Formation pressure prediction-based pre-drilling well overflow leakage risk pre-judgment method

Abstract

The invention discloses a pre-drilling well overflow leakage risk prejudgment method based on formation pressure prediction. According to the method, the formation pressure calculation influence attribute characteristics of the adjacent wells are obtained according to the adjacent well data of the target well, and after the formation pressure calculation influence attribute characteristics of the adjacent wells are quantitatively represented through numerical processing, the weights of the formation pressure calculation influence attribute characteristics are analyzed based on the water injection principle; then, based on a KNN algorithm, the similarity between the adjacent well formation pressure calculation influence attribute characteristic samples is analyzed, the adjacent well formation pressure calculation influence attribute characteristic samples are screened according to the similarity, distance weighting analysis is conducted through a distance weighting method, and the weight of each formation pressure calculation influence attribute characteristic in the target well relative to the adjacent well is determined; the target formation pressure judgment model is established, the formation pressure of the target formation is predicted through the target formation pressure judgment model, the probability of occurrence of overflow events and leakage events of the designated well section in the target well is accurately predicted, and a foundation is laid for safe development of oil and gas fields.

Description

Technical Field

[0001] The present invention relates to the technical field of safe development of oil and gas resources, and in particular to a method for predicting well spill risk before drilling based on formation pressure prediction. Background Art

[0002] There are many uncertainties in the formation during the drilling and completion process, and there are complex well control situations. In serious cases, it will lead to the occurrence of well control emergency situations or blowouts, which seriously threaten the production safety of oil and gas resource development. In addition, once a blowout occurs, the situation downhole will become more complicated in the least, causing varying degrees of damage to the oil and gas layers; in severe cases, the blowout will be out of control, which will not only cause serious damage to oil and gas resources, but may also cause fires, resulting in serious consequences such as machine destruction, death, and environmental pollution, causing huge economic losses and extremely adverse social impacts.

[0003] Blowout is the primary risk of well control. Accurately predicting the target formation pressure before drilling and predicting overflow and leakage events are the core ways to achieve well control risk prevention and control. Important geological information such as formation pressure and stratigraphic position of adjacent wells before drilling is an important reference for drilling geological designers. However, the adjacent well data itself has many attributes (including lithology, formation pressure, overflow and leakage, etc.) and they influence each other. At the same time, the reference value of different adjacent well data varies significantly. This makes how to correctly select effective attributes and efficient adjacent well samples, and scientifically handle the relationship between adjacent well samples and target cases, the key to calculating formation pressure and quantitative risk prediction in drilling geological design.

[0004] The existing methods for predicting formation pressure before drilling mainly adopt two approaches:

[0005] 1. Obtain the formation velocity by means of seismic exploration, and then obtain the formation pressure by inversion through calculation formulas such as Eaton trend line and Fillipone model;

[0006] 2. By collecting data from adjacent wells in the same geological structure, a regional formation pressure map is established to predict the target formation pressure range.

[0007] Both of the above methods have certain limitations. The first method requires the use of geophysical radar and requires complex data processing in the later stage. It has poor accuracy and high inversion cost. It is suitable for the formation pressure range of large-scale geological structures, but difficult to be applied to the formation pressure prediction of mature blocks. The second method requires the use of empirical analogy, and mostly gives the formation pressure range of the target layer after a simple statistics of the historical formation pressure of adjacent wells. It relies too much on experience and ignores the weight of different case samples themselves, and lacks reliability.

[0008] Therefore, it is urgent to propose a method for predicting drilling spill risk before drilling based on formation pressure prediction, overcome the limitations of the above two methods, use a large amount of adjacent well data and combine machine learning methods to accurately predict the formation pressure of the target formation before drilling, and accurately predict spill events or leakage events during the drilling process, laying the foundation for guiding the drilling operations of on-site operators. Summary of the invention

[0009] The present invention aims to solve the above-mentioned problem and proposes a method for predicting the risk of drilling spillage before drilling based on formation pressure prediction. Based on the adjacent well data of the target well, the accurate prediction of the formation pressure of the target formation before drilling is realized, and the overflow or leakage events that may occur during the drilling process are predicted in advance, thereby providing a guarantee for the safe development of oil and gas fields.

[0010] The present invention specifically adopts the following technical solutions:

[0011] A method for predicting the risk of leakage before drilling based on formation pressure prediction includes the following steps:

[0012] Step 1, select the target formation, obtain the neighboring well data of the target well where the target formation is located, obtain the attribute characteristics of each neighboring well that affect the formation pressure calculation, and quantitatively characterize the attribute characteristics of each neighboring well that affect the formation pressure calculation to obtain the attribute characteristics of the formation pressure calculation that are numerically processed;

[0013] Step 2: According to the quantitatively characterized influencing attribute characteristics of formation pressure calculation, weight analysis is performed on the influencing attribute characteristics of each formation pressure calculation based on the water injection principle to obtain the weight of each formation pressure calculation influencing attribute characteristics;

[0014] Step 3: Analyze the similarity of the attribute feature samples that affect the formation pressure calculation of adjacent wells based on the KNN algorithm, screen the attribute feature samples that affect the formation pressure calculation of adjacent wells, and obtain the screened attribute feature samples that affect the formation pressure calculation of adjacent wells;

[0015] Step 4: Based on the screened adjacent well formation pressure calculation influencing attribute feature samples, a distance-weighted analysis is performed on the adjacent well formation pressure calculation influencing attribute feature samples of the target well based on a distance-weighted method to determine the weight of each formation pressure calculation influencing attribute feature in the target well relative to the adjacent wells;

[0016] Step 5, establish a target formation pressure discrimination model, combine the weights of the attribute characteristics affecting formation pressure calculation in the target well relative to the weights of the adjacent wells, use the target formation pressure discrimination model to predict the formation pressure G of the target formation, and use the similarity of the attribute characteristic samples affecting formation pressure calculation in the adjacent wells after screening by the target formation pressure discrimination model to draw a radar chart, and determine the reliability Q of the formation pressure prediction;

[0017] Step 6, based on the formation pressure, the possibility of drilling risk is predicted before drilling, and the formation pressure prediction value G and prediction reliability Q of the target formation are used to determine the probability of overflow events or leakage events occurring in each well section in the drilling design of the target well.

[0018] Preferably, the step 1 specifically includes the following steps:

[0019] Step 1.1, obtaining the attribute characteristics affecting formation pressure calculation;

[0020] Select the target formation, determine the horizon of the target formation, obtain the well data of the neighboring wells around the target well where the target formation is located, and obtain the attribute characteristics affecting the formation pressure calculation of each neighboring well based on the neighboring well data;

[0021] The formation pressure calculation influencing attribute characteristics include layer structure influencing factors and well location geological design influencing factors, wherein the layer structure influencing factors include structural similarity and lithological similarity, and the well location geological design influencing factors include layer thickness, well-to-well distance, layer bottom boundary distance, lithological type index and structural type index;

[0022] Step 1.2, processing the stratigraphic structural influencing factors of each adjacent well;

[0023] The stratigraphic structural influencing factors of each adjacent well are standardized, quantitatively characterized, and the normalized values ​​of the stratigraphic structural influencing factors of each adjacent well are obtained;

[0024] Step 1.3, processing the influencing factors of well location geological design of each adjacent well;

[0025] Firstly, the strata corresponding to the target strata in each adjacent well are determined according to the horizon of the target strata, and based on the data of the adjacent wells, the lithology type index of the strata is divided according to the mineral composition and particle size, and the structural type index of the strata is divided according to the location of the geological structure where the strata is located and whether there is a fault interval, so as to determine the lithology type index and structural type index of the strata corresponding to the target strata in each adjacent well;

[0026] Secondly, based on the indicators of each lithology type and each structural type, combined with the layer thickness, distance between wells, and distance between layer bottom boundaries, the importance weight vector binary allocation scale value of each well location geological design influencing factor is calculated respectively, and a binary weight allocation judgment matrix is ​​constructed to quantitatively characterize the well location geological design influencing factors of each adjacent well;

[0027] Finally, the binary weight allocation judgment matrix is ​​solved by successively using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method, and the weight matrix of the well location geological design influencing factors obtained by the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method is obtained. The weight of the geological design influencing factors of each well location is obtained by calculating the average value.

