Drill bit design method and device and electronic equipment
By building a proxy model for predicting drill bit wear degree, optimizing the drill bit design process, the problems of high computational costs and long cycles in drill bit design are solved, and more efficient drill bit design and longer drill bit life are achieved.
Patent Information
- Application Number
- CN202510159328.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-02-13
- Publication Date
- 2025-07-04
AI Technical Summary
In the prior art, drill bit design requires strong nonlinear calculations, resulting in high calculation costs and long design cycles.
The drill bit wear degree prediction agent model is adopted, and the model is constructed through the nonlinear support vector regression method based on the initial and wear point cloud data of the drill bit sample data, and the point cloud data before the drill bit enters the well is optimized to reduce complex nonlinear calculations.
Reduces computing resource requirements and costs, accelerates the drill bit design process, improves design efficiency and drilling efficiency, and extends the drill bit life.
Smart Images
Figure CN120257500A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of oilfield drilling, and particularly to a bit design method, device, and electronic device. Background Art
[0002] With the continuous deepening of oil and gas field exploration and development work, the encountered formation structures are becoming more and more complex, and the drilling difficulty is increasing. In order to save the comprehensive drilling cost, higher requirements are put forward for the quality of the bit. In the prior art, generally, a bit-rock mechanics interaction model or a bit hydraulic structure model is used to design the structure of the bit.
[0003] However, when these two models are used for auxiliary calculations, strong non-linear calculations are required, resulting in high calculation costs and long design cycles. Summary of the Invention
[0004] The present invention provides a bit design method, device, and electronic device to solve the defect in the prior art that strong non-linear calculations are required for bit structure design, resulting in high calculation costs and long design cycles, and to reduce complex non-linear calculations, thereby reducing the demand and cost of computing resources, accelerating the bit design process, and reducing the design cycle.
[0005] According to one aspect of the present invention, there is provided a bit design method, including: Obtaining bit information related to a target bit, where the bit information at least includes first point cloud data before the bit enters the well; Optimizing the first point cloud data according to the first point cloud data and a bit wear prediction surrogate model to obtain optimized first point cloud data, and the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit; Wherein, the bit wear prediction surrogate model is at least trained based on the initial point cloud data before each bit sample data in the bit sample dataset enters the well and the worn point cloud data after the bit exits the well.
[0006] In addition, for the bit design method according to one aspect of the present invention, obtaining bit information related to a target bit includes: Determining the diameter of the target bit based on the target formation type encountered and the well opening stage; Obtaining bit information for drilling the target formation type based on the diameter of the target bit.
[0007] In addition, for the bit design method according to one aspect of the present invention, optimizing the first point cloud data according to the first point cloud data and a bit wear prediction surrogate model to obtain optimized first point cloud data, and the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit, includes: Input the first point cloud data into the drill bit wear degree prediction proxy model to obtain the predicted second point cloud data after the drill bit exits the well; Based on the first point cloud data and the predicted second point cloud data, obtain the predicted value of the first drill bit wear depth value corresponding to the drill bit information; Taking the predicted value of the first drill bit wear depth value as the optimization objective, use the gradient descent algorithm to optimize the first point cloud data; Iteratively update the predicted value of the first drill bit wear depth value. When the difference between the predicted values of the first drill bit wear depth value obtained in two consecutive iterations is less than the preset convergence threshold, the corresponding first point cloud data is the optimized first point cloud data, and the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit.
[0008] In addition, according to the drill bit design method of one aspect of the present invention, the drill bit wear degree prediction proxy model is trained according to the following steps: Divide the drill bit sample data set into a training data set and a verification data set. The training data set and the verification data set respectively contain several drill bit sample data; Based on the training data set, using the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, construct and train the drill bit wear degree prediction proxy model by using the non-linear support vector regression method; Use the verification data set to evaluate the prediction accuracy of the drill bit wear degree prediction proxy model; When the prediction accuracy meets the predetermined conditions, complete the optimization of the drill bit wear degree prediction proxy model.
[0009] In addition, according to the drill bit design method of one aspect of the present invention, based on the training data set, using the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, training the drill bit wear degree prediction proxy model by using the non-linear support vector regression method includes: Construct an optimization problem, and the optimization problem includes an optimization objective function and constraint conditions; The optimization objective function is expressed as: The constraint conditions are expressed as: Wherein, 、 respectively represent the feature vectors of the i-th training sample and the j-th training sample in the training set, including the initial point cloud data; is the true target value of the i-th training sample in the training set, including the worn point cloud data; is the parameter of the insensitive loss function, is the penalty parameter, is the Lagrange multiplier of the i-th training sample, is the Lagrange multiplier of the j-th training sample, is the kernel function; Solve the optimization problem using the training data set to obtain the optimal Lagrange multiplier and the optimal bias term; Based on the optimization problem, the optimal Lagrange multiplier and the optimal bias term, construct a prediction surrogate model for the drill bit wear degree.
[0010] In addition, according to the drill bit design method of an aspect of the present invention, based on the optimization problem, the optimal Lagrange multiplier and the optimal bias term, construct a prediction surrogate model for the drill bit wear degree, including: Based on the optimization problem, the optimal Lagrange multiplier and the optimal bias term, construct a regression function, and the regression function is a numerical representation of the prediction surrogate model for the drill bit wear degree; Among them, the regression function is expressed as: In the formula, is the predicted worn point cloud data output by the prediction surrogate model for the drill bit wear degree; b is the optimal bias term, is the kernel function, is the optimal Lagrange multiplier of the i-th training sample.
