A drilling method, apparatus, equipment and storage medium based on top drive spindle stress field reconstruction

Abstract

The application discloses a drilling method and device based on stress field reconstruction of a top drive spindle, equipment and a storage medium, relates to the technical field of drilling monitoring, and comprises the following steps: constructing a stress snapshot matrix based on finite element stress simulation data under different drilling conditions generated by a target finite element model of the top drive spindle which is constructed in advance; performing mean field separation processing and singular value decomposition processing on the stress snapshot matrix to obtain a mean field, a POD mode and POD mode coefficients corresponding to each drilling condition, training a Kriging proxy model to obtain a target coefficient prediction model; inputting real-time working condition parameters of a drilling site into the target coefficient prediction model to reconstruct a full-node Mises stress field of the top drive spindle under the real-time working condition parameters after obtaining POD prediction coefficients; and determining a corresponding drilling safety evaluation result based on the full-node Mises stress field to perform drilling operation. In this way, the top drive system can be safely operated during drilling.

Description

Technical Field

[0001] This invention relates to the field of drilling monitoring technology, and in particular to a drilling method, apparatus, equipment and storage medium based on top drive spindle stress field reconstruction. Background Technology

[0002] The top drive spindle is the core power transmission component in a top drive drilling rig. During deep drilling, the spindle simultaneously withstands high torque rotation, axial drilling pressure, drilling fluid internal pressure, and complex alternating loads. The distribution of its internal stress field directly affects the safety of drilling and completion operations and the service life of the equipment. Because the spindle operates in a high-speed rotating and enclosed assembly environment, sensors are difficult to place at critical critical sections. Measurement methods can only obtain limited response signals from near-field components such as bearing housings or the casing, making it impossible to directly acquire the full stress distribution within the spindle body. While traditional finite element simulation can accurately calculate the full-field stress of the entire spindle under specific operating conditions, the computational load for a single detailed solution is enormous, typically requiring tens of minutes or even hours, making it difficult to meet the online evaluation needs under conditions of frequent changes in drilling parameters. If a surrogate model is used to directly map operating parameters to the full-node stress field, the output dimension reaches hundreds of thousands or even millions, imposing stringent requirements on the quantity and quality of training samples. Under small sample conditions, overfitting or spatial consistency mismatch problems are highly likely to occur.

[0003] Therefore, there is an urgent need for a reconstruction scheme that balances the ability to characterize stress across the entire field with real-time computing efficiency, so as to provide reliable data support for the digital twin operation and maintenance of the top drive spindle during drilling. Summary of the Invention

[0004] In view of this, the purpose of this invention is to provide a drilling method, apparatus, equipment, and storage medium based on top drive spindle stress field reconstruction, which can provide reliable data support for the digital twin operation and maintenance of the top drive spindle during drilling. The specific solution is as follows: In a first aspect, this application discloses a drilling method based on top drive spindle stress field reconstruction, comprising: Based on the pre-built target finite element model of the top drive spindle, finite element stress simulation data under different drilling conditions are generated, and a stress snapshot matrix is ​​constructed based on the node data in the finite element stress simulation data. The stress snapshot matrix is ​​subjected to mean field separation and singular value decomposition to obtain the mean field, POD modes and POD mode coefficients corresponding to each drilling condition. The Kriging surrogate model is then trained using the condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model. The real-time operating parameters of the drilling site are input into the target coefficient prediction model to obtain the POD prediction coefficients. Based on the POD prediction coefficients, the mean field, and the POD mode, the full-node Mises stress field of the top drive spindle under the real-time operating parameters is reconstructed. The corresponding drilling safety assessment results are determined based on the full-node Mises stress field, so that drilling operations can be carried out based on the drilling safety assessment results.

[0005] Optionally, the step of generating finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and constructing a stress snapshot matrix based on the node data in the finite element stress simulation data, includes: Obtain the three-dimensional geometric model of the top drive spindle to obtain the target finite element model of the top drive spindle; Based on a preset depth interval, multiple drilling conditions are determined within the target drilling depth range, and the load boundary conditions corresponding to each drilling condition are determined. The load boundary conditions corresponding to each of the drilling conditions are applied to the target finite element model, and finite element solutions are performed to obtain finite element stress simulation data corresponding to each of the drilling conditions. Read the node number, node three-dimensional coordinates, and node Mises stress value of each node from the finite element stress simulation data corresponding to each drilling condition; A unified node order is determined using the node number and the node three-dimensional coordinates, and a node consistency check is performed on the finite element stress simulation data corresponding to each drilling condition; wherein, the node consistency check is to determine whether the number of nodes, node number, and node three-dimensional coordinates are consistent among the finite element stress simulation data corresponding to each drilling condition. If the number of nodes, the node number, and the three-dimensional coordinates of the nodes are consistent among the finite element stress simulation data, then the Mises stress values ​​of the nodes corresponding to each drilling condition are arranged into stress column vectors according to the unified node order, and the stress column vectors corresponding to each drilling condition are combined column by column to obtain a stress snapshot matrix.

[0006] Optionally, the step of performing mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes, and POD mode coefficients corresponding to each of the drilling conditions includes: The average value of the stress column vectors corresponding to each drilling condition in the stress snapshot matrix is ​​determined to obtain the mean field, and the centered stress snapshot matrix is ​​determined based on the mean field and the stress column vectors corresponding to each drilling condition. Singular value decomposition is performed on the centered stress snapshot matrix to obtain the candidate POD modes of each order and the singular values ​​corresponding to each candidate POD mode; The energy proportion of each candidate POD mode is determined based on the singular values ​​corresponding to each order of the candidate POD modes, and the POD mode is determined based on the energy proportion and the preset cumulative energy threshold. The drilling conditions in the centered stress snapshot matrix are projected onto the POD modes to obtain the POD mode coefficients corresponding to each drilling condition.

