Drill bit wear degree prediction model training method, prediction method, device and equipment

Abstract

The application provides a drill bit wear degree prediction model training method, a prediction method, an apparatus and equipment. The method obtains a pre-constructed drill bit wear degree prediction model, assigns values to at least one optimization coefficient in the model, constructs a training sample according to the drill bit wear degree prediction value of the model after the assignment, trains a drilling speed prediction model based on the training sample to obtain a trained drilling speed prediction model, reassigns values to the optimization coefficient until the trained drilling speed prediction model corresponding to the re-assigned optimization coefficient meets a prediction deviation requirement, determines the drill bit wear degree prediction model corresponding to the drilling speed prediction model meeting the prediction deviation requirement as a target drill bit wear degree prediction model, and predicts the wear degree of a target drill bit based on the model. The application obtains a drill bit wear degree prediction model for accurately predicting the wear degree of a drill bit, thereby realizing accurate prediction of the wear degree of a drill bit.

Description

Technical Field

[0001] This application relates to data processing technology, and in particular to a training method, prediction method, apparatus and equipment for a drill bit wear prediction model. Background Technology

[0002] In drilling operations, minimizing cost is typically the goal of engineering optimization. During the drilling process, the wear and tear of the drill bit significantly impacts costs. In practice, costs are predicted by forecasting drill bit wear, and the drilling strategy is then optimized based on these cost predictions. Therefore, predicting drill bit wear is crucial during the drilling process.

[0003] In related technologies, a drill bit wear prediction model is determined by classifying the wear level of the drill bit when it enters and exits the well during the drilling process, and the wear level of the drill bit during the drilling process is predicted based on the determined drill bit wear prediction model.

[0004] However, in related technologies, the wear level of drill bits during entry and exit from the well is manually classified, resulting in low accuracy of the obtained wear level classification. This leads to poor accuracy in drill bit wear prediction models based on the low accuracy of the wear level classification results. In other words, there is a technical problem in related technologies where there is a lack of methods for developing drill bit wear prediction models that can accurately predict the wear level of drill bits. Summary of the Invention

[0005] This application provides a drill bit wear prediction model training method, prediction method, apparatus, and equipment to solve the technical problem in the prior art that there is a lack of a method for accurately predicting drill bit wear.

[0006] In a first aspect, embodiments of this application provide a method for training a drill bit wear prediction model, comprising: acquiring a pre-constructed drill bit wear prediction model, wherein the pre-constructed drill bit wear prediction model includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit; assigning values ​​to at least one optimization coefficient, and constructing training samples for training a drilling speed prediction model based on the assigned drill bit wear prediction model, wherein each training sample includes at least a predicted value of drill bit wear and is marked with an actual drilling speed; training the preset drilling speed prediction model using the training samples to obtain a trained drilling speed prediction model; in response to the trained drilling speed prediction model not meeting the prediction deviation requirement, reassigning at least one optimization coefficient, and executing the step of constructing training samples for training a drilling speed prediction model based on the reassigned drill bit wear prediction model to obtain the corresponding trained drilling speed prediction model, until the trained drilling speed prediction model meets the prediction deviation requirement; determining the assigned drill bit wear prediction model corresponding to the drilling speed prediction model that meets the prediction deviation requirement as the target drill bit wear prediction model, wherein the target drill bit wear prediction model is used to predict the wear degree of the target drill bit.

[0007] Secondly, embodiments of this application provide a method for predicting drill bit wear, including:

[0008] Obtain the value of at least one state parameter corresponding to the target drill bit at the target prediction point; input the value of at least one state parameter into the target drill bit wear prediction model; use the target drill bit wear prediction model to calculate the predicted value of the drill bit wear at the target prediction point. The target drill bit wear prediction model is trained using the drill bit wear prediction model training method as described in the first aspect.

[0009] Thirdly, embodiments of this application provide a drill bit wear prediction model training device, the device comprising: a first acquisition module, configured to acquire a pre-constructed drill bit wear prediction model, the pre-constructed drill bit wear prediction model including at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit; a construction module, configured to assign values ​​to at least one optimization coefficient and construct training samples for training a drilling speed prediction model based on the assigned drill bit wear prediction model, each training sample including at least a predicted drill bit wear value and labeled with the actual drilling speed; and a second acquisition module, configured to train the preset drilling speed prediction model using the training samples to obtain training samples. The training process includes: a drill speed prediction model; a model optimization module, which, in response to the training drill speed prediction model not meeting the prediction deviation requirement, performs steps to reassign at least one optimization coefficient and construct training samples based on the reassigned drill bit wear prediction model to obtain the corresponding trained drill speed prediction model, until the trained drill speed prediction model meets the prediction deviation requirement; and a determination module, which determines the reassigned drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement as the target drill bit wear prediction model, which is used to predict the wear degree of the target drill bit.

[0010] Fourthly, embodiments of this application provide an electronic device, including: a processor, a memory communicatively connected to the processor, and a transceiver; the memory stores computer-executable instructions; the transceiver is used for sending and receiving data; the processor executes the computer-executable instructions stored in the memory to implement the method described in the first aspect or the second aspect.

[0011] This invention provides a method, method, apparatus, and device for training a drill bit wear prediction model. The method involves acquiring a pre-constructed drill bit wear prediction model, which includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit. The at least one optimization coefficient is assigned a value, and training samples are constructed based on the assigned drill bit wear prediction model to train a drilling speed prediction model. Each training sample includes at least a predicted drill bit wear value and is labeled with the actual drilling speed. The training samples are used to train the preset drilling speed prediction model to obtain a training result. The subsequent drilling speed prediction model, in response to the training model failing to meet the prediction deviation requirement, reassigns at least one optimization coefficient and executes the step of constructing training samples based on the reassigned drill bit wear prediction model to obtain the corresponding trained drilling speed prediction model, until the trained drilling speed prediction model meets the prediction deviation requirement. The drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement is determined as the target drill bit wear prediction model, which is used to predict the wear degree of the target drill bit. Compared with the method in related technologies that trains the drill bit wear prediction model by manually determining the wear degree level of the drill bit at the wellhead and the wellhead, the input parameters of the target drill bit wear prediction model constructed in this embodiment are the state parameters of the target drill bit during the drilling process. Based on these state parameters, a quantitative value of the drill bit wear degree prediction can be obtained. Therefore, the prediction results of the target drill bit wear prediction model obtained according to the method of this embodiment have high accuracy. Furthermore, in the method of this embodiment, at least one optimization coefficient in the drill bit wear prediction model is assigned a value. Based on the drill bit wear prediction value output by the corresponding drill bit wear prediction model, a drilling speed prediction model is trained. According to the judgment result of whether the trained drilling speed prediction model meets the prediction deviation requirement, at least one optimization coefficient in the drill bit wear prediction model is reassigned a value until the trained drilling speed prediction model obtained based on the reassigned optimization coefficient meets the prediction deviation requirement. This process is equivalent to adjusting at least one optimization coefficient in the drill bit wear prediction model according to the prediction accuracy of the drilling speed prediction model, thereby obtaining a target drill bit wear prediction model with high prediction accuracy. That is, this solution provides a target drill bit wear prediction model with high accuracy, solving the technical problem in related technologies where there is a lack of methods for accurately predicting drill bit wear.

[0012] It should be understood that the description in the foregoing summary section is not intended to limit the key or essential features of the embodiments of the present invention, nor is it intended to restrict the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 This is a schematic diagram of an application scenario of the drill bit wear prediction model training method and prediction method provided in the embodiments of the present invention;

[0015] Figure 2 A flowchart of a drill bit wear prediction model training method provided in an embodiment of the present invention;

[0016] Figure 3 A flowchart of a drill bit wear prediction method provided in another embodiment of the present invention;

[0017] Figure 4(a) is a schematic diagram of the drill bit speed value results provided according to an embodiment of the present invention;

[0018] Figure 4(b) is a schematic diagram of the drilling pressure value results provided according to an embodiment of the present invention;

[0019] Figure 4(c) is a schematic diagram of the displacement value of a drilling power pump provided according to an embodiment of the present invention;

[0020] Figure 4(d) is a schematic diagram of the drill bit rotation speed values ​​provided according to an embodiment of the present invention;

[0021] Figure 4(e) is a schematic diagram of the rock compressive strength values ​​provided according to an embodiment of the present invention;

[0022] Figure 4(f) is a schematic diagram of the drill bit wear degree value results provided according to an embodiment of the present invention;

[0023] Figure 5 This is a schematic diagram of the structure of a drill bit wear prediction model training device provided in an embodiment of the present invention;

[0024] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention;

[0025] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0026] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0027] First, let me explain the terms used in this application:

[0028] Model hyperparameters refer to configuration variables outside the model, which are usually determined according to user needs. For example, the standard deviation in a Gaussian process model, etc.

