively;

[0095] Step 501, the C ug- I ex The fluid identification chart is to select the I calculated in step 4 ex As the horizontal axis, C ug As the vertical axis, the gas content C of the rock per unit volume at the corresponding depth with the determined fluid properties in step 4 is ug and mining index I ex Values ​​plotted at C ug- I ex In the intersection diagram of the two coordinates, we can get Figure 2 C shown ug- I exRecord the fluid identification chart; when the reservoir is a gas layer, due to the neutron logging excavation effect, Decrease, and and This results in the porosity index I ex is greater than 1, the higher the gas saturation, the greater the value, and the water layer I ex is less than 1, it can be seen that the boundaries of the air layer and the water layer are "C ug >1、I ex >1" and "C ug <0.4”.

[0096] Step 502, further, calculate the I of the unknown A well data ex and C ug , and projected onto the plate, identifying the gas layer and water layer of well A according to the identification boundaries of the water layer and gas layer given in step 501;

[0097] Step 6, will fall on C ug -I ex The transitional fluid type samples in the non-gas and non-water regions in the image are input into the improved k-nearest neighbor calculation module (see the calculation process for details). Figure 3 ), complete the identification of transitional fluid properties, and combine with step 5 to realize the identification of different fluid properties such as gas layer, water layer, gas and water layer;

[0098] Step 601, C ug- I ex The transitional fluid type sample points in the fluid identification chart are balanced. ug- I ex For any sample x in the transitional fluid sample points in the non-gas and non-water zones on the fluid identification plate, calculate the closest sample w (e.g., gas and water in the same layer) and the closest sample v (e.g., gas-water layer) to the sample x according to formula (7). Calculate the distance difference between x and w and between x and v respectively. If the difference satisfies formula (8), the sample x is retained; otherwise, the sample is discarded. The calculation of all transitional fluid sample points is completed in a cycle to obtain the equalized transitional fluid sample point set.

[0099]

[0100]

[0101]

[0102] Among them, x i , x j They represent the C calculated according to formula (1) ug and I calculated according to formula (6) nsdFor all transitional fluid sample points, the distance between the two is represented by L p where p is a parameter of the distance formula, and in the art, p = 2. x i and x j are both vectors in the n-dimensional feature space. For problems in the art, n takes the value of 2;

[0103] ||x - w|| - ||x - v|| > rδ n (8)

[0104] S n = min{||x i - x j ||: l(x i ) ≠ l(x j )}

[0105] where δ n represents the minimum distance between two samples in w and v. r takes the value 0 < r < 1 for threshold adjustment of formula (8). In the art, r takes 0.75. l(x i ) represents the category of x i , and l(x j ) represents the category of x j , that is, they respectively represent different transitional fluid properties in the study area, such as gas-water co-layers, gas-bearing water layers, etc.;

[0106] Step 602, calculate the distance between a sample point of unknown fluid property at a certain depth and the sample points of known fluid properties after equalization: After the above equalization process, calculate the distance between a sample point of unknown fluid property at a certain depth and all sample points of known transitional fluid properties according to formula (7). If there are N sample points of all known fluid properties, then N distances can be calculated, and these N distances are sorted from small to large;

[0107] Step 603, classify the unknown samples according to the value of K to complete the identification of transitional fluid properties; Take the integer part of as the value of k in the improved k-nearest neighbor method. Take the first k distances sorted in step 602 above, count the number of data points in each category, and assign the sample point of the new unknown fluid property to the category with the largest number, that is, the fluid property of the unknown sample point is classified as the fluid property of the sample point with the largest number of statistics;

[0108] Step 7, repeat the above steps 2 to step 6, and the fluid properties at different depth positions can be analyzed to effectively identify different fluid properties such as gas layers, water layers, and gas-water co-layers in the reservoir of the target well and target interval. The results taking a well area in the Baodao area of the Qiongdongnan Basin in this embodiment as an example are shown in Figure 2The five-pointed black dots in the box in the figure are sample points of unknown fluid properties at 8 depth points. Finally, 1, 2, 4, and 7 are classified as poor gas layers, and 3, 5, 6, and 8 are classified as gas-bearing water layers. The MDT cable formation test sampling results confirm the above analysis results.