[0007] Optionally, training the Kriging surrogate model using the operating condition parameter values ​​and the POD modal coefficients to obtain the target coefficient prediction model includes: The operating condition parameter values ​​corresponding to each drilling condition are standardized to obtain the standardized operating condition parameter values ​​corresponding to each drilling condition. The standardized operating condition parameter values ​​corresponding to each of the drilling conditions are used as input features, and the POD modal coefficients of each order corresponding to each of the drilling conditions are used as output labels. Based on the input features and the output labels, the Kriging proxy model corresponding to each of the POD modal coefficients is trained to obtain the target coefficient prediction model corresponding to each order of the POD modal coefficients; Accordingly, the step of inputting real-time operating parameters from the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient includes: The real-time operating parameters are standardized, and the standardized real-time operating parameters are input into the target coefficient prediction model to obtain the POD prediction coefficients corresponding to each POD mode.

[0008] Optionally, the reconstructing of the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficients, the mean field, and the POD modes includes: The stress fluctuation vector corresponding to the real-time operating parameters is obtained by weighted summation of the POD prediction coefficients for each order of the POD modes. The stress fluctuation vector is superimposed with the mean field to obtain the full-node Mises stress vector of the top drive spindle under the real-time operating parameters. The stress values ​​of each node in the full-node Mises stress vector are associated with the node number and three-dimensional coordinates of the corresponding node in the target finite element model to obtain the full-node Mises stress field of the top drive spindle under the real-time operating parameters; wherein, each node stress value in the full-node Mises stress field corresponds to a fixed node position in the target finite element model.

[0009] Optionally, determining the corresponding drilling safety assessment results based on the full-node Mises stress field includes: Based on the three-dimensional coordinates of each node in the target finite element model and the node Mises stress values ​​of the corresponding nodes in the full node Mises stress field, a three-dimensional Mises stress cloud map of the top drive spindle is generated. Determine the node position corresponding to the maximum Mises stress value in the full-node Mises stress field; The nodes in the full node Mises stress field are sorted from largest to smallest based on the node Mises stress values, and a set of high-stress nodes is obtained by extracting a preset proportion of the nodes at the top of the sort. The target stress region and its spatial distribution of the top drive spindle are then determined based on the set of high-stress nodes. The corresponding drilling safety assessment results are determined based on the target stress region and its spatial distribution.

[0010] Optionally, the method further includes: If the real-time operating parameters exceed the preset training sample range threshold, or if the maximum stress relative error of the full-node Mises stress field exceeds the preset threshold, or if the spatial distribution of the target stress region deviates from the historical verification benchmark by more than the preset range difference, the finite element verification operation is triggered to obtain the verification stress field under the current real-time operating parameters. The verification stress field is checked for node consistency based on a preset node check sequence. If the node consistency check is passed, the verified stress field and its corresponding real-time operating parameters are added as new samples to the stress snapshot matrix. Jump to the step of performing mean field separation and singular value decomposition on the stress snapshot matrix.

[0011] Secondly, this application discloses a drilling apparatus based on top drive spindle stress field reconstruction, comprising: The data processing module is used to generate finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and to construct a stress snapshot matrix based on the node data in the finite element stress simulation data. The model training module is used to perform mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes and POD mode coefficients corresponding to each drilling condition, and to train the Kriging surrogate model using the condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model. The stress field reconstruction module is used to input the real-time operating parameters of the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient, and reconstruct the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficient, the mean field and the POD mode. The safety assessment result determination module is used to determine the corresponding drilling safety assessment result based on the full-node Mises stress field, so as to carry out drilling operations based on the drilling safety assessment result.

[0012] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is used to execute the computer program to implement the aforementioned drilling method based on top drive spindle stress field reconstruction.

[0013] Fourthly, this application discloses a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements the aforementioned drilling method based on top drive spindle stress field reconstruction.

[0014] As can be seen, in this application, finite element stress simulation data under different drilling conditions are generated based on a pre-constructed target finite element model of the top drive spindle, and a stress snapshot matrix is ​​constructed based on the node data in the finite element stress simulation data; the stress snapshot matrix is ​​subjected to mean field separation processing and singular value decomposition processing to obtain the mean field, POD modes, and POD mode coefficients corresponding to each drilling condition, and the Kriging surrogate model is trained using the operating condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model; the real-time operating condition parameters of the drilling site are input into the target coefficient prediction model to obtain the POD prediction coefficients, and the full-node Mises stress field of the top drive spindle under the real-time operating condition parameters is reconstructed based on the POD prediction coefficients, the mean field, and the POD modes; the corresponding drilling safety assessment result is determined based on the full-node Mises stress field so that drilling operations can be performed based on the drilling safety assessment result. In the offline phase, by constructing a stress snapshot matrix covering different drilling conditions and performing mean field separation and singular value decomposition, the high-dimensional full-node stress field is compressed into a small number of POD modal coefficients. This allows the Kriging surrogate model to learn only the mapping relationship between operating parameters and low-dimensional modal coefficients, thus avoiding the problems of overfitting or spatial consistency mismatch that easily occur under small sample conditions when directly predicting the full-node stress field due to excessively high output dimensionality. In the online phase, the real-time operating parameters can be input into the target coefficient prediction model to reconstruct the full-node Mises stress field of the top drive spindle without repeatedly performing time-consuming finite element solutions, thus balancing the ability to characterize the full-field stress and real-time computational efficiency. Furthermore, the drilling safety assessment results are determined based on the full-node Mises stress field, which can identify high-stress areas of the top drive spindle and provide a basis for drilling operations. This ensures the safe operation of the top drive system during drilling and provides reliable data support for the digital twin operation and maintenance of the top drive spindle. Attached Figure Description

[0015] 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 only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0016] Figure 1 This is a flowchart of a drilling method based on top drive spindle stress field reconstruction disclosed in this application; Figure 2 This is a flowchart of a finite element stress sample generation and snapshot matrix construction method disclosed in this application; Figure 3 This application discloses a flowchart of a POD-Kriging surrogate model training and stress field reconstruction process. Figure 4 This is a flowchart of an error verification and model update mechanism disclosed in this application; Figure 5 This is a flowchart of a specific drilling method based on top drive spindle stress field reconstruction disclosed in this application; Figure 6 This is a schematic diagram of a drilling rig structure based on top drive spindle stress field reconstruction disclosed in this application; Figure 7 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0018] A search of existing technologies revealed solutions involving strain field reconstruction, digital twins of bridges or machine tools, finite element model verification, and top drive structure analysis. However, none of these solutions addressed the full-node stress field of the top drive spindle in deep drilling under complex operating conditions, establishing a complete closed loop from finite element samples, POD (Proper Orthogonal Decomposition) order reduction, Kriging regression to real-time reconstruction and error correction. In actual drilling, the spindle bearing is subjected to multiple load couplings, and its internal stress state directly affects system safety. However, sensor monitoring can only reflect local points and cannot obtain the full-field distribution of critical areas such as stress relief grooves and threaded connection areas. Therefore, this application will specifically introduce a drilling method based on top drive spindle stress field reconstruction, which can solve the above problems.

