An intelligent design method for pressure-containing plugging system in ultra-deep fractured reservoirs

Abstract

The present invention discloses an intelligent design method for a pressure-bearing plugging system for an ultra-deep fractured reservoir, and belongs to the technical field of leakage control. The present invention includes establishing a high-precision plugging formula prediction model based on machine learning and multi-task learning, and then establishing a plugging formula characteristic parameter screening mathematical model with pressure bearing capacity and cumulative leakage as objective functions, obtaining the plugging formula characteristic parameters by solving the mathematical model, and finally matching the plugging materials according to the characteristic parameters, thereby determining the plugging formula. The present invention designs the plugging formula based on machine learning, which helps to improve the accuracy and speed of decision-making. The designed plugging formula has a high one-time plugging success rate, rapid response, and convenient operation. It can improve the plugging efficiency of drilling fluid leakage during drilling and reduce the time cost and economic loss caused by well leakage. At the same time, the prediction model of the present invention takes into account the influence of crack width on the plugging effect, and has strong adaptability and a wide range of applications.

Description

Technical Field

[0001] The present invention belongs to the technical field of well leakage control in the process of oil and gas reservoir exploration and development, also known as the field of leakage control technology, and also involves the field of data-driven machine learning technology, specifically an intelligent design method for a pressure plugging system for an ultra-deep fractured reservoir. Background Art

[0002] In deep and ultra-deep well drilling, lost circulation is a major engineering challenge, especially when it occurs in the reservoir section, which can seriously hinder the discovery, evaluation and efficient utilization of oil and gas reservoirs, thereby significantly reducing oil and gas production. When drilling deep fractured reservoirs, the loss of drilling fluid usually leads to the most serious reservoir damage during drilling, because solids and liquids in the working fluid will penetrate into the reservoir in large quantities, which may trigger various types of reservoir damage, such as solid plugging, fluid and stress-sensitive damage, and liquid phase closure damage. The plugging formula is the key to whether the reservoir fractures can be effectively plugged. At present, the design of plugging formulas often adopts empirical or semi-empirical methods. This method lacks scientific basis, resulting in a large amount of waste of plugging materials and unsatisfactory crack plugging quality. The plugging formula design method needs to be improved urgently. In addition, the reaction from the beginning of leakage to the formation of the plugging formula is not fast enough, which delays the working time. Therefore, it is of great significance to design a plugging formula with high plugging quality according to the characteristics of reservoir fractures. Summary of the invention

[0003] To solve the above problems, the purpose of the present invention is to provide an intelligent design method for a pressure-bearing plugging system for ultra-deep fractured reservoirs, which uses multi-task learning and machine learning to establish a high-precision plugging formula prediction model, and then establishes a mathematical optimization model with pressure-bearing capacity and cumulative leakage as objective functions. The characteristic parameters of the plugging formula are obtained by solving the mathematical optimization model, and finally, the plugging materials are screened according to the characteristic parameters of the plugging formula.

[0004] The present invention is achieved through the following technical solutions:

[0005] An intelligent design method for a pressure-containing plugging system for ultra-deep fractured reservoirs comprises the following steps:

[0006] S1. Determine the evaluation index of the plugging performance of the plugging formula and the characteristic parameters related to the evaluation index, wherein the evaluation index includes the pressure bearing capacity and the cumulative leakage amount of the plugging formula under a certain crack width condition, and the characteristic parameters include the crack width and the characteristic parameters of the plugging formula when determining the plugging performance of the plugging formula;

[0007] In some embodiments, the components of the plugging formula are divided into three categories: bridging materials, filling materials, and deformable reinforcing materials. Different characteristic parameters are determined for each component according to its functional difference. That is, the characteristic parameters of the plugging formula also include characteristic parameters of the bridging material, characteristic parameters of the filling material, and characteristic parameters of the deformable reinforcing material. For example, in some embodiments, the characteristic parameters of the bridging material include the amount of the bridging material and the content of various bridging materials (the proportion of the bridging material in the total amount of the plugging formula), the friction coefficient, D 90 (particle size corresponding to 90% of the cumulative distribution in the system), high temperature resistance, acid solubility and compression resistance;

[0008] The characteristic parameters of the filling material include the amount of the filling material and the content of various filling materials, friction coefficient, D 90 , high temperature resistance and acid solubility;

[0009] The characteristic parameters of the deformable reinforcement material include the quantity of the deformable reinforcement material and the content of various deformable reinforcement materials, friction coefficient, fiber aspect ratio, high temperature resistance and acid solubility.