[0029] Footage refers to the distance from the surface to the latest drilled point in the drilling operation, and is used to characterize the progress of the drilling operation.

[0030] Lithology refers to the performance parameters that reflect the characteristics of rocks, including rock composition, etc.

[0031] Wells that have been drilled and whose drilling operations have been completed in the target work area.

[0032] Drilled section: In the target operating area, the section of a well that has already been drilled during drilling operations.

[0033] Drill bit speed refers to the speed at which the drill bit rotates.

[0034] Mechanical drilling rate refers to the drilling depth of a well per unit time during drilling operations, and is used to characterize the speed of drilling operations.

[0035] The time difference of sound waves refers to the time it takes for a sound wave to travel a unit distance in the earth's strata.

[0036] International Association of Drilling Contractors (IADC) Wear Code: In drilling operations, the wear level is a grade of drill bit wear obtained according to the IADC drill bit wear grading standard to characterize the degree of drill bit wear.

[0037] Currently, in oil drilling, drill bits break up the bottom rock through compression and cutting, thereby connecting underground reservoir fluids with surface pipelines to develop oil and gas resources. In oil drilling operations, the goal of technical optimization is typically to minimize cost per meter. The lifespan or footage of a single drill bit directly affects the cost per meter of the section drilled by that bit, and the lifespan or footage of the drill bit is related to its wear level. Therefore, pre-drilling prediction, monitoring, and forecasting of drill bit wear are crucial for pre-drilling scheme design and optimization of drilling parameters. It is important to understand that, given a fixed drill bit model and size, drill bit wear is influenced by a combination of factors, such as drilling parameters (e.g., pressure on bit and bit rate of penetration) and geological characteristics (e.g., lithology and rock strength).

[0038] In related technologies, wear conditions of drill bits during well entry and exit are obtained through manual observation. The wear level at these times is then determined using the IADC (Independent Diagnostic and Advanced Data Classification) drill bit wear grading standard, with the IADC wear code used to indicate the wear level. Next, a recursive calculation model for drill bit wear, taking into account drilling parameters and formation characteristics, is established through experimental analysis. The relevant weights in the recursive calculation model are then calibrated using the IADC wear code, enabling the prediction of wear levels for the same type of drill bit in new wells or new wellbores.

[0039] However, in related technologies, the accuracy of obtaining IADC wear codes during drill bit entry and exit from the well through manual observation is low, which leads to low accuracy in the drill bit wear prediction results obtained from the IADC wear codes. In other words, the existing technology lacks a method for accurately predicting drill bit wear levels using a drill bit wear prediction model.

[0040] To address the lack of accurate drill bit wear prediction models in related technologies, it is necessary to determine the drill bit wear prediction model using state parameters of the target drill bit during the drilling process, rather than training the model based on drill bit wear level classifications obtained from manual observation. Specifically, a target drill bit wear prediction model is determined by using a pre-constructed drill bit wear prediction model that includes state parameters of the target drill bit during the drilling process. Furthermore, the pre-constructed drill bit wear prediction model includes at least one optimization coefficient. To optimize the pre-constructed model and obtain a target drill bit wear prediction model with high accuracy, the assignment of the aforementioned at least one optimization coefficient needs to be optimized. Specifically, by assigning values ​​to at least one optimization coefficient, training samples are constructed based on the drill bit wear prediction model's predicted drill bit wear values ​​and the corresponding actual drilling speeds. These training samples are then used to train a pre-defined drilling speed prediction model, resulting in a trained model. If the trained model does not meet the prediction deviation requirement, at least one optimization coefficient is re-assigned until the model meets the requirement. The assigned value of at least one optimization coefficient corresponding to the model that meets the deviation requirement is the optimal value for that coefficient. The drill bit wear prediction model containing this optimal value is then considered the target model. This process optimizes the assignment of at least one optimization coefficient in the pre-constructed drill bit wear prediction model, resulting in a target model with high prediction accuracy.

[0041] The following section introduces the application scenarios of the drill bit wear prediction model training method, prediction method, device, and equipment provided in this application.

[0042] Figure 1 This is a schematic diagram illustrating an application scenario of the drill bit wear prediction model training method and prediction method provided in this embodiment of the invention, such as... Figure 1 As shown, this application scenario includes a drilling operation area (drill well 101), a drill bit wear prediction model training device 104, and a drill bit wear prediction device 105. The drilling well 101 contains multiple predetermined measurement points 102, and also includes target prediction points 103 for which the wear degree of the target drill bit needs to be predicted.

[0043] Specifically, by assigning values ​​to at least one optimization coefficient in the pre-built drill bit wear prediction model, the state parameter values ​​of multiple predetermined measurement points 102 are input into the pre-built drill bit wear prediction model to obtain the corresponding drill bit wear prediction values. The drill bit wear prediction values ​​are combined with the actual drilling speed to construct training samples, and the preset drilling speed prediction model is trained to obtain the trained drilling speed prediction model. If the drill bit wear prediction model meets the prediction deviation requirement after training, it is determined whether the drill bit wear prediction model with the corresponding optimized coefficient is the target drill bit wear prediction model. The target drill bit wear prediction model is then sent to the drill bit wear prediction device 105. If the prediction deviation requirement is not met, at least one of the aforementioned optimized coefficients is re-assigned. Based on the drill bit wear prediction value output by the re-assigned drill bit wear prediction model, the steps of constructing training samples to train the drill bit prediction model and obtaining the corresponding trained drill bit wear prediction model are executed until the trained drill bit speed prediction model meets the prediction deviation requirement. The drill bit wear prediction model with the corresponding optimized coefficient is then determined as the target drill bit wear prediction model, and the target drill bit wear prediction model is then sent to the drill bit wear prediction device 105. The drill bit wear prediction device 105 stores the target drill bit wear prediction model trained by the drill bit wear prediction model training device 104, and receives the values ​​of the state parameters of the target prediction point. It inputs the values ​​of the state parameters of the target prediction point into the target drill bit wear prediction model to obtain the predicted value of the drill bit wear at the target prediction point.

[0044] It should be noted that this application can train and predict drill bit wear prediction models for various application scenarios. For example, it can be applied to oil drilling scenarios, as well as other scenarios that require drilling operations and prediction of drill bit wear during the drilling process.

[0045] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0046] Figure 2 This is a flowchart illustrating a drill bit wear prediction model training method according to an embodiment of the present invention. The subject of this method is a drill bit wear prediction model training device, such as… Figure 2 As shown, the method in this embodiment includes the following steps.

[0047] S101, Obtain a pre-built drill bit wear prediction model, which includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit.

[0048] In one embodiment, at least one state parameter during the drilling process of the target drill bit may include: state parameters of the target drill bit and geological state parameters. The state parameters of the target drill bit may include: drill pressure, drill bit speed, and the geological state parameters may include rock compressive strength, etc.

[0049] In one embodiment, there can be various types of pre-built drill bit wear prediction models. For example, a drill bit wear prediction model can be constructed based on drill pressure, drill bit speed, rock compressive strength, depth of measurement point, drill pressure coefficient corresponding to drill pressure, drill speed coefficient corresponding to drill bit speed, first model adjustment coefficient, and second model adjustment coefficient.

[0050] S102, assign values ​​to at least one optimization coefficient, and construct training samples for training the drilling speed prediction model based on the assigned drill bit wear prediction model. Each training sample includes at least the predicted value of drill bit wear and is marked with the actual drilling speed.

[0051] The actual drilling speed is the actual measured value of the mechanical drilling speed of the target drill bit.

[0052] In one embodiment, assigning a value to at least one optimization coefficient may include the following scheme: first, predetermine the value range of each optimization coefficient among at least one optimization coefficient, and then assign a value to at least one optimization coefficient according to a predetermined assignment strategy.

[0053] In one embodiment, constructing training samples for training the drilling speed prediction model based on the assigned drill bit wear prediction model can include the following approach: obtaining multiple drill bit wear prediction values ​​output by the assigned drill bit wear prediction model based on the values ​​of multiple sets of state parameters; and constructing training samples based on the multiple drill bit wear prediction values ​​and the corresponding multiple actual drilling speeds. Specifically, the values ​​of the multiple sets of state parameters and the multiple actual drilling speeds correspond one-to-one. The values ​​of a particular set of state parameters are obtained from state data collected at a certain measurement point during the drilling process of the target drill bit, and the actual drilling speed corresponding to that particular set of state parameters is the mechanical drilling speed measurement value collected at that measurement point.

[0054] S103, the preset drilling speed prediction model is trained using training samples to obtain the trained drilling speed prediction model.

[0055] In one optional embodiment, the preset drilling speed prediction model is a machine model. Specifically, there can be various types of preset drilling speed prediction models, such as neural network models, Gaussian process models, support vector machine models, extreme learning machine models, random forest models, and so on.