[0011] In addition, according to the drill bit design method of an aspect of the present invention, use the validation data set to evaluate the prediction accuracy of the prediction surrogate model for the drill bit wear degree, including: Input the initial point cloud data of each drill bit sample data in the validation data set into the prediction surrogate model for the drill bit wear degree to obtain the predicted worn point cloud data output by the prediction surrogate model for the drill bit wear degree; Based on the initial point cloud data, the predicted worn point cloud data, and the worn point cloud data of each drill bit sample data, obtain the predicted value and the actual value of the drill bit wear depth value corresponding to each drill bit sample data; Evaluate the prediction accuracy of the prediction surrogate model for the drill bit wear degree based on the predicted value and the actual value of the drill bit wear depth value of each drill bit sample data.
[0012] In addition, according to the drill bit design method of an aspect of the present invention, when the prediction accuracy meets the predetermined conditions, complete the optimization of the prediction surrogate model for the drill bit wear degree, including: When the fitting degree between the predicted value of the drill bit wear depth value corresponding to each drill bit sample data and the actual value of the drill bit wear depth value is greater than 80%, it is determined that the prediction accuracy meets the predetermined conditions, and the optimization of the prediction surrogate model for the drill bit wear degree is completed.
[0013] According to another aspect of the present invention, there is provided a drill bit design device, including: An acquisition module is configured to acquire drill bit information related to a target drill bit, wherein the drill bit information at least includes first point cloud data before the drill bit enters the well; an optimization module configured to optimize the first point cloud data according to the first point cloud data and the drill wear degree prediction agent model, and obtain the optimized first point cloud data, wherein the drill structure corresponding to the optimized first point cloud data is the drill structure of the target drill bit; The drill bit wear degree prediction agent model is at least trained based on the initial point cloud data before the drill bit enters the well and the wear point cloud data after the drill bit leaves the well of each drill bit sample data in the drill bit sample data set.
[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned drill bit design method when executing the computer program.
[0015] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the drill bit design method as described above is implemented.
[0016] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the drill bit design method as described above is implemented.
[0017] The drill bit design method, device and electronic device provided by the present invention optimize the first point cloud data before the drill bit enters the well through the first point cloud data before the drill bit enters the well and the drill bit wear degree prediction agent model, and can make targeted adjustments to the drill bit structure according to the actual drill bit wear condition and usage environment. Compared with the traditional drill bit-rock mechanics model or drill bit hydraulic structure model, this method reduces complex nonlinear calculations and reduces the demand and cost of computing resources, thereby accelerating the drill bit design process and shortening the design cycle. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0019] Figure 1 It is one of the flow charts of the drill bit design method provided by the present invention.
[0020] Figure 2 This is the second flow chart of the drill bit design method provided by the present invention.
[0021] Figure 3 It is the third schematic flow chart of the drill bit design method provided by the present invention.
[0022] Figure 4 It is the schematic structural diagram of the drill bit design device provided by the present invention.
[0023] Figure 5 It is the schematic structural diagram of the electronic device provided by the present invention. Detailed implementation manners
[0024] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without making creative efforts shall fall within the scope of protection of the present invention.
[0025] Figure 1 It is one of the schematic flow charts of the drill bit design method provided by the present invention. As Figure 1 shown, the method includes the following: Step 101, obtain drill bit information related to the target drill bit, where the drill bit information at least includes the first point cloud data before the drill bit enters the well.
[0026] In an embodiment of the present invention, the target drill bit refers to a drill bit designed for a specific drilling task, which can meet specific geological conditions. The drill bit information refers to various data and parameters related to the drill bit, at least including the first point cloud data before the drill bit enters the well. The first point cloud data refers to the initial geometric shape and surface features of the drill bit obtained by scanning or measuring techniques before the drill bit starts drilling operations, representing the state of the drill bit when it has not been used and has not been worn.
[0027] Exemplarily, the drill bit information may further include the second point cloud data after entering the well. The second point cloud data refers to the three-dimensional geometric shape and surface features of the drill bit surface obtained by scanning or measuring techniques after the drill bit is actually used. It is used to analyze the wear condition, damage condition and other changes of the drill bit after actual use.
[0028] Exemplarily, the bit information may further include bit identity information, usage information, and wear information. Among them, the identity information refers to information such as the manufacturer of the bit, the cobalt removal depth of the teeth given by the manufacturer, the tooth surface shape, and the recommended drilling parameters. The usage information of the bit refers to the drilling parameters during bit use, the formation parameters of the drilled formation, and logging and well logging data such as the mechanical drilling rate. The wear information of the bit refers to the surface point cloud data (i.e., the aforementioned first point cloud data) and the point cloud three-dimensional model before the bit enters the well, the surface point cloud data (i.e., the aforementioned second point cloud data) and the point cloud three-dimensional model after the bit exits the well, and the wear parameters and characteristics obtained by comparing the point cloud data before and after the bit enters the well, including the position, area, depth, and wear type of bit wear.