[0010] S2. Collect the evaluation index and characteristic parameters of each formula as a data set, and train a machine learning model based on multi-task learning to obtain a plugging formula prediction model, wherein the plugging formula prediction model is used to predict the evaluation index of the formula through the characteristic parameters of the plugging formula, that is, to predict the pressure bearing capacity and cumulative leakage of the formula through the characteristic parameters;

[0011] S3. Determine the leakage type of the leaking well, and establish a mathematical model for screening characteristic parameters of the plugging formula based on the pressure bearing capacity and cumulative leakage volume;

[0012] The objective function of the mathematical model for screening characteristic parameters of the plugging formula is:

[0013] f=Max(w1Y1-(1-w1)Y2)

[0014] In the formula, f is the target value, Max() means taking the maximum value of the function in the brackets; Y1 is the pressure bearing capacity; Y2 is the cumulative leakage; w1 is the evaluation weight of the pressure bearing capacity, the induced rupture leakage is 0.4, the crack extension leakage is 0.5, and the natural fracture leakage is 0.7;

[0015] The constraint conditions of the mathematical model for screening characteristic parameters of the plugging formula are:

[0016] Y1,Y2=M(X)

[0017] x i (min)≤x i ≤x i (max),x i ∈X

[0018] Among them, M() is the plugging formula prediction model; x i is the value of the i-th characteristic parameter in the plugging formula x, and X is the set of characteristic parameters of the plugging formula x; i (max), x i (min) are the upper limit and lower limit of the ith characteristic parameter respectively. The specific value can be preset manually or the extreme value can be taken from the data set in step S2.

[0019] S4, substituting the fracture width of the lost circulation well into the plugging formula characteristic parameter screening mathematical model and solving the mathematical model to obtain the plugging formula characteristic parameters;

[0020] The plugging formula characteristic parameter screening mathematical model of the present invention is a mathematical model involving an optimization problem. For such an optimization problem, it is difficult to solve with a traditional algorithm. A heuristic algorithm can be used to solve the optimization problem to obtain a stable and reliable optimization result. In some embodiments, the plugging formula characteristic parameter screening mathematical model can be solved using an improved particle swarm algorithm in the prior art. For the specific solution process, reference can be made to "An Efficient Improved Particle Swarm Optimization Algorithm" by Li Hongliang.

[0021] As a specific implementation of the present invention, the constraint conditions of the plugging formula characteristic parameter screening mathematical model also include:

[0022] y=ae bm

[0023] Where m is the particle size D of the bridging material 90 The ratio of crack width to y is the content of bridging material; a and b are constants related to the friction coefficient, and their values ​​can be determined by the design chart of high-efficiency bridging plugging material dosage ( Figure 5 ) is obtained; further, the plate can be digitized to facilitate computer processing of data.

[0024] S5. Determine the plugging formula based on the plugging formula characteristic parameters.

[0025] In some embodiments, the plugging formula can be determined manually based on the characteristic parameters, but this requires experience to obtain better results. In other embodiments, matching can be performed through a similarity calculation model to determine the components of the plugging formula. For example, each component of each plugging formula in the data set is taken as an individual, and a plugging material performance parameter database is established. According to the cosine similarity, individuals are screened from the plugging material performance parameter database as components of the plugging formula. During the specific screening, the higher the cosine similarity, the higher the similarity between the two components. The individual with the highest similarity to the plugging formula characteristic parameters determined in step S4 can be selected as the component of the plugging formula. Of course, issues such as economy can also be considered, so as to screen the ideal components of the plugging formula from multiple individuals with high similarity.

[0026] The calculation formula of the cosine similarity is:

[0027]

[0028] In the formula, a i is the value of the i-th characteristic parameter of a component in the plugging formula, A is the set of characteristic parameters of a component in the plugging formula, A=(a1, a2,…, a n ), n is the total number of characteristic parameters; b i is the value of the i-th characteristic parameter of a certain individual in the plugging material performance parameter database, B is the set of characteristic parameters of a certain individual in the plugging material performance parameter database, B = (b1, b2, ..., b n ).