[0056] In one optional embodiment, training the model of the preset drilling speed prediction model with training samples to obtain the trained drilling speed prediction model may include the following scheme: training the model of the preset drilling speed prediction model with the prediction deviation value of the drilling speed prediction model satisfying a predetermined deviation value as the optimization objective to obtain the trained drilling speed prediction model.

[0057] S104, in response to the trained drilling speed prediction model not meeting the prediction deviation requirement, at least one optimization coefficient is reassigned, and the step of constructing training samples for training the drilling speed prediction model based on the reassigned drill bit wear prediction model is executed until the corresponding trained drilling speed prediction model is obtained, until the trained drilling speed prediction model meets the prediction deviation requirement.

[0058] In one optional embodiment, determining whether the trained drilling speed prediction model meets the prediction deviation requirement may include the following approach: obtaining the prediction error of the trained drilling speed prediction model when assigning different values ​​to at least one optimization coefficient, and determining whether the trained drilling speed prediction model meets the prediction deviation requirement based on the prediction error.

[0059] S105, the drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement is determined as the target drill bit wear prediction model. The target drill bit wear prediction model is used to predict the wear degree of the target drill bit.

[0060] In an optional embodiment, determining the drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement as the target drill bit wear prediction model includes: determining the assignment of at least one optimization coefficient corresponding to the drill speed prediction model that meets the prediction deviation requirement, and determining the drill bit wear prediction model corresponding to the assignment of the at least one optimization coefficient as the target drill bit wear prediction model.

[0061] In this optional embodiment, a drill bit wear prediction model is constructed, including at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit. Values ​​are assigned to at least one optimization coefficient in the drill bit wear prediction model, and a drilling speed prediction model is trained based on the drill bit wear prediction value output by the corresponding drill bit wear prediction model. Depending on whether the trained drilling speed prediction model meets the prediction deviation requirement, at least one optimization coefficient in the drill bit wear prediction model is reassigned until the trained drilling speed prediction model obtained based on the reassigned optimization coefficient meets the prediction deviation requirement. This process is equivalent to adjusting at least one optimization coefficient in the drill bit wear prediction model according to the prediction accuracy of the drilling speed prediction model, thereby obtaining a target drill bit wear prediction model with high prediction accuracy. Compared to related technologies that train a drill bit wear prediction model based on manually determined wear levels during drill bit entry and exit from the well, and then predict drill bit wear based on these wear levels, the target drill bit wear prediction model constructed in this embodiment uses state parameters of the target drill bit during the drilling process as input parameters. Based on these state parameters, a quantitative value for the predicted drill bit wear can be obtained. Therefore, the accuracy of the drill bit wear prediction result obtained by the target drill bit wear prediction model in this embodiment is high. In other words, this solution provides a target drill bit wear prediction model with high accuracy, solving the technical problem in related technologies where there is a lack of methods for accurately predicting drill bit wear.

[0062] In an optional embodiment, step S102 of the above embodiment, assigning values ​​to at least one optimization coefficient and constructing training samples for training the drilling speed prediction model based on the assigned drill bit wear prediction model, may include the following scheme:

[0063] S201, assign values ​​to at least one optimization coefficient to obtain a drill bit wear prediction model after assignment.

[0064] In one embodiment, assigning values ​​to at least one optimization coefficient may include the following approach: assigning values ​​to the corresponding optimization coefficients according to a predetermined value range for each optimization coefficient. The value ranges for each optimization coefficient may be the same or different, and can be determined based on empirical data. There are various methods for assigning values ​​to at least one optimization coefficient according to a predetermined assignment strategy; for example, the corresponding optimization coefficient can be randomly selected from its corresponding value range for assignment.

[0065] S202, based on the value of at least one state parameter of the target drill bit at multiple predetermined measurement points, and the drill bit wear degree prediction model after assignment, determine the predicted value of the drill bit wear degree at multiple predetermined measurement points.

[0066] In one optional embodiment, the multiple predetermined measurement points include at least multiple measurement points for collecting status parameters when the target drill bit is performing drilling operations in a drilled well and / or a drilled section. Here, a drilled well refers to a well where drilling operations have been completed, and a drilled section refers to a well where drilling operations are currently underway.

[0067] In one alternative embodiment, there can be multiple methods for selecting multiple predetermined measurement points. For example, multiple predetermined measurement points can be determined in the drilled well and / or drilled section by random selection, multiple predetermined measurement points can be determined according to a predetermined interval, or multiple predetermined measurement points can be determined according to other measurement point selection strategies.

[0068] In one optional embodiment, multiple state parameters of multiple predetermined measurement points are obtained, and the multiple state parameters of multiple predetermined measurement points are input into the drill bit wear prediction model after assignment to obtain the drill bit wear prediction value of multiple predetermined measurement points.

[0069] S203, construct training samples based on the predicted values ​​of drill bit wear at each predetermined measurement point and the corresponding actual drilling speed.

[0070] In one optional embodiment, the actual drilling speed at each predetermined measurement point is obtained, the actual drilling speed at each predetermined measurement point is used as a label, the corresponding predicted value of drill bit wear is used as the model input parameter of the drilling speed prediction model, and training samples are constructed based on the actual drilling speed at each predetermined measurement point and the corresponding predicted value of drill bit wear.

[0071] In an optional embodiment, the actual drilling speed at each predetermined measurement point is used as a label, and the following model input parameters for the drilling speed prediction model are obtained: the predicted value of the drill bit wear degree, drilling speed, and drilling pressure at each predetermined measurement point, as well as the rock compressive strength, drilling power pump displacement, and drilling fluid density at each predetermined measurement point. Training samples are constructed based on the values ​​of the model input parameters at each predetermined measurement point and the model input parameters.

[0072] In this optional embodiment, based on the drill bit wear prediction model with at least one optimized coefficient assigned a value, and the values ​​of at least one state parameter of the target drill bit at multiple predetermined measurement points, predicted values ​​of drill bit wear at multiple predetermined measurement points are obtained. Training samples for training the drilling speed prediction model are then constructed based on the predicted values ​​of drill bit wear at the predetermined measurement points and the corresponding actual drilling speed. Since the predicted values ​​of drill bit wear are related to the assigned values ​​of at least one optimized coefficient, the accuracy of the prediction result of the trained drilling speed prediction model is related to the predicted values ​​of drill bit wear. The assigned value of at least one optimized coefficient affects the accuracy of the prediction result of the drilling speed prediction model. By constructing training samples for training the drilling speed prediction model based on the predicted values ​​of drill bit wear at predetermined measurement points, the assigned values ​​of at least one optimized coefficient can be linked to the accuracy of the prediction result of the drilling speed prediction model. This allows for the determination of the optimal assigned value of at least one optimized coefficient based on the accuracy of the prediction result of the drilling speed prediction model, thereby improving the accuracy of the prediction result of the determined target drill bit wear prediction model.

[0073] In an optional embodiment, in step S102 of the above embodiment, the preset drilling speed prediction model is trained using training samples to obtain the trained drilling speed prediction model. This can include the following approach:

[0074] S301, assign values ​​to at least one hyperparameter of the preset drilling speed prediction model a predetermined number of times to obtain multiple hyperparameter assignment drilling speed prediction models corresponding to the predetermined number of hyperparameter assignments.

[0075] In one optional embodiment, assigning values ​​to at least one hyperparameter of a preset drilling speed prediction model a predetermined number of times may include the following approach: determining initial values ​​for at least one hyperparameter of the preset drilling speed prediction model based on empirical data; updating the initial values ​​of at least one hyperparameter a predetermined number of times based on multiple differences corresponding to each of the at least one hyperparameter, thereby obtaining the predetermined number of times the at least one hyperparameter of the trained drilling speed prediction model has been assigned values. That is, by assigning values ​​to at least one hyperparameter of the preset drilling speed prediction model a predetermined number of times, multiple sets of hyperparameter values ​​are obtained, wherein the number of sets of hyperparameter values ​​is equal to the predetermined number of times the at least one hyperparameter has been assigned values.

[0076] S302, for each of the multiple hyperparameter-assigned drilling speed prediction models: use training samples to train the hyperparameter-assigned drilling speed prediction model to obtain the first hyperparameter-assigned drilling speed prediction model after training, and determine the prediction deviation value of the first hyperparameter-assigned drilling speed prediction model after training.

[0077] In one embodiment, a training set and a test set are determined from the training samples according to a predetermined sample size ratio. A hyperparameter-assigned drilling speed prediction model is trained using the training set to obtain a trained first hyperparameter-assigned drilling speed prediction model. The trained first hyperparameter-assigned drilling speed prediction model is tested using the test set to obtain the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model.

