Hot dry rock reservoir tiny crack fine description method integrating multi-source heterogeneous data

Abstract

The invention provides a hot dry rock reservoir micro fracture fine description method integrating multi-source heterogeneous data. The hot dry rock reservoir micro fracture fine description method comprises the following steps that S1, joint fracture investigation is carried out; s2, performing fracture logging on an existing drilling well of the hot dry rock, analyzing the fracture development direction, opening degree, type and occurrence of a reservoir section, identifying the position and scale of a fracture development favorable area, counting the number of fractures, and preliminarily studying and judging the occurrence of primary and secondary fractures in combination with the rock core fracture development condition; s3, carrying out seismic inversion on high-precision seismic exploration data implemented in the hot dry rock site by adopting a Bayesian formula, and analyzing the possibility of fracture development at different positions; performing data dimension reduction, and forming a reservoir inversion initial data set after dimension reduction; multiplying each data of the reservoir inversion initial data set by a plurality of groups of 0-1 random numbers to form a plurality of groups of porosity initial data; and S4, carrying out numerical simulation on the initial porosity data, calculating the concentration of the tracer agent of the production well, and inverting parameters according to the difference between a predicted value and an actual observed value.

Description

Technical Field

[0001] The present invention belongs to the field of geothermal resource exploration and development, and particularly relates to a method for finely depicting microfractures in a hot dry rock reservoir by integrating multi-source heterogeneous data. Background Art

[0002] Hot dry rock generally refers to a high-temperature rock mass with a temperature greater than 180 °C, buried thousands of meters deep, and having no fluid or only a small amount of underground fluid (dense and impermeable) inside. It has the characteristics of huge resource potential and almost no pollution in development and utilization, and is a new type of energy with great strategic value. The hot dry rock reservoir itself is dense. The development process usually involves drilling several wells first, forming an artificial fracture network by fracturing low-porosity and low-permeability rocks, and guiding the design of subsequent drilling targets and reservoir construction plans according to the characteristics of the main development direction, position, occurrence, etc. of the fractures. At present, the main methods for identifying deep reservoir structures include surface investigation, physical exploration, borehole identification, microseismic monitoring, etc. The related inventions proposed are also mainly based on the above methods, usually involving in-depth processing of a certain method or combined analysis of several methods.

[0003] Surface investigation mainly takes the surface outcrops of the reservoir with the same age, lithology, and genesis as the research object, analyzes the development law of pores, holes, and fractures in surface rocks, and infers the deep reservoir geological structure. However, inferring deep fractures through the surface has a large error. On the one hand, because the in-situ stress conditions of deep reservoirs and surface outcrops are very different, the fracture opening directions, opening degrees, and occurrences may be very different. On the other hand, due to the influence of weathering, the physical and chemical properties of surface outcrops may change to a large extent during the long geological age, resulting in differences in the fracture development law from the deep part.

[0004] High-precision physical exploration is to artificially generate electrical, magnetic, and seismic signals on the surface and infer the reservoir structure by capturing the corresponding reflected signals. This method, especially the high-precision seismic method, has a high accuracy in identifying large and medium-sized structural planes such as detachment surfaces, faults, and stratigraphic interfaces. Hot dry rock masses are generally dense, and the water-conducting fractures are below the meter scale. It is difficult for geophysical exploration to correctly identify small-scale, weakly reflected signals.

[0005] Borehole identification includes drilling cores and fracture logging. Both of these methods reveal the fracture development characteristics at the drilling scale and have relatively high accuracy. However, this method can only explore the situation around the wellbore. The distribution of natural reservoir weak surfaces is uneven, and the heterogeneity of reservoir fracture development is further enhanced after artificial fracturing transformation. Therefore, it is difficult to expand the study of the fracture development law of the entire geothermal field by this method.

[0006] Microseismic monitoring is to monitor the formation disturbance caused by the release of rock fracture stress during artificial fracturing, and to infer the nature and development of fractures by analyzing their location, time series, magnitude and origin. The effect of this technology in actual operation is relatively poor. On the one hand, the generation of microseismic events may not be due to the formation disturbance caused by the fracturing fluid entering the rock fractures, but may also be due to the continuous injection of the fracturing fluid, which causes the formation stress to change and the relative displacement of the distal rock mass to generate microseismic events. On the other hand, the accuracy of microseismic monitoring is relatively low, and the positioning error is generally greater than 50m.

