A method and system for integrated monitoring of formation pressure while drilling

Abstract

The present invention discloses an integrated monitoring method and system for formation pressure measurement and recording while drilling. The monitoring method includes the following steps: selecting and importing logging data; removing abnormal data from the logging data, standardizing, and making logging data samples; selecting n logging parameters, performing quality evaluation and data standardization on the logging parameters, and eliminating the influence of dimensions; starting from the n logging parameters, performing dimensionality reduction processing on the logging parameters through principal component calculation to obtain m logging principal component parameters; establishing an RNN neural network learning model in combination with the m logging principal component parameters; putting the logging data samples into training and correcting the learning model. The present invention combines logging with logging. Through logging technology, it can quantitatively measure parameters while drilling into the well. Combined with logging technology, it can make a qualitative evaluation of the formation pressure downhole, and has the characteristics of real-time monitoring and intuitive evaluation.

Description

Technical Field

[0001] The present invention relates to the technical field of drilling and logging exploration and development of oil, natural gas, etc., and in particular to a method and system for integrated monitoring of formation pressure while drilling. Background Art

[0002] In recent years, with the continuous increase in drilling and logging projects, logging while drilling technology has made great progress. However, high-pressure gas well drilling has always been a major problem that needs to be overcome in the exploration and development of natural gas, oil, etc. in various countries and regions. Due to the lack of analysis of drilling formation pressure parameters, it often leads to a lack of prediction and response measures for the possibility of unknown accidents such as blowouts in the actual operation of drilling operations. This has resulted in the current drilling and logging industry being unable to predict and quickly deal with sudden blowout accidents. This has resulted in a blowout accident once it occurs, which will bring great danger to the safety of construction personnel and will also cause certain environmental pollution.

[0003] Usually, the method of pressure monitoring while drilling is usually used to monitor formation pressure in drilling projects. The current methods of pressure monitoring while drilling mainly include standard drilling speed method, sigma method, DC index method, rock strength method, etc. These methods involve many intermediate parameters and complex calculation process, which makes it difficult to use them on engineering sites.

[0004] At present, there are relatively few patents on the pressure monitoring system while drilling. The patent document with the publication number CN101025084 discloses a method for predicting the pore pressure of the bottom layer under the drill bit while drilling. The method takes the wells to be drilled and the wells already drilled in the same block as the processing objects, and uses the formation density logging data, the acoustic time difference logging data and the natural gamma logging data to analyze and establish a formation pressure calculation model to predict the pore pressure of the current formation encountered. However, these are all static and single analyses, lacking integrated and dynamic logging pressure monitoring while drilling. The patent document with the publication number CN1966934 discloses a method for predicting the collapse pressure and fracture pressure of the formation under the drill bit while drilling. It only has the joint prediction of seismic and logging, does not fully apply the comprehensive logging information available while drilling, and lacks timeliness and authenticity in monitoring the formation pressure. Summary of the invention

[0005] The technical problem to be solved by the present invention is that the existing while-drilling pressure monitoring technology lacks timeliness and comprehensiveness in monitoring formation pressure. The present invention provides a while-drilling formation pressure measurement and recording integrated monitoring method and system to solve the above problems. The method combines well logging with mud logging. Through the well logging technology, it is possible to quantitatively measure parameters while drilling into the well. Combined with the mud logging technology, it is possible to make a qualitative evaluation of the downhole formation pressure. It has the characteristics of real-time monitoring and intuitive evaluation.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for integrated monitoring of formation pressure while drilling, comprising the following steps:

[0008] S1. Select and import logging data;

[0009] S2. Remove abnormal data from the collected logging data, standardize them, and prepare logging data samples;

[0010] S3. Select n logging parameters, perform quality evaluation and data standardization on the logging parameters, and eliminate the influence of dimension; starting from the n logging parameters, perform dimension reduction processing on the logging parameters through principal component calculation to obtain m logging principal component parameters;

[0011] Specifically, starting from n logging parameters, their frequencies are analyzed separately, and the relationships between the above parameters are comprehensively analyzed. The logging parameters are reduced in dimension through principal component calculation to obtain m logging principal component parameters.

[0012] S4. Establishing a RNN neural network learning model by combining m logging principal component parameters;

[0013] S5. Put the logging data samples obtained in step S2 into training to correct the learning model.

[0014] Further optionally, the logging data includes logging engineering parameters and logging formation pressure basic data.

[0015] Further optionally, in step S2, abnormal data is removed from the collected logging data, and the 3σ rule is used to determine whether the data is abnormal.