[0078] S303, determine the hyperparameter-assigned drilling speed prediction model corresponding to the first hyperparameter-assigned drilling speed prediction model after training with the smallest prediction deviation value as the first drilling speed prediction model.

[0079] In one embodiment, the prediction deviation values ​​of the first hyperparameter-assigned drilling speed prediction models trained in step S302, which are respectively corresponding to multiple hyperparameter-assigned drilling speed prediction models, are compared, and the hyperparameter-assigned drilling speed prediction model corresponding to the trained first hyperparameter-assigned drilling speed prediction model with the smallest prediction deviation value is determined as the first drilling speed prediction model.

[0080] S304, The first drilling speed prediction model is trained using training samples to obtain the trained drilling speed prediction model.

[0081] It is understandable that the first drilling speed prediction model is trained using training samples to obtain the trained drilling speed prediction model, but this embodiment does not limit this.

[0082] In this optional embodiment, multiple trained first hyperparameter-assigned drilling speed prediction models are obtained through multiple hyperparameter-assigned drilling speed prediction models, and the prediction deviation values ​​of the multiple trained first hyperparameter-assigned drilling speed prediction models are obtained. The hyperparameter-assigned drilling speed prediction model corresponding to the trained first hyperparameter-assigned drilling speed prediction model with the smallest prediction deviation value is determined as the first drilling speed prediction model, thereby obtaining the first drilling speed prediction model with the highest prediction accuracy.

[0083] In an optional embodiment, in step S302, the prediction bias value of the trained first hyperparameter assignment drilling speed prediction model is determined. This may include the following scheme:

[0084] S401, determine multiple sample sets and test sets from the training samples, and the proportion of the number of samples in each sample set and test set in the multiple sample sets and test sets meets the predetermined proportion of the number of samples.

[0085] In one embodiment, the ratio of the number of samples in the test set and the training set is set according to user needs. For example, to improve the accuracy of the prediction bias value of the trained first hyperparameter-assigned drilling speed prediction model, the proportion of the number of samples in the test set can be increased; similarly, to improve the prediction accuracy of the trained first hyperparameter-assigned drilling speed prediction model, the proportion of the number of samples in the training set can be increased. In one embodiment, the ratio of the number of samples in the test set to the number of samples in the training set is greater than or equal to 5.

[0086] S402, for each of the multiple sample sets and test sets: use the sample set to train the hyperparameter-assigned drilling speed prediction model to obtain the first hyperparameter-assigned drilling speed prediction model after training, and obtain the prediction bias value of the first hyperparameter-assigned drilling speed prediction model after training based on the test set.

[0087] In one embodiment, obtaining the prediction bias value of the trained first hyperparameter-assigned drilling speed prediction model based on the test set may include the following scheme: inputting data other than the actual drilling speed from the test set into the trained first hyperparameter-assigned drilling speed prediction model to obtain multiple drilling speed prediction values; obtaining the root mean square error between the multiple drilling speed prediction values ​​and the corresponding actual drilling speed in the test set; and determining the root mean square error as the prediction bias value of the trained first hyperparameter-assigned drilling speed prediction model based on the test set.

[0088] S403, obtain the average prediction deviation of the first hyperparameter assignment drilling speed prediction model after training, corresponding to multiple sets of sample sets and test sets, respectively.

[0089] In one embodiment, the average prediction deviation is obtained by averaging the prediction deviation values ​​of multiple trained first hyperparameter assignment drilling speed prediction models corresponding to multiple sample sets and test sets, respectively.

[0090] S404, determine the average prediction deviation as the prediction deviation value of the drill speed prediction model after training, based on the first hyperparameter assignment.

[0091] It is understandable that there are various methods to determine the average prediction deviation as the prediction deviation value of the drill speed prediction model after training the first hyperparameter assignment, and this embodiment does not limit this method.

[0092] In this optional embodiment, the average prediction deviation of multiple trained first hyperparameter-assigned drilling speed prediction models corresponding to multiple sample sets and test sets is obtained, and this average prediction deviation is determined as the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model. Therefore, the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model can be accurately obtained.

[0093] As an optional implementation, determining whether the trained drilling speed prediction model meets the prediction bias requirement includes:

[0094] S501, determine the prediction error of the drill speed prediction model after reassigning at least one optimization coefficient and the prediction error of the training model corresponding to the different assigned optimization coefficients.

[0095] In one optional embodiment, assigning values ​​to at least one optimization coefficient may include the following approach: within a predetermined value range for each optimization coefficient, reassigning each optimization coefficient according to a predetermined assignment strategy. For example, changing the assignment of each optimization coefficient by increasing a predetermined difference, or changing the assignment of each optimization coefficient by decreasing a predetermined difference.

[0096] In one optional embodiment, for each of the multiple trained drilling speed prediction models corresponding to different assigned optimization coefficients: an error calculation sample is determined, wherein the error calculation sample can be a portion of the samples randomly determined from the training samples, or it can be all the training samples; the data from the error calculation sample, excluding the actual drilling speed, is input into the trained drilling speed prediction model to obtain a set of drilling speed prediction values; based on the obtained set of drilling speed prediction values ​​and the corresponding actual drilling speed, a set of measurement point prediction errors is obtained; the prediction error of the trained drilling speed prediction model is determined based on the obtained set of measurement point prediction errors. There are various methods for determining the prediction error of the trained drilling speed prediction model based on the obtained set of measurement point prediction errors. For example, the average of the aforementioned set of measurement point prediction errors can be calculated, and the average value of the obtained prediction errors can be determined as the prediction error of the trained drilling speed prediction model.

[0097] S502, based on the prediction error and the preset error threshold, determine whether the trained drilling speed prediction model meets the prediction deviation requirements.

[0098] In one optional embodiment, if the prediction error is less than a preset error threshold, the trained drilling speed prediction model is determined to meet the prediction deviation requirement; if the prediction error is greater than or equal to the preset error threshold, the trained drilling speed prediction model is determined to not meet the prediction deviation requirement.

[0099] In this optional embodiment, by determining the prediction error of the trained drilling speed prediction model, and based on the prediction error and a preset error threshold, it is determined whether the trained drilling speed prediction model meets the prediction deviation requirements. Therefore, a quantitative analysis of whether the trained drilling speed prediction model meets the prediction deviation requirements can be performed, improving the accuracy of the judgment result regarding whether the trained drilling speed prediction model meets the prediction deviation requirements.

[0100] As an optional embodiment, if there are multiple target drill bits, and these multiple target drill bits are simultaneously drilling in the same target well, the drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement is determined as the target drill bit wear prediction model. This can include the following schemes:

[0101] S601, obtain multiple prediction deviation values ​​of the drilling speed prediction model that meets the prediction deviation requirements for multiple target drill bits respectively.

[0102] In one embodiment, for each of the multiple target drill bits, steps S501 to S503 are executed to determine a trained drilling speed prediction model that meets the prediction deviation requirements. Steps S301 to S304 and steps S401 to S403 are executed to obtain multiple prediction deviation values ​​corresponding to the multiple target drill bits.

[0103] S602, determine the sum of the inverses of multiple prediction bias values.

[0104] In an optional embodiment, the summation of the reciprocals of the multiple prediction deviation values ​​obtained in step S601 is obtained.

[0105] S603, for each of the multiple target drill bits: based on the predicted deviation value of the target drill bit and the superposition value of multiple predicted deviation values, the optimization coefficient adjustment value of the target drill bit is obtained. Based on the optimization coefficient adjustment value, the assignment of at least one optimization coefficient in the drill bit wear degree prediction model corresponding to the drill speed prediction model that meets the predicted deviation requirements is adjusted to obtain the target drill bit wear degree prediction model.

[0106] In one embodiment, the optimization coefficient adjustment value of the target drill bit is obtained based on the predicted deviation value of the target drill bit and the sum of multiple predicted deviation values. This may include the following approach: obtaining the ratio of the reciprocal of the predicted deviation value of the target drill bit to the sum of multiple predicted deviation values, and obtaining the optimization coefficient adjustment value of the target drill bit based on this ratio.

[0107] In one embodiment, based on the optimization coefficient adjustment value, the assignment of at least one optimization coefficient in the drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement is adjusted to obtain the target drill bit wear prediction model. This can include the following scheme: calculating the product of the optimization coefficient adjustment value and the corresponding optimization coefficient assignment value, determining the product as the adjusted assignment value of the corresponding optimization coefficient, and determining the target drill bit wear prediction model based on the adjusted assignment value of the corresponding optimization coefficient.