[0028] Preferably, in step 1.2, the calculation formula for the standardization process is as shown in formula (1):

[0029]

[0030] In the formula, max is the maximum value of the sample data to be processed; min is the minimum value of the sample data to be processed; y is the original value of the sample data; and x is the normalized value of the sample data.

[0031] Preferably, the binary distribution scale value of the importance weight vector is determined based on expert experience, and the specific determination process is:

[0032] Randomly select index A and index B from the factors affecting well location geological design. When index A is as important as index B, the importance weight vector of index A is assigned a binary scale value C. AB is 0.5, the importance weight vector of indicator B is assigned a binary scale value C BA is 0.5;

[0033] When indicator A is slightly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.55, and the importance weight vector of index j is assigned a binary scale value C BA is 0.45;

[0034] When indicator A is significantly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.6, and the importance weight vector of index j is assigned a binary scale value C BA is 0.4;

[0035] When indicator A is more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.7, and the importance weight vector of index j is assigned a binary scale value C BA is 0.3;

[0036] When indicator A is extremely important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.8, and the importance weight vector of index j is assigned a binary scale value C BA is 0.2;

[0037] According to the binary distribution scale value of the importance weight vector of each well location geological design influencing factor, a binary weight distribution judgment matrix C is constructed for each well location geological design influencing factor, as shown in formula (2):

[0038]

[0039] In the formula, C ijis the binary allocation scale value of the importance weight vector of the i-th row and j-th column in the binary weight allocation judgment matrix; i is the row number of the binary weight allocation judgment matrix, i=1, 2, ..., n, j is the column number of the binary weight allocation judgment matrix, j=1, 2, ..., n, n is the total number of rows of the binary weight allocation judgment matrix;

[0040] Since the importance weight vector of indicator A is binary assigned scale value C AB The importance weight vector of index B is assigned to the scale value C BA There is complementarity between them, and the importance weight vector of indicator A is binary assigned scale value C AB It is related to the weight of indicator A and the weight of indicator B, as shown in formula (3):

[0041]

[0042] Where W A is the weight of indicator A, W B is the weight of indicator B;

[0043] Therefore, the weights of the factors affecting well location geological design can be determined by solving the binary weight allocation judgment matrix.

[0044] Preferably, when the reciprocal row sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is firstly reciprocated to construct a weight vector calculation matrix D. The weight vector calculation matrix Among them, d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector,

[0045] According to the weight vector calculation matrix D, the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method is obtained. In the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method, the absolute weight T of the index A is A for:

[0046]

[0047] In the formula, R A It is the intermediate calculation amount of indicator A.

[0048] Preferably, when the reciprocal column sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is firstly reciprocated to construct a weight vector calculation matrix D. The weight vector calculation matrix Among them, d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector,

[0049] According to the weight vector calculation matrix D, the weight matrix of the factors affecting the well location geological design is obtained. In the weight matrix of the factors affecting the well location geological design, the absolute weight of indicator B is T B for:

[0050]

[0051] In the formula, CB is the intermediate calculation amount of indicator B;

[0052] The absolute weight T of indicator B B Perform normalization processing to obtain the weight matrix of the factors affecting the well location geological design determined by the inverse series sum method;

[0053] The normalization calculation formula is:

[0054]

[0055] Where W B is the weight of indicator B.

[0056] Preferably, when the row-column geometric mean comparison method is used to solve the binary weight allocation judgment matrix C, the geometric mean of each row and the geometric mean of each column in the binary weight allocation judgment matrix C are calculated respectively, as shown in formula (7) and formula (8):

[0057]

[0058]

[0059] Where Gi is the geometric mean of the i-th row in the binary weight allocation judgment matrix, and Gj is the geometric mean of the j-th column in the binary weight allocation judgment matrix;

[0060] Divide the geometric mean of the same row number and column number to get the absolute weight T of indicator A A for:

[0061]

[0062] The absolute weight T of indicator A A After normalization processing, the weight matrix of the influencing factors of well location geological design determined by the row-column geometric mean comparison method is obtained.

[0063] Preferably, in step 2, weights are allocated to the attribute characteristics affecting formation pressure calculation based on the water injection principle, as shown in formula (10):

[0064]

[0065] In the formula, C Wis the total capacity of the attribute characteristics affecting formation pressure calculation, which is used to describe the amount of information contained in the attribute characteristics affecting formation pressure calculation; m is the dimension of the attribute characteristics affecting formation pressure calculation, μ k Calculate the mean of the influencing attribute characteristics for the k-th formation pressure, ω k Calculate the weight of the attribute characteristics for the k-th formation pressure, σ k is the standard deviation of the attribute feature affecting the k-th formation pressure calculation, k is the number of the attribute feature affecting the formation pressure calculation;

[0066] According to formula (10), the Lagrangian function is constructed and the partial derivative of the weight affecting the attribute characteristics of the formation pressure is calculated, as shown in formula (11):

[0067]

[0068] Where λ is the Lagrangian operator,

[0069] Based on the fact that the sum of the weights of the attribute features affecting the calculation of the formation pressure is 1, combined with the threshold ε, the formula (11) is constrained, and the weights of the attribute features affecting the calculation of the formation pressure are calculated as:

[0070]

[0071] In the formula, ε is the threshold value,

[0072] Preferably, in step 3, a complex KNN distance similarity calculation model is established, as shown in formula (13):

[0073]

[0074] In the formula, D(x,y) is the distance similarity between sample y and sample x, and x k Calculate the characteristic value affecting the attribute characteristics for the k-th formation pressure in sample x, y k Calculate the characteristic value affecting the attribute characteristics for the k-th formation pressure in sample y;

[0075] The distance similarity between two samples is calculated using the complex KNN distance similarity calculation model, and the calculated distance similarity result is compared with the preset accuracy value e;

[0076] If the distance similarity D(x,y) between sample y and sample x is greater than the preset precision value e, it is determined that sample y is not similar to sample x and sample y is removed;

[0077] If the distance similarity D(x, y) between sample y and sample x does not exceed the preset precision value e, it is determined that sample y is similar to sample x and sample y is retained.

[0078] Preferably, the precision value e is between 1 and 1.4.

[0079] Preferably, in step 4, the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well are calculated respectively, and weighted processing is performed based on the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well, so as to determine the weight of the p-th attribute characteristics influencing formation pressure calculation in the target well relative to the adjacent well q, as shown in formula (14):

[0080]

[0081] In the formula, ω pq Calculate the weight of the influencing attribute feature for the pth formation pressure in the target well relative to the adjacent well q, D pq is the Euclidean distance between the attribute characteristics affecting formation pressure calculation of the target well and the attribute characteristics affecting formation pressure calculation of the adjacent well, N pq The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent wells.