[0029] Compared with the prior art, at least one embodiment of the present invention has the following beneficial effects:

[0030] (1) The present invention designs plugging formulas based on machine learning, which helps to improve the accuracy and speed of decision-making and improves plugging efficiency.

[0031] (2) The prediction model of the present invention takes into account the influence of crack width on plugging effect, and has strong adaptability and a wide range of applications.

[0032] (3) The present invention classifies and characterizes the components of the plugging formula, thereby improving the accuracy of model prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 It is a design flow chart of the intelligent design method of the ultra-deep fractured reservoir pressure plugging system in Example 1;

[0034] Figure 2 It is a schematic diagram of the plugging formula prediction model;

[0035] Figure 3aThis is the evaluation result diagram of the plugging formula prediction model without feature enhancement processing;

[0036] Figure 3b This is the evaluation result diagram of the plugging formula prediction model using feature enhancement processing;

[0037] Figure 4 It is the solution flow chart of the improved particle algorithm;

[0038] Figure 5 It is based on the design drawing of adding high-efficiency bridging plugging materials;

[0039] Figure 6 It is a distributed modulus design plate based on dense filling;

[0040] Figure 7 This is a diagram showing the effect of an experimental verification of the leak-proof formula. DETAILED DESCRIPTION

[0041] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are 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 creative work are within the scope of protection of the present invention.

[0042] Example 1

[0043] An intelligent design method for pressure-bearing plugging system in ultra-deep fractured reservoirs, such as Figure 1 As shown, the following steps are included:

[0044] S1. Determine the evaluation index of the plugging performance of the plugging formula and the characteristic parameters related to the evaluation index, wherein the evaluation index includes the pressure bearing capacity and cumulative leakage of the plugging formula under a certain crack width condition, which can be obtained through an indoor plugging formula pressure plugging experiment; In this embodiment, the plugging formula is divided into three types according to the different roles played by bridging, filling, and deformed reinforcement in the crack plugging layer, namely, bridging materials, filling materials, and deformed reinforcement materials. The determined characteristic parameters include the crack width and formula characteristic parameters when determining the plugging performance of the plugging formula, wherein the formula characteristic parameters include the characteristic parameters of the bridging material, the characteristic parameters of the filling material, and the characteristic parameters of the deformed reinforcement material. The characteristic parameters of the bridging material include the amount of the bridging material and the content of various bridging materials, the friction coefficient, D 90 , high temperature resistance, acid solubility and compressive resistance; the characteristic parameters of the filling material include the amount of filling material and the content of various filling materials, friction coefficient, D 90, high temperature resistance and acid solubility; the characteristic parameters of the deformed reinforcement materials include the number of deformed reinforcement materials and the content of various deformed reinforcement materials, friction coefficient, fiber aspect ratio, high temperature resistance and acid solubility.

[0045] In this embodiment, an indoor leak-blocking formula pressure-bearing leak-blocking experiment was conducted on more than 200 kinds of leak-blocking formulas, and characteristic parameters of each leak-blocking formula were obtained.

[0046] S2. The evaluation index and characteristic parameters of each formula are collected as a data set, and a plugging formula prediction model is obtained by training a machine learning model based on multi-task learning. The plugging formula prediction model is used to predict the evaluation index of the plugging formula through the characteristic parameters, that is, to predict the pressure bearing capacity and cumulative leakage of the plugging formula through the characteristic parameters of the plugging formula and the width of the cracks that need to be plugged by the plugging formula;

[0047] In this embodiment, a total of 108 performance parameters of plugging materials are collected. The collected formulas include those containing only one of bridging materials, filling materials, and deformed reinforcing bar materials, as well as those containing a combination of two of the three, and also include the three coexisting situations. Therefore, it can be considered that the collected plugging formula types are comprehensive and diverse, which is conducive to the model learning more comprehensive information during training to improve the accuracy of prediction. In this embodiment, the characteristic parameters in the data set are sequentially subjected to feature enhancement processing and normalization processing, wherein, during feature enhancement processing, the crack width and component content are not enhanced. During feature enhancement processing, all values ​​of this type of characteristic parameters are multiplied by a random scalar at the same time, and a random deviation is added, and finally a new data sequence with an unchanged overall change trend is obtained. Data normalization is a routine operation and will not be described in detail here.