[0108] In this optional embodiment, if multiple target drill bits are drilling simultaneously in the same target well, the corresponding optimization coefficient adjustment value is obtained based on the multiple prediction deviation values ​​of the drilling speed prediction model that meets the prediction deviation requirements for each target drill bit. Based on the corresponding optimization coefficient adjustment value, the assignment of at least one optimization coefficient in the drill bit wear prediction model corresponding to the assigned value of the drilling speed prediction model that meets the prediction deviation requirements for each target drill bit is adjusted. It should be understood that in the same drilling operation scenario, if multiple target drill bits are drilling simultaneously in the same target well, the wear degree of the multiple target drill bits is less than the wear degree of a single target drill bit operating alone. The solution of this optional embodiment is equivalent to using the optimization coefficient adjustment value as a weight to further optimize the target drill bit wear prediction model obtained based on a single target drill bit. The resulting target drill bit wear prediction model for each target drill bit takes into account the influence of other target drill bits operating together with each target drill bit, thus improving the accuracy of the prediction results of the determined target drill bit wear prediction model.

[0109] In one alternative embodiment, according to any of the above alternative embodiments or the methods of the embodiments, the state parameters include the target rock compressive strength before determining the predicted values ​​of drill bit wear at multiple predetermined measurement points. The following scheme is also included:

[0110] S701, acquire the time difference of multiple target acoustic waves at multiple predetermined measurement points.

[0111] In one alternative embodiment, the target acoustic time difference is obtained by measuring the acoustic time difference at multiple predetermined measurement points using a logging instrument.

[0112] S702, based on the acoustic transit time of multiple targets and the pre-constructed correspondence between rock compressive strength and acoustic transit time, determines the compressive strength of multiple target rocks at multiple predetermined measurement points.

[0113] In one optional embodiment, a pre-constructed correspondence between rock compressive strength and acoustic transit time is built based on the acoustic transit time and the coefficients of the rock compressive strength calculation model fitted from empirical data. Multiple target acoustic transit times are then input into the pre-constructed correspondence between rock compressive strength and acoustic transit time to obtain multiple corresponding rock compressive strengths. Optionally, the obtained multiple rock compressive strengths can be denoised to obtain highly accurate target rock compressive strengths. Various methods can be used to denoise the multiple rock compressive strengths; for example, Gaussian smoothing can be used.

[0114] In this optional embodiment, the compressive strength of multiple target rocks at multiple predetermined measurement points can be quickly obtained by using the multiple target acoustic transit times at multiple predetermined measurement points and the pre-constructed correspondence between rock compressive strength and acoustic transit times.

[0115] Figure 3 This is a flowchart illustrating a drill bit wear prediction method according to an embodiment of the present invention. The method in this application is implemented by a drill bit wear prediction device, such as… Figure 3 As shown, the method in this embodiment includes the following steps:

[0116] S801, obtain the value of at least one state parameter of the target drill bit at the target prediction point.

[0117] In one embodiment, at least one state parameter obtained includes: the depth of the target predicted point, the drill pressure of the target drill bit, the drill speed of the target drill bit, and the compressive strength of the target rock. Optionally, the compressive strength of the target rock is obtained according to the method of steps S701 to S702.

[0118] S802, input the value of at least one state parameter into the target drill bit wear prediction model.

[0119] In an optional embodiment, the following state parameters are input into the target drill bit wear prediction model: the depth of the target prediction point, the drill pressure of the target drill bit, the drill speed of the target drill bit, and the compressive strength of the target rock.

[0120] S803, the target drill bit wear prediction model is used to calculate the predicted value of the drill bit wear at the target prediction point. The target drill bit wear prediction model is trained using any of the above drill bit wear prediction model training methods.

[0121] In an optional embodiment, the predicted value of the wear degree of the target drill bit at the target prediction point is obtained by inputting the value of at least one state parameter into the target drill bit wear degree prediction model.

[0122] In this optional embodiment, by inputting the value of at least one state parameter corresponding to the target drill bit at the target prediction point into the target drill bit wear prediction model with high accuracy in predicting the wear degree of the drill bit, a high-accuracy target drill bit wear degree prediction result can be obtained.

[0123] In one optional embodiment, the drill bit wear prediction method may include the following steps:

[0124] S901, acquire the target geological parameters of the drilled section corresponding to the target drill bit and the target state parameters of the target drill bit. Specifically, this includes the following scheme:

[0125] S9011 collects logging data from multiple predetermined measurement points in the drilled well or drilled section, and obtains the target rock compressive strength S based on the logging data from multiple predetermined measurement points.

[0126] The logging data specifically includes sonic transit time, and the rock compressive strength S at the predetermined measurement point is obtained according to the following method. log :

[0127] S log = a × (304.8 / Δt) b

[0128] Where Δt is the acoustic transit time, in µs / ft; a and b are the calculation coefficients for rock compressive strength, which can be determined by historical rock compressive strength and corresponding historical acoustic transit time data obtained from rock mechanics tests.

[0129] The rock compressive strength S was obtained using the Gaussian smoothing method shown below. log Perform smoothing:

[0130]

[0131] Where μ represents the rock compressive strength S within a certain smooth window. log The average value is expressed in MPa; σ represents the rock compressive strength S within the corresponding smoothing window. log The standard deviation of ; S is the target rock compressive strength obtained after smoothing, in MPa.

[0132] The obtained rock compressive strength S was smoothed using a Gaussian smoothing method. log Smoothing can improve the compressive strength S of the rock. log The purpose of noise reduction is to obtain a high accuracy and usability of the target rock compressive strength S.

[0133] S9012 collects logging data from multiple predetermined measurement points in the drilled section.

[0134] Specifically, the logging data includes the depth h of multiple predetermined measurement points, the drilling pressure W, the drill bit speed N of the target drill bit, the displacement Q of the drilling power pump, and the drilling fluid density ρ.

[0135] It is important to understand that the multiple predetermined measurement points in steps S9011 and S9012 may not be consistent. In this case, it is necessary to use the depth h in the logging data as a column and employ interpolation to process the multiple target rock compressive strengths S at the multiple predetermined measurement points in step S9011, obtaining multiple target rock compressive strengths S corresponding to the multiple predetermined measurement points in S9012. Based on the logging data and the multiple target rock compressive strengths S after interpolation, a data table is generated with the depth h in the logging data as the first column.

[0136] By iterating through all depths h in the data table and combining them with the cuttings logging dataset constructed based on empirical data, lithological data corresponding to each depth h are obtained. The cuttings logging dataset contains data at multiple depths, along with the corresponding lithological data for each depth.

[0137] In one embodiment, after acquiring lithological data corresponding to each depth h, the acquired lithological data is classified into four categories according to a predetermined classification strategy. These four categories of lithology include: sandstone, mudstone, carbonate rocks, and igneous rocks. The classification method includes replacing "mudstone," "siltstone," and "fine sandstone" with "sand-mudstone," and replacing "argillaceous limestone" and "dolomite" with "carbonate rocks." This classification reduces the number of lithological data types.

[0138] Write the lithological data into a data table to obtain a merged data table containing the following data: depth h, drilling pressure W, target drill bit speed N, drilling power pump displacement Q, drilling fluid density ρ, target rock compressive strength S, and lithological data. The merged data table is shown in Table 1.

[0139] Table 1

[0140]

[0141]

[0142] In this optional embodiment, the target drill bit's state parameters include drill bit state parameters and geological state parameters. The drill bit state parameters include the target drill bit's drilling speed N, drilling power pump displacement Q, drilling fluid density ρ, and drilling pressure W. The target geological parameters include depth h, lithological data, and the target rock compressive strength S.

[0143] S902, a pre-built model for predicting drill bit wear.

[0144] The following method was used to obtain a model for predicting drill bit wear in mudstone formations:

[0145]

[0146] Where V1 is the predicted value of drill bit wear corresponding to the mudstone layer, and c sd c w,sd c n,sd c v,sd For the optimization coefficients of the drill bit wear prediction model corresponding to the mudstone layer, specifically, c sd c w,sd c n,sd c v,sd These are the first model coefficient corresponding to the mudstone layer, the drill pressure coefficient of the drill bit, the drill speed coefficient of the target drill bit, and the second model coefficient, respectively. sd cw,sd c n,sd c v,sd This can be determined using a heuristic algorithm; specifically, in this optional embodiment, c sd c w,sd c n,sd c v,sd Obtained through steps S903 to S907.

[0147] The following method was used to obtain a model for predicting drill bit wear in sandstone formations:

[0148]

[0149] Where V2 is the predicted value of drill bit wear corresponding to the sandstone formation, and c sa c w,sa c n,sa c v,sa For the optimization coefficients of the drill bit wear prediction model corresponding to sandstone formations, specifically, c sa c w,sa c n,sa c v,sa These are, respectively, the first model adjustment coefficient corresponding to the sandstone formation, the drill pressure coefficient corresponding to the drill pressure, the drill rate coefficient corresponding to the drill bit speed, and the second model adjustment coefficient, c. sa c w,sa , c v,sa Optimization needs to be achieved through heuristic algorithms. Specifically, in this optional embodiment, c sa c w,sa , c v,sa Obtained through steps S903 to S907.