[0082] Preferably, the Euclidean distance between the attribute characteristics influencing the formation pressure calculation of the target well and the attribute characteristics influencing the formation pressure calculation of the adjacent well is as shown in formula (15):

[0083]

[0084] In the formula, D(X p ,X q ) is the Euclidean distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well, X p Calculate the influencing attribute characteristic matrix for the formation pressure of the target well, X q Calculate the attribute characteristic matrix affecting the formation pressure of adjacent wells; x qr Calculate the influencing attribute characteristic value for the rth formation pressure in the adjacent well, x pr is the r-th formation pressure calculation influencing attribute characteristic value in the target well, where r is the sequence number of the formation pressure calculation influencing attribute characteristic;

[0085] The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well is shown in formula (16):

[0086]

[0087] Where N pis the Mahalanobis distance between the p-th formation pressure calculation influencing attribute feature of the target well and the p-th formation pressure calculation influencing attribute feature of the adjacent well, x p Calculate the observation vector affecting the attribute characteristics for the p-th formation pressure in the target well, x p =(x p1 ,x p2 ,…,x pm ) T ;μ x The arithmetic mean vector of the sample values ​​of the attribute characteristics that affect the calculation of the formation pressure of the adjacent wells, μ x =(μ x1 ,μ x2 ,…,μ xn ) T ; ∑ is the covariance matrix of the attribute feature samples affecting the formation pressure calculation of the target well,

[0088] Preferably, in step 5, the target formation pressure discrimination model is:

[0089]

[0090] Where G is the predicted value of formation pressure, ω δ is the screening result of the attribute feature samples that affect the calculation of the formation pressure of adjacent wells, and I is the attribute feature sample matrix that affects the calculation of the formation pressure of adjacent wells.

[0091] Preferably, in step 6, the drilling risk type and occurrence probability are predicted before drilling according to the formation pressure prediction value G and prediction reliability Q of the target formation;

[0092] If G-ρ h -Δρ>0, it is predicted that the target formation will have an overflow event, and the probability of an overflow event is R1, as shown in formula (18):

[0093]

[0094] If h +Δρ-G>C′, then the target formation is predicted to have a leakage event, and the probability of a leakage event is R2, as shown in formula (19):

[0095]

[0096] In the formula, ρ h is the design density of the static column of drilling fluid, Δρ is the additional pressure equivalent density, and C′ is the formation fracture strength equivalent density.

[0097] The present invention has the following beneficial effects:

[0098] (1) The present invention predicts the formation pressure of the target formation based on the data of adjacent wells, without the need to use measurement methods such as geophysical exploration and logging while drilling, thereby reducing the cost of obtaining formation pressure and having stronger applicability.

[0099] (2) The present invention takes geological structure (structural similarity, formation thickness, formation bottom depth, lithology similarity) and well-to-well distance and other attribute factors as influencing attributes for formation pressure calculation. By quantifying structural similarity and lithology similarity, the present invention overcomes the previous formation pressure prediction method that only considers well spacing or rock properties, and avoids the problem of reference data misleading formation pressure prediction due to subjective judgment.

[0100] (3) The present invention combines the water injection iteration theory with the improved KNN algorithm using distance weighting, introduces the water injection principle to process the weights of the attributes that affect the formation pressure prediction, changes the previous practice of equally weighting the attributes, and corrects the weight relationship between the attributes. At the same time, based on the similarity of samples between adjacent wells, the screening of attribute feature samples that affect the formation pressure calculation of adjacent wells is achieved, ensuring the accuracy of the prediction of the formation pressure of the target formation.

[0101] (4) The present invention improves the method of assigning weights of attribute feature samples that affect formation pressure calculation in adjacent wells, and uses distance weighting to process the weights of attribute features that affect formation pressure calculation in adjacent wells, thereby improving the accuracy of formation pressure prediction and laying a foundation for predicting overflow events and leakage events in the formation during drilling. BRIEF DESCRIPTION OF THE DRAWINGS

[0102] Figure 1 The present invention is a flow chart of a method for predicting the risk of well spillage before drilling based on formation pressure prediction.

[0103] Figure 2 This is the radar map of the adjacent wells of well A.

[0104] Figure 3 Shown is a schematic diagram of the weight of well A.

[0105] Figure 4 Schematic diagram of the weight of well B.

[0106] Figure 5 This is the radar map of the adjacent wells of well B. DETAILED DESCRIPTION

[0107] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments:

[0108] Example 1

[0109] The present invention proposes a method for predicting the risk of leakage before drilling based on formation pressure prediction. Figure 1 As shown, the specific steps include:

[0110] Step 1, select the target formation, obtain the neighboring well data of the target well where the target formation is located, obtain the attribute characteristics of the formation pressure calculation influencing each neighboring well, and quantitatively characterize the attribute characteristics of the formation pressure calculation influencing each neighboring well to obtain the attribute characteristics of the formation pressure calculation influencing each neighboring well after numerical processing, which specifically includes the following steps:

[0111] Step 1.1, obtaining the attribute characteristics affecting formation pressure calculation;

[0112] Select the target formation, determine the position of the target formation, obtain the well data of the neighboring wells around the target well where the target formation is located, and obtain the attribute characteristics affecting the formation pressure calculation of each neighboring well based on the neighboring well data.

[0113] The formation pressure calculation influencing attribute characteristics include layer structure influencing factors and well location geological design influencing factors, wherein the layer structure influencing factors include structural similarity and lithological similarity, and the well location geological design influencing factors include layer thickness, well-to-well distance, layer bottom boundary distance, lithological type index and structural type index.

[0114] Step 1.2, processing the stratigraphic structural influencing factors of each adjacent well;

[0115] By standardizing the stratigraphic structural influence factors of each adjacent well, the stratigraphic structural influence factors of each adjacent well are quantitatively characterized, and the normalized values ​​of the stratigraphic structural influence factors of each adjacent well are obtained. After the standardization, the stratigraphic structural influence factors of each adjacent well are taken as [0,1].

[0116] The calculation formula for the standardization process in this embodiment is shown in formula (1):

[0117]

[0118] In the formula, max is the maximum value of the sample data to be processed; min is the minimum value of the sample data to be processed; y is the original value of the sample data; and x is the normalized value of the sample data.

[0119] Step 1.3, processing the influencing factors of well location geological design of each adjacent well;

[0120] Firstly, the strata corresponding to the target strata in each adjacent well are determined according to the horizon of the target strata, and based on the data of the adjacent wells, the lithology type index of the strata is divided according to the mineral composition and particle size, and the structural type index of the strata is divided according to the location of the geological structure where the strata is located and whether there is a fault interval, so as to determine the lithology type index and structural type index of the strata corresponding to the target strata in each adjacent well;

[0121] Secondly, based on the indicators of each lithology type and each structural type, combined with the layer thickness, distance between wells, and distance between layer bottom boundaries, the importance weight vector binary allocation scale value of each well location geological design influencing factor is calculated respectively, and a binary weight allocation judgment matrix is ​​constructed to quantitatively characterize the well location geological design influencing factors of each adjacent well;

[0122] Finally, the binary weight allocation judgment matrix is ​​solved by successively using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method, and the weight matrix of the well location geological design influencing factors obtained by the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method is obtained. The weight of the geological design influencing factors of each well location is obtained by calculating the average value.