[0019] See Figure 1 As shown in the figure, this application discloses a drilling method based on top drive spindle stress field reconstruction, including: Step S11: Generate finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and construct a stress snapshot matrix based on the node data in the finite element stress simulation data.

[0020] In this embodiment, the step of generating finite element stress simulation data under different drilling conditions based on a pre-constructed target finite element model of the top drive spindle, and constructing a stress snapshot matrix based on the node data in the finite element stress simulation data, includes: acquiring a three-dimensional geometric model of the top drive spindle to obtain a target finite element model of the top drive spindle; determining multiple drilling conditions within a target drilling depth range based on a preset depth interval, and determining the load boundary conditions corresponding to each drilling condition; applying the load boundary conditions corresponding to each drilling condition to the target finite element model respectively, and performing finite element solution to obtain finite element stress simulation data corresponding to each drilling condition; and reading the node data of each node in the finite element stress simulation data corresponding to each drilling condition. The system identifies node numbers, node 3D coordinates, and node Mises stress values. A unified node order is determined using the node numbers and node 3D coordinates, and a node consistency check is performed on the finite element stress simulation data corresponding to each drilling condition. The node consistency check involves determining whether the number of nodes, node numbers, and node 3D coordinates are consistent across the finite element stress simulation data corresponding to each drilling condition. If the number of nodes, node numbers, and node 3D coordinates are consistent, the node Mises stress values ​​corresponding to each drilling condition are arranged into stress column vectors according to the unified node order, and these stress column vectors are combined column-wise to obtain a stress snapshot matrix.

[0021] Specifically, such as Figure 2As shown, when obtaining the three-dimensional geometric model of the top drive spindle, this application models it based on the actual structure and service load of the top drive spindle, and retains structural features related to stress concentration, such as stress relief grooves, threaded areas, and bolt hole areas. Material parameters are set accordingly: the spindle material is 4340 steel, Young's modulus is 205000 MPa, and Poisson's ratio is 0.29. Hexahedral C3D8R elements are used for mesh generation. Considering the large stress gradient at stress concentration points, mesh refinement or local detailing is performed on local areas with high stress. These local areas can specifically be the stress relief grooves, threaded connection areas, main bearing mating areas, bolt hole areas, or torque transmission areas of the top drive spindle. Simultaneously, a nonlinear elastic connector is used to simulate the bearing's support effect on the spindle, reflecting the elastic deformation when the spindle contacts the bearing, making the support boundary closer to the actual assembly state. Furthermore, to achieve a balance between computational accuracy and computational scale, this application determines the final number of elements to be 491947 and the number of nodes to be 525602 through mesh independence analysis. It should be noted that the numerical value is only used to illustrate the data scale of this embodiment and does not constitute a limitation on the scope of protection of this application. For different models of top drive spindles or different mesh strategies, the number of nodes and the number of elements may be different, but when constructing sample data later, it must be ensured that all samples come from the same mesh or have a strictly corresponding node order; otherwise, the data in the same row of the stress snapshot matrix will no longer correspond to the same spatial location.

[0022] In this embodiment, when determining the load boundary conditions corresponding to each drilling condition, the longitudinal load of the drill string, drilling fluid pressure, drive torque, bearing load torque, and drill string load torque are applied according to the spindle mechanical model. Specifically, based on the drilling depth, the corresponding loads are calculated or read. The drilling fluid pressure is applied to the relevant surfaces of the spindle internal channel, the drive torque and load torque are applied to the corresponding torque transmission positions, and the longitudinal load is applied along the spindle axis. The condition parameter values ​​corresponding to each drilling condition specifically include five items: drilling depth, drilling fluid pressure, longitudinal load, load torque, and drive torque. In actual operation, this embodiment uses an 8000m drilling depth as the background. Within the first 7600m depth range, 19 independent finite element simulations are conducted at 400m intervals, and the obtained data are used for training. Furthermore, within the subsequent 400m depth range, two simulations are conducted at 200m intervals, and the obtained data are used to verify the predictive performance of the surrogate model under subsequent actual conditions.

[0023] In this embodiment, each finite element result file corresponds to an independent working condition. For 19 sets of training samples, each Excel finite element result file contains node numbers, three columns of node coordinates, and node Von Mises stress values. The working condition parameters for each working condition are parsed from the finite element result file name or working condition record. The reason for using node numbers and node 3D coordinates to jointly determine a unified node order is that the node numbers (NodeID) derived from the finite element method are not globally unique labels. If sorting is done solely by node numbers, stress values ​​at different spatial locations may be incorrectly combined, causing subsequent order reduction and reconstruction to lose physical meaning. Therefore, this embodiment uses NodeID and X, Y, and Z coordinates to jointly determine the node order and checks whether the number of nodes, node labels, and node coordinates of all samples are consistent. If the number of nodes, coordinates, or order of nodes for a certain working condition are inconsistent with the baseline sample, then that working condition must not be included in subsequent POD training. After the data is read, the system generates a sample parameter table, a node coordinate table, a single sample stress vector, and a stress snapshot matrix. For 19 sets of training samples, a stress snapshot matrix consisting of 19 columns of stress vectors is formed. Each row of the matrix corresponds to the stress change of the same spatial node under different working conditions, and each column corresponds to the full node stress field under a working condition.

[0024] Step S12: Perform mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes, and POD mode coefficients corresponding to each drilling condition. Then, use the condition parameter values ​​and the POD mode coefficients to train the Kriging surrogate model to obtain the target coefficient prediction model.

