Wellbore stability prediction model training, wellbore stability prediction method and device

Abstract

This invention discloses a wellbore stability prediction model training method and apparatus. The method includes: based on acquired pressure-controlled drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data from adjacent wells; determining the formation pressure input vector, bottom hole pressure input vector, and risk type input vector respectively; obtaining a formation pressure prediction model based on the formation pressure profile data and formation pressure input vector; obtaining a bottom hole pressure prediction model based on the bottom hole pressure input vector and bottom hole pressure data; obtaining an engineering parameter neural network based on the drilling risk type data and risk type input vector; and connecting the formation pressure prediction model, bottom hole pressure prediction model, and engineering parameter neural network in parallel to generate a wellbore stability prediction model. This method can accurately predict the types of risks present in the wellbore.

Description

Technical Field

[0001] This invention relates to the field of oil and gas well drilling technology, and in particular to a wellbore stability prediction model training, wellbore stability prediction method and device. Background Technology

[0002] Currently, the focus of oil and gas exploration and development is shifting towards deep and ultra-deep unconventional shale oil and gas formations. However, due to the deep burial of oil and gas structures and the characteristics of the burial environment, such as high temperature, complex formation pressure systems, and narrow safe drilling fluid density windows, improper drilling techniques can easily lead to complex wellbore pressure instability issues such as lost circulation, overflow, and gas intrusion. In severe cases, this can result in the abandonment of the entire well. Therefore, accurately predicting the type of drilling risk and determining whether the wellbore is stable is a key technology for achieving safe and efficient drilling in deep, high-temperature, high-pressure, and complex formations.

[0003] Traditional methods for predicting drilling risk types or wellbore stability primarily rely on real-time monitoring of logging parameters using integrated logging instruments to diagnose complex downhole risks. However, threshold setting depends on the professional experience of technicians and is highly subjective, resulting in low accuracy, high false alarm rates, and high missed alarm rates in predicting drilling risk types. While existing technologies that apply artificial intelligence to predict wellbore stability or drilling risk types eliminate human judgment and reduce subjectivity, they do not consider the relationship between formation pressure, bottom hole pressure, and drilling parameters. Therefore, existing methods for predicting wellbore stability or drilling risk types suffer from poor stability and reliability. Summary of the Invention

[0004] In view of the above problems, the present invention is proposed to provide a wellbore stability prediction model training, wellbore stability prediction method and apparatus that overcome or at least partially solve the above problems.

[0005] In a first aspect, embodiments of the present invention provide a training method for a wellbore stability prediction model, comprising:

[0006] Acquire pressure-controlled drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data from adjacent wells;

[0007] Based on the formation pressure profile data and the well logging data, a formation pressure input vector is determined. Based on the formation pressure profile data and the formation pressure input vector, a first training sample is obtained. The first training sample is input into a pre-built first neural network for training to obtain a formation pressure prediction model.

[0008] Based on the controlled-pressure drilling data and the bottom hole pressure data, a bottom hole pressure input vector is determined. Based on the bottom hole pressure input vector and the bottom hole pressure data, a second training sample is obtained. The second training sample is then input into a pre-built second neural network for training to obtain a bottom hole pressure prediction model.

[0009] Based on the drilling risk type data and the drilling data, a risk type input vector is determined. Based on the drilling risk type data and the risk type input vector, a third training sample is obtained. The third training sample is input into a pre-built third neural network for training to obtain an engineering parameter neural network.

[0010] The formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are connected in parallel to generate a wellbore stability prediction model, which is used to output the risk type of the adjacent well.

[0011] In one embodiment, at least one type of well logging data is provided.

[0012] The step of determining the formation pressure input vector based on the formation pressure profile data and the well logging data includes:

[0013] Calculate the correlation strength between the formation pressure profile data and each type of well logging data to obtain the formation pressure correlation strength corresponding to each type of well logging data;

[0014] Well logging data with a formation pressure correlation strength greater than or equal to a preset formation pressure correlation strength threshold are used as formation pressure input data. Feature extraction is performed on the formation pressure input data to obtain a formation pressure input vector.

[0015] In one embodiment, calculating the correlation strength between the formation pressure profile data and each type of well logging data to obtain the formation pressure correlation strength corresponding to each type of well logging data includes:

[0016] The correlation strength between the formation pressure profile data and the well logging data is calculated using the following formula;

[0017] ;

[0018] In the above formula, This represents formation pressure profile data. Represents well logging data, express covariance, express variance express variance express The correlation coefficient.

[0019] In one embodiment, at least one type of pressure-controlled drilling data is provided.

[0020] The step of determining the bottom hole pressure input vector based on the controlled drilling data and the bottom hole pressure data includes:

[0021] Calculate the correlation strength between the bottom hole pressure data and each type of controlled-pressure drilling data to obtain the bottom hole pressure correlation strength corresponding to each type of controlled-pressure drilling data;

[0022] Controlled-pressure drilling data with a bottom-hole pressure correlation strength greater than or equal to a preset bottom-hole pressure correlation strength threshold are used as bottom-hole pressure input data. Feature extraction is performed on the bottom-hole pressure input data to obtain the bottom-hole pressure input vector.