[0123] In this embodiment, the binary distribution scale value of the importance weight vector is determined based on expert experience, and the specific determination process is as follows:

[0124] Randomly select two indicators from the factors affecting well location geological design, record them as indicator A and indicator B. When indicator A is as important as indicator B, the importance weight vector of indicator A is binary assigned scale value C. AB is 0.5, the importance weight vector of indicator B is assigned a binary scale value C BA is 0.5.

[0125] When indicator A is slightly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.55, and the importance weight vector of index j is assigned a binary scale value C BA is 0.45.

[0126] When indicator A is significantly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.6, and the importance weight vector of index j is assigned a binary scale value C BA is 0.4.

[0127] When indicator A is more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.7, and the importance weight vector of index j is assigned a binary scale value C BA is 0.3.

[0128] When indicator A is extremely important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.8, and the importance weight vector of index j is assigned a binary scale value C BA is 0.2.

[0129] According to the binary distribution scale value of the importance weight vector of each well location geological design influencing factor, a binary weight distribution judgment matrix C is constructed for each well location geological design influencing factor, as shown in formula (2):

[0130]

[0131] In the formula, C ij is the binary allocation scale value of the importance weight vector in the i-th row and j-th column of the binary weight allocation judgment matrix; i is the row number of the binary weight allocation judgment matrix, i=1, 2,…, n, j is the column number of the binary weight allocation judgment matrix, j=1, 2,…, n, n is the total number of rows of the binary weight allocation judgment matrix.

[0132] Since the importance weight vector of indicator A is binary assigned scale value C AB The importance weight vector of index B is assigned to the scale value C BA There is complementarity between them, and the importance weight vector of indicator A is binary assigned scale value C AB It is related to the weight of indicator A and the weight of indicator B, as shown in formula (3):

[0133]

[0134] Where W A is the weight of indicator A, W B is the weight of indicator B.

[0135] Therefore, the weights of the factors affecting well location geological design can be determined by solving the binary weight allocation judgment matrix.

[0136] In this embodiment, the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method are used in turn to solve the binary weight allocation judgment matrix, and the weight matrix of the well location geological design influencing factors obtained by using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method is obtained.

[0137] in,

[0138] When the reciprocal row sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is first reciprocated to construct the weight vector calculation matrix D. The weight vector calculation matrix D = (d ij ) n×n , where d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector,

[0139] According to the weight vector calculation matrix D, the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method is obtained. In the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method, the absolute weight T of the index A is A for:

[0140]

[0141] In the formula, R AIt is the intermediate calculation amount of indicator A.

[0142] When the reciprocal column sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is first reciprocated to construct the weight vector calculation matrix D. The weight vector calculation matrix D = (d ij ) n×n , where d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector,

[0143] According to the weight vector calculation matrix D, the weight matrix of the factors affecting the well location geological design is obtained. In the weight matrix of the factors affecting the well location geological design, the absolute weight of indicator B is T B for:

[0144]

[0145] In the formula, C B is the intermediate calculation amount of indicator B;

[0146] The absolute weight T of indicator B B Perform normalization processing to obtain the weight matrix of the factors affecting the well location geological design determined by the inverse series sum method;

[0147] The normalization calculation formula is:

[0148]

[0149] Where W B is the weight of indicator B.

[0150] When the row-column geometric mean comparison method is used to solve the binary weight allocation judgment matrix C, the geometric mean of each row and the geometric mean of each column in the binary weight allocation judgment matrix C are calculated respectively, as shown in formula (7) and formula (8):

[0151]

[0152]

[0153] Where Gi is the geometric mean of the i-th row in the binary weight allocation judgment matrix, and Gj is the geometric mean of the j-th column in the binary weight allocation judgment matrix;

[0154] Divide the geometric mean of the same row number and column number to get the absolute weight T of indicator A A for:

[0155]

[0156] The absolute weight T of indicator A AAfter normalization processing, the weight matrix of the influencing factors of well location geological design determined by the row-column geometric mean comparison method is obtained.

[0157] Step 2: According to the quantitatively characterized formation pressure calculation influencing attribute characteristics, based on the water injection principle, weight analysis is performed on the attribute characteristics influencing the formation pressure calculation to obtain the weights of the attribute characteristics influencing the formation pressure calculation.

[0158] Since the conventional KNN algorithm is used to calculate similarity, each attribute is considered equally important, ignoring the inclination towards important attributes, which deviates from the expert thinking. However, there are many attributes affecting formation pressure calculation that need to be considered in formation pressure prediction, and in the decision-making process of the many attributes affecting formation pressure calculation, the relationship between the attributes affecting formation pressure calculation is fuzzy. Only the recursive feature elimination method is used to eliminate redundant attributes, which cannot achieve the purpose of assigning weights to important attributes, resulting in the inability to truly restore the thinking process of experts when selecting. Therefore, how to reasonably simplify and weight the summarized influencing factors in order to better achieve the purpose of learning the comprehensive experience of experts and imitating the comprehensive behavior of experts is a key issue that needs to be considered in solving the calculation of formation pressure based on adjacent well information.

[0159] In view of the problem that the existing KNN algorithm can only regard the effects of attributes affecting formation pressure calculation as equally important when calculating similarity, the present invention proposes a method using the iterative water injection principle to obtain the optimal attribute subset with weights, effectively reduce the interference of redundant attributes to case classification, and retain the effective attributes to the maximum extent.

[0160] In this embodiment, the weights of the attribute characteristics affecting formation pressure calculation are first allocated based on the water injection principle, as shown in formula (10):

[0161]

[0162] In the formula, C W is the total capacity of the attribute characteristics affecting formation pressure calculation, which is used to describe the amount of information contained in the attribute characteristics affecting formation pressure calculation; m is the dimension of the attribute characteristics affecting formation pressure calculation, μ k Calculate the mean of the influencing attribute characteristics for the k-th formation pressure, ω k Calculate the weight of the attribute characteristics for the k-th formation pressure, σ k is the standard deviation of the attribute feature affecting the k-th formation pressure calculation, and k is the number of the attribute feature affecting the formation pressure calculation.

[0163] Then, according to formula (10), the Lagrangian function is constructed and the partial derivative of the weight affecting the attribute characteristics of the formation pressure is calculated, as shown in formula (11):

[0164]

[0165] Where λ is the Lagrangian operator,

[0166] Finally, based on the fact that the sum of the weights of the attribute features affecting the calculation of the formation pressure is 1, combined with the threshold ε, formula (11) is constrained, and the weights of the attribute features affecting the calculation of the formation pressure are calculated as:

[0167]

[0168] In the formula, ε is the threshold value,

[0169] Step 3: Analyze the similarity of the attribute feature samples that affect the formation pressure calculation of adjacent wells based on the KNN algorithm, screen the attribute feature samples that affect the formation pressure calculation of adjacent wells, and obtain the screened attribute feature samples that affect the formation pressure calculation of adjacent wells.

[0170] In order to solve the problem that the distortion of characteristic samples of some attributes affecting formation pressure calculation in multiple adjacent well data can easily lead to misjudgment, and at the same time, in order to improve the calculation efficiency of the KNN algorithm, it is necessary to perform similarity analysis on the attributes affecting formation pressure calculation in adjacent wells, and eliminate some characteristic samples of attributes affecting formation pressure calculation in adjacent wells, thereby weakening the dependence of the KNN algorithm on the number of samples to a certain extent and enhancing the robustness of the model.