[0150] A model for predicting drill bit wear in carbonate formations is obtained using the following method:

[0151]

[0152] Where V3 is the predicted value of drill bit wear corresponding to carbonate rock formations, and c cb c w,cb c n,cb c v,cb For the optimization coefficients of the drill bit wear prediction model corresponding to carbonate rock formations, specifically, c cb c w,cb c n,cb c v,cb These are, respectively, the first model adjustment coefficient corresponding to carbonate rock formations, the drill pressure coefficient corresponding to drill pressure, the drill rate coefficient corresponding to drill bit speed, and the second model adjustment coefficient, c. cbc w,cb c n,cb c v,cb Optimization needs to be achieved through heuristic algorithms. Specifically, in this optional embodiment, c cb c w,cb c n,cb c v,cb Obtained through steps S903 to S907.

[0153] A model for predicting drill bit wear in igneous formations is obtained using the following method:

[0154]

[0155] Where V4 is the predicted value of drill bit wear for the target drill bit corresponding to the igneous rock strata, and c ig c w,ig c n,ig c v,ig For the optimization coefficients of the drill bit wear prediction model corresponding to igneous rock formations, specifically, c ig c w,ig c n,ig c v,ig These are, respectively, the first model adjustment coefficient corresponding to igneous rock formations, the drill pressure coefficient corresponding to drill pressure, the drill rate coefficient corresponding to drill bit speed, and the second model adjustment coefficient, c. ig c w,ig c n,ig c v,ig Optimization needs to be achieved through heuristic algorithms. Specifically, in this optional embodiment, c ig c w,ig c n,ig c v,ig Obtained through steps S903 to S907.

[0156] S903, assign a value to at least one optimization coefficient.

[0157] First, define the value range for each optimization coefficient:

[0158] c sd ,c sa ,c cb ,c ig ∈[0.001,10]

[0159] c w,sd ,c w,sa ,c w,cb ,c w,ig ∈[0.01,10]

[0160] c n,sd ,c n,sa ,c n,cb,c n,ig ∈[0.01,10]

[0161] c v,sd ,c v,sa ,c v,cb ,c v,ig ∈[0.01,10]

[0162] Within the range of values ​​for each optimization coefficient, values ​​are assigned to the corresponding optimization coefficients according to a predetermined assignment strategy. For example, the assigned value for each optimization coefficient can be randomly selected within its corresponding range.

[0163] Referring to Table 1, the lithology corresponding to the prediction point can be determined based on the depth of the prediction point, and the corresponding drill bit wear prediction model can be determined based on the lithology. Based on the assigned values ​​of the optimization coefficients in the drill bit wear prediction model, the drill bit wear prediction model corresponding to each lithology can be determined.

[0164] S904, obtain a drilling speed prediction model that meets the prediction deviation requirements.

[0165] S9041, Construct a drilling rate prediction model.

[0166] Specifically, a drilling speed prediction model is established based on a machine learning model and takes into account the predicted value of drill bit wear. The machine learning model used to establish the drilling speed prediction model can be of various types, such as neural network models, Gaussian process models, support vector machine models, extreme learning machine models, random forest models, and so on.

[0167] The input parameters of the drilling speed prediction model include at least: the drilling speed N and drilling pressure W of the target drill bit at each predetermined measurement point, as well as the target rock compressive strength S, the displacement Q of the drilling power pump, and the drilling fluid density ρ at each predetermined measurement point.

[0168] The drilling speed prediction model can be expressed as:

[0169] rop=f(W,N,S,Q,ρ,V)

[0170] Wherein, rop is the predicted drilling speed in m / h; f represents the machine learning model; and V is the predicted wear level of the target drill bit, including V1, V2, V3 or V4 in step S902.

[0171] S9042, determine the drill rate prediction model after training.

[0172] The values ​​of the following state parameters obtained from multiple predetermined measurement points are input into the drill bit wear prediction model, which is optimized according to step S903, to obtain the predicted wear value V of the target drill bit at each predetermined measurement point. The drill bit speed N, drilling pressure W, target rock compressive strength S, drilling power pump displacement Q, and drilling fluid density ρ at multiple predetermined measurement points are also obtained. The predicted wear value V at each predetermined measurement point, along with the drill bit speed N, drilling pressure W, target rock compressive strength S, drilling power pump displacement Q, and drilling fluid density ρ, are used as model input parameters.

[0173] Obtain the actual drilling speed of the target drill bit at each predetermined measurement point, and use the actual drilling speed as a label for the corresponding wear degree prediction value.

[0174] Based on the above model input parameters and corresponding labels, training samples are constructed and divided into multiple parts. Preferably, the sample scores are greater than or equal to 5 parts.

[0175] Based on empirical data, a set of hyperparameters for the preset drilling speed prediction model are assigned values. It is important to understand that hyperparameters are the model hyperparameters used when building and training the drilling speed prediction model using a machine model. Taking the Gaussian process model used to build the drilling speed prediction model as an example, the model hyperparameters include the model's standard deviation.

[0176] A predetermined number of samples are selected as the test set, and the remaining samples are used as the training set. The predetermined drilling speed prediction model is trained using the training set to obtain the first hyperparameter-assigned drilling speed prediction model after training. The root mean square error of the prediction result of the first hyperparameter-assigned drilling speed prediction model after training is then calculated.

[0177] According to the predetermined selection strategy, a predetermined number of samples are selected from the multiple training samples as the test set, and the other samples are the training set. The above operation is repeated to obtain multiple trained first hyperparameter-assigned drilling speed prediction models corresponding to the set of hyperparameters, and the root mean square error of the prediction results of the trained first hyperparameter-assigned drilling speed prediction models is calculated.

[0178] The average root mean square error is taken from multiple hyperparameters to obtain the average root mean square error corresponding to the hyperparameter set. This average root mean square error is used as the prediction deviation value of the hyperparameter assignment drilling speed prediction model corresponding to the hyperparameter set.

[0179] The hyperparameters of the preset drilling speed prediction model are assigned values ​​a predetermined number of times to obtain multiple sets of hyperparameters. The above operation is repeated to obtain the prediction deviation values ​​of the drilling speed prediction model corresponding to the hyperparameter assignment values ​​of the multiple sets of hyperparameters.

[0180] The drilling speed prediction model corresponding to the minimum prediction deviation value is determined as the first drilling speed prediction model, and the hyperparameter combination corresponding to the first drilling speed prediction model is the optimal hyperparameter combination.

[0181] The hyperparameters of the drilling rate prediction model can be optimized using random search, grid search, or Bayesian search methods. Specifically, in this embodiment, the objective function for obtaining the optimal combination of hyperparameters can be expressed as:

[0182] θ = arg min(RMSE(f) θ -rop θ ))

[0183] Where θ represents the hyperparameter; f θ To assign the first hyperparameter of the drill speed prediction model trained with θ as the hyperparameter, rop θ This represents the actual drilling speed. RMSE(f) θ -rop θ ) represents the root mean square error (RMSE) between the predicted value and the corresponding actual drilling speed of the first hyperparameter-assigned drilling speed prediction model trained with θ as a hyperparameter. θ -rop θ )) indicates that the root mean square error between the predicted value and the corresponding actual drilling speed of the drill speed prediction model with the first hyperparameter assignment after training with θ as the hyperparameter is minimized.

[0184] The first drilling speed prediction model is trained using a training set that has not been divided into sub-sets, resulting in the trained drilling speed prediction model.

[0185] S905 is a composite model that obtains the drill bit wear prediction model and the trained drill speed prediction model.

[0186] After obtaining the trained drilling speed prediction model according to steps S901 to S904, the drill bit wear prediction model and the trained drilling speed prediction model are concatenated to obtain a composite model. In this composite model, the drill bit wear prediction value output by the drill bit wear prediction model is used as the input parameter value of the trained drilling speed prediction model. Heuristic algorithms are used to sparsely optimize the drill bit wear model. These heuristic algorithms include, but are not limited to, genetic algorithms, particle swarm optimization algorithms, and simulated annealing algorithms.

[0187] The fitness function, fitness, for sparse optimization of the drill bit wear model using a heuristic algorithm, is defined as follows:

[0188]

[0189] Through multiple iterations until predetermined iteration conditions are met, the particle or population with the highest fitness is selected as the result of fitting the drill bit wear model for this round, and the determined drill bit wear prediction model and the trained drill speed prediction model are saved. The predetermined iteration conditions include stopping iteration after a predetermined number of iterations is met. Specifically, the drill bit wear prediction value output by the drill bit wear prediction model is used as the input parameter value of the trained drill speed prediction model, and at least one optimization coefficient in the drill bit wear prediction model is optimized using the prediction accuracy of the trained drill speed prediction model.