[0048] When training the model, 80% of the data set is used as the training set, and the remaining 20% ​​is used as the validation set. Figure 2 As shown in the figure, the prediction model is divided into a shared layer and two task (pressure bearing capacity and cumulative leakage) private layers. The shared layer has 1 input layer and 4 hidden layers, and the number of neurons is 48, 128, 64, 32, and 16 respectively; the private layer of the two tasks has 1 hidden layer and one output layer, and the number of neurons is 8 and 1 respectively; Adam is used as the optimization function, the learning rate is 0.01, the batch size is 4, and the number of training rounds is 100.

[0049] In addition, in order to determine the accuracy of the plugging formula prediction models established in different ways, this embodiment also sets a comparative example without feature enhancement processing. The plugging formula prediction models established in two ways are used to calculate the validation set respectively. The results are shown in Figure 3. When the feature enhancement model is not used, the evaluation index result of the validation set of task 1 (pressure bearing capacity) is: R 2=0.87, RMSE = 1.66, sMAPE = 0.54, the evaluation index results of the validation set of task 2 (cumulative loss) are: R 2 =0.82, RMSE = 93.01, sMAPE = 0.31; when the feature enhancement model is used, the evaluation index results of the validation set of task 1 (pressure bearing capacity) are: R 2 =0.91, RMSE = 1.41, sMAPE = 0.52, the evaluation index results of the validation set of task 2 (cumulative loss) are: R 2 =0.86, RMSE=83.05, sMAPE=0.21. This shows that the multi-task learning plugging formula high-precision proxy model has good performance, and the feature enhancement method can improve the model prediction accuracy. Therefore, the plugging formula prediction model with feature enhancement is used for subsequent operations.

[0050] S3. Determine the leakage type of the leakage well, and establish a mathematical model for screening characteristic parameters of the plugging formula with pressure bearing capacity and cumulative leakage as the target; in this embodiment, the leakage type is fracture extension leakage, and the relative weights of pressure bearing capacity and cumulative leakage are 0.5 and 0.5 respectively. Therefore, the objective function of the mathematical model for screening characteristic parameters of the plugging formula is:

[0051] f=Max(0.5Y1-0.5Y2)

[0052] In the formula, f is the target value; Y1 is the pressure bearing capacity; Y2 is the cumulative leakage;

[0053] The constraint conditions of the mathematical model for screening characteristic parameters of the plugging formula include:

[0054] Y1,Y2=M(X)

[0055] x i (min)≤x i ≤x i (max),x i ∈X

[0056] Among them, M() is the plugging formula prediction model; x i is the value of the i-th characteristic parameter in the plugging formula x, and X is the set of characteristic parameters of each component in the plugging formula x; i (max), x i (min) are the upper and lower limits of the ith characteristic parameter respectively, and the extreme value is taken from the data set in step S2 when taking the specific value.

[0057] The constraint conditions of the mathematical model for screening characteristic parameters of the plugging formula bridging material in this embodiment also include:

[0058] y=ae bm

[0059] Where m is the ratio of material particle size to crack width; y is the amount of bridging material added; a and b are constants related to the friction coefficient, and their values ​​can be determined by the design chart of efficient bridging plugging material addition ( Figure 5 ) is obtained, and this embodiment performs numerical processing on it. First, four bridging material addition curves with different friction coefficients are determined. When the friction coefficient is between the two, the bridging material content is obtained by interpolation.

[0060] When the friction coefficient of the bridging material is 1.2, y = 9.7184e -6.936x ;

[0061] When the friction coefficient of the bridging material is 1.0, y = 16.893e -7.69x ;

[0062] When the friction coefficient of the bridging material is 0.8, y = 25.564e -7.824x ;

[0063] When the friction coefficient of the bridging material is 0.6, y = 109.36e -9.73x .