[0029] Further, obtaining bit information related to the target bit includes: Determining the diameter of the target bit based on the target formation type encountered and the well opening stage; Obtaining bit information for drilling the target formation type based on the diameter of the target bit.
[0030] Specifically, "drilling encounter" refers to the situation where the drill bit encounters a specific target layer during the drilling process. The target formation type refers to the formation type where the drill bit is damaged during drilling in a certain block, and it is qualitatively determined based on the understanding of logging and mud logging data. The open hole stage is the specific stage during the drilling process. The target formation type can be one or a combination of two or more of hard and dense formations, hard and highly abrasive formations, and hard and dense mudstone formations. A hard and dense formation means that the rock-forming minerals in the formation rock have high hardness, fine particles, a general particle size of 0.01 - 0.2 mm, high quartz content, siliceous cementation, a dense structure, a large bonding force between particles, an indentation hardness of up to 5000 MPa, and a uniaxial compressive strength of up to 150 MPa. A hard and highly abrasive formation means that the rock-forming minerals quartz and feldspar in the formation rock have high content, relatively coarse particles, a cementing material of calcareous or argillaceous, a compressive strength between 130 - 160 MPa, and a drillability of the rock of 8 - 10 levels. A hard and dense mudstone formation means that in addition to containing montmorillonite, the mudstone forming the formation also contains a large amount of quartz components (generally 40% - 60%). Among the components of the mudstone, the particles are extremely fine, mostly with a particle size of 200 mesh, with a dense structure and strong elastoplasticity. When it is necessary to optimize the drill bit at the specific open hole stage of the target formation type, according to the target formation type and the open hole stage encountered by the drill bit during the drilling process, select the applicable wellbore structure design specification. According to this wellbore structure design specification, determine the diameter of the target drill bit. Subsequently, collect drill bit information that is the same as or similar to the diameter of the target drill bit and has encountered the target formation type during the drilling process. The above technical solution for obtaining the relevant drill bit information of the target drill bit can significantly improve the pertinence and adaptability of drill bit selection, optimize drill bit design, enhance performance, reduce drilling costs and risks, enhance the safety and reliability of drilling operations, and support data-driven decision-making. These technical effects are of great significance for improving drilling efficiency, reducing costs, and ensuring drilling safety.
[0031] Step 102: Optimize the first point cloud data according to the first point cloud data and the drill bit wear prediction surrogate model to obtain the optimized first point cloud data, and the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit; Among them, the drill bit wear prediction surrogate model is at least trained based on the initial point cloud data of each drill bit sample data in the drill bit sample dataset before the drill bit enters the well and the worn point cloud data after the drill bit exits the well.
[0032] In one embodiment of the present invention, the initial point cloud data before the drill bit enters the well is the same as the first point cloud data, which refers to the three-dimensional geometric shape data and surface features of the drill bit when it is not in use, representing the state of the drill bit before use and without wear. The worn point cloud data after the drill bit exits the well is the same as the second point cloud data, which refers to the three-dimensional geometric data and surface features of the surface wear of the drill bit after actual use. The drill bit wear degree prediction proxy model can be, but is not limited to, a model constructed based on machine learning or numerical simulation technology for predicting the wear condition of the drill bit during actual use. By analyzing the initial point cloud data of the drill bit before entering the well and the worn point cloud data of the drill bit after exiting the well, the model establishes a mathematical model of the wear law. Optimizing the first point cloud data using the first point cloud data and the drill bit wear degree prediction proxy model can adjust the first point cloud data of the drill bit according to the wear law during actual use of the drill bit, and then use the drill bit structure corresponding to the optimized first point cloud data as the drill bit structure of the target drill bit. This optimization can reduce the wear rate of the drill bit during use, extend the service life of the drill bit, and improve the drilling efficiency. For example, by optimizing the shape and distribution of the cutting teeth of the drill bit, it can wear more evenly when facing different rock abrasiveness.
[0033] In addition, the training data of the drill bit wear degree prediction proxy model not only includes the initial point cloud data before the drill bit enters the well and the worn point cloud data after exiting the well, but also the following data can be combined to further improve the prediction accuracy and adaptability: Drilling parameters: such as weight on bit, rotary speed, displacement, use of speed-up tools, etc. These parameters directly affect the wear condition of the drill bit. Research shows that the relative wear rate of the cutting teeth is proportional to a certain power of parameters such as weight on bit and cutting speed.
[0034] Formation parameters: such as lithology parameters, rock strength, thickness of the drilled formation, etc. Lithology parameters include basic strength parameters such as the elastic modulus, Poisson's ratio, and fracture toughness of the whole rock, and also include the difference in characteristic mineral parameters of the target formation designed for the drill bit. These parameters reflect the wear risks faced by the drill bit under different geological conditions. For example, the increase in rock abrasiveness will lead to an increase in the relative wear rate of the cutting teeth.
[0035] Exemplarily, when using drilling parameters and formation parameters as the training data of the drill bit wear degree prediction proxy model, the drilling parameters and formation parameters are standardized to have zero mean and unit variance to reduce the influence of data distribution differences.
[0036] By incorporating the above parameters into the training of the drill bit wear degree prediction proxy model, the drill bit wear degree prediction proxy model can more accurately predict the wear condition of the drill bit under different geological conditions and drilling parameters, thereby optimizing the drill bit design, reducing the wear rate, and extending the drill bit life.