[0171] The complex KNN distance similarity calculation model established in this embodiment is shown in formula (13):

[0172]

[0173] In the formula, D(x,y) is the distance similarity between sample y and sample x, and x k Calculate the characteristic value affecting the attribute characteristics for the k-th formation pressure in sample x, y k Calculate the eigenvalues ​​affecting the attribute characteristics for the k-th formation pressure in sample y.

[0174] The complex KNN distance similarity calculation model is used to calculate the distance similarity between two samples, and the calculated distance similarity result is compared with the preset accuracy value e:

[0175] If the distance similarity D(x,y) between sample y and sample x is greater than the preset precision value e, it is determined that sample y is not similar to sample x and sample y is removed;

[0176] If the distance similarity D(x, y) between sample y and sample x does not exceed the preset precision value e, it is determined that sample y is similar to sample x and sample y is retained.

[0177] Step 4: Based on the screened adjacent well formation pressure calculation influencing attribute feature samples, a distance-weighted analysis is performed on the adjacent well formation pressure calculation influencing attribute feature samples of the target well based on the distance-weighted method to determine the weights of the attribute features influencing formation pressure calculation in the target well relative to the adjacent wells.

[0178] Since the conventional KNN classification method is too dependent on the number of samples, and can only simply compare the labels of the target to be decided and the K neighboring well samples, it is impossible to calculate its specific value. Therefore, the distance weighting method is adopted in this embodiment, which can not only strengthen the weight of reliable samples and reduce the influence of unreliable samples, but also redefine its discriminant function to avoid the occurrence of the situation where the neighboring sample points of different categories are equal and cannot be judged, which weakens the dependence of the model on the number of samples to a certain extent and improves the accuracy and robustness of the model.

[0179] In this embodiment, the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well are calculated respectively, and weighted processing is performed based on the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well, so as to determine the weight of the attribute characteristics influencing formation pressure calculation of the pth target well relative to the adjacent well q, as shown in formula (14):

[0180]

[0181] In the formula, ω pq Calculate the weight of the influencing attribute feature for the pth formation pressure in the target well relative to the adjacent well q, D pq is the Euclidean distance between the attribute characteristics affecting formation pressure calculation of the target well and the attribute characteristics affecting formation pressure calculation of the adjacent well, N pq The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent wells.

[0182] Among them, the Euclidean distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well is as shown in formula (15):

[0183]

[0184] In the formula, X p Calculate the influencing attribute characteristic matrix for the formation pressure of the target well, X q Calculate the attribute characteristic matrix affecting the formation pressure of adjacent wells; x qr Calculate the influencing attribute characteristic value for the rth formation pressure in the adjacent well, x pr is the attribute characteristic value affecting the calculation of the rth formation pressure in the target well, and r is the sequence number of the attribute characteristic affecting the calculation of the formation pressure.

[0185] The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well is shown in formula (16):

[0186]

[0187] In the formula, x p Calculate the observation vector affecting the attribute characteristics for the p-th formation pressure in the target well, x p =(x p1 ,x p2 ,…,x pm ) T ;μ x The arithmetic mean vector of the sample values ​​of the attribute characteristics that affect the calculation of the formation pressure of the adjacent wells, μ x =(μ x1 ,μ x2 ,…,μ xm ) T ; ∑ is the covariance matrix of the attribute feature samples affecting the formation pressure calculation of the target well,

[0188] Step 5, establish a target formation pressure discrimination model, combine the weights of the attribute characteristics affecting the formation pressure calculation in the target well relative to the adjacent wells, use the target formation pressure discrimination model to predict the formation pressure G of the target formation, and use the similarity of the attribute characteristic samples affecting the formation pressure calculation of each adjacent well after screening by the target formation pressure discrimination model to draw a radar chart, and use the ratio between the shaded area and the polygon area in the radar chart to determine the reliability Q of the formation pressure prediction.

[0189] The target formation pressure discrimination model in this embodiment is:

[0190]

[0191] Where G is the predicted value of formation pressure, ω δ is the screening result of the attribute feature samples that affect the calculation of the formation pressure of adjacent wells, and I is the attribute feature sample matrix that affects the calculation of the formation pressure of adjacent wells.

[0192] Step 6, based on the formation pressure, the possibility of drilling risk is predicted before drilling, and the formation pressure prediction value G and prediction reliability Q of the target formation are used to determine the probability of overflow events or leakage events occurring in each well section in the drilling design of the target well.

[0193] If G-ρ h -Δρ>0, it is predicted that the target formation will have an overflow event, and the probability of an overflow event is R1, as shown in formula (18):

[0194]

[0195] If h +Δρ-G>C′, then the target formation is predicted to have a leakage event, and the probability of a leakage event is R2, as shown in formula (19):

[0196]

[0197] In the formula, ρ h is the design density of the static column of drilling fluid, Δρ is the additional pressure equivalent density, and C′ is the formation fracture strength equivalent density.

[0198] Example 2

[0199] This embodiment takes Well A as an example, and adopts the method for predicting the risk of leakage before drilling based on formation pressure prediction in Embodiment 1 to predict the formation pressure of a specified formation of Well A, specifically including the following steps:

[0200] Step 1, the neighboring wells of well A include well A1, well A2, well A3, well A4, and well A5. The neighboring well data of well A are obtained, and the formation pressure calculation influencing attribute characteristics of each neighboring well are obtained according to the well data, as shown in Table 1. The formation pressure calculation influencing attribute characteristics include layer structure influencing factors and well location geological design influencing factors, wherein the layer structure influencing factors include structural similarity and lithology similarity, and the well location geological design influencing factors include layer thickness, well-to-well distance, layer bottom boundary distance, lithology type index and structural type index.

[0201] Table 1 Summary of attribute characteristics affecting formation pressure calculation

[0202] hashtag Bottom depth (m) Layer thickness (m) Lithology structure Target distance (m) Formation pressure coefficient A1 4941 149 Limestone, white cloud Category I 250×0.8 1.14 A2 5413.5 216.5 Dolomites Category I 2.5×250 1.12 A3 4704.5 103.5 Limestone, dolomite Category II 500 1.17 A4 5101 188 Limestone, dolomite Category I 3.5×250 1.18 A5 5232 197 Limestone, dolomite Category I 5.5×250 1.19 A 4856 98 Limestone, dolomite Category I 0 ——

[0203] The stratigraphic structural influencing factors of each adjacent well are standardized and quantitatively characterized.

[0204] Then, the influencing factors of the well location geological design of each adjacent well are processed. In this embodiment, sandstone, conglomerate, gray dolomite, limestone, and mud shale are recorded as index 1, index 2, index 3, index 4, and index 5 in turn, and the binary weight distribution judgment matrix of lithology is obtained as follows:

[0205]

[0206] The binary weight distribution judgment matrix of lithology is solved by using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method respectively.

[0207] Among them, the lithology type index weight matrix obtained by the reciprocal row sum method is:

[0208] (0.099448 0.165552 0.316114 0.241800 0.165552)T

[0209] The lithology type index weight matrix obtained by the inverse column sum method is:

[0210] (0.149965 0.183134 0.276576 0.218125 0.172199) T

[0211] The lithology type index weight matrix obtained by using the row-column geometric mean comparison method is:

[0212] (0.105298 0.170097 0.327892 0.238720 0.157993) T

[0213] It can be seen from the calculation results that the weight vectors under the above three solutions are similar, which also confirms that the consistency level of the binary weight allocation judgment matrix C of lithology is good.