[0023] In one embodiment, calculating the correlation strength between the bottom hole pressure data and each type of controlled-pressure drilling data to obtain the bottom hole pressure correlation strength corresponding to each type of controlled-pressure drilling data includes:

[0024] The correlation strength between bottom hole pressure data and controlled pressure drilling data is calculated using the following formula:

[0025] ;

[0026] In the above formula, This indicates bottom hole pressure data. This represents controlled-pressure drilling data. express covariance, express variance express variance express The correlation coefficient.

[0027] In one embodiment, at least one type of drilling risk type data is provided; at least one type of drilling data is provided.

[0028] The step of determining the risk type input vector based on the drilling risk type data and the drilling data includes:

[0029] Calculate the correlation strength between the drilling risk type data and each type of drilling data to obtain the risk type correlation strength between each type of drilling risk data and each type of drilling data;

[0030] Drilling data with a risk type correlation strength greater than or equal to a preset risk type correlation strength threshold are used as risk type input data. Feature extraction is performed on the risk type input data to obtain a risk type input vector.

[0031] In one embodiment, calculating the correlation strength between the drilling risk type data and each type of drilling data, to obtain the correlation strength between the risk type data of each drilling risk type and each type of drilling data, includes:

[0032] The following formula is used to calculate the correlation strength between drilling risk type and drilling data:

[0033] ;

[0034] In the above formula, Indicates the type of drilling risk. Represents drilling data, express covariance, express variance express variance express The correlation coefficient.

[0035] Secondly, embodiments of the present invention provide a method for predicting wellbore stability, comprising:

[0036] Acquire controlled-pressure drilling data, logging data, and drilling data for wells with the risk type to be predicted;

[0037] Based on the logging data of the wells of the type of risk to be predicted, a formation pressure input vector is generated; based on the pressure-controlled drilling data of the wells of the type of risk to be predicted, a bottom hole pressure input vector is generated; and based on the drilling data of the wells of the type of risk to be predicted, a risk type input vector is generated.

[0038] The formation pressure input vector, the bottom hole pressure input vector, and the risk type input vector are input into the wellbore stability prediction model to obtain the risk type of the well to be predicted.

[0039] The wellbore stability prediction model is obtained through the training method of the aforementioned wellbore stability prediction model.

[0040] Thirdly, embodiments of the present invention provide a training apparatus for a wellbore stability prediction model, comprising:

[0041] The first acquisition module is used to acquire controlled pressure drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data of adjacent wells;

[0042] The formation pressure prediction model training module is used to determine the formation pressure input vector based on the formation pressure profile data and the well logging data, obtain a first training sample based on the formation pressure profile data and the formation pressure input vector, and input the first training sample into a pre-built first neural network for training to obtain the formation pressure prediction model.

[0043] The bottom hole pressure prediction model training module is used to determine the bottom hole pressure input vector based on the controlled drilling data and the bottom hole pressure data, obtain a second training sample based on the bottom hole pressure input vector and the bottom hole pressure data, and input the second training sample into a pre-built second neural network for training to obtain the bottom hole pressure prediction model.

[0044] The engineering parameter neural network training module is used to determine the risk type input vector based on the drilling risk type data and the drilling data, obtain a third training sample based on the drilling risk type data and the risk type input vector, and input the third training sample into the pre-built third neural network for training to obtain the engineering parameter neural network.

[0045] The first generation module is used to connect the formation pressure prediction model, the bottom hole pressure prediction model and the engineering parameter neural network in parallel to generate a wellbore stability prediction model. The wellbore stability prediction model is used to output the risk type of the adjacent well.

[0046] Fourthly, embodiments of the present invention provide a wellbore stability prediction device, comprising:

[0047] The second acquisition module is used to acquire controlled-pressure drilling data, logging data, and drilling data for wells of the risk type to be predicted;

[0048] The second generation module is used to generate a formation pressure input vector based on the logging data of the wells of the type of risk to be predicted, a bottom hole pressure input vector based on the pressure-controlled drilling data of the wells of the type of risk to be predicted, and a risk type input vector based on the drilling data of the wells of the type of risk to be predicted.

[0049] The prediction module is used to input the formation pressure input vector, the bottom hole pressure input vector, and the risk type input vector into the wellbore stability prediction model to obtain the risk type of the well to be predicted.

[0050] The wellbore stability prediction model is obtained through the training method of the aforementioned wellbore stability prediction model.

[0051] Fifthly, embodiments of the present invention provide a computer storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned training method for the wellbore stability prediction model or the aforementioned method for predicting wellbore stability.

[0052] In a sixth aspect, embodiments of the present invention provide a computing device, a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the aforementioned training method for the wellbore stability prediction model or the aforementioned method for predicting wellbore stability.