[0109] In summary, the method of the present invention can significantly improve the accuracy of identifying the fluid properties of gas layers, especially distinguishing the properties of transitional fluids. It has strong adaptability and can effectively solve the problem of identifying gas reservoir types with complex fluid properties, playing an important role in exploration and development.

[0110] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, from all points of view, the embodiments should be regarded as illustrative and non-restrictive. The scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes that fall within the meaning and range of equivalents of the claims are included in the present invention. Any reference signs in the claims should not be construed as limiting the claim to which they relate.

[0111] In addition, it should be understood that although this specification is described in terms of implementation methods, not every implementation method contains only one independent technical solution. This narrative method of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment can also be appropriately combined to form other implementation methods that can be understood by those skilled in the art.

Claims

1. The improved k-nearest neighbor algorithm for identifying gas reservoir fluid properties by integrating recorded and measured parameter constraints is characterized by: The steps include: Step 1: Obtain gas logging and well logging data in the target depth interval; Step 2: Calculate the gas content of underground rock per unit volume; Step 3: Calculate the acoustic time difference porosity, neutron porosity and density porosity, and calculate the mining index; the mining index is calculated as follows: the acoustic time difference Δt at the corresponding logging depth, the compensated neutron porosity, and the density porosity are calculated. Bulk density ρ b The logging value and the mud content V at the depth point calculated in step 301 sh Substituting into formulas (3), (4), and (5), the apparent acoustic time difference porosity can be calculated: Neutron porosity and apparent porosity Then calculate the mining index I according to formula (6): ex , dimensionless; Among them, Δt, ρ b , the units of GR are us / m, dimensionless, g / cm 3 , API; Δt ma , ρ ma are the rock logging values ​​of acoustic wave time difference, compensated neutron and volume density of rock skeleton, with the same units as above; Δt f , ρ f are the logging values ​​of the fluid acoustic wave time difference, compensated neutron, and volume density respectively; V sh , I ex dimensionless; Step 4: Select the gas logging and well logging values ​​at the depth corresponding to the test production data to calculate the gas content of the rock per unit volume and the excavation index; Step 5: Construct a fluid identification map, calculate the excavation index and rock gas content per unit volume corresponding to the unknown well, and project them onto the fluid identification map to identify gas and water layers; Step 6: Input the transitional fluid type samples that fall in the non-gas area and the non-water area in the identification plate into the improved k-nearest neighbor calculation module to complete the transitional fluid property identification; Step 7: Repeat steps 2 to 6 to identify different fluid properties in the target well and target layer.