[0007] Tracer test is a method in which a specific chemical substance is injected from one well as an injection well, and one or several wells are used as production wells, and the inter-well connectivity is analyzed by monitoring the change of the produced concentration. This method does not directly reflect the reservoir structure, but the size, scale, direction, aperture, etc. of deep fractures all affect the concentration of tracers in the production wells, that is, the tracer breakthrough curve can also reflect the deep permeability to a certain extent. Due to the strong anisotropy and heterogeneity of deep reservoirs, the inversion of tracer curves involves highly nonlinear equations, and there will be a multi-solution problem when identifying the deep reservoir structure only based on surface data such as flow rate, pressure, and concentration.

[0008] The above methods can all invert the reservoir structure to a certain extent, but there are often problems such as strong multi-solution and insufficient accuracy. Summary of the Invention

[0009] In order to solve the above problems, the purpose of the present invention is to provide a method for fine characterization of microfractures in hot dry rock reservoirs by integrating multi-source heterogeneous data. By newly proposing a parameter inversion algorithm and effectively coupling joint fracture investigation, geophysical exploration and numerical simulation results using artificial intelligence methods, the integration of shallow-medium-deep and field-laboratory data of hot dry rock is successfully realized, and the reservoir geological structure is finely inverted.

[0010] In order to achieve the above purpose, the present invention adopts the following technical solutions:

[0011] A method for fine characterization of microfractures in hot dry rock reservoirs by integrating multi-source heterogeneous data, comprising the following steps:

[0012] S1: Conduct joint fracture investigation and develop outcrops of the same lithology around the development site;

[0013] S2: Conduct fracture logging on existing hot dry rock wells, focus on analyzing the fracture development direction, aperture, type and occurrence in the reservoir section, identify the location and scale of favorable fracture development areas, count the number of fractures and draw a strike rose diagram, and preliminarily judge the occurrence of main and secondary fractures in combination with the fracture development in the core.

[0014] S3: Perform seismic inversion on the high-precision seismic exploration data obtained within the hot dry rock site using Bayes' formula, identify post-stack seismic anisotropy based on the previous geological understanding, and analyze the possibility of fracture development at different locations. Since the data obtained from seismic inversion is large in quantity and has many meaningless numbers, perform data dimensionality reduction on it using cluster analysis. After dimensionality reduction, form the initial reservoir inversion dataset. Multiply each data in the initial reservoir inversion dataset by several groups of random numbers between 0 and 1 to form several groups of initial porosity data.

[0015] S4: Perform numerical simulation on several groups of initial porosity data using the equivalent continuous medium theory, calculate the tracer concentration in the production well, and invert the parameters based on the difference between the predicted value and the actual observed value.

[0016] Preferably, in step S1, the joint fracture investigation focuses on analyzing the morphology, type, origin, length, aperture, and filling minerals of the joint fractures.

[0017] Preferably, in step S1, for sedimentary rocks as the thermal reservoir lithology, rock masses with the same sedimentation and diagenesis processes should be investigated; for igneous rocks, rock masses with the same lithofacies type, intrusion sequence, and structural components should be investigated; for metamorphic rocks, rock masses with the same metamorphic facies and metamorphic mineral assemblages, and deformation and metamorphism processes should be investigated.

[0018] Preferably, in step S2, the number of fractures is counted in units of 10 m and a strike rose diagram is drawn.

[0019] Preferably, in the data dimensionality reduction in step S3, half of the maximum possibility is used as the threshold. For values less than this threshold, relatively large clustering radii and sample numbers are set, and for values greater than this threshold, relatively small clustering radii and sample numbers are set.

[0020] Preferably, the high-precision seismic exploration data in step S3 is one or more of three-dimensional seismic, vertical seismic profile, time-lapse seismic, and cross-well seismic data.

[0021] Preferably, step S4 includes:

[0022] 1) Set the assimilation number Na, and vectorize several groups of initial porosity data generated from the previous geological understanding to form a prior parameter set. In the formula, Ne represents the number of parameters in one assimilation.

[0023] 2) Conduct numerical simulation:

[0024]

[0025] Among them, g() represents the numerical simulation to calculate the tracer concentration in the production well. represents the predicted value.