[0053] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:

[0054] The training method for a wellbore stability prediction model provided in this invention acquires information such as controlled-pressure drilling data, logging data, drilling data, formation pressure profile data, and drilling risk types from adjacent wells. Specifically, based on the formation pressure profile data and logging data, a formation pressure input vector is determined. A first training sample is obtained based on the formation pressure profile data and the formation pressure input vector. The first training sample is used to train a pre-built first neural network to obtain a formation pressure prediction model. Based on the bottom-hole pressure data and controlled-pressure drilling data, a bottom-hole pressure input vector is generated. A second training sample is obtained based on the bottom-hole pressure data and the bottom-hole pressure input vector. The second training sample is used to train a pre-built second neural network to obtain a bottom-hole pressure prediction model. Based on the drilling risk type data and drilling data, a risk type input vector is determined. A third training sample is obtained based on the drilling risk type data and the risk type input vector. The third training sample is used to train a pre-built third neural network to obtain a... An engineering parameter neural network is used to generate a wellbore stability prediction model by connecting the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network in parallel. In this embodiment of the invention, the formation pressure prediction model, based on well logging data that reflects geophysical information, can calculate and predict the formation pressure profile. The bottom hole pressure prediction model, based on pressure-controlled drilling data that reflects the flow information of drilling fluids, can achieve efficient prediction of bottom hole pressure. Considering that the cause of drilling risk is the imbalance between bottom hole pressure and formation pressure, the wellbore stability prediction model generated by connecting the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network in parallel considers both the balance between wellbore pressure and formation pressure and the impact of changes in engineering parameters caused by construction. This improves the stability and generalization ability of the wellbore stability prediction model and ensures the accuracy of the output results.

[0055] The wellbore stability prediction method provided in this invention generates a formation pressure input vector, a bottom hole pressure input vector, and a risk type input vector based on the obtained controlled-pressure drilling data, logging data, and drilling data of the well to be predicted for the risk type. Then, it calls the wellbore stability prediction model and outputs the risk type of the well to be predicted for the risk type. This reduces drilling costs, improves drilling quality and efficiency, enhances drilling safety and controllability, and ensures drilling safety.

[0056] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0057] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0058] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0059] Figure 1 This is a flowchart of the training method for the wellbore stability prediction model in Embodiment 1 of the present invention;

[0060] Figure 2 This is a schematic diagram of the wellbore stability prediction model in Embodiment 1 of the present invention;

[0061] Figure 3 This is a figure from Embodiment 2 of the present invention;

[0062] Figure 4 This is a flowchart of the wellbore stability prediction method in Embodiment 2 of the present invention;

[0063] Figure 5 This is a schematic diagram of the structure of the training device for the wellbore stability prediction model in an embodiment of the present invention;

[0064] Figure 6 This is a schematic diagram of the wellbore stability prediction device in an embodiment of the present invention. Detailed Implementation

[0065] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0066] To address the problems existing in the prior art, embodiments of the present invention provide a wellbore stability prediction model training method and apparatus for predicting wellbore stability.

[0067] Example 1

[0068] Embodiment 1 of the present invention provides a training method for a wellbore stability prediction model, the process of which is as follows: Figure 1 As shown, it includes the following steps:

[0069] Step S1: Obtain pressure-controlled drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data from adjacent wells;

[0070] Step S2: Determine the formation pressure input vector based on the formation pressure profile data and well logging data. Obtain the first training sample based on the formation pressure profile data and formation pressure input vector. Input the first training sample into the pre-built first neural network for training to obtain the formation pressure prediction model.

[0071] Step S3: Determine the bottom hole pressure input vector based on the controlled pressure drilling data and bottom hole pressure data. Obtain the second training sample based on the bottom hole pressure input vector and bottom hole pressure data. Input the second training sample into the pre-built second neural network for training to obtain the bottom hole pressure prediction model.

[0072] Step S4: Based on the drilling risk type data and drilling data, determine the risk type input vector. Based on the drilling risk type data and risk type input vector, obtain the third training sample. Input the third training sample into the pre-built third neural network for training to obtain the engineering parameter neural network.

[0073] Step S5: Connect the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network in parallel to generate a wellbore stability prediction model. The wellbore stability prediction model is used to output the risk type of adjacent wells.

[0074] In some optional embodiments, in order to ensure the accuracy of the trained wellbore stability prediction model, data from adjacent wells are generally selected as the base data. The adjacent wells can be completed wells or well sections in the well. After processing the data from the adjacent wells, training samples are obtained to train the pre-built model.

[0075] Among them, there are many types of controlled pressure drilling data during the drilling process, such as well depth data, standpipe pressure data, inlet flow rate data, outlet flow rate data, inlet density data, outlet density data, back pressure pump flow rate data, additional back pressure data, wellhead regulating pressure data, tripping speed data, etc.

[0076] There are many types of well logging data, such as sonic logging data, gamma logging data, density logging data, and caliper logging data.

[0077] There are many types of drilling data, such as: drilling pressure data, pump pressure data, displacement data, hook load data, etc.

[0078] There are many types of formation pressure profile data, such as: collapse pressure data, fracture pressure data, formation pore pressure data, leakage pressure data, etc.

[0079] There are many types of drilling risk data, such as gas intrusion risk, overflow risk, well kick risk, blowout risk, lost circulation risk, wellbore collapse risk, etc., which will not be listed one by one in the embodiments of this invention.

[0080] In some optional embodiments, in order to ensure the integrity of the training model data, outlier and missing value processing is performed on the obtained pressure-controlled drilling data, logging data, drilling data, formation pressure profile data and drilling risk type of adjacent wells. The advantage of doing so is that it improves the quality of the training data and ensures the accuracy of the wellbore stability prediction model after training.