[0190] S906, Repeat steps S904 to S905 until the drill rate prediction model trained in the composite model meets the prediction deviation requirements.

[0191] Specifically, repeat steps S904 to S905 until the prediction error of the trained drilling speed prediction model in the composite model is less than a predetermined error threshold, then stop the iteration and fitting. Store the corresponding target drill bit wear prediction model and the trained drilling speed prediction model in a computer storage medium for later use.

[0192] When there are multiple target drill bits and they are operating simultaneously in the same well, repeat the above steps to obtain the trained drilling speed prediction model and drill bit wear prediction model for each target drill bit.

[0193] A set of target drill bit wear prediction models M for multiple target drill bits wear It can be represented as:

[0194] M wear ={M wear,1 M wear,2 M wear,3 ,…,M wear,i …,M wear,K}

[0195] Among them, M wear,1 M wear,2 M wear,3 M wear,i M wear,K These represent the target drill bit wear prediction models for the 1st, 2nd, 3rd, i-th, and Kth target drill bits, respectively.

[0196] The set M of trained drilling rate prediction models for multiple target drill bits rop It can be represented as:

[0197] M rop ={M rop,1 M rop,2 M rop,3 ,…,M rop,i …,M rop,K}

[0198] Among them, M rop,1 M rop,2 M rop,3 M rop,i M rop,K These represent the assigned drill bit wear prediction models for the 1st, 2nd, 3rd, i-th, and Kth target drill bits, respectively, which meet the prediction deviation requirements.

[0199] Assign optimization coefficient adjustment values ​​to the models in each model set in the above model. For example, the optimization coefficient adjustment value for the i-th target drill bit can be expressed as w. i .

[0200] Therefore, the set M of target drill bit wear prediction models for multiple target drill bits is... wear It can be represented as:

[0201]

[0202] The set M of trained drilling rate prediction models for multiple target drill bits rop It can be represented as:

[0203]

[0204] When predicting the wear level of a target drill bit during drilling operations, an optimization coefficient adjustment value w can be set. i The initial value is w i = 1 / K.

[0205] S907, predicts the wear level of the target drill bit.

[0206] During the target drill bit operation, the optimization coefficient adjustment value w can be obtained using the following method. i :

[0207]

[0208] Among them, RMSE i Let be the prediction deviation value of the drilling speed prediction model for the i-th target drill bit that meets the prediction deviation requirement. Specifically, this deviation value is the root mean square error between the predicted value and the actual drilling speed of the drilling speed prediction model for the i-th target drill bit that meets the prediction deviation requirement.

[0209] Adjust the value w according to the optimization coefficient. i The target drill bit wear prediction model and the trained drilling speed prediction model for multiple target drill bits are adjusted to obtain the target drill bit wear prediction model for each target drill bit.

[0210] Figures 4(a) to 4(f)These are schematic diagrams showing the results of obtaining drill bit speed, drilling pressure, drilling power pump displacement, drill bit rotation speed, rock compressive strength, and drill bit wear degree according to the method of this embodiment. In these diagrams, "already drilled" represents the data corresponding to predetermined measurement points in the drilled section, "future" represents the measured data corresponding to the target prediction point, and "predicted" represents the predicted data corresponding to the target prediction point. Referring to Figure 4(a), the deviation between the actual drill bit speed data and the predicted drill bit speed data is small; that is, according to the method of this application, combined with... Figures 4(b) to 4(f) The small deviation between the drill bit speed prediction results obtained from the data and the actual drilling speed indicates that the accuracy of the drilling speed prediction model obtained through the application method is high.

[0211] In this embodiment, a drill bit wear prediction model is pre-constructed based on empirical data, and a drilling speed prediction model is constructed based on a machine learning model. Based on the state parameters of the target drill bit during the drilling process, and with the goal of minimizing the root mean square error of the drilling speed prediction model, the drill bit wear prediction model and the drilling speed prediction model are trained alternately. Finally, a drill bit wear prediction model that can accurately predict drill bit wear is obtained. Furthermore, multi-target drill bits and their corresponding drill bit wear prediction models are obtained. This embodiment effectively solves the problems in related technologies, such as the lack of a drill bit wear prediction model that can accurately predict drill bit wear, and the inability to accurately predict drill bit wear. It achieves accurate prediction of the wear of the drill bit to be drilled, which is helpful for pre-drilling drill bit wear prediction and monitoring and predicting wear while drilling, effectively improving the digital and information-based control level of oil drilling.

[0212] Figure 5 This is a schematic diagram of a drill bit wear prediction model training device provided in an embodiment of the present invention. Figure 5 As shown, the drill bit wear prediction model training device 50 provided in this embodiment includes: a first acquisition module 501, a construction module 502, a second acquisition module 503, a model optimization module 504, and a determination module 505.

[0213] The system comprises the following modules: a first acquisition module 501, used to acquire a pre-constructed drill bit wear prediction model, which includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit; a construction module 502, used to assign values ​​to at least one optimization coefficient and construct training samples for training the drilling speed prediction model based on the assigned drill bit wear prediction model, wherein each training sample includes at least a predicted drill bit wear value and is labeled with the actual drilling speed; a second acquisition module 503, used to train the pre-constructed drilling speed prediction model using the training samples to obtain the trained drilling speed prediction model; and a model optimization module. Block 504 is used to respond to the fact that the trained drilling speed prediction model does not meet the prediction deviation requirement, to perform the steps of reassigning at least one optimization coefficient, and to perform the steps of constructing training samples for training the drilling speed prediction model based on the reassigned drill bit wear prediction model until the corresponding trained drilling speed prediction model is obtained, until the trained drilling speed prediction model meets the prediction deviation requirement; the determination module 505 is used to determine the reassigned drill bit wear prediction model corresponding to the drilling speed prediction model that meets the prediction deviation requirement as the target drill bit wear prediction model, and the target drill bit wear prediction model is used to predict the wear degree of the target drill bit.

[0214] The drill bit wear prediction model training device provided in this embodiment can execute... Figure 2 The technical solutions of the method embodiments shown are similar in principle and in effect, and will not be described again here.

[0215] Optionally, the construction module 502 is specifically used to: assign values ​​to at least one optimization coefficient to obtain a drill bit wear prediction model after assignment; determine the drill bit wear prediction values ​​at multiple predetermined measurement points based on the values ​​of at least one state parameter of the target drill bit at multiple predetermined measurement points and the drill bit wear prediction model after assignment; and construct training samples for the drill bit wear prediction values ​​and corresponding actual drilling speeds at each predetermined measurement point.

[0216] Optionally, the second acquisition module 503 is specifically used for: assigning values ​​to at least one hyperparameter of a preset drilling speed prediction model a predetermined number of times to obtain multiple hyperparameter-assigned drilling speed prediction models corresponding to multiple sets of hyperparameter assignments; for each hyperparameter-assigned drilling speed prediction model among the multiple hyperparameter-assigned drilling speed prediction models: training the hyperparameter-assigned drilling speed prediction model using training samples to obtain a trained first hyperparameter-assigned drilling speed prediction model, and determining the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model; determining the hyperparameter-assigned drilling speed prediction model corresponding to the trained first hyperparameter-assigned drilling speed prediction model with the smallest prediction deviation value as the first drilling speed prediction model; and training the first drilling speed prediction model using training samples to obtain a trained drilling speed prediction model.

[0217] Optionally, the second acquisition module 503 is further specifically used for: determining multiple sets of sample sets and test sets from the training samples, wherein the proportion of the number of samples in each set of sample sets and test sets meets a predetermined proportion requirement; for each set of sample sets and test sets: using the sample set to train a hyperparameter-assigned drilling speed prediction model to obtain a trained first hyperparameter-assigned drilling speed prediction model, and obtaining the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model based on the test set; obtaining the average prediction deviation of the multiple sets of trained first hyperparameter-assigned drilling speed prediction models corresponding to the multiple sets of sample sets and test sets respectively; and determining the average prediction deviation as the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model.

[0218] Optionally, the model optimization module 504 is specifically used to: determine the prediction error of the trained drilling speed prediction model corresponding to the different assigned optimization coefficients after reassigning at least one optimization coefficient; determine the prediction error of the trained drilling speed prediction model as the assignment changes; and determine whether the trained drilling speed prediction model meets the prediction deviation requirements based on the prediction error and a preset error threshold.

[0219] Optionally, the determining module 505 is specifically used to: if there are multiple target drill bits, and multiple target drill bits are drilling simultaneously in the same target well, obtain multiple prediction deviation values ​​of the drilling speed prediction model that meets the prediction deviation requirements corresponding to the multiple target drill bits respectively; determine the superposition value of the multiple prediction deviation values; for each of the multiple target drill bits: based on the prediction deviation value of the target drill bit and the superposition value of the multiple prediction deviation values, obtain the optimization coefficient adjustment value of the target drill bit; adjust the assignment value of at least one optimization coefficient in the drill bit wear degree prediction model corresponding to the drilling speed prediction model that meets the prediction deviation requirements based on the optimization coefficient adjustment value, and obtain the target drill bit wear degree prediction model.