[0016] Optionally, the abnormality of logging parameters is determined by the following formula:

[0017]

[0018] In formula (1), i ranges from 1 to n, n represents the number of logging data samples, and X i Represents the i-th logging data.

[0019] Further optionally, in step S2, the well logging data after the abnormal data is removed is standardized by a z-score standardization method, which is suitable for the case where the maximum and minimum values ​​of the parameters are unknown.

[0020] Alternatively, the i logging data are imported into an Excel spreadsheet to calculate the arithmetic mean and standard deviation S of each logging parameter. ij , the conversion function is as follows:

[0021]

[0022]

[0023] In formula (2) and formula (3), X represents the conversion function, i represents the logging parameter number, ranging from 1 to n; j represents the logging parameter index number, ranging from 1 to n; X ij represents the jth index value of the i-th logging parameter, Z ij represents the jth index value of the i-th logging parameter after standardization, S ij Represents the standard deviation of the jth index value of the ith logging parameter. Represents the arithmetic mean of the jth logging parameter index.

[0024] The main component calculation of logging parameters is used for dimensionality reduction. Mud logging can record more than 30 drilling parameters in real time through sensors, and more than 300 logging engineering parameters can be derived through calculation. The dimensionality reduction process here is of great significance. The process includes selecting i logging engineering parameters and performing main component calculation.

[0025] Further optionally, in step S3, for the standardized parameter value Z ij , the correlation coefficient calculation formula is:

[0026]

[0027] From this, the correlation coefficient matrix can be obtained as R;

[0028] The characteristic roots are obtained as λ1,λ2,…λ n ; The eigenvectors are α1, α2, …, α n ;

[0029] Then the nth principal component is: F n =α1*S X1 +α2*S X2 +…+α n *S Xn ;

[0030] The proportion of the principal component is: F = (α1*S X1 +α2*S X2 +…+α n *S Xn ) / (α1+α2+…+α n )

[0031] where α1, α2, ..., αn are the eigenvectors corresponding to the eigenvalues ​​of the covariance matrix ∑ of X, S x1 ,S x2 ,...,S xn The original variables are standardized, and the principal component proportions of each parameter are calculated and ranked to determine m logging principal component parameters;

[0032] Among them, k ranges from 1 to n, n represents the number of logging data samples, r ij represents the correlation coefficient of the jth index value of the i-th logging parameter, Z ik represents the kth index value of the i-th logging parameter after standardization, Z kj Represents the jth index value of the kth logging parameter after standardization.

[0033] Further optionally, in step S4, the m logging principal component parameters are combined with the RNN neural network learning method to establish a model as follows:

[0034] a (t) =b+Wh (t-1) +Ux (t)

[0035] h (t) =tanh(a (t) )

[0036] ο (t) =c+Vh (t)

[0037]

[0038] Among them, b and c are the bias vectors of the parameters, W, U, and V are weight matrices corresponding to the connections from input to hidden and hidden to output respectively;

[0039] Among them, x (t) =[M1 M2...M m ] T , where M1,M2,...,M m Represents the principal component logging engineering parameters.

[0040] Combine the m logging principal component parameters with the RNN neural network to build the model part. RNN is a recurrent neural network with a cyclic memory function, which can link adjacent data to obtain dynamic and standard prediction values. The following parameters are mainly set here. Set the training parameters, select the neural network as the learning method, and set the following: learning efficiency (0.1), maximum allowable error rate (0.000001), function alpha value (0.8), number of hidden layer neurons (15), and number of iterations (10000).

[0041] The logging data samples are put into the training part. If the formation pressure corresponding to the same well depth of the selected logging object and the logging object is similar, no special processing is required. If there is a large difference between the two, the well depth data of the logging object should be adjusted according to the pressure profile, and other logging engineering parameters should be adjusted accordingly to obtain more accurate analysis results.

[0042] An integrated monitoring system for formation pressure measurement and recording while drilling, comprising

[0043] The logging parameter processing module is used to import the logging data, remove abnormal data from the logging data, standardize the logging data, and make logging data samples;

[0044] The logging parameter processing module is used to evaluate the quality and standardize the data of n logging parameters to eliminate the influence of dimension; and is used to reduce the dimension of the logging parameters by principal component calculation to obtain m logging principal component parameters.

[0045] The bottom pressure monitoring module while drilling is used to establish the RNN neural network learning model by combining m logging principal component parameters, and put the logging data samples into training to correct the learning model.

[0046] Further optionally, it is used to implement the above-mentioned integrated monitoring method for formation pressure measurement and recording while drilling.