[0081] In some optional embodiments, at least one type of logging data is used. In step S2 above, the formation pressure input vector is determined based on the formation pressure profile data and the logging data, which can be achieved in the following way:

[0082] (1) Calculate the correlation strength between the formation pressure profile data and each type of logging data to obtain the formation pressure correlation strength corresponding to each type of logging data;

[0083] Specifically, the correlation strength between formation pressure profile data and well logging data is calculated using the following formula;

[0084] ;

[0085] In the above formula, This represents formation pressure profile data. Represents well logging data, express covariance, express variance express variance express The correlation coefficient.

[0086] (2) Well logging data with formation pressure correlation intensity greater than or equal to the preset formation pressure correlation intensity threshold are used as formation pressure input data. Feature extraction is performed on the formation pressure input data to obtain the formation pressure input vector. There are various ways to perform feature extraction to generate the vector, and this embodiment of the invention does not limit this.

[0087] In practice, well logging data can influence formation pressure profile data. For each well, the correlation strength between formation pressure profile data and each type of well logging data is calculated to obtain the formation pressure correlation strength corresponding to each type of well logging data. The formation pressure correlation strength characterizes the degree to which well logging data influences formation pressure profile data. From several types of well logging data, well logging data with a formation pressure correlation strength greater than or equal to a preset formation pressure correlation strength threshold are selected and retained as formation pressure input data. In other words, based on the formation pressure correlation strength corresponding to each type of well logging data, several types of well logging data with a strong influence on formation pressure profile data are selected from multiple types of well logging data. These selected types of well logging data are used as formation pressure input data. Then, feature extraction is performed on the formation pressure input data to obtain the formation pressure input vector. That is to say, the formation pressure input vector of the well corresponds to the formation pressure profile data of that well.

[0088] Then, feature extraction is performed on the formation pressure profile data to generate a vector of formation pressure profile data. The formation pressure input vector and the corresponding vector of formation pressure profile data are used as the first training sample. The formation pressure input vector and the corresponding vector of formation pressure profile data of a large number of adjacent wells are obtained through the above method to obtain a large number of first training samples and a first training sample set. The first training sample set is input into the pre-built first neural network for training to obtain the formation pressure prediction model.

[0089] Formation pressure prediction models can calculate and predict formation pressure profiles of target wells based on well logging data that reflects geophysical information.

[0090] In actual exploration and development, due to the influence of various factors, the formation pressure input data obtained from different exploration blocks may be different. For example, the formation pressure input data of Block A consists of sonic logging data, gamma logging data, density logging data, and resistivity logging data, while the formation pressure input data of Block B consists of sonic logging data, gamma logging data, and density logging data.

[0091] In some optional embodiments, at least one type of controlled-pressure drilling data is used. In step S3 above, the bottom-hole pressure input vector is determined based on the controlled-pressure drilling data and the bottom-hole pressure data, which can be achieved in the following way:

[0092] (1) Calculate the correlation strength between the bottom hole pressure data and each type of controlled pressure drilling data to obtain the bottom hole pressure correlation strength corresponding to each type of controlled pressure drilling data;

[0093] Specifically, the correlation strength between bottom hole pressure data and controlled pressure drilling data is calculated using the following formula:

[0094] ;

[0095] In the above formula, This indicates bottom hole pressure data. This represents controlled-pressure drilling data. express covariance, express variance express variance express The correlation coefficient.

[0096] (2) Use the controlled pressure drilling data with a bottom pressure correlation strength greater than or equal to the preset bottom pressure correlation strength threshold as bottom pressure input data, extract features from the bottom pressure input data, and obtain the bottom pressure input vector.

[0097] The bottom hole pressure can be obtained through actual measurement or other methods, and this embodiment of the invention does not limit this.

[0098] Controlled pressure drilling data reflects the flow information of drilling fluids and other fluids within the wellbore annulus. For each well, the correlation strength between bottom-hole pressure data and each type of controlled pressure drilling data is calculated to obtain the bottom-hole pressure correlation strength corresponding to each type of controlled pressure drilling data. The bottom-hole pressure correlation strength characterizes the degree to which controlled pressure drilling data affects bottom-hole pressure data. Controlled pressure drilling data with a bottom-hole pressure correlation strength greater than or equal to a preset bottom-hole pressure correlation strength threshold are selected and retained as bottom-hole pressure input data. In other words, based on the bottom-hole pressure correlation strength corresponding to each type of controlled pressure drilling data, several types of controlled pressure drilling data with a strong influence on bottom-hole pressure data are selected from a variety of controlled pressure drilling data. These selected types of controlled pressure drilling data are used as bottom-hole pressure input data. Then, feature extraction is performed on the bottom-hole pressure input data to obtain the bottom-hole pressure input vector. That is, the bottom-hole pressure input vector of a well corresponds to the bottom-hole pressure data of that well.

[0099] Then, feature extraction is performed on the bottom hole pressure data to generate a vector of bottom hole pressure data. The bottom hole pressure input vector and the corresponding bottom hole pressure data vector are used as the second training sample. The bottom hole pressure input vector and the corresponding bottom hole pressure data vector of a large number of adjacent wells are obtained in the above way to obtain a large number of second training samples, which are used as the second training sample set. The second training sample set is input into the pre-built second neural network for training to obtain the bottom hole pressure prediction model.

[0100] Bottomhole pressure prediction models can calculate the bottomhole pressure data of a well based on controlled-pressure drilling data, such as well depth data, standpipe pressure data, inlet flow rate data, outlet flow rate data, inlet density data, outlet density data, back pressure pump flow rate data, additional back pressure data, wellhead regulating pressure data, tripping speed data, etc.