2. The method according to claim 1, wherein The steps include: Step 1: Perform gas logging and conventional logging in the target layer of the target well to obtain gas logging and logging data in the target depth interval; Specifically, the gas logging data includes C1, C2, C3, iC4 and nC4 and iC5 and nC5 content curves; the logging data includes acoustic time difference Δt, compensated neutron Bulk density ρ b Well logging and natural gamma ray (GR) logging curves; Among them, C1 represents methane, C2 represents ethane, C3 represents propane, iC4 represents isobutane, nC4 represents normal butane, iC5 represents isopentane, and nC5 represents normal butane; Step 2: record the gas logging C1, C2, C3, iC4 and nC4 and iC5 and nC5 measurement values ​​and the drill bit diameter D, drilling fluid density ρ, displacement drilling Q, drilling time t, drilling fluid outlet temperature T out , geothermal gradient G t Temperature and pressure parameters, wellbore vertical depth H, and deviation compressibility coefficient Z of each component n Substitute the parameters into formula (1) to calculate the gas content of underground rock per unit volume C ug ; C ug =12159∑(Z n C n )tQ(HG t +273.15) / [ρH(T out +273.15)πD 2 ] (1) Where D is the drill diameter, mm; T out is the drilling fluid outlet temperature, °C; H is the vertical depth of the wellbore, m; ρ is the drilling fluid density, g / cm 3 ; G t is the geothermal gradient, ℃ / m; Q is the displacement, including the sum of the circulation displacement and the booster pump displacement, L / min; t is the drilling time, min / m; C n , n=1, 2, 3, 4, 5, respectively represent CH4, C2H6, C3H8, iC4H 10 、nC4H 10 、iC5H 12 、nC5H 12 , represents the gas measured value of each hydrocarbon component, ppm; the deviation compression coefficient Z of each component n , n=1,2,3,4,5; Step 3: The acoustic time difference Δt and compensation Substance, volume density ρ b The acoustic transit time pore size is calculated by well logging and natural gamma logging curves respectively. degree, neutron porosity and density porosity And further calculate the mining index I ex ; Step 4: Select the gas logging and typical logging values ​​at the depth corresponding to the test production data, and calculate the gas content C of the rock per unit volume in the underground according to steps 2 and 3. ug and mining index I ex ; Step 5, build C ug -I ex Fluid Identification Plate, select I ex is the horizontal axis, C ug As the vertical coordinate, calculate the I corresponding to the unknown well ex and C ug , and projected to C ug -I ex In the fluid identification chart, if the data falls into the gas layer or water layer area, it will be identified as the gas layer or water layer accordingly; Step 6, will fall on C ug -I ex The transitional fluid type samples of the non-gas and non-water areas in the image are input into the improved k-nearest neighbor calculation module to complete the identification of transitional fluid properties. Combined with step 5, the identification of fluid properties of gas layers, water layers, gas-water layers, and gas-water layers can be realized.

3. The method according to claim 1, wherein Repeat steps 2 to 6 above to analyze the fluid properties at different depths to effectively identify the fluid properties of the reservoir gas layer, water layer, gas-water layer, and gas-water layer in the target well and target layer.

4. The method according to claim 2, wherein In step 3, the logging value at a certain depth point of the natural gamma ray logging curve GR is calculated according to formula (2)-1 to obtain the mud content V corresponding to the depth. sh , where GCUR is the Hillch index, where I GR Calculated by formula (2)-2, GR max With GR min Take the natural gamma logging values ​​corresponding to the large set of mudstone layers and the large set of sandstone layers of the target well respectively; I GR =(GR-GR min ) / (GR max -GR min ) (2)-2。 5. The method according to claim 2, wherein In step 4, the test production data refers to the gas test and MDT sampling data. Based on the gas test and MDT sampling data, the deterministic fluid properties can be directly obtained, including gas layers, water layers, gas-water layers, and gas-water layers. The gas logging and typical logging values ​​at the corresponding depth of the deterministic fluid properties are selected and substituted into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point ex value; When the test production data is insufficient, select the gas logging and typical logging values ​​corresponding to the depth point where the well logging interpretation conclusion is consistent with the well logging interpretation conclusion, and substitute them into the formula (1) shown in step 2 to obtain the underground unit volume rock gas content C at the corresponding depth point. ug Substitute the value into the formula (6) shown in step 3 to obtain the excavation index I of the corresponding depth point ex value.

6. The method according to claim 2, wherein The specific method of step 5 is as follows: Step 501, the C ug -I ex The fluid identification chart is to select the I calculated in step 4 ex As the horizontal axis, C ug As the vertical axis, the gas content C of the rock per unit volume at the corresponding depth with the determined fluid properties in step 4 is ug and mining index I ex Values ​​plotted at C ug -I ex In the intersection diagram of the two coordinates, we can get C ug -I ex Record the fluid identification chart; when the reservoir is a gas layer, the porosity index I ex Greater than 1, water layer I ex Less than 1, the boundaries of the air layer and water layer are "C ug >1、I ex >1" and "C ug <0.4”; Step 502, further, calculate the I of the unknown well data ex and C ug , and projected onto the plate, and according to the identification boundaries of the water layer and gas layer given in step 501, the gas layer and water layer of the unknown well are identified.