[0025] In this embodiment, the step of performing mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes, and POD mode coefficients corresponding to each drilling condition includes: determining the average value of the stress column vectors corresponding to each drilling condition in the stress snapshot matrix to obtain the mean field, and determining a centered stress snapshot matrix based on the mean field and the stress column vectors corresponding to each drilling condition; performing singular value decomposition on the centered stress snapshot matrix to obtain candidate POD modes of each order and the singular values ​​corresponding to each order of candidate POD modes; determining the energy proportion of each order of candidate POD modes based on the singular values ​​corresponding to each order of candidate POD modes, and determining the POD mode based on the energy proportion and a preset cumulative energy threshold; and projecting each drilling condition in the centered stress snapshot matrix onto the POD mode to obtain the POD mode coefficients corresponding to each drilling condition.

[0026] In actual operation, such as Figure 3As shown, the average Mises stress field of 19 training samples can be calculated first as the mean field. Then, the average stress field is subtracted from the stress column vector of each sample to form a centered stress snapshot matrix. This process allows subsequent decomposition to focus on the fluctuation of the stress field as it changes with the working conditions. Then, SVD (Singular Value Decomposition) decomposition is performed on the centered stress snapshot matrix, or snapshot-based POD decomposition can be used to obtain candidate POD modes, singular values, and sample mode coefficients that can represent the spatial variation characteristics of the principal axis stress field. When determining the POD modes, 0.999 can be used as a preset cumulative energy threshold. The calculation results show that the energy proportion of the first-order POD mode is 0.935918, and the cumulative energy of the first two-order POD modes is 0.999441, which exceeds the set cutoff threshold. Therefore, the main model retains two POD modes. It should be noted that the cutoff threshold is the cumulative energy retention threshold for POD modes; in this embodiment, it is set to 0.999. The cumulative energy of the first-order POD mode is 0.935918, and that of the first two orders is 0.999441. Therefore, the first two POD modes are retained, and the remaining modes are truncated. This result shows that under the current 19 drilling working condition paths, the Mises stress variation of the top drive spindle in the entire field has obvious low-dimensional characteristics. Through this processing, the high-dimensional full-node Mises stress field of 525,602 nodes is transformed into a small number of POD mode coefficients, which reduces the difficulty of proxy modeling under small sample conditions while maintaining the spatial correspondence of nodes. It should be noted that the two POD modes mentioned in this embodiment are only selected under the current sample path. When the sample range, structure type, or mesh changes, the number of modes can be re-determined based on the cumulative energy and verification error. The cumulative energy threshold can be a preset threshold or adaptively determined based on the model verification error, and a fixed number of samples, fixed number of nodes, or fixed number of modes is not a necessary limitation. It should be noted that in the figure, "mean field + "×coefficient" means: ; Where x is the real-time operating condition parameter, The mean stress field of the training samples, Here is the POD mode basis matrix. These are the POD modal coefficients predicted by the Kriging model. The full-node Mises stress field obtained from the reconstruction.

[0027] In this embodiment, training the Kriging surrogate model using the operating condition parameter values ​​and the POD modal coefficients to obtain the target coefficient prediction model includes: standardizing the operating condition parameter values ​​corresponding to each drilling condition to obtain standardized operating condition parameter values ​​corresponding to each drilling condition; using the standardized operating condition parameter values ​​corresponding to each drilling condition as input features, and using the POD modal coefficients of each order corresponding to each drilling condition as output labels; and training the Kriging surrogate model corresponding to each POD modal coefficient based on the input features and the output labels to obtain the target coefficient prediction model corresponding to each order of the POD modal coefficients. Specifically, in this embodiment, five operating condition parameters—drilling depth, drilling fluid pressure, longitudinal load, load torque, and drive torque—are used as input features. These input features are standardized according to the statistics used in the training phase to eliminate the influence of differences in the dimensions and numerical ranges of each parameter on model training. The POD modal coefficients corresponding to each drilling condition are used as output labels, and a Kriging regression model is trained for each POD modal coefficient. For example, two POD modes are retained, meaning two Kriging regression models are trained separately. In other embodiments, a Gaussian process regression model equivalent to Kriging can also be trained. Considering the small number of finite element samples, this embodiment uses leave-one-out cross-validation to validate the model. That is, N-1 sets of samples are used for training each time, and the remaining 1 set is used for validation, repeating this process N times to obtain the cross-validation prediction results for all training samples. Alternatively, small-sample K-fold cross-validation can be used to evaluate the modal coefficient prediction error and the overall reconstruction error. After training is completed, the standardized parameters, Kriging model, mean stress field, POD modes, and node coordinate library are saved for direct use in the online reconstruction phase.

[0028] Step S13: Input the real-time operating parameters of the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient, and reconstruct the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficient, the mean field and the POD mode.

[0029] In this embodiment, inputting real-time operating parameters from the drilling site into the target coefficient prediction model to obtain POD prediction coefficients includes: standardizing the real-time operating parameters and inputting the standardized real-time operating parameters into the target coefficient prediction model to obtain POD prediction coefficients corresponding to each POD mode. Specifically, in this embodiment, the real-time operating parameters can come from drilling operation records, top drive control systems, or can be converted from on-site pressure, torque, load, and drilling records. After access, the system checks whether the input includes five items: drilling depth, drilling fluid pressure, longitudinal load, load torque, and drive torque. It checks whether the units of each field are consistent with those in the training phase, and performs time synchronization, missing value processing, outlier removal, unit unification, and normalization. Then, it transforms the data according to the standardized parameters saved in the training phase, thereby organizing the on-site operating data into an input vector with the same order and units as in the training phase. When the input condition falls near the training condition path, the system enters the stress field reconstruction process; when the input parameters exceed the training sample range or the field is abnormal, the system generates an extrapolation risk warning, outputs a data quality identifier and passes it to the subsequent review process to prevent the surrogate model from being used directly under unsuitable conditions.