[0101] In actual exploration and development, due to the influence of various factors, the bottom hole pressure input data obtained in different exploration blocks may be different. For example, the bottom hole pressure input data of Block A consists of well depth data, inlet density data, inlet flow rate data, back pressure pump flow rate data, and additional back pressure data, while the bottom hole pressure input data of Block B consists of several other types of controlled pressure drilling data.

[0102] In some optional embodiments, there is at least one type of drilling risk data; there is at least one type of drilling data, and in step S4 above, the risk type input vector is determined based on the drilling risk type and the drilling data, which can be achieved in the following way:

[0103] (1) Calculate the correlation strength between drilling risk type data and each type of drilling data to obtain the correlation strength between each type of drilling risk data and each type of drilling data;

[0104] Specifically, the correlation strength between drilling risk type data and drilling data is calculated using the following formula:

[0105] ;

[0106] In the above formula, Data indicating drilling risk type, Represents drilling data, express covariance, express variance express variance express The correlation coefficient.

[0107] (2) Drilling data with a risk type correlation strength greater than or equal to the preset risk type correlation strength threshold are used as risk type input data. Features are extracted from the risk type input data to obtain the risk type input vector.

[0108] Drilling data can reflect changes in drilling parameters. For each well, the correlation strength between drilling risk type data and each type of drilling data is calculated to obtain the risk type correlation strength corresponding to each type of drilling data. The risk type correlation strength characterizes the degree to which drilling data affects the drilling risk type. Drilling data with a risk type correlation strength greater than or equal to a preset risk type correlation strength threshold are selected and retained as risk type input data. In other words, based on the risk type correlation strength corresponding to each type of drilling data, several types of drilling data with a strong influence on the drilling risk type data are selected from multiple types of drilling data. These selected types of drilling data are used as bottom hole pressure input data. Then, feature extraction is performed on the risk type input data to obtain the risk type input vector. That is to say, the risk type input vector of a well corresponds to the risk type that exists in that well.

[0109] Then, feature extraction is performed on the risk type data to generate risk type data vectors. The risk type input vector and the corresponding risk type data vector are used as third training samples. A large number of adjacent well risk type input vectors and corresponding risk type data vectors are obtained through the above method to obtain a large number of third training samples, resulting in a third training sample set. The third training sample set is input into a pre-built third neural network for training to obtain an engineering parameter neural network, which fully considers the impact of changes in engineering parameters during drilling on well stability.

[0110] Similarly, in the actual exploration and development process, due to the influence of various factors, the risk type input data obtained for different exploration blocks may be different. For example, the risk type input data for Block A consists of drilling speed data, riser pressure data, hook load data, outlet flow rate data, inlet flow rate data, and total pool volume data, while the risk type input data for Block B consists of several other types of drilling data.

[0111] In some optional embodiments, the third training sample includes the risk type input vector of the adjacent well and the corresponding risk type data vector. The risk type data vector can be obtained by one-hot vector encoding to obtain the vector of the number of risk types. Specifically, based on the number n of drilling risk types, an n-dimensional row vector is set. Each column in the row vector represents a type of drilling risk. When the drilling risk occurs, the column representing the current drilling risk is recorded as 1, and the other columns are recorded as 0. The drilling risk type data is encoded in this way.

[0112] In some optional embodiments, the imbalance between the bottom hole pressure and the formation pressure during drilling leads to wellbore instability and poses a risk. In step S5 above, considering the balance between the bottom hole pressure and the formation pressure, as well as the impact of changes in engineering parameters caused by construction on wellbore stability, the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are connected in parallel, such as... Figure 2 As shown, a wellbore stability prediction model is generated. Figure 2 In this method, the formation pressure input vector is used as the input to the formation pressure prediction model, the bottom hole pressure input vector is used as the input to the bottom hole pressure prediction model, and the risk type input vector is used as the input to the engineering parameter neural network. The output results of the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are obtained respectively. The output of the formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are used as input, and the drilling risk type data is used as output. The formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are fused in parallel to generate a wellbore stability prediction model, which effectively improves the stability and generalization ability of the wellbore stability prediction model.

[0113] To ensure the accuracy of the wellbore stability prediction model, an error loss function is constructed, and the parameters in the wellbore stability prediction model are solved using the backpropagation algorithm. Based on the value of the error loss function, the parameters in the wellbore stability prediction model are adjusted until the value of the loss function is less than a preset threshold, at which point the accuracy of the wellbore stability prediction model is considered to have reached its maximum.

[0114] Example 2

[0115] Embodiment 2 of the present invention provides a method for predicting wellbore stability, the process of which is as follows: Figure 4 As shown, it includes the following steps:

[0116] Step S41: Obtain controlled pressure drilling data, logging data, and drilling data for wells of the risk type to be predicted;

[0117] Step S42: Generate a formation pressure input vector based on the logging data of the well with the risk type to be predicted, generate a bottom hole pressure input vector based on the pressure control drilling data of the well with the risk type to be predicted, and generate a risk type input vector based on the drilling data of the well with the risk type to be predicted.

[0118] Step S43: Input the formation pressure input vector, bottom hole pressure input vector, and risk type input vector into the wellbore stability prediction model to obtain the risk type of the well to be predicted;

[0119] The wellbore stability prediction model is obtained through the training method of the aforementioned wellbore stability prediction model.