7. The method according to claim 2, wherein The improved k-nearest neighbor calculation module described in step 6 includes three main processes: I-Balanced processing of transitional fluid type sample points: The improved k-nearest neighbor classification described in step 6 usually adopts the majority voting classification principle. Since different transitional fluid samples often have an imbalanced number problem, that is, the number of transitional fluid sample points of different types is very different, the k-nearest neighbor Gaussian fuzzy formula classification effect is prone to overfitting or underfitting. Therefore, C is randomly selected. ug -I ex For any sample x in the transitional fluid sample points in the non-gas zone and non-water zone in the fluid identification plate, calculate the closest sample x to the gas-water layer w and the gas-water layer v according to formula (7). Calculate the distance difference between x and w and between x and v respectively. If it satisfies formula (8), the sample x is retained. Otherwise, the sample is eliminated. The calculation of all transitional fluid sample points is completed in a cycle to obtain the transitional fluid sample point set after equalization. Among them, x i , x j They represent the gas content C of the underground rock per unit volume calculated according to formula (1) ug and the mining index I calculated according to formula (6) ex All transitional fluid sample points, the distance between them is L p Indicates that p is the parameter of the distance formula, x i and x j are all vectors in n-dimensional feature space; ||x-w||-||x-v||>rδ n (8) d n =min{||x i -x j ||:1(X i )≠1(x j )} Among them, δ n represents the minimum distance between two samples in w and v, r ranges from 0 < r < 1, which is used to adjust the threshold of formula (8), l(x i ) represents the category of x i , l(x j ) represents the category of x j , that is, they respectively represent different transitional fluid properties; II-Calculate the distance between a sample point of unknown fluid properties at a certain depth and a sample point of known fluid properties after equalization: After the above equalization process, calculate the distance between a sample point of unknown fluid properties at a certain drilling depth and all sample points of known transitional fluid properties according to formula (7). If there are N sample points of all known fluid properties, then N distances can be calculated and sorted from small to large. III-Determine the fluid properties of the unknown sample point: Take the N sorted samples in step II The integer part of is used as the k value of the improved k-nearest neighbor method, and the number of data points in each category is counted. The new sample points of unknown fluid properties are assigned to the category with the largest number, that is, the fluid properties of the unknown sample points are classified as the fluid properties to which the sample points with the largest number of statistics belong.

8. A system for identifying gas reservoir fluid properties according to the method of any one of claims 1 to 7, characterized in that: The steps include: Data acquisition module: used to obtain gas logging and well logging data in the target depth interval; Module for calculating gas content of underground rock per unit volume; Mining index calculation module: used to calculate the sonic transit time porosity, neutron porosity and density porosity, and to calculate the mining index; Module for calculating underground rock gas content per unit volume and excavation index: used to select gas logging and well logging values ​​corresponding to the depth of test production data to calculate the underground rock gas content per unit volume and excavation index; Fluid identification chart construction module: used to calculate the excavation index and rock gas content per unit volume corresponding to the unknown well, and project them into the fluid identification chart to identify gas and water layers; Transitional fluid property identification module: used to input transitional fluid type samples falling in the non-gas area and non-water area of ​​the identification plate into the improved k-nearest neighbor calculation module to complete the transitional fluid property identification; Circular calculation module: used to repeatedly calculate the mining index module, the underground unit volume rock gas content and mining index calculation module, the fluid identification plate construction module, and the transitional fluid property identification module, ultimately realizing the identification of different fluid properties in the target well and target layer.

9. The system according to claim 8, wherein It also includes a display module that displays intermediate results and final results through a screen or a mobile terminal.

10. A device comprising the system according to claim 8 or 9, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program, and the computer program code implements the method according to any one of claims 1 to 7.

Images