[0064] S4. Substitute the crack width of 5 mm of the leaking well into the mathematical model for screening characteristic parameters of the plugging formula, and use the improved particle swarm algorithm to solve the mathematical model for screening characteristic parameters of the plugging formula to obtain the characteristic parameters of the plugging formula, namely, the characteristic parameters of the bridging material, the characteristic parameters of the filling material and the characteristic parameters of the deformed reinforcement material, see Table 1 for details.

[0065] The specific process of using the improved particle swarm algorithm to solve the mathematical model for filtering the characteristic parameters of the plugging formula is as follows: Figure 4 As shown, the specific steps are:

[0066] S41, classifying the particle composition in the particle swarm algorithm;

[0067] S42, each particle swarm independently performs a standard particle swarm optimization process;

[0068] S43. Within each group, evaluate the performance of all particles and save the optimal solution of each group;

[0069] S44, after a predetermined iteration cycle, the groups exchange information and exchange optimal solutions;

[0070] S45. Examine diversity within and between groups and, if diversity falls below a certain threshold, intervene as necessary to increase diversity;

[0071] S46. Update the global optimal solution based on the information provided by each group;

[0072] S47, checking whether the termination condition of the algorithm is met and whether the maximum number of iterations is reached;

[0073] S48. Return the global optimal output solution.

[0074] Among them, the grouping rule of the S41 initial particle swarm is:

[0075] Assume that the particle swarm consists of m particles, and divide it into three sub-populations S1, S2, and S3. The number of particles in each sub-population is m1, m2, and m3, where m = m1 + m2 + m3. Each sub-population uses a different evolution equation. Population S1 uses a fast-converging evolution equation to enhance local optimization capabilities; population S2 uses a "social model" (C1 = 0); and population S3 uses an evolution equation with global search capabilities. The specific evolution equations used are as follows: V i+1 =ωV i +C1R1(P best,i -X i )+C2R2(G best -X i )

[0076] ω=0.3

[0077] V i+1 =ωV i +C2R2(G best -X i )

[0078] ω=0.6

[0079] V i+1 =ωV i +C1R1(P best,i -X i )+C2R2(G best -X i )

[0080] ω=0.95

[0081] i represents the current iteration step, V i+1 and V i denote the particle speeds corresponding to the next iteration step and the current iteration step, respectively; C1 and C2 denote learning factors; R1 and R2 are random numbers between 0 and 1; P best,i Indicates that the current corresponding position is a candidate solution, G best represents the global optimal solution achieved by comparing the optimal solutions of all individuals; w represents the inertia weight, X i Indicates the particle position corresponding to the current iteration step.

[0082] Table 1 Characteristic parameters of plugging formula

[0083]

[0084] S5. Determine the components of the plugging formula based on the characteristic parameters of the plugging formula, specifically including taking each component of each plugging formula in the data set as an individual, establishing a plugging material performance parameter database, and selecting the individuals with the highest similarity from the plugging material performance parameter database as the components of the plugging formula according to the cosine similarity.

[0085] The calculation formula of the cosine similarity is:

[0086]

[0087] In the formula, a i is the value of the i-th characteristic parameter of a component in the plugging formula, A is the set of characteristic parameters of a component in the plugging formula, A=(a1, a2,…, a n ), n is the total number of characteristic parameters; b i is the value of the i-th characteristic parameter of a certain individual in the plugging material performance parameter database, B is the set of characteristic parameters of a certain individual in the plugging material performance parameter database, B = (b1, b2, ..., b n )

[0088] After calculation, the components of the screened plugging formula are shown in Table 2.

[0089] Table 2 Statistical table of cosine similarity matching results of plugging material performance parameter database

[0090]

[0091] In this embodiment, for LCM-Z13, LCM-Z27, LCM-Z30 particle size distribution range [D 90 ,D 10 ] are [6344.52, 3534.16], [2833.25, 1532.31], [2130.12, 425.31] respectively, and the particle size distribution range of the main bridging material LCM-Z32 is [8344.52, 7134.32]. It does not conform to the principle of continuous distribution of particle sizes of all materials. It is necessary to optimize the particle size distribution range. The amount of filling materials and particle size distribution can be adjusted by combining the difference compensation method and the close packing theory. This belongs to the existing technology. For example, Xu Chengyuan's "Research on the Design Method of Deep Fracture Reservoir Plugging Formula Based on Efficient Bridging and Dense Filling" has an introduction. The specific optimization steps include:

[0092] S51. Use the “difference compensation method” to optimize the particle size distribution of the filling material to ensure that the particle sizes of all materials are continuously distributed, that is, each component is divided into a continuous distribution body:

[0093] Assume that there are n components of plugging materials in the plugging formula, and the particle size distribution ranges of each component are:

[0094] [x1,x2],[x3,x4],[x5,x6],…,[x n-1 ,x n ]

[0095] The particle size distribution after the difference compensation method is as follows:

[0096]

[0097] Among them, x1 and x n are the maximum and minimum particle sizes in the system, respectively.

[0098] After optimizing the particle size distribution range, the particle size of LCM-Z32, LCM-Z13, LCM-Z27, and LCM-Z30 [D 90 , D 10 ] are: [8344.52, 6344.52], [6344.52, 3183.70], [3183.70, 1831.15], [1831.15, 425.31].

[0099] S52, determine the particle size ratio index η. In this embodiment, the crack width to particle size ratio λ (ratio of crack width to maximum bridging material particle size (D90)) is 0.6. Under the premise of meeting the maximum stacking efficiency, according to the distribution modulus design chart based on dense filling ( Figure 6 ), find the horizontal coordinate corresponding to the lowest porosity point of the plugging layer, that is, the particle size index η is 0.5.

[0100] S53. Under the framework of the close packing theory, the amount of filler material to be added needs to take into account the particle size distribution and density of the material. For each material, use the following formula to calculate its volume share:

[0101]

[0102] in is the volume percentage of the material, %; D i90 is the maximum particle size in the interval, mm; D i10 is the minimum particle size in the interval, mm; D L is the maximum particle size of the particles, mm; D s is the minimum particle size of the particles, mm; η is the particle size ratio index.

[0103] According to the above formula, the dosage of LCM-Z30 is designed as: D i90 =1831.15, D i10 =425.31, D L=8344.52, D s =425.31, from which the designed volume percentage of LCM-Z30 is calculated as:

[0104]

[0105] The designed volume percentage of LCM-Z27 is: D i90 =3183.70, D i10 =1831.15,D L =8344.52, D s =425.31, from which the designed volume percentage of LCM-Z27 is calculated as:

[0106]

[0107] The designed volume percentage of LCM-Z13 is: D i90 =6344.52, D i10 =3183.70, D L =8344.52, D s =425.31, from which the designed volume percentage of LCM-Z13 is calculated as:

[0108]

[0109] S54. Convert the volume ratio of the filling material into the mass ratio. When the density of the plugging materials is the same, the mass volume ratio of the plugging materials is:

[0110]

[0111] Where: C i is the mass volume percentage of material i, %; C m is the mass proportion of the material with the largest particle size, %; is the volume percentage of material i, %; is the volume proportion of the material with the largest particle size, %.

[0112] When the density of the plugging material is different, the mass volume ratio of the plugging material is:

[0113]

[0114] Where: i is the density of material i, g / cm 3 ρ m is the density of the largest particle size material, g / cm 3 .

[0115] In this embodiment, the bridging material LCM-32 is the material with the largest particle size in the system. The mass fraction of LCM-32 can be obtained by solving the model to be 1.7%, the volume fraction of LCM-32 is 16.53%, and the density is 1.05g / cm3. According to the formula, the volume fraction is converted into the mass fraction. For the filling material LCM-30, the volume fraction is 31.27%, the density is 1.53g / cm3, and the mass fraction of LCM-30 is obtained as follows:

[0116]

[0117] Similarly, the mass fractions of LCM-27 and LCM-13 are 0.02278 and 0.01842, respectively. This forms a complete intelligent design method for high-pressure plugging system in ultra-deep fractured formations, and the final plugging formula is base slurry + 1.7% LCM-Z32 + 2% LCM-Z9 + 2.5% LCM-Z21 + 1.8% LCM-Z13 + 2.3% LCM-Z27 + 2.2% LCM-Z30 + 4% LCM-K2.