[0120] In some optional embodiments, step S42 above can be implemented, for example, in the following manner:

[0121] Based on the formation pressure input vector of the wellbore stability prediction model, several types of logging data corresponding to the formation pressure input vector are selected from the logging data of the wells with the risk type to be predicted. For example, the selected logging data are sonic logging data, gamma logging data, density logging data, and resistivity logging data. Feature extraction is performed on the several types of logging data corresponding to the formation pressure input vector to obtain the formation pressure input vector of the wells with the risk type to be predicted.

[0122] Based on the bottom hole pressure input vector of the wellbore stability prediction model, several types of controlled pressure drilling data corresponding to the bottom hole pressure input vector are selected from the controlled pressure drilling data of the wells with the risk type to be predicted. For example, the selected controlled pressure drilling data are well depth data, inlet density data, inlet flow rate data, back pressure pump flow rate data, and additional back pressure composition data. Feature extraction is performed on the several types of controlled pressure drilling data corresponding to the bottom hole pressure input vector to obtain the bottom hole pressure input vector of the well with the risk type to be predicted.

[0123] Based on the risk type input vector of the wellbore stability prediction model, several types of drilling data corresponding to the risk type input vector are selected from the drilling data of the wells whose risk type is to be predicted. For example, the selected drilling data are drilling speed data, riser pressure data, hook load data, outlet flow rate data, inlet flow rate data, and total pool volume data. Feature extraction is performed on the several types of drilling data corresponding to the risk type input vector to obtain the risk type input vector of the wells whose risk type is to be predicted.

[0124] In some optional embodiments, in step S42 above, the extracted formation pressure input vector, bottom hole pressure input vector, and risk type input vector are input into the wellbore stability prediction model. The wellbore stability prediction model can output one or more types of risks such as gas intrusion risk, overflow risk, well kick risk, blowout risk, lost circulation risk, and wellbore collapse risk. Accordingly, based on the output results, corresponding measures are taken to ensure safety during the exploration and development process. If the wellbore stability prediction model does not output risk type results, it indicates that the well with the predicted risk type has good stability and there is currently no risk.

[0125] Based on the same inventive concept, embodiments of the present invention also provide a training device for a wellbore stability prediction model. This device can be installed in [a specific location / location], and its structure is as follows: Figure 5 As shown, it includes:

[0126] The first acquisition module 51 is used to acquire controlled-pressure drilling data, logging data, formation pressure profile data and drilling risk type of adjacent wells;

[0127] The formation pressure prediction model training module 52 is used to determine the formation pressure input vector based on the formation pressure profile data and well logging data. The formation pressure profile data and formation pressure input vector are used as the first training samples. The first training samples are input into the pre-built first neural network for training to obtain the formation pressure prediction model.

[0128] The bottom hole pressure prediction model training module 53 is used to obtain the bottom hole pressure data of adjacent wells based on the controlled pressure drilling data, determine the bottom hole pressure input vector based on the controlled pressure drilling data and the bottom hole pressure data, use the bottom hole pressure input vector and the bottom hole pressure data as the second training sample, and input the second training sample into the pre-built second neural network for training to obtain the bottom hole pressure prediction model.

[0129] The engineering parameter neural network training module 54 is used to determine the risk type input vector based on the drilling risk type and drilling data, and to use the drilling risk type and risk type input vector as the third training sample. The third training sample is then input into the pre-built third neural network for training to obtain the engineering parameter neural network.

[0130] The first generation module 55 is used to connect the formation pressure prediction model, the bottom hole pressure prediction model and the engineering parameter neural network in parallel to generate a wellbore stability prediction model. The wellbore stability prediction model is used to output the risk type of adjacent wells.

[0131] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0132] Based on the same inventive concept, embodiments of the present invention also provide a wellbore stability prediction device, which can be installed in the wellbore, and the structure of the device is as follows: Figure 6 As shown, it includes:

[0133] The second acquisition module 61 is used to acquire controlled-pressure drilling data, logging data and drilling data of wells with the risk type to be predicted;

[0134] The second generation module 62 is used to generate a formation pressure input vector based on the logging data of the well to be predicted risk type, generate a bottom hole pressure input vector based on the pressure control drilling data of the well to be predicted risk type, and generate a risk type input vector based on the drilling data of the well to be predicted risk type.

[0135] Prediction module 63 is used to input the formation pressure input vector, bottom hole pressure input vector and risk type input vector into the wellbore stability prediction model to obtain the risk type of the well to be predicted.

[0136] The wellbore stability prediction model is obtained through the training method of the aforementioned wellbore stability prediction model.

[0137] Regarding the apparatus in the above embodiments, the specific manner in which each module performs its operation has been described in detail in the embodiments related to the method, and will not be elaborated upon here.

[0138] Based on the same inventive concept, embodiments of the present invention also provide a computer storage medium storing computer-executable instructions, which, when executed by a processor, implement the aforementioned training method for the wellbore stability prediction model or the aforementioned method for predicting wellbore stability.

[0139] Based on the same inventive concept, embodiments of the present invention also provide a computing device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the aforementioned training method for the wellbore stability prediction model or the aforementioned prediction method for wellbore stability.