[0214] Finally, the weight of each lithology type index is obtained by calculating the average value in this embodiment:

[0215] (0.118237 0.172928 0.306861 0.232882 0.165248) T

[0216] The classification of lithology evaluation indicators in this embodiment is shown in Table 2.

[0217] Table 2 Classification of lithology evaluation indicators

[0218] Lithology classification sandstone Conglomerate Gray dolomite Limestone Shale Weight 0.118237 0.172928 0.306861 0.232882 0.165248

[0219] The structural types in the area where Well A is located are anticline center, anticline edge, and fault barrier position. The structural types are divided according to the structural position and fault barrier, as shown in Table 3.

[0220] Table 3 Structural classification table

[0221] Tectonic location Whether there is a fault Construct similarity classification Center of anticline none 1 Center of anticline have 2 Anticline edge none 3 Anticline edge have 4

[0222] In this embodiment, the binary weight distribution type judgment matrix for constructing similarity is:

[0223]

[0224] The constructed binary weight allocation judgment matrix is ​​solved by using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method respectively;

[0225] Among them, the structural type index weight matrix obtained by the reciprocal row sum method is:

[0226] (0.477273 0.315789 0.206897) T

[0227] The structural type index weight matrix obtained by the reciprocal column sum method is:

[0228] (0.410224 0.322943 0.266833) T

[0229] The structural type index weight matrix obtained by using the row and column geometric mean comparison method is:

[0230] (0.477913 0.314769 0.207318) T

[0231] From the calculation results, it can be seen that the weight vectors under the above three solutions are similar, which also confirms that the consistency level of the constructed binary weight allocation judgment matrix C is good.

[0232] Finally, this embodiment obtains the weight of each structural type index by calculating the average value:

[0233] (0.455137 0.317834 0.227016) T

[0234] That is, the results of the quantitative evaluation of the structural classification of the area where Well A is located in this embodiment are shown in Table 4.

[0235] Table 4 Classification of quantitative evaluation of construct similarity variables

[0236] Classification Center of anticline Anticline edge Partition layer location Weight 0.455137 0.317834 0.227016

[0237] In summary, the results of quantitative characterization of the attribute characteristics affecting the formation pressure calculation of each adjacent well in this embodiment are obtained, as shown in Table 5.

[0238] Table 5 Quantitative characterization results of adjacent well information

[0239] hashtag Bottom depth (m) Layer thickness (m) Lithology structure Target distance (m) A1 0.3336 0.4304 0.1183 0.4551 0.1455 A2 1 1 0.7231 0.3421 0.4545 A3 0 0.0464 0.0765 0.0784 0.3636 A4 0.5592 0.7595 1 0.2432 0.6364 A5 0.7440 0.8354 0.0876 1 1

[0240] Step 2: Based on the water injection principle, weight analysis is performed on the attribute characteristics that affect the calculation of formation pressure, and the weights of the attribute characteristics that affect the calculation of formation pressure are obtained. The weight distribution result of the attribute characteristics that affect the calculation of formation pressure in this embodiment is calculated using formula (10): (0.1176, 6.83, 37.32, 37.32, 6.77).

[0241] Step 3: Analyze the similarity of the attribute feature samples that affect the formation pressure calculation of adjacent wells based on the KNN algorithm, screen the attribute feature samples that affect the formation pressure calculation of adjacent wells, and obtain the screened attribute feature samples that affect the formation pressure calculation of adjacent wells.

[0242] In this embodiment, formula (13) is used to calculate the similarity of the attributes affecting the formation pressure calculation of each adjacent well in Table 1, and the similarity is (0.83, 0.27, 0.42, 0.75, 0.67). After screening, there are no bad characteristic samples of the attributes affecting the formation pressure calculation.

[0243] Step 4, based on the selected adjacent well formation pressure calculation influencing attribute feature samples, respectively calculate the Euclidean distance and Mahalanobis distance between the target well formation pressure calculation influencing attribute feature and the adjacent well formation pressure calculation influencing attribute feature, perform distance weighted analysis on the adjacent well formation pressure calculation influencing attribute feature samples of the target well, and determine the weight of each formation pressure calculation influencing attribute feature in the target well relative to each adjacent well. In this embodiment, the weights of each formation pressure calculation influencing attribute feature in the target well A relative to each adjacent well are (73%, 2%, 3%, 14%, 8%).

[0244] Step 5, establish a target formation pressure discrimination model, combine the weights of the attribute characteristics affecting formation pressure calculation in the target well relative to the weights of the adjacent wells, and use the target formation pressure discrimination model to predict the formation pressure G of the target formation. In this embodiment, the predicted formation pressure of the target formation is 1.15. At the same time, the radar chart is drawn based on the similarity of the attribute characteristic samples affecting formation pressure calculation in the adjacent wells after screening by the target formation pressure discrimination model, as shown in Figure 2 As shown, the reliability Q of the formation pressure prediction is determined by using the ratio between the shaded area and the polygon area in the radar map.

[0245] The influence attribute weight calculation is calculated based on the formation pressure of the adjacent wells of well A, such as Figure 3 As shown in the figure, the weights of the attributes affecting the formation pressure calculation of well A's adjacent wells are lithology similarity, structural continuity, vertical depth, ground spacing, and layer thickness from high to low. Among them, adjacent well A1 is most similar to target well A, accounting for most of the coefficient weights and should be the focus of reference, which is consistent with the actual situation.

[0246] The overall similarity of samples from adjacent wells of Well A is 0.4076, and the similarity of samples from adjacent wells is not much different, indicating that the adjacent well samples taken are basically reliable. The predicted formation pressure of Well A is 1.15, which is consistent with the actual formation pressure of the remaining sections of the well, and is better than the 1.12-1.19 value designed for drilling engineering. This indicates that no overflow or leakage events will occur during the drilling of this section of the formation. The actual drilling history report also proves that no underground complex conditions occur in this section of the well, thereby verifying the accuracy of the pre-drilling drilling risk prediction based on formation pressure prediction of the present invention.

[0247] Example 3

[0248] This embodiment takes the formation pressure prediction of Well B as an example, adopts the pre-drilling drilling spill risk prediction method based on formation pressure prediction proposed in Embodiment 1, and compares and verifies it with the actual design.

[0249] In this embodiment, the formation pressure calculation influencing attributes of each adjacent well are obtained based on the adjacent well data of well B, as shown in Table 6.

[0250] Table 6 Summary of attribute characteristics affecting formation pressure calculation

[0251]

[0252] The weights of the influencing attributes are calculated by calculating the formation pressure of the adjacent wells of well B, such as Figure 4 As shown, ( Figure 4 Schematic diagram of the weights of well B) shows that the weights of the attributes affecting formation pressure calculation of its adjacent wells are vertical depth, ground spacing, lithology similarity / structural continuity, and layer thickness, which are completely different from the weights of the characteristic samples of the attributes affecting formation pressure calculation of the adjacent wells of well A. This shows that the weights of the attributes affecting formation pressure association change dynamically with the characteristic sample data of the attributes affecting formation pressure calculation, which further verifies the accuracy of the calculation of the weights of the attributes affecting formation pressure calculation by the method of the present invention.

[0253] The target formation pressure discrimination model is used to predict the formation pressure in this embodiment, and the predicted value is 1.047. At the same time, the radar chart is drawn by using the similarity of the attribute feature samples affected by the formation pressure calculation of each adjacent well after screening using the target formation pressure discrimination model, as shown in FIG. Figure 5 As shown in the figure, the overall similarity of the attribute characteristic samples affecting the formation pressure calculation of well B's adjacent wells is 0.1125, and the similarity of the adjacent well samples is quite different. Wells B1 and B3 have the largest similarity and can be used as reference. The reliability of the formation pressure prediction value of this well is relatively low, and a conservative operation method should be adopted.