[0037] Each bit sample data in the above bit sample data set may include bit information related to the target bit. That is, the acquisition method of the bit sample data set may also refer to the acquisition method of the bit information related to the target bit.
[0038] In summary, according to the technical solution provided by the embodiment of the present invention, by optimizing the first point cloud data before the bit enters the well through the first point cloud data before the bit enters the well and the bit wear degree prediction proxy model, it is possible to make targeted adjustments to the bit structure according to the actual bit wear situation and use environment. Compared with the traditional bit-rock mechanics action model or bit hydraulic structure model, this method reduces complex non-linear calculations, reduces the demand and cost of computing resources, thereby accelerating the bit design process and reducing the design cycle. Further, by incorporating drilling parameters and formation parameters into model training, this method can more accurately predict the wear situation of the bit under different geological conditions and drilling parameters, thereby optimizing the bit design, reducing the wear rate, and extending the bit life.
[0039] Figure 2 It is the second flow schematic diagram of the bit design method provided by the present invention.
[0040] As Figure 2 shown, the bit wear degree prediction proxy model is trained according to the following steps: Step S201: Divide the bit sample data set into a training data set and a validation data set, and the training data set and the validation data set respectively include several bit sample data.
[0041] In an embodiment of the present invention, each sample data in the bit sample data set is randomly assigned to the training data set and the validation data set, and the training data set and the validation data set respectively include several bit sample data. Exemplarily, it can be divided according to a certain ratio, such as 70% for training and 30% for validation.
[0042] Step S202: Based on the training data set, with the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, use the non-linear support vector regression method to construct and train the bit wear degree prediction proxy model.
[0043] In an embodiment of the present invention, with the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, input the initial point cloud data and the worn point cloud data in the training data set into the non-linear SVR model, and by adjusting the model parameters, make the model accurately predict the worn point cloud data, thereby completing the training of the bit wear degree prediction proxy model.
[0044] Further, based on the training dataset, using the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, a non-linear support vector regression method is used to train the drill bit wear degree prediction surrogate model, including: Construct an optimization problem, which includes an optimization objective function and constraints; The optimization objective function is expressed as: The constraints are expressed as: Where, and respectively represent the feature vectors of the i-th training sample and the j-th training sample in the training set, including the initial point cloud data; is the true target value of the i-th training sample in the training set, including the worn point cloud data; is the parameter of the insensitive loss function, is the penalty parameter, is the Lagrange multiplier of the i-th training sample, is the Lagrange multiplier of the j-th training sample, is the kernel function, and n is the total number of training samples.
[0045] Exemplarily, the kernel function formula is: Where, is used to adjust the scaling degree of the inner product between the feature vectors of each training sample in the training set, and its value range is 0 to 1; is a constant term, which plays a biasing role in the kernel function, causing the value of the kernel function to be translated up and down as a whole, and its value range is 0 to 2; represents the degree of the polynomial, and its value is an integer greater than 2.
[0046] The above-mentioned feature vector refers to a vector representation extracted from the point cloud data and used to describe the data distribution and change direction.
[0047] Exemplarily, it is also possible to use the initial point cloud data, drilling parameters, and formation parameters in each drill bit sample data in the training set as the independent variables and the worn point cloud data as the dependent variable, and use a non-linear support vector regression method to train the drill bit wear degree prediction surrogate model. Furthermore, the feature vectors of the above-mentioned training samples can also include the initial point cloud data, drilling parameters, and formation parameters.
[0048] Solve the optimization problem using the training dataset to obtain the optimal Lagrange multiplier and the optimal bias term; Based on the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms, a prediction surrogate model for the bit wear degree is constructed.
[0049] Specifically, by using the training set data to solve the above optimization problem, the optimal Lagrange multipliers can be obtained. Using the obtained optimal Lagrange multipliers, the optimal bias terms are obtained. These parameters are used to construct a prediction surrogate model for the bit wear degree, and this model can be used to predict the wear condition of new bit samples.
[0050] Exemplarily, by using the training data set to solve the optimization problem, the optimal Lagrange multipliers that satisfy the constraint conditions are found. The training samples corresponding to the optimal Lagrange multipliers are used as support vectors and substituted into the regression function to obtain the optimal bias terms.
[0051] The formula of the regression function is as follows: where b is the optimal bias term, is the predicted wear point cloud data corresponding to the training sample , i is the index variable, is any training sample in the training sample set, and n is the total number of training samples.
[0052] In an embodiment of the present invention, to solve the above optimization problem, it can be solved by various methods, including but not limited to the sequential minimal optimization (SMO) algorithm, the interior point method, etc., to find the optimal Lagrange multipliers and the optimal bias terms.
[0053] Furthermore, based on the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms, a regression function is constructed, and the regression function is the numerical representation of the prediction surrogate model for the bit wear degree.
[0054] Here, the regression function is constructed by using the kernel function of the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms, and the regression function is the numerical representation of the prediction surrogate model for the bit wear degree.
[0055] Among them, the regression function is expressed as: In the formula, is the predicted wear point cloud data output by the prediction surrogate model for the bit wear degree; b is the optimal bias term, is the kernel function, is the optimal Lagrange multiplier of the i-th training sample, and n is the total number of training samples.