[0026] 3) Calculate the predicted value covariance Covariance between parameter values ​​and predicted values Use the following formula to find The pseudo-inverse of is:

[0027]

[0028] Where U and V are column orthogonal matrices, and ∑ is a diagonal matrix;

[0029] 4) Calculate the Kalman gain:

[0030]

[0031] 5) Analyze the Kalman gain and determine the further update logic:

[0032]

[0033]

[0034] 6) Update parameter values ​​and adjust parameters according to upper and lower limits.

[0035] The present invention has three main advantages. First, it integrates existing data to form a systematic geological understanding. By proposing a new parameter inversion algorithm, artificial intelligence methods are used to effectively couple joint and fissure investigation, geophysical exploration and numerical simulation results, and successfully achieve the integration of shallow-medium-deep and field-indoor data of hot dry rocks, and finely invert the reservoir geological structure. Second, this method is implemented on the basis of existing drilling and geophysical exploration, with relatively small workload and low cost, but with high efficiency. Third, this method can be continuously updated as geological understanding continues to deepen, and gradually approach the real reservoir structure. In actual mining, the work accuracy can be gradually improved as the exploration level increases, and it is highly compatible with actual work.

[0036] The beneficial effects of the present invention are:

[0037] The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data provided by the present invention effectively makes up for the deficiencies of the two methods in reservoir inversion by establishing a set of data inversion algorithms to couple conventional methods with tracer experiments, and finely characterizes the distribution of reservoir pores, caves and fractures, providing technical support for subsequent drilling target design and reservoir transformation. The present invention is applied to the fine characterization of pores, caves and fractures in hot dry rock reservoirs, identifying the development laws of deep geological structures, revealing reservoir permeability, reducing development and utilization costs, and serving resource development and utilization. The present invention can also be used in the field of high-temperature oil and gas and shale gas development. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 This is a typical joint and crack diagram of a preferred embodiment of the present invention.

[0039] Figure 2 This is a plan view of the drilling trajectory of a preferred embodiment of the present invention.

[0040] Figure 3 The fracture logging diagram of a preferred embodiment of the present invention shows the natural gas fracture density map of Well GH01, Well G002 and Well GH03.

[0041] Figure 4 This is a comparison diagram of a vertical seismic profile before (a) and after (b) processing according to a preferred embodiment of the present invention.

[0042] Figure 5 Comparison between the predicted curve and the observed curve (a) and the reservoir pore distribution diagram (b) of a preferred embodiment of the present invention.

[0043] Figure 6 This is a comparison diagram of fracture inversion and post-pressure temperature measurement in Well GH02 and Well GH03 according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to avoid the problems of strong multi-solution and insufficient precision in the existing technology, mutual verification and joint inversion can be used for research. Based on this, the present invention proposes a method that couples ground investigation, physical exploration, borehole identification, microseismic monitoring and tracer test, and proposes a speech algorithm suitable for low permeability reservoirs to determine the development of pores and fractures in hot dry rocks.

[0045] The present invention is applied to the fine characterization of pores and cracks in hot dry rock reservoirs, to ascertain the development law of deep geological structures, to reveal reservoir permeability, to reduce development and utilization costs, and to serve resource development and utilization. The present invention can also be used in the field of high-temperature oil and gas and shale gas development.

[0046] The cracks in hot dry rock reservoirs are deeply buried and small in scale. Conventional reservoir crack detection methods are difficult to accurately identify, and the tracer test inversion of reservoir structure often falls into multiple solutions. The present invention establishes a set of data inversion algorithms to couple conventional methods with tracer tests, effectively making up for the shortcomings of the two methods in reservoir inversion, and finely depicts the distribution of reservoir pores and cracks, providing technical support for subsequent drilling target design and reservoir transformation.

[0047] The method for finely characterizing micro-fractures in hot dry rock reservoirs by integrating multi-source heterogeneous data of the present invention comprises the following steps:

[0048] S1: Conduct joint and fissure investigation and develop outcrops of the same lithology around the site; during the investigation, focus on analyzing the morphology, type, origin, length, aperture, and filling minerals of the joints and fissures.

[0049] S2: Conduct fracture logging on existing dry hot rock wells, focusing on analyzing the fracture development directions, apertures, types, and attitudes in the reservoir section, identifying the locations and scales of favorable fracture development areas, counting the number of fractures in units of 10 m and drawing a strike rose diagram, and preliminarily judging the attitudes of the main and secondary fractures in combination with the fracture development in the core.