[0030] In this embodiment, the reconstructing of the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficients, the mean field, and the POD modes includes: weighted summation of each order of the POD modes based on the POD prediction coefficients to obtain the stress fluctuation vector corresponding to the real-time operating parameters; superimposing the stress fluctuation vector with the mean field to obtain the full-node Mises stress vector of the top drive spindle under the real-time operating parameters; associating the stress values ​​of each node in the full-node Mises stress vector with the node number and three-dimensional coordinates of the corresponding node in the target finite element model to obtain the full-node Mises stress field of the top drive spindle under the real-time operating parameters; wherein, each node stress value in the full-node Mises stress field corresponds to a fixed node position in the target finite element model. Specifically, in this embodiment, the system first calls the trained Kriging model to predict the POD coefficients corresponding to the current operating condition, then combines the predicted POD coefficients with the POD mode matrix, and superimposes the mean stress field to obtain the full-node Mises stress vector under the current operating condition. Then, the stress vector is combined with the node number (NodeID) and X, Y, Z coordinates to form a nodal stress result table containing the node number, three-dimensional coordinates, and predicted Mises stress, so that each predicted stress value corresponds to a fixed node position in the principal axis finite element model; this result table can be saved as a CSV file or read by a three-dimensional visualization program for subsequent display and analysis.

[0031] Step S14: Determine the corresponding drilling safety assessment results based on the full-node Mises stress field, so as to carry out drilling operations based on the drilling safety assessment results.

[0032] In this embodiment, determining the corresponding drilling safety assessment result based on the full-node Mises stress field includes: generating a three-dimensional Mises stress cloud map of the top drive spindle based on the three-dimensional coordinates of each node in the target finite element model and the node Mises stress value of the corresponding node in the full-node Mises stress field; determining the node position corresponding to the maximum Mises stress value in the full-node Mises stress field; sorting each node in the full-node Mises stress field from largest to smallest based on the node Mises stress value, extracting a preset proportion of nodes at the top of the sort to obtain a high-stress node set, and determining the target stress region and its spatial distribution of the top drive spindle based on the high-stress node set; and determining the corresponding drilling safety assessment result based on the target stress region and its spatial distribution.

[0033] Specifically, in this embodiment, PyVista or a similar 3D visualization tool is used to generate a 3D point cloud or visualization result of the spindle from the node coordinates and predicted stress values. The output format may include a point cloud VTK (Visualization Toolkit) file, a 3D stress cloud map, and time history curves of key parts. The reconstructed result is then pasted back onto the finite element node coordinates of the top drive spindle. When extracting the high-stress node set, the preset ratio can be Top 1% or Top 5%, that is, the nodes whose Mises stress values ​​rank in the top 1% or top 5% constitute the high-stress node set, and the overlap between the high-stress region and the finite element verification result is calculated. In actual operation, the stress state near the stress relief groove region, thread region, and bolt hole region is the focus. The local maximum stress or the distribution of high-stress nodes in the above regions can be output separately, along with the coordinates of the maximum stress node, the range of the high-stress region, the potential danger area, and the warning level, providing a basis for judging the possible fatigue failure risk location of the spindle.

[0034] Furthermore, during the model validation or finite element verification stage, the reliability of the reconstruction can be evaluated based on at least one of the following indicators: Top-K node set, overlap of the top percentage of high-stress regions, maximum relative stress error, distance to the maximum stress location, or overall L2 relative error. In this embodiment, two sets of finite element simulations with 200m intervals within a subsequent 400m depth range are used as validation data to compare the differences between the predicted stress field and the finite element stress field. Simultaneously, the average overall L2 relative error, MAE, RMSE, MAPE, maximum Mises relative error, and Jaccard indexes for the top 1% and top 5% high-stress regions are statistically analyzed using leave-one-out cross-validation results from 19 sets of training samples. The evaluation results show that the average overall L2 relative error is 0.058740, the average MAE is 1.048204, the average RMSE is 1.337271, the average MAPE is 5.842365%, the average maximum Mises relative error is 6.644054%, and the average Jaccard indexes for the top 1% and top 5% high-stress regions are 0.914839 and 0.917073, respectively. These results indicate that the overall relative error remains at a low level, and the Jaccard indexes for high-stress regions all exceed 0.91, suggesting that the POD-Kriging model can maintain the overall stress distribution trend of the principal axis well near the current training path and can provide a reference for identifying high-stress regions. It is important to emphasize that the above indicators are used to evaluate the predictive ability of the current model near existing working paths and do not mean that the model can replace finite element simulation under any working condition. For long-distance extrapolation working conditions, working conditions with significant changes in structural boundary conditions, or abnormal on-site working conditions, verification through finite element analysis is still necessary.

[0035] In this embodiment, the method further includes: if the real-time operating parameters exceed a preset training sample range threshold, or if the maximum relative stress error of the full-node Mises stress field exceeds a preset threshold, or if the spatial distribution of the target stress region deviates from the historical verification benchmark by a preset range difference, a finite element verification operation is triggered to obtain the verification stress field under the current real-time operating parameters; a node consistency check is performed on the verification stress field based on a preset node check order; if the node consistency check passes, the verification stress field and its corresponding real-time operating parameters are added as new samples to the stress snapshot matrix; and the process jumps to the step of performing mean field separation processing and singular value decomposition processing on the stress snapshot matrix.