[0056] Step 203: Evaluate the prediction accuracy of the drill bit wear degree prediction surrogate model using the validation dataset.
[0057] In an embodiment of the present invention, a part of the data that did not participate in model training (i.e., the validation dataset) is used to test the prediction performance of the drill bit wear degree prediction surrogate model. The purpose of doing this is to test the model's prediction ability for new data and ensure that the model is not overfitted, that is, it performs well on the training data but poorly on new data. Exemplarily, the methods for evaluating model fitness may include but are not limited to using statistical metrics such as mean squared error (MSE), root mean squared error (RMSE), or mean absolute error (MAE) to measure the difference between the model's predicted values and the actual values. These metrics can provide a quantitative assessment of the model's prediction accuracy.
[0058] Furthermore, input the initial point cloud data of each drill bit sample data in the validation dataset into the drill bit wear degree prediction surrogate model to obtain the predicted worn point cloud data output by the drill bit wear degree prediction surrogate model; Based on the initial point cloud data, predicted worn point cloud data, and worn point cloud data of each drill bit sample data, obtain the predicted value and actual value of the drill bit wear depth value corresponding to each drill bit sample data; Evaluate the prediction accuracy of the drill bit wear degree prediction surrogate model based on the predicted value and actual value of the drill bit wear depth value of each drill bit sample data.
[0059] Specifically, substitute the initial point cloud data of each drill bit sample data in the validation dataset (i.e., the point cloud data before the drill bit enters the well) into the drill bit wear degree prediction surrogate model to obtain the predicted worn point cloud data after the drill bit exits the well output by the model. For each drill bit sample data, compare its initial point cloud data with the predicted worn point cloud data after the drill bit exits the well corresponding to it. This comparison usually involves calculating the difference between the two point clouds, and these differences reflect the wear situation predicted by the model. According to the comparison result of the initial point cloud data and the predicted worn point cloud data, solve the predicted value of the drill bit wear depth value corresponding to each drill bit sample data. This predicted value represents the model's estimate of the drill bit wear degree. Similarly, compare the initial point cloud data of each drill bit sample data with the worn point cloud data after the drill bit exits the well corresponding to it (i.e., the actually measured worn point cloud data). This step is to obtain the actual value of the wear depth value. According to the comparison result of the initial point cloud data and the actual worn point cloud data, solve the actual value of the drill bit wear depth value corresponding to each drill bit sample data by the same or similar method. This actual value represents the true wear degree of the drill bit. Finally, compare the predicted value of the drill bit wear depth value corresponding to each drill bit sample data with its corresponding actual value. Evaluate the prediction accuracy of the drill bit wear degree prediction surrogate model by calculating error metrics (such as mean squared error, mean absolute error, etc.).
[0060] Exemplarily, the drill bit wear depth value can be calculated by the following formula: where x is the initial point cloud data of the drill bit sample data, is the predicted wear point cloud data of the drill bit sample data.
[0061] Step 204, when the prediction accuracy meets the predetermined condition, complete the optimization of the drill bit wear degree prediction proxy model.
[0062] In an embodiment of the present invention, after obtaining the predicted value of the drill bit wear depth value corresponding to each drill bit sample data and the actual value of the drill bit wear depth value, compare the predicted value of the drill bit wear depth value corresponding to each drill bit sample data with the actual value of the drill bit wear depth value. When each comparison result meets the set condition, it is determined that the prediction accuracy of the drill bit wear degree prediction proxy model meets the predetermined condition, thereby completing the construction of the drill bit wear degree prediction proxy model.
[0063] Further, when the prediction accuracy meets the predetermined condition, completing the optimization of the drill bit wear degree prediction proxy model includes: when the fitting degree of the predicted value of the drill bit wear depth value corresponding to each drill bit sample data and the actual value of the drill bit wear depth value is greater than 80%, it is determined that the prediction accuracy meets the predetermined condition, and the optimization of the drill bit wear degree prediction proxy model is completed.
[0064] Specifically, when the fitting degree of the predicted value of the drill bit wear depth value corresponding to each drill bit sample data and the actual value of the drill bit wear depth value is greater than 80%, it can be determined that the prediction accuracy of the drill bit wear degree prediction proxy model meets the predetermined condition, thereby completing the optimization of the drill bit wear degree prediction proxy model.
[0065] Further, if the fitting degree of the predicted value of the drill bit wear depth value corresponding to at least one drill bit sample data and the actual value of the drill bit wear depth value is not greater than 80%, then it is necessary to re-fit the regression function by adjusting the and value of until the fitting degree of the out-of-hole drill bit point cloud data predicted by the regression function meets the requirements, and the optimization of the drill bit wear degree prediction proxy model is completed.
[0066] In summary, according to the technical solution provided by the embodiments of the present disclosure, through the optimization problem and the non-linear support vector regression method, the model can find the optimal regression hyperplane in the high-dimensional space, thereby improving the prediction accuracy of the drill bit wear degree. The model is evaluated using the validation data set to ensure that the model not only performs well on the training data, but also maintains good prediction performance on unseen data, enhancing the generalization ability of the model. Further, by evaluating the model on an independent validation data set, overfitting phenomena, i.e., the situation where the model performs excellently on the training data but poorly on new data, can be detected and reduced in a timely manner. The drill bit wear degree prediction proxy model constructed by the above method can accurately predict the wear degree of the drill bit, providing a scientific basis for drill bit design and drilling operations, thereby potentially extending the service life of the drill bit, reducing the drilling cost, and improving the drilling efficiency.