[0050] S3: Use Bayes' formula to invert the high-precision seismic exploration data such as three-dimensional seismic, vertical seismic profile, time-lapse seismic, and cross-well seismic implemented in the dry hot rock site, identify the post-stack seismic anisotropy based on the previous geological understanding, and analyze the possibility of fracture development at different locations. The data obtained from seismic inversion is large in quantity and has many meaningless numbers, so clustering analysis is used to reduce the data dimension. Take half of the maximum possibility as the threshold, set a relatively large clustering radius and number of samples for the values less than the threshold, and set a relatively small clustering radius and number of samples for the values greater than the threshold to form an initial reservoir inversion dataset after dimension reduction; multiply each data in the initial reservoir inversion dataset by several groups of random numbers of 0-1 to form several groups of initial porosity data.

[0051] S4: Use the equivalent continuous medium theory to conduct numerical simulations on several groups of initial porosity data, calculate the tracer concentration in the production well, and invert the parameters according to the difference between the predicted value and the actual observed value.

[0052] The steps include:

[0053] 1) Set the assimilation number Na, and vectorize several groups of initial porosity data generated from the previous geological understanding to form a prior parameter set where Ne represents the number of parameters in one assimilation;

[0054] 2) Conduct numerical simulations:

[0055]

[0056] where g() represents calculating the tracer concentration in the production well through numerical simulation, represents the predicted value;

[0057] 3) Calculate the covariance of the predicted value The parameter value and the covariance of the predicted value Use the following formula to find the pseudo-inverse of:

[0058]

[0059] where U and V are column orthogonal matrices, and ∑ is a diagonal matrix;

[0060] 4) Calculate the Kalman gain:

[0061]

[0062] 5) Analyze the Kalman gain and determine the further update logic:

[0063]

[0064] 6) Update parameter values ​​and adjust parameters according to upper and lower limits.

[0065] Example 1

[0066] Taking the Qinghai Gonghe hot dry rock test mining site as an example, the steps include:

[0067] S1: First, conduct a joint and fissure investigation on the Dangjiasi granite body that erupted from the same source and same period of the Gonghe hot dry rock in Qinghai. Granite is an igneous rock, and the rock bodies with the same lithofacies type, intrusion level, and structural components should be investigated. Figure 1 As shown in the figure, shear joints are generally developed. This type of joint has a stable occurrence, the joint surface is straight and smooth and extends far, mostly cutting through quartz, feldspar crystals and clay minerals. The spacing between the same group of cracks is relatively uniform, and some joints form conjugate "X"-shaped joints in two groups with different directions. A small number of tension joints are developed, and the joint surface is short, filled with quartz, calcite veins, etc. The main cracks in the Dangjia Temple rock mass are shear joints, and the secondary cracks are tension joints. Joints and cracks of different properties and different periods cut and intersect.

[0068] S2: Conduct fracture logging on existing wells drilled in hot dry rock to analyze the direction, aperture, type, and occurrence of fractures in the reservoir section, identify the location and scale of favorable fracture development areas, count the number of fractures and draw a trend rose diagram, and preliminarily determine the occurrence of major and minor fractures based on the development of fractures in the core. Figure 2 As shown in the figure, the curves in the figure are the plan views of the drilling well trajectories of Well GH01 (abbreviated as GH01 in the figure), Well GH02 (abbreviated as GH02 in the figure), Well GH03 (abbreviated as GH03 in the figure), Well GH04 (abbreviated as GH04 in the figure), and Well GH05 (abbreviated as GH05 in the figure). The circles show the distribution of microseismic events, where the larger the circle and the brighter the color, the greater the magnitude. Since the reservoir is heterogeneous and there are more fractures that penetrate the reservoir, directional wells are used instead of vertical wells based on microseismic events. Carry out fracture test reinterpretation, such as Figure 3 As shown in the figure, there are dense northeast and northwest natural fractures in the Gonghe hot dry rock test mining site; logging data show that the spatial differences in the development of natural fractures are large, and there are fewer fractures in the target reservoir section (the target reservoir section is 3500-4000m). The density of natural fractures in wells GH01 and GH02 is generally less than 10 / 10 meters; except for the bottom hole section (depth of about 3850m) in well GH03, there are almost no natural fractures, and fractures are only developed in a few sections. Figure 4As shown in the figure, the existing geological understanding was integrated, and the Gonghe vertical seismic profile was reprocessed. The comparison diagrams before and after processing are shown in Figures (a) before processing and (b) after processing. Before processing, since prior knowledge was not incorporated, only anisotropy detection was performed on the reflection signals of the vertical seismic profile. However, the reflection signals may change due to factors such as seismic signal excitation and reception conditions, noise factors, incident angle changes, wave interference, lithofacies changes, and stratigraphic structure changes. Therefore, it is difficult to identify the development law of deep geothermal reservoir fractures. After processing, the regularity of fracture distribution was improved, and the distribution position and scale of the liquid injection channels could be clearly observed.