[0254] Since the predicted value of the formation pressure in this embodiment is 1.047, which differs by 0.01 from the actual drilling formation pressure of the Carboniferous formation in well B, it is verified that the method of the present invention can accurately predict the formation pressure using the data of adjacent wells.

[0255] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the terms used herein are only for describing specific embodiments, and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0256] In the present invention, the directions or positional relationships indicated by terms such as "upper", "lower", "bottom", "top", etc. are based on the directions or positional relationships shown in the accompanying drawings. They are relational words determined only for the convenience of describing the structural relationships of the various parts or elements of the present invention. They do not specifically refer to any part or element in the present invention and cannot be understood as limitations on the present invention.

[0257] In the present invention, terms such as "connected" and "connection" should be understood in a broad sense, indicating that the connection can be fixed, integral or detachable; it can be directly connected or indirectly connected through an intermediate medium. Relevant scientific research or technical personnel in this field can determine the specific meaning of the above terms in the present invention according to specific circumstances, and they should not be understood as limiting the present invention.

[0258] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A method for predicting the risk of leakage before drilling based on formation pressure prediction, characterized in that: The specific steps include: Step 1, select the target formation, obtain the neighboring well data of the target well where the target formation is located, obtain the attribute characteristics of each neighboring well that affect the formation pressure calculation, and quantitatively characterize the attribute characteristics of each neighboring well that affect the formation pressure calculation to obtain the attribute characteristics of the formation pressure calculation that are numerically processed; Step 2: According to the quantitatively characterized influencing attribute characteristics of formation pressure calculation, weight analysis is performed on the influencing attribute characteristics of each formation pressure calculation based on the water injection principle to obtain the weight of each formation pressure calculation influencing attribute characteristics; Step 3: Analyze the similarity of the attribute feature samples that affect the formation pressure calculation of adjacent wells based on the KNN algorithm, screen the attribute feature samples that affect the formation pressure calculation of adjacent wells, and obtain the screened attribute feature samples that affect the formation pressure calculation of adjacent wells; Step 4: Based on the screened adjacent well formation pressure calculation influencing attribute feature samples, a distance-weighted analysis is performed on the adjacent well formation pressure calculation influencing attribute feature samples of the target well based on a distance-weighted method to determine the weight of each formation pressure calculation influencing attribute feature in the target well relative to the adjacent wells; Step 5, establish a target formation pressure discrimination model, combine the weights of the attribute characteristics affecting formation pressure calculation in the target well relative to the weights of the adjacent wells, use the target formation pressure discrimination model to predict the formation pressure G of the target formation, and use the similarity of the attribute characteristic samples affecting formation pressure calculation in the adjacent wells after screening by the target formation pressure discrimination model to draw a radar chart, and determine the reliability Q of the formation pressure prediction; Step 6, based on the formation pressure, the possibility of drilling risk is predicted before drilling, and the formation pressure prediction value G and prediction reliability Q of the target formation are used to determine the probability of overflow events or leakage events occurring in each well section in the drilling design of the target well.

2. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 9, characterized in that: The step 1 specifically includes the following steps: Step 1.1, obtaining the attribute characteristics affecting formation pressure calculation; Select the target formation, determine the horizon of the target formation, obtain the well data of the neighboring wells around the target well where the target formation is located, and obtain the attribute characteristics affecting the formation pressure calculation of each neighboring well based on the neighboring well data; The formation pressure calculation influencing attribute characteristics include layer structure influencing factors and well location geological design influencing factors, wherein the layer structure influencing factors include structural similarity and lithological similarity, and the well location geological design influencing factors include layer thickness, well-to-well distance, layer bottom boundary distance, lithological type index and structural type index; Step 1.2, processing the stratigraphic structural influencing factors of each adjacent well; The stratigraphic structural influencing factors of each adjacent well are standardized, quantitatively characterized, and the normalized values ​​of the stratigraphic structural influencing factors of each adjacent well are obtained; Step 1.3, processing the influencing factors of well location geological design of each adjacent well; Firstly, the strata corresponding to the target strata in each adjacent well are determined according to the horizon of the target strata, and based on the data of the adjacent wells, the lithology type index of the strata is divided according to the mineral composition and particle size, and the structural type index of the strata is divided according to the location of the geological structure where the strata is located and whether there is a fault interval, so as to determine the lithology type index and structural type index of the strata corresponding to the target strata in each adjacent well; Secondly, based on the indicators of each lithology type and each structural type, combined with the layer thickness, distance between wells, and distance between layer bottom boundaries, the importance weight vector binary allocation scale value of each well location geological design influencing factor is calculated respectively, and a binary weight allocation judgment matrix is ​​constructed to quantitatively characterize the well location geological design influencing factors of each adjacent well; Finally, the binary weight allocation judgment matrix is ​​solved by successively using the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method, and the weight matrix of the well location geological design influencing factors obtained by the reciprocal row sum method, the reciprocal column sum method and the row-column geometric mean comparison method is obtained. The weight of the geological design influencing factors of each well location is obtained by calculating the average value.

3. The method for predicting the risk of leakage before drilling based on formation pressure prediction according to claim 2, characterized in that: In step 1.2, the calculation formula for standardization is as shown in formula (1): In the formula, max is the maximum value of the sample data to be processed; min is the minimum value of the sample data to be processed; y is the original value of the sample data; and x is the normalized value of the sample data.

4. The method for predicting the risk of leakage before drilling based on formation pressure prediction according to claim 2, characterized in that: The binary distribution scale value of the importance weight vector is determined based on expert experience, and the specific determination process is as follows: Randomly select index A and index B from the factors affecting well location geological design. When index A is as important as index B, the importance weight vector of index A is assigned a binary scale value C. AB is 0.5, the importance weight vector of indicator B is assigned a binary scale value C BA is 0.5; When indicator A is slightly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.55, and the importance weight vector of index j is assigned a binary scale value C BA is 0.45; When indicator A is significantly more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.6, and the importance weight vector of index j is assigned a binary scale value C BA is 0.4; When indicator A is more important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.7, and the importance weight vector of index j is assigned a binary scale value C BA is 0.3; When indicator A is extremely important than indicator B, the importance weight vector of indicator i is assigned a binary scale value C AB is 0.8, and the importance weight vector of index j is assigned a binary scale value C BA is 0.2; According to the binary distribution scale value of the importance weight vector of each well location geological design influencing factor, a binary weight distribution judgment matrix C is constructed for each well location geological design influencing factor, as shown in formula (2): In the formula, C ij Binary allocation scale value for the importance weight vector of row i and column j in the binary weight allocation judgment matrix; i is the row number of the binary weight allocation judgment matrix, i=1, 2, ..., n, j is the column number of the binary weight allocation judgment matrix, j=1, 2, ..., n, n is the total number of rows of the binary weight allocation judgment matrix; Since the importance weight vector of indicator A is binary assigned scale value C AB The importance weight vector of index B is assigned to the scale value C BA There is complementarity between them, and the importance weight vector of indicator A is binary assigned scale value C AB It is related to the weight of indicator A and the weight of indicator B, as shown in formula (3): Where W A is the weight of indicator A, W B is the weight of indicator B; Therefore, the weights of the factors affecting well location geological design can be determined by solving the binary weight allocation judgment matrix.