[0067] Exemplarily, when the training data set of the drill bit wear degree prediction proxy model includes the initial point cloud data, drilling parameters, and formation parameters of each drill bit sample data, it can also be to input the initial point cloud data, drilling parameters, and formation parameters of each drill bit sample data into the drill bit wear degree prediction proxy model to obtain the predicted worn point cloud data output by the drill bit wear degree prediction proxy model. According to the initial point cloud data, worn point cloud data, and predicted worn point cloud data of each drill bit, the predicted value and true value of the drill bit wear depth value or wear volume corresponding to each drill bit sample data are solved, and the prediction accuracy of the drill bit wear degree prediction proxy model is evaluated using the predicted value and true value of the drill bit wear depth value or wear volume.
[0068] In summary, by comparing the actual data and the predicted data, the prediction proxy model can be continuously optimized, thereby improving the accuracy of predicting the drill bit wear degree.
[0069] Figure 3 It is the third flow diagram of the drill bit design method provided by the present invention.
[0070] As Figure 3 shown, according to the first point cloud data and the drill bit wear degree prediction proxy model, the first point cloud data is optimized to obtain the optimized first point cloud data, and the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit, including: Step 301: Input the first point cloud data into the drill bit wear degree prediction proxy model to obtain the predicted second point cloud data after the drill bit exits the well.
[0071] In an embodiment of the present invention, after the construction of the drill bit wear degree prediction proxy model is completed, the first point cloud data in the drill bit information related to the target drill bit is input into the drill bit wear degree prediction proxy model to obtain the predicted second point cloud data corresponding to the drill bit sample data output by the drill bit wear degree prediction proxy model.
[0072] Step 302: Based on the first point cloud data and the predicted second point cloud data, obtain the predicted value of the first drill bit wear depth corresponding to the drill bit information.
[0073] In an embodiment of the present invention, by comparing the first point cloud data and the predicted point cloud data, the predicted value of the first drill bit wear depth of the drill bit containing the drill bit information in the current prediction process can be obtained.
[0074] Step S303: Using the predicted value of the first drill bit wear depth as the optimization target, optimize the first point cloud data using the gradient descent algorithm.
[0075] In an embodiment of the present invention, taking the predicted value of the first drill bit wear depth as the optimization target, substitute the predicted second point cloud data into the gradient descent algorithm to optimize the first point cloud data.
[0076] Specifically, the calculation principle of the gradient descent algorithm is as follows: Among them, The initial value of is the first point cloud data, is the updated first point cloud data, is the point cloud data during the update, is the learning rate, which controls the step size of each update, is the gradient of the predicted value of the first drill bit wear depth, which can be calculated using the central difference method in the computer. d is the predicted value of the first drill bit wear depth, and its initial value is obtained through The initial value of and its corresponding predicted second point cloud data.
[0077] Exemplarily, the gradient of the predicted value of the first drill bit wear depth can be calculated according to the following formula: Among them, is a very small positive number, usually taking 10 -6 ~10 -8 , represents the change rate of the predicted value of the first drill bit wear depth with respect to the point cloud data during the update when other variables remain unchanged, represents multiple independent variables required for calculating the drill bit wear depth value, including all the point cloud data after each update, represents the point cloud data during the update among the multiple independent variables required for calculating the predicted value of the first drill bit wear depth .
[0078] Exemplarily, when the feature vectors of the training samples used by the drill bit wear degree prediction proxy model include initial point cloud data, drilling parameters, and formation parameters, the above-mentioned multiple independent variables can also be the drilling parameters, formation parameters, and all point cloud data after each update.
[0079] Step 303: Iteratively update the predicted value of the first drill bit wear depth value. When the difference between the predicted values of the first drill bit wear depth value obtained from two consecutive iterations is less than a preset convergence threshold, the corresponding first point cloud data is the optimized first point cloud data, and the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit.
[0080] In one embodiment of the present invention, the first point cloud data, that is, The updated first point cloud data calculated from the initial value is input into the regression function, and the predicted value of the first drill bit wear depth value is calculated again, and the gradient is recalculated according to the gradient descent algorithm. Repeat the above method to iteratively update the predicted value of the first drill bit wear depth value. When the difference between the predicted values of the first drill bit wear depth value obtained from two consecutive iterations is less than the preset convergence threshold and the wear amount index (i.e., the first drill bit wear depth value) no longer decreases significantly, at this time, it is the optimized first point cloud data, and the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit.
[0081] In actual operation, the selection of the convergence threshold depends on the specific nature of the problem, the required accuracy, and the computing resources. Setting the threshold too small may cause the algorithm to require more iteration times, increasing the computing cost; setting it too large may cause the algorithm to stop prematurely before reaching the optimal solution, thus affecting the model performance. It can be specifically set according to actual needs.