[0140] Unless otherwise specifically stated, terms such as processing, calculation, operation, determination, display, etc., may refer to the actions and / or processes of one or more processing or computing systems or similar devices that represent the manipulation and conversion of data representing physical (e.g., electronic) quantities within the registers or memory of the processing system into other data similarly representing physical quantities within the memory, registers, or other such information storage, transmission, or display devices of the processing system. Information and signals can be represented using any of a variety of different techniques and methods. For example, data, instructions, commands, information, signals, bits, symbols, and chips mentioned throughout the above description can be represented by voltage, current, electromagnetic waves, magnetic fields or particles, light fields or particles, or any combination thereof.

[0141] It should be understood that the specific order or hierarchy of steps in the disclosed process is an example of an exemplary method. Based on design preferences, it should be understood that the specific order or hierarchy of steps in the process may be rearranged without departing from the scope of this disclosure. The appended method claims provide elements of various steps in an exemplary order and are not intended to limit the scope to the specific order or hierarchy described.

[0142] In the detailed description above, various features are combined together in a single embodiment to simplify this disclosure. This approach to disclosure should not be construed as reflecting an intention that embodiments of the claimed subject matter require more features than are explicitly stated in each claim. Rather, as reflected in the appended claims, the invention is presented with fewer features than all of the features in a single disclosed embodiment. Therefore, the appended claims are hereby explicitly incorporated into the detailed description, with each claim representing a separate preferred embodiment of the invention.

[0143] Those skilled in the art will also understand that the various illustrative logic blocks, modules, circuits, and algorithm steps described in conjunction with the embodiments herein can be implemented as electronic hardware, computer software, or a combination thereof. To clearly illustrate the interchangeability between hardware and software, the various illustrative components, blocks, modules, circuits, and steps described above are generally described in terms of their functionality. Whether such functionality is implemented as hardware or software depends on the specific application and the design constraints imposed on the overall system. Those skilled in the art can implement the described functionality in alternative ways for each specific application; however, such implementation decisions should not be construed as departing from the scope of this disclosure.

[0144] The steps of the methods or algorithms described in conjunction with the embodiments herein can be directly embodied in hardware, software modules executed by a processor, or a combination thereof. The software modules can reside in RAM memory, flash memory, ROM memory, EPROM memory, EEPROM memory, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium well known in the art. An exemplary storage medium is connected to the processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium can also be a component of the processor. The processor and storage medium can reside in an ASIC. The ASIC can reside in a user terminal. Alternatively, the processor and storage medium can exist as discrete components in the user terminal.

[0145] For software implementation, the techniques described in this application can be implemented using modules (e.g., procedures, functions, etc.) that perform the functions described in this application. This software code can be stored in memory units and executed by a processor. The memory units can be implemented within the processor or outside the processor; in the latter case, they are communicatively coupled to the processor via various means, as is well known in the art.

[0146] The foregoing description includes examples of one or more embodiments. It is certainly impossible to describe all possible combinations of components or methods in order to describe the above embodiments, but those skilled in the art will recognize that further combinations and arrangements of the various embodiments are possible. Therefore, the embodiments described herein are intended to cover all such changes, modifications, and variations that fall within the scope of the appended claims. Furthermore, the term "comprising" as used in the specification or claims is interpreted in a manner similar to the term "including," as interpreted when used as a conjunction in the claims. Additionally, the use of any term "or" in the specification of the claims is intended to mean "non-exclusive or."

Claims

1. A training method for a wellbore stability prediction model, characterized in that, include: Acquire pressure-controlled drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data from adjacent wells; Based on the formation pressure profile data and the well logging data, a formation pressure input vector is determined. Based on the formation pressure profile data and the formation pressure input vector, a first training sample is obtained. The first training sample is input into a pre-built first neural network for training to obtain a formation pressure prediction model. Based on the controlled-pressure drilling data and the bottom hole pressure data, a bottom hole pressure input vector is determined. Based on the bottom hole pressure input vector and the bottom hole pressure data, a second training sample is obtained. The second training sample is then input into a pre-built second neural network for training to obtain a bottom hole pressure prediction model. Based on the drilling risk type data and the drilling data, a risk type input vector is determined. Based on the drilling risk type data and the risk type input vector, a third training sample is obtained. The third training sample is input into a pre-built third neural network for training to obtain an engineering parameter neural network. The formation pressure prediction model, the bottom hole pressure prediction model, and the engineering parameter neural network are connected in parallel to generate a wellbore stability prediction model, which is used to output the risk type of the adjacent well.

2. The method as described in claim 1, characterized in that, At least one type of well logging data is mentioned; The step of determining the formation pressure input vector based on the formation pressure profile data and the well logging data includes: Calculate the correlation strength between the formation pressure profile data and each type of well logging data to obtain the formation pressure correlation strength corresponding to each type of well logging data; Well logging data with a formation pressure correlation strength greater than or equal to a preset formation pressure correlation strength threshold are used as formation pressure input data. Feature extraction is performed on the formation pressure input data to obtain a formation pressure input vector.

3. The method as described in claim 2, characterized in that, The calculation of the correlation strength between the formation pressure profile data and each type of well logging data, to obtain the formation pressure correlation strength corresponding to each type of well logging data, includes: The correlation strength between the formation pressure profile data and the well logging data is calculated using the following formula; ； In the above formula, This represents formation pressure profile data. Represents well logging data, express covariance, express variance express variance express The correlation coefficient.