[0047] The present invention has the following advantages and beneficial effects:

[0048] The present invention provides an integrated monitoring method for formation pressure measurement and logging while drilling, which is mainly divided into five parts: standardization of logging data; establishment of logging parameter samples; selection of logging parameter processing objects (n logging parameters), and dimension reduction of logging parameters through principal component calculation to obtain m logging principal component parameters; establishment of a model by combining m logging principal component parameters with RNN neural network; and inputting logging data samples into the model for training and correction. The above-obtained downhole formation pressure monitoring results are analyzed, and the drilling and logging parameters are adjusted according to the real-time monitoring situation to ensure the safety of drilling and logging.

[0049] The principal component analysis of drilling and logging parameters of the present invention performs dimension reduction through statistical analysis among n multiple parameters, calculates and obtains m principal components of logging parameters, and simplifies the subsequent analysis process.

[0050] The present invention can achieve the monitoring result of downhole formation pressure by combining logging data with mud logging data during the drilling process, selecting object parameters for modeling and analysis through the RNN neural network model, and adjusting the drilling and logging parameters according to the real-time monitoring situation. It can meet the requirements of most downhole formation pressure real-time monitoring while drilling, and achieve the purpose of preventing blowouts and ensuring drilling safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0052] Figure 1 It is an overall flow chart of the integrated monitoring method for formation pressure measurement and recording while drilling of the present invention.

[0053] Figure 2 This is a flow chart of the integrated monitoring method for formation pressure measurement and recording while drilling of Example 2. DETAILED DESCRIPTION

[0054] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0055] In the following description, a large number of specific details are set forth in order to provide a thorough understanding of the present invention. However, it is apparent to one of ordinary skill in the art that these specific details are not necessarily employed to practice the present invention. In other embodiments, in order to avoid obscuring the present invention, well-known structures, circuits, materials, or methods are not specifically described.

[0056] Throughout the specification, references to "one embodiment," "an embodiment," "an example," or "an example" mean that a particular feature, structure, or characteristic described in conjunction with the embodiment or example is included in at least one embodiment of the present invention. Therefore, the phrases "one embodiment," "an embodiment," "an example," or "an example" appearing in various places throughout the specification do not necessarily all refer to the same embodiment or example. In addition, particular features, structures, or characteristics may be combined in one or more embodiments or examples in any suitable combination and / or sub-combination. In addition, it will be appreciated by those of ordinary skill in the art that the figures provided herein are for illustrative purposes and that the figures are not necessarily drawn to scale. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0057] Example 1

[0058] This embodiment provides an integrated monitoring method for formation pressure measurement and recording while drilling, such as Figure 1 and Figure 2 The specific steps are as follows:

[0059] S1. Select and import well logging data; the well logging data includes well logging engineering parameters and well logging formation pressure basic data.

[0060] S2. Remove abnormal data from the collected logging data, standardize them, and prepare logging data samples;

[0061] The collected logging data is removed from the abnormal data, and the 3σ rule is used to determine whether the data is abnormal. The abnormality of the logging parameters is determined by the following formula:

[0062]

[0063] In formula (1), i ranges from 1 to n, n represents the number of logging data samples, and Xi Represents the i-th logging data.

[0064] The logging data after removing abnormal data are standardized by the z-score standardization method; the i logging data are imported into the Excel table to calculate the arithmetic mean and standard deviation S of each logging parameter ij , the conversion function is as follows:

[0065]

[0066]

[0067] In formula (2) and formula (3), X represents the conversion function, i represents the logging parameter number, and the range is x j 1~n; j represents the logging parameter index number, ranging from 1 to n; X ij represents the jth index value of the i-th logging parameter, Z ij represents the jth index value of the i-th logging parameter after standardization, S ij Indicates the standard deviation of the jth index value of the ith logging parameter. Indicates the arithmetic mean of the jth logging parameter index.

[0068] S3. Select n logging parameters, perform quality evaluation and data standardization on the logging parameters, and eliminate the influence of dimension; starting from the n logging parameters, perform dimension reduction processing on the logging parameters through principal component calculation to obtain m logging principal component parameters;

[0069] For the standardized parameter value Z ij , the correlation coefficient calculation formula is:

[0070]

[0071] From this, the correlation coefficient matrix can be obtained as R;

[0072] The characteristic roots are obtained as λ1,λ2,…λ n ; The eigenvectors are α1, α2, …, α n ;

[0073] Then the nth principal component is: F n =α1*S X1 +α2*S X2 +…+α n *S Xn ;

[0074] The proportion of the principal component is: F = (α1*S X1 +α2*S X2 +…+α n *S Xn ) / (α1+α2+…+αn )

[0075] where α1, α2, ..., αn are the eigenvectors corresponding to the eigenvalues ​​of the covariance matrix ∑ of X, S x1 ,S x2 ,...,S xn The original variables are standardized, and the principal component proportions of each parameter are calculated and ranked to determine m logging principal component parameters;

[0076] Among them, k ranges from 1 to n, n represents the number of logging data samples, r ij represents the correlation coefficient of the jth index value of the i-th logging parameter, Z ik represents the kth index value of the i-th logging parameter after standardization, Z kj Represents the jth index value of the kth logging parameter after standardization.