[0082] Exemplarily, the first point cloud data of the bit information, the drilling parameters, and the formation parameters are input into the bit wear degree prediction proxy model to obtain the predicted second point cloud data output by the bit wear degree prediction proxy model. Based on the first point cloud data and the predicted second point cloud data, the predicted value of the first bit wear depth value corresponding to the bit information can be solved. By combining the gradient descent algorithm and the bit wear degree prediction proxy model, the predicted value of the first bit wear depth value is iteratively updated. Each time it is updated, the drilling parameters and formation parameters input into the bit wear degree prediction proxy model remain unchanged. Combining the current bit structure parameter design experience and research understanding as the constraint of the bit structure parameter change range, the first point cloud data before the bit enters the well is modified according to the set calculation step (for example, by adding or subtracting a small number from the position of the bit surface point cloud coordinate teeth to change the bit structure parameters), and the bit point cloud data after the bit exits the well (i.e., the predicted second point cloud data) corresponding to the changed bit structure parameters is iteratively calculated. Then, the bit wear depth value (i.e., the predicted value of the first bit wear depth value) is obtained by comparing the bit point cloud data after the bit exits the well with the corresponding modified surface point cloud data before the bit enters the well. When the calculated wear depth value meets the convergence condition of the optimization algorithm, the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit.
[0083] Exemplarily, when solving the structural point cloud vector set of the bit (i.e., the first point cloud data), the constraint parameter values are set based on the existing design experience and research understanding. For example: the tooth inclination angle for drilling in hard rock formations is 20° - 40°, the tooth surface shape is flat, concave, convex, or stepped, the number of blade wings is 7 - 12, the blade wing inclination angle is 15° - 30°, and the blade wing shape is spoon-shaped or wedge-shaped, etc.
[0084] In summary, according to the technical solution provided by the embodiments of the present invention, by iteratively optimizing and accurately predicting the bit wear degree, the efficiency and accuracy of bit design can be significantly improved, thereby improving the overall performance and economic benefits of drilling operations.
[0085] Furthermore, the second point cloud data after the bit exits the well can be determined according to the bit wear area range. Similarly, the first point cloud data before the bit enters the well is determined according to the bit range corresponding to this second point cloud data. For example: if only the outer row of teeth of the bit shows wear that affects drilling, the bit point cloud data can be only the point cloud data corresponding to the outer row of teeth. The geometric structure range of the bit corresponding to the first point cloud data and the second point cloud data can be determined according to actual needs. Preferably, the cutting teeth and part of the blade wings of the bit are selected. In this way, according to the finally obtained optimized first point cloud data, the optimal arrangement, shape, etc. of the cutting teeth and blade wings can be obtained to improve the wear resistance and drilling efficiency of the bit.
[0086] Furthermore, the worn point cloud data of the drill bit after it exits the well can be determined according to the range of the worn area of the drill bit. Similarly, the initial point cloud data before the drill bit enters the well is determined according to the drill bit range corresponding to the worn point cloud data. For example: if only the outer row of teeth of the drill bit shows wear that affects drilling, the point cloud data of the drill bit can be only the point cloud data corresponding to the outer row of teeth. The geometric structure range of the drill bit corresponding to the initial point cloud data and the worn point cloud data can be determined according to actual needs. Preferably, the cutting teeth and part of the blade of the drill bit are selected. In this way, when training the drill bit wear degree prediction proxy model, the point cloud data corresponding to the worn area of the drill bit can be used for training. Similarly, when applying the model, the first point cloud data related to the worn area of the drill bit (i.e., the point cloud data without wear) is used. Based on the finally obtained optimized first point cloud data, the optimal arrangement and shape of the cutting teeth and the blade can be obtained to improve the wear resistance and drilling efficiency of the drill bit.
[0087] Figure 4 is a schematic structural diagram of the drill bit design device provided by the present invention.
[0088] As Figure 4 shown, the drill bit design device 400 includes: an acquisition module 401, and an optimization module 403.
[0089] Furthermore, the acquisition module 401 is configured to acquire drill bit information related to the target drill bit, and the drill bit information at least includes the first point cloud data before the drill bit enters the well; Furthermore, the optimization module 402 is configured to optimize the first point cloud data according to the first point cloud data and the drill bit wear degree prediction proxy model, and acquire the optimized first point cloud data. The drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit; Among them, the drill bit wear degree prediction proxy model is at least trained based on the initial point cloud data before the drill bit enters the well and the worn point cloud data after the drill bit exits the well in each drill bit sample data in the drill bit sample dataset.
[0090] Figure 5 Illustrates a schematic physical structure diagram of an electronic device. As Figure 5 shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540. Among them, the processor 510, the communication interface 520, and the memory 530 complete mutual communication through the communication bus 540. The processor 510 can call the logical instructions in the memory 530 to execute the above-mentioned drill bit design method.
[0091] In addition, when the logical instructions in the above-mentioned memory 530 can be implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.
[0092] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the above-mentioned drill bit design method.
[0093] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the above-mentioned drill bit design method.
[0094] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. A person of ordinary skill in the art can understand and implement it without creative labor.
[0095] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the technical solution, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disks, optical discs, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments.
[0096] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A drill bit design method, characterized in that, Including: Obtaining bit information related to a target bit, where the bit information at least includes first point cloud data before the bit enters the well; Optimizing the first point cloud data according to the first point cloud data and a bit wear degree prediction proxy model, and obtaining optimized first point cloud data, where the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit; Among them, the bit wear degree prediction proxy model is at least trained based on the initial point cloud data before the bit enters the well and the worn point cloud data after the bit exits the well in each bit sample data in the bit sample data set.