[0036] Specifically, in this embodiment, as Figure 4As shown, since the training samples are established along the drilling depth variation path, there is a strong correlation between drilling depth and some load parameters. Therefore, the POD-Kriging model is mainly suitable for interpolation prediction near the training working condition path and is not suitable as a general model that can completely replace finite element simulation under any working condition. The above-mentioned verification mechanism is used to limit the reasonable application range of the model near the training working condition path. It should be noted that the preset training sample range threshold is the sample collection range determined based on the actual working condition of the target training data. For example, taking the actual working condition of an 8000m drilling depth as the background, 19 independent finite element simulations are carried out at 400m intervals within the first 7600m depth range, and this data is used for training. Then, the first 7600m depth range is the preset training sample range threshold. The real-time working condition parameters obtained beyond the first 7600m are considered to exceed the preset training sample range threshold. The verification error threshold is determined based on the cross-validation results of the 19 training samples and the verification results of the independent working conditions at 7800m and 8000m, and combined with the engineering allowable error of the principal axis stress prediction, this embodiment takes the maximum Mises stress relative error of 5%. The target stress region is the area where the node Mises stress is not less than 90% of the maximum stress under this working condition; the preset range difference is determined based on the normal spatial fluctuation of high-stress regions in historical verification results. In this embodiment, the proportion of non-overlapping predicted regions and historical verification regions is taken as 10%. When judging the prediction deviation, the prediction result of the surrogate model can be compared with the monitoring conversion value of measurable locations, the verification value of key points, the results of historical similar working conditions, or the periodic finite element verification results. When the deviation between the two exceeds the preset threshold, at least one of the following operations is performed: supplementing finite element samples, updating POD modes, retraining the Kriging surrogate model, adjusting the warning threshold, or outputting finite element recalculation instructions. If the new working condition deviates significantly from the training sample path, the prediction uncertainty is high, or the high-stress area identification result is inconsistent with engineering experience, the system will not take the prediction result as the final finite element conclusion, but will prompt for finite element verification. The verified finite element result will be checked for node consistency according to the aforementioned rules. After the check is passed, sample data will be added, and POD order reduction and Kriging training will be re-executed accordingly. This forms a closed loop of "finite element sample library - POD order reduction - Kriging modal coefficient regression - real-time data-driven reconstruction - error threshold correction", which helps to reduce the risk of extrapolation misuse of surrogate models.

[0037] As can be seen, in this embodiment, finite element stress simulation data under different drilling conditions are generated based on a pre-constructed target finite element model of the top drive spindle, and a stress snapshot matrix is ​​constructed based on the node data in the finite element stress simulation data; the stress snapshot matrix is ​​subjected to mean field separation and singular value decomposition to obtain the mean field, POD modes, and POD mode coefficients corresponding to each drilling condition; the Kriging surrogate model is trained using the operating condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model; the real-time operating condition parameters of the drilling site are input into the target coefficient prediction model to obtain the POD prediction coefficients; and the full-node Mises stress field of the top drive spindle under the real-time operating condition parameters is reconstructed based on the POD prediction coefficients, the mean field, and the POD modes; the corresponding drilling safety assessment result is determined based on the full-node Mises stress field so that drilling operations can be performed based on the drilling safety assessment result. That is, as Figure 5 As shown, an offline finite element simulation was used to construct a stress sample library covering typical drilling conditions. The high-dimensional full-field stress was compressed into low-dimensional modal coefficients using POD (Position-Order Decomposition) model, and then the Kriging model was used to establish the mapping relationship between drilling parameters and these coefficients. In the online phase, only real-time operating parameters need to be input to quickly reconstruct the Mises stress field of all nodes, with a response speed several orders of magnitude faster than direct finite element solutions. The reconstruction results can be used to identify high-stress areas and assess the spindle safety status, providing a basis for real-time adjustment of drilling parameters, thereby reducing non-productive time while ensuring drilling safety.

[0038] refer to Figure 6 The present application also discloses a drilling apparatus based on top drive spindle stress field reconstruction, comprising: The data processing module 11 is used to generate finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and to construct a stress snapshot matrix based on the node data in the finite element stress simulation data. The model training module 12 is used to perform mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes and POD mode coefficients corresponding to each drilling condition, and to train the Kriging surrogate model using the condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model. The stress field reconstruction module 13 is used to input the real-time operating parameters of the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient, and reconstruct the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficient, the mean field and the POD mode. The safety assessment result determination module 14 is used to determine the corresponding drilling safety assessment result based on the full-node Mises stress field, so as to carry out drilling operations based on the drilling safety assessment result.

[0039] As can be seen, in this embodiment, a stress sample library covering typical drilling conditions is constructed through offline finite element simulation. The high-dimensional full-field stress is compressed into low-dimensional modal coefficients using POD (Position-Order Reduction), and then the Kriging model is used to establish the mapping relationship between drilling parameters and these coefficients. In the online stage, only real-time operating parameters need to be input to quickly reconstruct the Mises stress field of all nodes, with a response speed several orders of magnitude faster than direct finite element solutions. The reconstruction results can be used to identify high-stress areas and assess the spindle safety status, providing a basis for real-time adjustment of drilling parameters, thereby reducing non-productive time while ensuring drilling safety.

[0040] In some specific embodiments, the data processing module 11 may specifically include: The model building unit is used to obtain the three-dimensional geometric model of the top drive spindle to obtain the target finite element model of the top drive spindle; The boundary condition determination unit is used to determine multiple drilling conditions within the target drilling depth range based on a preset depth interval, and to determine the load boundary conditions corresponding to each drilling condition. The simulation data acquisition unit is used to apply the load boundary conditions corresponding to each of the drilling conditions to the target finite element model and perform finite element solution to obtain the finite element stress simulation data corresponding to each of the drilling conditions. The data reading unit is used to read the node number, three-dimensional coordinates of the node, and Mises stress value of the node in the finite element stress simulation data corresponding to each drilling condition. The node determination unit is used to determine a unified node order using the node number and the node three-dimensional coordinates, and to perform a node consistency check on the finite element stress simulation data corresponding to each drilling condition; wherein, the node consistency check is to determine whether the number of nodes, node number and node three-dimensional coordinates are consistent among the finite element stress simulation data corresponding to each drilling condition. The matrix creation unit is used to arrange the Mises stress values ​​of each drilling condition into stress column vectors according to the unified node order if the number of nodes, node numbers and node three-dimensional coordinates of the finite element stress simulation data are consistent, and to combine the stress column vectors of each drilling condition into a stress snapshot matrix.

[0041] In some specific embodiments, the model training module 12 may specifically include: The snapshot matrix determination unit is used to determine the average value of the stress column vectors corresponding to each drilling condition in the stress snapshot matrix to obtain the mean field, and to determine the centered stress snapshot matrix based on the mean field and the stress column vectors corresponding to each drilling condition. The singular value determination unit is used to perform singular value decomposition on the centered stress snapshot matrix to obtain the candidate POD modes of each order and the singular values ​​corresponding to each candidate POD mode. The mode determination unit is used to determine the energy proportion of each candidate POD mode based on the singular values ​​corresponding to each order of the candidate POD modes, and to determine the POD mode based on the energy proportion and a preset cumulative energy threshold. The coefficient determination unit is used to project each of the drilling conditions in the centered stress snapshot matrix onto the POD mode to obtain the POD mode coefficients corresponding to each drilling condition.