[0077] S4. Establishing a RNN neural network learning model by combining m logging principal component parameters;

[0078] The model established by combining m logging principal component parameters with RNN neural network learning method is as follows:

[0079] a (t) =b+Wh (t-1) +Ux (t)

[0080] h (t) =tanh(a (t) )

[0081] ο (t) =c+Vh (t)

[0082]

[0083] Among them, b and c are the bias vectors of the parameters, W, U, and V are weight matrices corresponding to the connections from input to hidden and hidden to output respectively;

[0084] Among them, x (t) =[M1 M2 ... M m ] T , where M1,M2,...,M m Represents the principal component logging engineering parameters.

[0085] S5. Put the logging data samples obtained in step S2 into training to correct the learning model.

[0086] The logging data samples are put into the training part. If the formation pressure corresponding to the same well depth of the selected logging object and the logging object is similar, no special processing is required. If there is a large difference between the two, the well depth data of the logging object should be adjusted according to the pressure profile, and other logging engineering parameters should be adjusted accordingly to obtain more accurate analysis results.

[0087] Example 2

[0088] This embodiment provides an integrated monitoring method for formation pressure measurement and recording while drilling, such as Figure 2 As shown, based on the method steps of Example 1, the implementation steps combined with actual engineering applications are as follows:

[0089] Step 1: Select logging engineering parameters and formation pressure data, remove outliers and standardize data, and establish data samples.

[0090] Step 2: Select the logging object and set the logging related parameters. Specifically, select the wells and related curves to be processed, set the parameter selection mode to self-selected parameters, set the modeling method to estimate, and set the depth extraction range to the entire well section. Next, select the predicted well and known downhole engineering parameters. For example, select the Keshen 205 well, select the 11 engineering parameters such as hook load, well diameter, drilling time, speed, drilling pressure, drill torque, well depth, pool volume, drill speed, outlet flow, and pump stroke. After all settings are completed, click Next.

[0091] Step 3: The above 11 logging parameters are processed by principal component calculation for dimensionality reduction, and the quality of the above 11 drilling and logging parameters is evaluated, and the data is standardized to eliminate the influence of dimension. Starting from the parameters such as turntable speed, drilling time, and hook load, their frequencies are analyzed separately, and the relationship between the above parameters is comprehensively analyzed. After calculation, the main components are determined to be the six engineering parameters of hook load, drilling time, speed, drilling pressure, well depth, and pool volume.

[0092] Step 4: Combine the six engineering parameters of hook load, drilling time, rotation speed, drilling pressure, well depth, and pool volume with the RNN neural network learning method to establish a model. The formula is as follows:

[0093] a (t) =b+Wh (t-1) +Ux (t)

[0094] h (t) =tanh(a (t) )

[0095] ο (t) =c+Vh (t)

[0096]

[0097] Among them, x (t) =[DEPTH DT RS WOB PV HL] T , DEPTH is the well depth, DT is the drilling time, RS is the rotation speed, WOB is the bit pressure, PV is the pool volume, and HL is the hook load.

[0098] Since the well logging object and Keshen 205 well are adjacent wells in this embodiment and have the same pressure profile, no special treatment is made for the well depth. Then set the training parameters, select the neural network as the learning method, and set the following: learning efficiency (0.9), maximum allowable error rate (0.000001), function alpha value (0.8), number of hidden layer neurons (15), number of iterations (10000), click prediction, and the prediction result of Keshen 205 well is obtained.

[0099] Step 5: Analyze and predict the downhole formation pressure monitoring results, and adjust the drilling and logging parameters according to the real-time monitoring situation to ensure the safety of drilling and logging.

[0100] Example 3

[0101] This embodiment provides an integrated monitoring system for formation pressure measurement and recording while drilling, which mainly includes a logging parameter processing module, a logging parameter processing module and a bottom pressure monitoring module while drilling. The functions of each module are described as follows:

[0102] The logging parameter processing module is used to import logging data, remove abnormal data from the logging data, standardize the logging data, and produce logging data samples.