2. The drill bit design method according to claim 1, characterized in that, The obtaining of the bit information related to the target bit includes: Determining the diameter of the target bit based on the target formation type encountered and the well opening stage; Obtaining the bit information for drilling the target formation type based on the diameter of the target bit.
3. The drill bit design method according to claim 1, characterized in that The optimizing the first point cloud data according to the first point cloud data and the bit wear degree prediction proxy model to obtain optimized first point cloud data, where the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit, includes: Inputting the first point cloud data into the bit wear degree prediction proxy model to obtain predicted second point cloud data after the bit exits the well; Based on the first point cloud data and the predicted second point cloud data, obtaining a predicted value of the first bit wear depth value corresponding to the bit information; Taking the predicted value of the first bit wear depth value as the optimization target, and using the gradient descent algorithm to optimize the first point cloud data; Iteratively updating the predicted value of the first bit wear depth value. When the difference between the predicted values of the first bit wear depth value obtained in two consecutive iterations is less than a preset convergence threshold, the corresponding first point cloud data is the optimized first point cloud data, and the bit structure corresponding to the optimized first point cloud data is the bit structure of the target bit.
4. The drill bit design method according to claim 1, characterized in that, The bit wear degree prediction proxy model is trained according to the following steps: Dividing the bit sample data set into a training data set and a validation data set, where the training data set and the validation data set respectively contain a number of the bit sample data; Based on the training data set, using the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, constructing and training the bit wear degree prediction proxy model by using the non-linear support vector regression method; Evaluating the prediction accuracy of the bit wear degree prediction proxy model by using the validation data set; When the prediction accuracy meets a predetermined condition, completing the optimization of the bit wear degree prediction proxy model.
5. The drill bit design method according to claim 4, characterized in that The training of the bit wear degree prediction proxy model by using the non-linear support vector regression method based on the training data set, with the initial point cloud data as the independent variable and the worn point cloud data as the dependent variable, includes: Constructing an optimization problem, where the optimization problem includes an optimization objective function and constraint conditions; The optimization objective function is expressed as: The constraint conditions are expressed as: Among them, and respectively represent the feature vectors of the i-th training sample and the j-th training sample in the training set, including the initial point cloud data; is the true target value of the i-th training sample in the training set, including the worn point cloud data; is the parameter of the insensitive loss function, is the penalty parameter, is the Lagrange multiplier of the i-th training sample, is the Lagrange multiplier of the j-th training sample, is the kernel function; Solving the optimization problem by using the training data set to obtain the optimal Lagrange multipliers and the optimal bias term; Based on the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms, construct the prediction surrogate model for the drill bit wear degree.
6. The drill bit design method according to claim 5, characterized in that The constructing of the prediction surrogate model for the drill bit wear degree based on the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms includes: Based on the optimization problem, the optimal Lagrange multipliers, and the optimal bias terms, construct a regression function, where the regression function is a numerical representation of the prediction surrogate model for the drill bit wear degree; wherein, the regression function is expressed as: wherein, is the predicted worn point cloud data output by the drill bit wear degree prediction proxy model; b is the optimal bias term, is the kernel function, is the optimal Lagrange multiplier of the i-th training sample.
7. The drill bit design method according to claim 6, characterized in that, The evaluating of the prediction accuracy of the prediction surrogate model for the drill bit wear degree by using the validation data set includes: Input the initial point cloud data of each drill bit sample data in the validation data set into the prediction surrogate model for the drill bit wear degree, and obtain the predicted worn point cloud data output by the prediction surrogate model for the drill bit wear degree; Based on the initial point cloud data, the predicted worn point cloud data, and the worn point cloud data of each drill bit sample data, obtain the predicted value and the actual value of the drill bit wear depth value corresponding to each drill bit sample data; Evaluate the prediction accuracy of the prediction surrogate model for the drill bit wear degree based on the predicted value and the actual value of the drill bit wear depth value of each drill bit sample data.
8. The drill bit design method according to claim 7, characterized in that The completing of the optimization of the prediction surrogate model for the drill bit wear degree when the prediction accuracy meets a predetermined condition includes: When the fitting degree between the predicted value and the actual value of the drill bit wear depth value corresponding to each drill bit sample data is greater than 80%, determine that the prediction accuracy meets the predetermined condition, and complete the optimization of the prediction surrogate model for the drill bit wear degree.
9. A drill bit design device, characterized in that, Includes: An acquisition module, configured to acquire drill bit information related to a target drill bit, where the drill bit information at least includes the first point cloud data before the drill bit enters the well; An optimization module, configured to optimize the first point cloud data according to the first point cloud data and the prediction surrogate model for the drill bit wear degree, and obtain the optimized first point cloud data, where the drill bit structure corresponding to the optimized first point cloud data is the drill bit structure of the target drill bit; wherein, the prediction surrogate model for the drill bit wear degree is at least trained based on the initial point cloud data before the drill bit enters the well and the worn point cloud data after the drill bit exits the well of each drill bit sample data in the drill bit sample data set.
10. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the drill bit design method according to any one of claims 1-8.