[0042] In some specific embodiments, the model training module 12 may specifically include: The parameter value determination unit is used to standardize the working condition parameter values ​​corresponding to each of the drilling working conditions to obtain the standardized working condition parameter values ​​corresponding to each of the drilling working conditions. The tag generation unit is used to take the standardized operating condition parameter values ​​corresponding to each of the drilling conditions as input features and take the POD modal coefficients of each order corresponding to each of the drilling conditions as output tags. The model training unit is used to train the Kriging proxy model corresponding to each of the POD modal coefficients based on the input features and the output labels, so as to obtain the target coefficient prediction model corresponding to each order of the POD modal coefficients; Accordingly, the step of inputting real-time operating parameters from the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient includes: The coefficient prediction unit is used to standardize the real-time operating parameters and input the standardized real-time operating parameters into the target coefficient prediction model to obtain the POD prediction coefficients corresponding to each POD mode.

[0043] In some specific embodiments, the stress field reconstruction module 13 may specifically include: The stress fluctuation vector determination unit is used to perform a weighted summation of the POD modes of each order based on the POD prediction coefficients to obtain the stress fluctuation vector corresponding to the real-time operating parameters. The full node vector determination unit is used to superimpose the stress fluctuation vector with the mean field to obtain the full node Mises stress vector of the top drive spindle under the real-time operating parameters. The stress field reconstruction unit is used to associate the stress values ​​of each node in the full-node Mises stress vector with the node number and three-dimensional coordinates of the corresponding node in the target finite element model, so as to obtain the full-node Mises stress field of the top drive spindle under the real-time operating parameters; wherein, each node stress value in the full-node Mises stress field corresponds to a fixed node position in the target finite element model.

[0044] In some specific embodiments, the security assessment result determination module 14 may specifically include: The node visualization unit is used to generate a three-dimensional Mises stress cloud map of the top drive spindle based on the three-dimensional coordinates of each node in the target finite element model and the node Mises stress value of the corresponding node in the full node Mises stress field. A node location determination unit is used to determine the node location corresponding to the maximum Mises stress value in the full node Mises stress field. The node sorting unit is used to sort the nodes in the full node Mises stress field from largest to smallest based on the node Mises stress value, so as to extract the nodes with a preset proportion of the sorted nodes to obtain a high-stress node set, and to determine the target stress region and its spatial distribution of the top drive spindle based on the high-stress node set. The safety assessment result determination unit is used to determine the corresponding drilling safety assessment result based on the target stress area and its spatial distribution.

[0045] In some specific embodiments, the drilling apparatus based on top drive spindle stress field reconstruction may further include: The real-time judgment module is used to trigger a finite element verification operation to obtain the verification stress field under the current real-time operating parameters if the real-time operating parameters exceed the preset training sample range threshold, or the maximum stress relative error of the full node Mises stress field exceeds the preset threshold, or the deviation between the spatial distribution of the target stress region and the historical verification benchmark exceeds the preset range difference. The consistency check module is used to perform node consistency checks on the verification stress field based on a preset node check sequence. The sample update module is used to add the verification stress field and its corresponding real-time operating parameters as new samples to the stress snapshot matrix if the node consistency check is passed. The step jump module is used to jump to the step of performing mean field separation processing and singular value decomposition processing on the stress snapshot matrix.

[0046] Furthermore, embodiments of this application also disclose an electronic device, Figure 7This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0047] Figure 7 This is a schematic diagram of the structure of an electronic device 20 provided in an embodiment of this application. Specifically, the electronic device 20 may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the drilling method based on top drive spindle stress field reconstruction disclosed in any of the foregoing embodiments. Alternatively, the electronic device 20 in this embodiment may specifically be an electronic computer.

[0048] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0049] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored thereon can include operating system 221, computer program 222, etc., and the storage method can be temporary storage or permanent storage.

[0050] The operating system 221 is used to manage and control the various hardware devices on the electronic device 20 and the computer program 222, which may be Windows Server, Netware, Unix, Linux, etc. In addition to including a computer program capable of performing the drilling method based on top drive spindle stress field reconstruction executed by the electronic device 20 as disclosed in any of the foregoing embodiments, the computer program 222 may further include computer programs capable of performing other specific tasks.

[0051] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned drilling method based on top drive spindle stress field reconstruction. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0052] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0053] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0054] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0055] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0056] The technical solutions provided in this application have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A drilling method based on top drive spindle stress field reconstruction, characterized in that, include: Based on the pre-built target finite element model of the top drive spindle, finite element stress simulation data under different drilling conditions are generated, and a stress snapshot matrix is ​​constructed based on the node data in the finite element stress simulation data. The stress snapshot matrix is ​​subjected to mean field separation and singular value decomposition to obtain the mean field, POD modes and POD mode coefficients corresponding to each drilling condition. The Kriging surrogate model is then trained using the condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model. The real-time operating parameters of the drilling site are input into the target coefficient prediction model to obtain the POD prediction coefficients. Based on the POD prediction coefficients, the mean field, and the POD mode, the full-node Mises stress field of the top drive spindle under the real-time operating parameters is reconstructed. The corresponding drilling safety assessment results are determined based on the full-node Mises stress field, so that drilling operations can be carried out based on the drilling safety assessment results.