4. The method as described in claim 1, characterized in that, At least one type of controlled-pressure drilling data is mentioned; The step of determining the bottom hole pressure input vector based on the controlled drilling data and the bottom hole pressure data includes: Calculate the correlation strength between the bottom hole pressure data and each type of controlled-pressure drilling data to obtain the bottom hole pressure correlation strength corresponding to each type of controlled-pressure drilling data; Controlled-pressure drilling data with a bottom-hole pressure correlation strength greater than or equal to a preset bottom-hole pressure correlation strength threshold are used as bottom-hole pressure input data. Feature extraction is performed on the bottom-hole pressure input data to obtain the bottom-hole pressure input vector.

5. The method as described in claim 4, characterized in that, The calculation of the correlation strength between the bottom hole pressure data and each type of controlled pressure drilling data, to obtain the bottom hole pressure correlation strength corresponding to each type of controlled pressure drilling data, includes: The correlation strength between bottom hole pressure data and controlled pressure drilling data is calculated using the following formula: ； In the above formula, This indicates bottom hole pressure data. This represents controlled-pressure drilling data. express covariance, express variance express variance express The correlation coefficient.

6. The method as described in claim 1, characterized in that, The drilling risk type data is at least one type; the drilling data is at least one type; The step of determining the risk type input vector based on the drilling risk type data and the drilling data includes: Calculate the correlation strength between the drilling risk type data and each type of drilling data to obtain the risk type correlation strength between each type of drilling risk data and each type of drilling data; Drilling data with a risk type correlation strength greater than or equal to a preset risk type correlation strength threshold are used as risk type input data. Feature extraction is performed on the risk type input data to obtain a risk type input vector.

7. The method as described in claim 6, characterized in that, The calculation of the correlation strength between the drilling risk type data and each type of drilling data, to obtain the correlation strength between the risk type data of each type of drilling risk and each type of drilling data, includes: The following formula is used to calculate the correlation strength between drilling risk type and drilling data: ； In the above formula, Indicates the type of drilling risk. Represents drilling data, express covariance, express variance express variance express The correlation coefficient.

8. A method for predicting wellbore stability, characterized in that, include: Acquire controlled-pressure drilling data, logging data, and drilling data for wells with the risk type to be predicted; Based on the logging data of the wells of the type of risk to be predicted, a formation pressure input vector is generated; based on the pressure-controlled drilling data of the wells of the type of risk to be predicted, a bottom hole pressure input vector is generated; and based on the drilling data of the wells of the type of risk to be predicted, a risk type input vector is generated. The formation pressure input vector, the bottom hole pressure input vector, and the risk type input vector are input into the wellbore stability prediction model to obtain the risk type of the well to be predicted. The wellbore stability prediction model is obtained by the training method of the wellbore stability prediction model according to any one of claims 1-7.

9. A training device for a wellbore stability prediction model, characterized in that, include: The first acquisition module is used to acquire controlled pressure drilling data, logging data, drilling data, formation pressure profile data, bottom hole pressure data, and drilling risk type data of adjacent wells; The formation pressure prediction model training module is used to determine the formation pressure input vector based on the formation pressure profile data and the well logging data, obtain a first training sample based on the formation pressure profile data and the formation pressure input vector, and input the first training sample into a pre-built first neural network for training to obtain the formation pressure prediction model. The bottom hole pressure prediction model training module is used to determine the bottom hole pressure input vector based on the controlled drilling data and the bottom hole pressure data, obtain a second training sample based on the bottom hole pressure input vector and the bottom hole pressure data, and input the second training sample into a pre-built second neural network for training to obtain the bottom hole pressure prediction model. The engineering parameter neural network training module is used to determine the risk type input vector based on the drilling risk type data and the drilling data, obtain a third training sample based on the drilling risk type data and the risk type input vector, and input the third training sample into the pre-built third neural network for training to obtain the engineering parameter neural network. The first generation module is used to connect the formation pressure prediction model, the bottom hole pressure prediction model and the engineering parameter neural network in parallel to generate a wellbore stability prediction model. The wellbore stability prediction model is used to output the risk type of the adjacent well.

10. A device for predicting wellbore stability, characterized in that, include: The second acquisition module is used to acquire controlled-pressure drilling data, logging data, and drilling data for wells of the risk type to be predicted; The second generation module is used to generate a formation pressure input vector based on the logging data of the well to be predicted risk type, generate a bottom hole pressure input vector based on the pressure control drilling data of the well to be predicted risk type, and generate a risk type input vector based on the drilling data of the well to be predicted risk type. The prediction module is used to input the formation pressure input vector, the bottom hole pressure input vector, and the risk type input vector into the wellbore stability prediction model to obtain the risk type of the well to be predicted. The wellbore stability prediction model is obtained by the training method of the wellbore stability prediction model according to any one of claims 1-7.

11. A computer storage medium, characterized in that, The computer storage medium stores computer-executable instructions, which, when executed by a processor, implement the training method for the wellbore stability prediction model according to any one of claims 1-7 or the wellbore stability prediction method according to claim 8.

12. A computing device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the training method of the wellbore stability prediction model according to any one of claims 1-7 or the wellbore stability prediction method according to claim 8.

Images