[0118] S6. In order to verify the performance of the designed plugging formula, an indoor plugging formula pressure plugging test was carried out. The results are as follows Figure 7 As shown, this plugging formula has a good plugging effect.

[0119] Table 3 Results of plugging formula design

[0120]

[0121]

[0122] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the embodiments of the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. An intelligent design method for a pressure-bearing plugging system for ultra-deep fractured reservoirs, characterized in that: The following steps are involved: S1. Determine the evaluation index of the plugging performance of the plugging formula and the characteristic parameters related to the evaluation index, wherein the evaluation index includes the pressure bearing capacity and the cumulative leakage amount of the plugging formula under a certain crack width condition, and the characteristic parameters include the crack width and the characteristic parameters of the plugging formula when determining the plugging performance of the plugging formula; S2. Collect the evaluation index and the characteristic parameters of each formula as a data set, and obtain a plugging formula prediction model based on a machine learning model trained with multi-task learning, wherein the plugging formula prediction model is used to predict the evaluation index of the plugging formula through the characteristic parameters; S3. Determine the leakage type of the leaking well, and establish a mathematical model for screening characteristic parameters of the plugging formula based on the pressure bearing capacity and cumulative leakage volume; The objective function of the mathematical model for screening characteristic parameters of the plugging formula is: f=Max(w1Y1-(1-w1)Y2) In the formula, f is the target value, Max() means taking the maximum value of the function in the brackets; Y1 is the pressure bearing capacity; Y2 is the cumulative leakage; w1 is the evaluation weight of the pressure bearing capacity, among which the induced fracture type leakage is 0.4, the fracture extension type leakage is 0.5, and the natural fracture type leakage is 0.7; The constraint conditions of the mathematical model for screening characteristic parameters of the plugging formula include: Y1,Y2=M(X) x i (min)≤x i ≤x i (max),x i ∈X Among them, M() is the plugging formula prediction model; x i is the value of the i-th characteristic parameter in the plugging formula x, and X is the set of characteristic parameters of the plugging formula x; i (max), x i (min) are the upper and lower limits of the i-th characteristic parameter respectively; S4, substituting the fracture width of the lost circulation well into the plugging formula characteristic parameter screening mathematical model and solving it to obtain the plugging formula characteristic parameter; S5, determining a plugging formula for the lost circulation well based on the plugging formula characteristic parameters; The components of the plugging formula are divided into three categories: bridging material, filling material and deformable reinforcing bar material. The characteristic parameters of the plugging formula include the characteristic parameters of the bridging material, the characteristic parameters of the filling material and the characteristic parameters of the deformable reinforcing bar material. The characteristic parameters of the bridging material include the amount of bridging material and the content of various bridging materials, friction coefficient, D 90 , high temperature resistance, acid solubility and compression resistance; The characteristic parameters of the filling material include the amount of the filling material and the content of various filling materials, friction coefficient, D 90 , high temperature resistance and acid solubility; The characteristic parameters of the deformable reinforcement material include the quantity of the deformable reinforcement material and the content of various deformable reinforcement materials, friction coefficient, fiber aspect ratio, high temperature resistance and acid solubility.

2. The intelligent design method for a pressure-bearing plugging system for ultra-deep fractured reservoirs according to claim 1 is characterized in that: In step S5, each component of each plugging formula in the data set is taken as an individual, a plugging material performance parameter database is established, and individuals are selected from the plugging material performance parameter database as components of the plugging formula according to cosine similarity; The calculation formula of the cosine similarity is: In the formula, a i is the value of the i-th characteristic parameter of a component in the plugging formula, A is the set of characteristic parameters of a component in the plugging formula, A=(a1, a2,…, a n ), n is the total number of characteristic parameters; b i is the value of the i-th characteristic parameter of a certain individual in the plugging material performance parameter database, B is the set of characteristic parameters of a certain individual in the plugging material performance parameter database, B = (b1, b2, ..., b n ).

3. The intelligent design method for ultra-deep fractured reservoir pressure plugging system according to claim 1 is characterized in that: In step S4, an improved particle swarm algorithm is used to solve the mathematical model for screening characteristic parameters of the plugging formula.

Images