[0069] S3: Use Bayes' formula to invert the Gonghe vertical seismic profile and invert the reservoir geological structure. Based on the previous geological understanding, identify the post-stack seismic anisotropy and analyze the possibility of fracture development at different positions. The data obtained from seismic inversion is large in quantity and has many meaningless numbers. Cluster analysis is used to reduce the data dimension. Half of the maximum possibility is used as the threshold. For values less than the threshold, relatively large clustering radii and sample numbers are set, and for values greater than the threshold, relatively small clustering radii and sample numbers are set. After dimension reduction, an initial reservoir inversion dataset is formed. Multiply each data in the initial inversion dataset by several groups of random numbers between 0 and 1 to form several groups of initial porosity data.

[0070] Using the aforementioned S4 step, numerical simulations are performed on several groups of initial porosity data using the equivalent continuous medium theory to calculate the tracer concentration in the production wells, and the parameters are inverted based on the difference between the predicted value and the actual observed value.

[0071] As Figure 5 shown in Figure a, after 37,000 inversions, the error value between the simulated curve and the measured curve has been lower than the threshold. At this time, it is considered that the inversion structure is relatively similar to the predicted structure. As Figure 5 shown in Figure b, the positions of Wells GH01 and GH02 are respectively shown by red pentagrams. It can be seen from the figure that the porosity around Well GH02 reaches the maximum value in the region (~10 -2 %), and it is speculated that chemical stimulation may have been used to dissolve the granite, forming a regional high-permeability region.

[0072] As Figure 6 shown in the figure, draw the temperature measurement diagram after fracturing to verify the accuracy of the method. Taking Wells GH02 and GH03 as examples, perform steady-state temperature measurement on Wells GH02 and GH03. In the absence of cold water interference, the formation temperature almost shows a linear upward trend. Inject cold water for fracturing testing. At the positions where there are seepage channels, the temperature drops suddenly due to the continuous outflow of cold water, and the mutation point is the liquid injection channel of the reservoir. By comparing the reservoir fracture probability volume with the temperature measurement after fracturing, it is found that all fracture probability volumes correspond to temperature mutations. Among them, the fracture probability volume often appears above the temperature mutation point. The main reason is that the temperature probe has a time delay and cannot quickly capture the temperature change, and the fracture probability volume is the real liquid injection channel.

[0073] As can be seen from this method, in the past, when arranging the trajectory of hot dry rock drilling wells, due to the low regularity of methods such as 3D seismic ( Figure 4 a), the distribution of microseismic events ( Figure 2 ) was often used as a reference. Microseismic events are spatially discrete, and it is difficult to evaluate the connectivity between them. This method effectively identifies fracture connectivity channels by newly proposing an artificial intelligence algorithm for parameter inversion ( Figure 5 a). Subsequently, parameter updates and model refinement can be carried out according to the continuous deepening of geological understanding, gradually approaching the true geological reservoir. Moreover, the inversion content is mainly carried out indoors, with advantages such as low cost and high efficiency, and has a wide application range. The present invention provides a method for finely characterizing microfractures in a hot dry rock reservoir that integrates multi-source heterogeneous data, forms a systematic geological understanding by integrating existing data, effectively couples joint fracture surveys, geophysical exploration, and numerical simulation results by newly proposing a parameter inversion algorithm using artificial intelligence methods, and successfully realizes the integration of shallow-medium-deep and field-indoor data of hot dry rock, and finely inverses the geological structure of the reservoir; this method is implemented based on existing drilling and geophysical exploration, with relatively less workload and low cost, but has high efficiency; this method can be continuously updated as geological understanding deepens, gradually approaching the true reservoir structure. In actual exploitation, the working accuracy can be gradually improved as the exploration degree increases, and it has a strong fit with actual work.