[0103] The logging parameter processing module is used to evaluate the quality and standardize the data of n logging parameters to eliminate the influence of dimension; and is used to reduce the dimension of the n logging parameters through principal component calculation to obtain m logging principal component parameters.

[0104] The bottom pressure monitoring module while drilling is used to establish the RNN neural network learning model by combining m logging principal component parameters, and put the logging data samples into training to correct the learning model.

[0105] It is specifically used to implement an integrated monitoring method for formation pressure measurement and recording while drilling provided in Example 1.

[0106] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for integrated monitoring of formation pressure while drilling, characterized in that: The following steps are involved: S1. Select and import logging data; S2. Remove abnormal data from the collected logging data, standardize it, and make logging data samples. Z ij Represents the standardized i The logging parameters j Index values, S3. Select n logging parameters, perform quality evaluation and data standardization on the logging parameters, and eliminate the influence of dimension; starting from the n logging parameters, perform dimension reduction processing on the logging parameters through principal component calculation to obtain m logging principal component parameters; S4. Establishing a RNN neural network learning model by combining m logging principal component parameters; S5. Put the logging data samples obtained in step S2 into training to correct the learning model; In step S3, for the standardized Z ij , the correlation coefficient calculation formula is: （4）； From this, the correlation coefficient matrix can be obtained as R; The characteristic roots are obtained as λ1,λ2,…λ n ; the eigenvectors are α1, α2,…, α n ; Then the nth principal component is: F n =α1*S X1 +α2*S X2 +…+α n *S Xn ; The proportion of the main component is: F=(α1*S X1 +α2*S X2 +…+α n *S Xn ) / ( α1+α2+…+α n ) where α1, α2, ..., αn are the eigenvectors corresponding to the eigenvalues ​​of the covariance matrix ∑ of X, S x1 ,S x2 ,...,S xn The original variables are standardized, and the principal component proportions of each parameter are calculated and ranked to determine m logging principal component parameters; Among them, the range of k is 1~n , n represents the sample size of logging data, r ij Indicates i Well logging parameters j The correlation coefficient of the indicator values ​​is Z ik Represents the standardized i The logging parameters k The index value, Z kj Represents the standardized k The logging parameters j Index value; In step S4, the m logging principal component parameters are combined with the RNN neural network learning method to establish a model as follows: ； ； ； ； Among them, b and c are the bias vectors of the parameters, W, U, and V are weight matrices corresponding to the connections from input to hidden and hidden to output respectively; in, , where M1,M2,...,M m Represents the principal component logging engineering parameters.

2. The integrated monitoring method for formation pressure measurement and recording while drilling according to claim 1, characterized in that: The logging data include logging engineering parameters and basic logging formation pressure data.

3. The integrated monitoring method for formation pressure measurement and recording while drilling according to claim 1, characterized in that: In step S2, abnormal data is removed from the collected logging data, and the 3σ rule is used to determine whether the data is abnormal.

4. The integrated monitoring method for formation pressure measurement and recording while drilling according to claim 3 is characterized in that: The abnormality of logging parameters is judged by the following formula: （1）； In formula (1), i Range is 1~n , n represents the sample size of logging data, Indicates i Well logging data.

5. The integrated monitoring method for formation pressure measurement and recording while drilling according to claim 1, characterized in that: In step S2, the well logging data after the abnormal data is removed is standardized by using the z-score standardization method.

6. The integrated monitoring method for formation pressure measurement and recording while drilling according to claim 5, characterized in that: Will i Import the logging data into the Excel table and calculate the arithmetic mean and standard deviation of each logging parameter S ij , the conversion function is as follows: （2）； （3）； In formula (2) and formula (3), X represents the conversion function, i Indicates the logging parameter number, ranging from 1~n;j Indicates the logging parameter index number, ranging from 1~n;X ij Indicates i The logging parameters j The index value, Z ij Represents the standardized i The logging parameters j The index value, S ij Indicates i The logging parameters j The standard deviation of the indicator values; Indicates j The arithmetic mean of the logging parameter indicators.

7. The monitoring system of the integrated monitoring method for formation pressure measurement and recording while drilling according to claim 1, characterized in that: include The logging parameter processing module is used to import the logging data, remove abnormal data from the logging data, standardize the logging data, and make logging data samples; The logging parameter processing module is used to evaluate the quality and standardize the data of n logging parameters to eliminate the influence of dimension; and is used to reduce the dimension of the logging parameters by principal component calculation to obtain m logging principal component parameters. The bottom pressure monitoring module while drilling is used to establish the RNN neural network learning model by combining m logging principal component parameters, and put the logging data samples into training to correct the learning model.

Images