[0220] Optionally, the drill bit wear prediction device further includes a fourth acquisition module, specifically used for: acquiring multiple target acoustic transit times at multiple predetermined measurement points; and determining the target rock compressive strength at multiple predetermined measurement points based on the multiple target acoustic transit times and the pre-constructed correspondence between rock compressive strength and acoustic transit times.

[0221] Figure 6 This is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present invention. Figure 6 As shown, the electronic device 60 includes: a processor 602, a memory 601 communicatively connected to the processor, and a transceiver 603.

[0222] The system includes a memory that stores computer-executed instructions; a transceiver for sending and receiving data; and a processor that executes the computer-executed instructions stored in the memory to implement the drill bit wear prediction model training or drill bit wear prediction method provided in the corresponding embodiment.

[0223] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0224] The modules described as separate components may or may not be physically separate. Similarly, the components shown as modules may or may not be physical modules; they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0225] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the drill bit wear prediction model training method or drill bit wear prediction method provided in any embodiment of this invention.

[0226] This invention also provides a computer-readable storage medium storing a computer program thereon, which is executed by a processor to implement the drill bit wear prediction model training method or drill bit wear prediction method provided in any embodiment of this invention.

[0227] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The integrated module can be implemented in hardware or in the form of hardware plus software functional modules.

[0228] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0229] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0230] Furthermore, although the operations are described in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, although several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0231] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A training method for a drill bit wear prediction model, characterized in that, include: A pre-constructed drill bit wear prediction model is obtained, wherein the pre-constructed drill bit wear prediction model includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit; the optimization coefficient includes a first model coefficient corresponding to the mudstone layer, a drill pressure coefficient of the drill pressure, a drill speed coefficient of the target drill bit, and a second model coefficient; The at least one optimization coefficient is assigned a value, and a training sample is constructed based on the drill bit wear prediction model after the assignment to train the drilling speed prediction model. Each training sample includes at least the drill bit wear prediction value and is marked with the actual drilling speed. The preset drilling speed prediction model is trained using training samples to obtain the trained drilling speed prediction model. In response to the trained drilling speed prediction model not meeting the prediction deviation requirement, the at least one optimization coefficient is reassigned, and the step of constructing training samples for training the drilling speed prediction model based on the reassigned drill bit wear prediction model is executed until the corresponding trained drilling speed prediction model is obtained, until the trained drilling speed prediction model meets the prediction deviation requirement. The drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirement is determined as the target drill bit wear prediction model. The target drill bit wear prediction model is used to predict the wear degree of the target drill bit.

2. The method according to claim 1, characterized in that, Assign values ​​to at least one optimization coefficient, and construct training samples for training the drilling speed prediction model based on the drill bit wear prediction model after the assignment, including: The at least one optimization coefficient is assigned a value to obtain a drill bit wear prediction model after the assignment. Based on the value of at least one state parameter of the target drill bit at multiple predetermined measurement points, and the drill bit wear degree prediction model after assignment, the predicted value of the drill bit wear degree at multiple predetermined measurement points is determined. Training samples were constructed based on the predicted values ​​of drill bit wear at each predetermined measurement point and the corresponding actual drilling speed.

3. The method according to claim 1, characterized in that, The step of training a preset drilling speed prediction model using training samples to obtain a trained drilling speed prediction model includes: At least one hyperparameter of the preset drilling speed prediction model is assigned a predetermined number of times to obtain multiple hyperparameter assignment drilling speed prediction models corresponding to multiple sets of hyperparameter assignments. For each of the multiple hyperparameter-assigned drilling speed prediction models: the hyperparameter-assigned drilling speed prediction model is trained using the training samples to obtain the first hyperparameter-assigned drilling speed prediction model after training, and the prediction deviation value of the first hyperparameter-assigned drilling speed prediction model after training is determined. The hyperparameter-assigned drilling speed prediction model corresponding to the training model with the smallest deviation from the prediction value is determined as the first drilling speed prediction model. The first drilling speed prediction model is trained using the training samples to obtain the trained drilling speed prediction model.

4. The method according to claim 3, characterized in that, Determining the prediction bias of the first hyperparameter-assigned drilling speed prediction model after training includes: Multiple sample sets and test sets are determined from the training samples, and the proportion of the number of samples in each sample set and test set meets the predetermined proportion of the number of samples. For each of the multiple sample sets and test sets: use the sample set to train the hyperparameter-assigned drilling speed prediction model to obtain the trained first hyperparameter-assigned drilling speed prediction model, and obtain the prediction deviation value of the trained first hyperparameter-assigned drilling speed prediction model based on the test set. Obtain the average prediction deviation of multiple trained drilling speed prediction models with first hyperparameter assignments that correspond to the multiple sample sets and test sets, respectively; The average prediction deviation is determined as the prediction deviation value of the first hyperparameter assignment drilling speed prediction model after training.

5. The method according to claim 1, characterized in that, Determine whether the trained drilling speed prediction model meets the prediction bias requirements, including: The prediction error of the trained drilling speed prediction model corresponding to the at least one optimization coefficient after reassignment is determined, and the prediction error is determined according to the prediction error and the preset error threshold.

6. The method according to claim 1, characterized in that, The target drill bit is multiple, and multiple target drill bits simultaneously perform drilling operations in the same target well. The step of determining the drill bit wear degree prediction model after assigning values ​​to the drilling speed prediction model that meets the prediction deviation requirements is the target drill bit wear degree prediction model, including: Obtain multiple prediction deviation values ​​for the drilling speed prediction models that meet the prediction deviation requirements, corresponding to the multiple target drill bits respectively; Determine the sum of the multiple prediction deviation values; For each of the plurality of target drill bits: based on the predicted deviation value of the target drill bit and the superposition value of the plurality of predicted deviation values, the optimization coefficient adjustment value of the target drill bit is obtained. Based on the optimization coefficient adjustment value, the assignment of at least one optimization coefficient in the drill bit wear degree prediction model corresponding to the drill speed prediction model that meets the predicted deviation requirements is adjusted to obtain the target drill bit wear degree prediction model.

7. The method according to claim 2, characterized in that, The state parameters include the compressive strength of the target rock. Before determining the predicted values ​​of drill bit wear at multiple predetermined measurement points, the method further includes: Acquire the time differences of multiple target acoustic waves at the multiple predetermined measurement points; Based on the multiple target acoustic transit times and the pre-constructed correspondence between rock compressive strength and acoustic transit times, the target rock compressive strength at the multiple predetermined measurement points is determined respectively.

8. A method for predicting drill bit wear, characterized in that, include: Obtain the value of at least one state parameter of the target drill bit at the target prediction point; Input the value of at least one state parameter into the target drill bit wear prediction model; The target drill bit wear prediction model is used to calculate the predicted value of the drill bit wear at the target prediction point. The target drill bit wear prediction model is trained using the drill bit wear prediction model training method as described in any one of claims 1 to 7.

9. A training device for a drill bit wear prediction model, characterized in that, The device includes: The first acquisition module is used to acquire a pre-constructed drill bit wear prediction model, which includes at least one state parameter and at least one optimization coefficient during the drilling process of the target drill bit; the optimization coefficient includes a first model coefficient corresponding to the mudstone layer, a drill pressure coefficient, a drilling speed coefficient of the target drill bit, and a second model coefficient. A construction module is used to assign values ​​to at least one optimization coefficient and construct training samples for training the drilling speed prediction model based on the assigned drill bit wear prediction model. Each training sample includes at least the predicted value of drill bit wear and is marked with the actual drilling speed. The second acquisition module is used to train the preset drilling speed prediction model using training samples to obtain the trained drilling speed prediction model. The model optimization module is used to respond to the fact that the trained drilling speed prediction model does not meet the prediction deviation requirement, and to perform the steps of reassigning the at least one optimization coefficient, and constructing training samples for training the drilling speed prediction model based on the reassigned drill bit wear prediction model until the corresponding trained drilling speed prediction model is obtained, until the trained drilling speed prediction model meets the prediction deviation requirement. The determination module is used to determine the drill bit wear prediction model corresponding to the drill speed prediction model that meets the prediction deviation requirements as the target drill bit wear prediction model. The target drill bit wear prediction model is used to predict the wear degree of the target drill bit.

10. An electronic device, comprising: A processor, a memory communicatively connected to the processor, and a transceiver; The memory stores computer-executed instructions; The transceiver is used for sending and receiving data; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 8.

Images