[0074] The implementation method of the present invention has been described in detail above, but these are only examples for easy understanding and should not be regarded as limiting the scope of the present invention. Similarly, any professional in the relevant field can make various possible equivalent changes or substitutions according to the technical solution of the present invention and the description of its preferred embodiments, but all these changes or substitutions should fall within the protection scope of the claims of the present invention.

Claims

1. A method for fine characterization of micro fractures in hot dry rock reservoirs based on multi-source heterogeneous data, characterized in that: The following steps are involved: S1: Conduct joint and fissure investigation and develop outcrops of the same lithology around the site; S2: Conduct fracture logging on existing wells drilled in hot dry rocks, focusing on analyzing the direction, aperture, type, and occurrence of fractures in the reservoir section, identifying the location and scale of favorable fracture development areas, counting the number of fractures and drawing a trend rose diagram, and preliminarily determining the occurrence of major and minor fractures based on the fracture development of the core; S3: The Bayesian formula is used to perform seismic inversion on the high-precision seismic exploration data implemented in the hot dry rock site, and the post-stack seismic anisotropy is identified based on the previous geological knowledge, and the possibility of fracture development in different positions is analyzed; the amount of data obtained by seismic inversion is large and there are many meaningless numbers, so cluster analysis is used to reduce the data dimension, and the initial data set for reservoir inversion is formed after the dimension reduction; each data of the initial reservoir inversion data set is multiplied by several groups of 0-1 random numbers to form several groups of initial porosity data; S4: Numerical simulation is performed on several groups of initial porosity data using equivalent continuum theory to calculate the tracer concentration in the production wells, and the parameters are inverted based on the difference between the predicted value and the actual observed value.

2. The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to claim 1 is characterized in that: The joint and fissure investigation in step S1 focuses on analyzing the morphology, type, origin, length, aperture, and filling minerals of the joints and fissures.

3. The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to claim 1 is characterized in that: In the step S1, if the reservoir lithology is sedimentary rock, the rock mass with the same sedimentation and diagenesis process is investigated; for igneous rock, the rock mass with the same lithofacies type, intrusive secondary, and structural component is investigated; for metamorphic rock, the rock mass with the same metamorphic phase and metamorphic mineral combination, deformation and metamorphic process is investigated.

4. The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to claim 1 is characterized in that: The step S2 counts the number of cracks in units of 10 m and draws a trend rose diagram.

5. The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to claim 1 is characterized in that: In the step S3, the data dimension is reduced by taking half of the maximum possibility as a threshold, and values ​​smaller than the threshold are set with a relatively large cluster radius and number of samples, while values ​​larger than the threshold are set with a relatively small cluster radius and number of samples.

6. The method for fine characterization of micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to claim 1 is characterized in that: The high-precision seismic exploration data in step S3 is one or more of three-dimensional seismic data, vertical seismic profile, time-lapse seismic data and cross-well seismic data.

7. The method for finely characterizing micro-fractures in hot dry rock reservoirs based on integrated multi-source heterogeneous data according to any one of claims 1 to 6, characterized in that: The S4 step includes: 1) Set the assimilation number Na, and vectorize several groups of initial porosity data generated by previous geological knowledge to form a priori parameter set Where Ne represents the number of parameters in one assimilation; 2) Carry out numerical simulation: Where g() represents the tracer concentration of the production well calculated by numerical simulation, represents the predicted value; 3) Calculate the predicted value covariance Covariance between parameter values ​​and predicted values Use the following formula to find The pseudo-inverse of is: Where U and V are column orthogonal matrices, and ∑ is a diagonal matrix; 4) Calculate the Kalman gain: 5) Analyze the Kalman gain and determine the further update logic: exceeds_count=sum(delta_m(:)>1|delta_m(:)<0); total_elements=numel(delta_m); if exceeds_count>0.1*total_elements %Handle the situation where delta_m is greater than 0 delta_m_positive=delta_m>0; % positive value mask exceeds_positive=delta_m>10*m_pred&delta_m_positive; % judged to be greater than 10 times the original value delta_m(exceeds_positive)=10*m_pred(exceeds_positive); % adjusted to 10 times the original value % Handle the situation where delta_m is less than 0 delta_m_negative=delta_m<0; exceeds_negative=delta_m<-10*m_pred&delta_m_negative; delta_m(exceeds_negative)=-10*m_pred(exceeds_negative); 6) Update parameter values ​​and adjust parameters according to upper and lower limits.

Images