5. The method for predicting the risk of leakage before drilling based on formation pressure prediction according to claim 4, characterized in that: When the reciprocal row sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is first reciprocated to construct the weight vector calculation matrix D. The weight vector calculation matrix D = (d ij ) n×n , where d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector, According to the weight vector calculation matrix D, the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method is obtained. In the weight matrix of the influencing factors of the well location geological design determined by the inverse row sum method, the absolute weight T of the index A is A for: In the formula, R A It is the intermediate calculation amount of indicator A.

6. The method for predicting the risk of leakage before drilling based on formation pressure prediction according to claim 4, characterized in that: When the reciprocal column sum method is used to solve the binary weight allocation judgment matrix C, the binary weight allocation judgment matrix C is first reciprocated to construct the weight vector calculation matrix D. The weight vector calculation matrix D = (d ij ) n×n , where d ij Calculate the index value of the i-th row and j-th column in the matrix D for the weight vector, According to the weight vector calculation matrix D, the weight matrix of the factors affecting the well location geological design is obtained. In the weight matrix of the factors affecting the well location geological design, the absolute weight of indicator B is T B for: In the formula, CB is the intermediate calculation amount of indicator B; The absolute weight T of indicator B B By performing normalization processing, a weight matrix of factors affecting well location geological design determined by the inverse series sum method is obtained; The normalization calculation formula is: Where W B is the weight of indicator B.

7. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 4, characterized in that: When the row-column geometric mean comparison method is used to solve the binary weight allocation judgment matrix C, the geometric mean of each row and the geometric mean of each column in the binary weight allocation judgment matrix C are calculated respectively, as shown in formula (7) and formula (8): Where Gi is the geometric mean of the i-th row in the binary weight allocation judgment matrix, and Gj is the geometric mean of the j-th column in the binary weight allocation judgment matrix; Divide the geometric mean of the same row number and column number to get the absolute weight T of indicator A A for: The absolute weight T of indicator A A After normalization processing, the weight matrix of the influencing factors of well location geological design determined by the row-column geometric mean comparison method is obtained.

8. The method for predicting the risk of leakage before drilling based on formation pressure prediction according to claim 4, characterized in that: In step 2, weights are allocated to the attribute characteristics affecting formation pressure calculation based on the water injection principle, as shown in formula (10): In the formula, C W is the total capacity of the attribute characteristics affecting formation pressure calculation, which is used to describe the amount of information contained in the attribute characteristics affecting formation pressure calculation; m is the dimension of the attribute characteristics affecting formation pressure calculation, μ k Calculate the mean of the influencing attribute characteristics for the k-th formation pressure, ω k Calculate the weight of the attribute characteristics for the k-th formation pressure, σ k is the standard deviation of the attribute feature affecting the k-th formation pressure calculation, k is the number of the attribute feature affecting the formation pressure calculation; According to formula (10), the Lagrangian function is constructed and the partial derivative of the weight affecting the attribute characteristics of the formation pressure is calculated, as shown in formula (11): Where λ is the Lagrangian operator, Based on the fact that the sum of the weights of the attribute features affecting the calculation of the formation pressure is 1, combined with the threshold ε, the formula (11) is constrained, and the weights of the attribute features affecting the calculation of the formation pressure are calculated as: In the formula, ε is the threshold value, 9. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 8, characterized in that: In step 3, a complex KNN distance similarity calculation model is established, as shown in formula (13): In the formula, D(x,y) is the distance similarity between sample y and sample x, and x k Calculate the characteristic value affecting the attribute characteristics for the k-th formation pressure in sample x, y k Calculate the characteristic value affecting the attribute characteristics for the k-th formation pressure in sample y; The distance similarity between two samples is calculated using the complex KNN distance similarity calculation model, and the calculated distance similarity result is compared with the preset accuracy value e; If the distance similarity D(x,y) between sample y and sample x is greater than the preset precision value e, it is determined that sample y is not similar to sample x and sample y is removed; If the distance similarity D(x, y) between sample y and sample x does not exceed the preset precision value e, it is determined that sample y is similar to sample x and sample y is retained.

10. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 9, characterized in that: The precision value e ranges from 1 to 1.

4.

11. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 9, characterized in that: In step 4, the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well are calculated respectively, and weighted processing is performed based on the Euclidean distance and the Mahalanobis distance between the attribute characteristics influencing formation pressure calculation of the target well and the attribute characteristics influencing formation pressure calculation of the adjacent well, so as to determine the weight of the attribute characteristics influencing formation pressure calculation of the pth target well relative to the adjacent well q, as shown in formula (14): In the formula, ω pq Calculate the weight of the influencing attribute feature for the pth formation pressure in the target well relative to the adjacent well q, D pq is the Euclidean distance between the attribute characteristics affecting formation pressure calculation of the target well and the attribute characteristics affecting formation pressure calculation of the adjacent well, N pq The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent wells.

12. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 11, characterized in that: The Euclidean distance between the attribute characteristics that affect the formation pressure calculation of the target well and the attribute characteristics that affect the formation pressure calculation of the adjacent well is shown in formula (15): In the formula, D(X p ,X q ) is the Euclidean distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well, X p Calculate the influencing attribute characteristic matrix for the formation pressure of the target well, X q Calculate the attribute characteristic matrix affecting the formation pressure of adjacent wells; x qr Calculate the influencing attribute characteristic value for the rth formation pressure in the adjacent well, x pr is the r-th formation pressure calculation influencing attribute characteristic value in the target well, where r is the sequence number of the formation pressure calculation influencing attribute characteristic; The Mahalanobis distance between the attribute characteristics affecting the formation pressure calculation of the target well and the attribute characteristics affecting the formation pressure calculation of the adjacent well is shown in formula (16): Where N p is the Mahalanobis distance between the p-th formation pressure calculation influencing attribute feature of the target well and the p-th formation pressure calculation influencing attribute feature of the adjacent well, x p Calculate the observation vector affecting the attribute characteristics for the p-th formation pressure in the target well, x p =(x p1 ,x p2 ,…,x pm ) T ;μ x The arithmetic mean vector of the sample values ​​of the attribute characteristics that affect the calculation of the formation pressure of the adjacent wells, μ x =(μ x1 ,μ x2 ,…,μ xm ) T ; ∑ is the covariance matrix of the attribute feature samples affecting the formation pressure calculation of the target well, 13. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 11, characterized in that: In step 5, the target formation pressure discrimination model is: Where G is the predicted value of formation pressure, ω δ is the screening result of the attribute feature samples that affect the calculation of the formation pressure of adjacent wells, and I is the attribute feature sample matrix that affects the calculation of the formation pressure of adjacent wells.

14. The method for predicting leakage risk of drilling before drilling based on formation pressure prediction according to claim 13, characterized in that: In step 6, based on the formation pressure prediction value G and prediction reliability Q of the target formation, the drilling risk type and occurrence probability are predicted before drilling; If G-ρ h -Δρ>0, it is predicted that the target formation will have an overflow event, and the probability of an overflow event is R1, as shown in formula (18): If h +Δρ-G>C′, then the target formation is predicted to have a leakage event, and the probability of a leakage event is R2, as shown in formula (19): In the formula, ρ h is the design density of the static column of drilling fluid, Δρ is the additional pressure equivalent density, and C′ is the formation fracture strength equivalent density.

Images