2. The drilling method based on top drive spindle stress field reconstruction according to claim 1, characterized in that, The method generates finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and constructs a stress snapshot matrix based on the node data in the finite element stress simulation data, including: Obtain the three-dimensional geometric model of the top drive spindle to obtain the target finite element model of the top drive spindle; Based on a preset depth interval, multiple drilling conditions are determined within the target drilling depth range, and the load boundary conditions corresponding to each drilling condition are determined. The load boundary conditions corresponding to each of the drilling conditions are applied to the target finite element model, and finite element solutions are performed to obtain finite element stress simulation data corresponding to each of the drilling conditions. Read the node number, node three-dimensional coordinates, and node Mises stress value of each node from the finite element stress simulation data corresponding to each drilling condition; A unified node order is determined using the node number and the node three-dimensional coordinates, and a node consistency check is performed on the finite element stress simulation data corresponding to each drilling condition; wherein, the node consistency check is to determine whether the number of nodes, node number, and node three-dimensional coordinates are consistent among the finite element stress simulation data corresponding to each drilling condition. If the number of nodes, the node number, and the three-dimensional coordinates of the nodes are consistent among the finite element stress simulation data, then the Mises stress values ​​of the nodes corresponding to each drilling condition are arranged into stress column vectors according to the unified node order, and the stress column vectors corresponding to each drilling condition are combined column by column to obtain a stress snapshot matrix.

3. The drilling method based on top drive spindle stress field reconstruction according to claim 1, characterized in that, The process of performing mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes, and POD mode coefficients corresponding to each of the drilling conditions includes: The average value of the stress column vectors corresponding to each drilling condition in the stress snapshot matrix is ​​determined to obtain the mean field, and the centered stress snapshot matrix is ​​determined based on the mean field and the stress column vectors corresponding to each drilling condition. Singular value decomposition is performed on the centered stress snapshot matrix to obtain the candidate POD modes of each order and the singular values ​​corresponding to each candidate POD mode; The energy proportion of each candidate POD mode is determined based on the singular values ​​corresponding to each order of the candidate POD modes, and the POD mode is determined based on the energy proportion and the preset cumulative energy threshold. The drilling conditions in the centered stress snapshot matrix are projected onto the POD modes to obtain the POD mode coefficients corresponding to each drilling condition.

4. The drilling method based on top drive spindle stress field reconstruction according to claim 1, characterized in that, The step of training the Kriging surrogate model using the operating condition parameter values ​​and the POD modal coefficients to obtain the target coefficient prediction model includes: The operating condition parameter values ​​corresponding to each drilling condition are standardized to obtain the standardized operating condition parameter values ​​corresponding to each drilling condition. The standardized operating condition parameter values ​​corresponding to each of the drilling conditions are used as input features, and the POD modal coefficients of each order corresponding to each of the drilling conditions are used as output labels. Based on the input features and the output labels, the Kriging proxy model corresponding to each of the POD modal coefficients is trained to obtain the target coefficient prediction model corresponding to each order of the POD modal coefficients; Accordingly, the step of inputting real-time operating parameters from the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient includes: The real-time operating parameters are standardized, and the standardized real-time operating parameters are input into the target coefficient prediction model to obtain the POD prediction coefficients corresponding to each POD mode.

5. The drilling method based on top drive spindle stress field reconstruction according to claim 1, characterized in that, The process of reconstructing the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficients, the mean field, and the POD modes includes: The stress fluctuation vector corresponding to the real-time operating parameters is obtained by weighted summation of the POD prediction coefficients for each order of the POD modes. The stress fluctuation vector is superimposed with the mean field to obtain the full-node Mises stress vector of the top drive spindle under the real-time operating parameters. The stress values ​​of each node in the full-node Mises stress vector are associated with the node number and three-dimensional coordinates of the corresponding node in the target finite element model to obtain the full-node Mises stress field of the top drive spindle under the real-time operating parameters; wherein, each node stress value in the full-node Mises stress field corresponds to a fixed node position in the target finite element model.

6. The drilling method based on top drive spindle stress field reconstruction according to any one of claims 1 to 5, characterized in that, The drilling safety assessment results determined based on the full-node Mises stress field include: Based on the three-dimensional coordinates of each node in the target finite element model and the node Mises stress values ​​of the corresponding nodes in the full node Mises stress field, a three-dimensional Mises stress cloud map of the top drive spindle is generated. Determine the node position corresponding to the maximum Mises stress value in the full-node Mises stress field; The nodes in the full node Mises stress field are sorted from largest to smallest based on the node Mises stress values, and a set of high-stress nodes is obtained by extracting a preset proportion of the nodes at the top of the sort. The target stress region and its spatial distribution of the top drive spindle are then determined based on the set of high-stress nodes. The corresponding drilling safety assessment results are determined based on the target stress region and its spatial distribution.

7. The drilling method based on top drive spindle stress field reconstruction according to claim 6, characterized in that, Also includes: If the real-time operating parameters exceed the preset training sample range threshold, or if the maximum stress relative error of the full-node Mises stress field exceeds the preset threshold, or if the spatial distribution of the target stress region deviates from the historical verification benchmark by more than the preset range difference, the finite element verification operation is triggered to obtain the verification stress field under the current real-time operating parameters. The verification stress field is checked for node consistency based on a preset node check sequence. If the node consistency check is passed, the verified stress field and its corresponding real-time operating parameters are added as new samples to the stress snapshot matrix. Jump to the step of performing mean field separation and singular value decomposition on the stress snapshot matrix.

8. A drilling apparatus based on top drive spindle stress field reconstruction, characterized in that, include: The data processing module is used to generate finite element stress simulation data under different drilling conditions based on the pre-built target finite element model of the top drive spindle, and to construct a stress snapshot matrix based on the node data in the finite element stress simulation data. The model training module is used to perform mean field separation and singular value decomposition on the stress snapshot matrix to obtain the mean field, POD modes and POD mode coefficients corresponding to each drilling condition, and to train the Kriging surrogate model using the condition parameter values ​​and the POD mode coefficients to obtain the target coefficient prediction model. The stress field reconstruction module is used to input the real-time operating parameters of the drilling site into the target coefficient prediction model to obtain the POD prediction coefficient, and reconstruct the full-node Mises stress field of the top drive spindle under the real-time operating parameters based on the POD prediction coefficient, the mean field and the POD mode. The safety assessment result determination module is used to determine the corresponding drilling safety assessment result based on the full-node Mises stress field, so as to carry out drilling operations based on the drilling safety assessment result.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor for executing the computer program to implement the drilling method based on top drive spindle stress field reconstruction as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program, which, when executed by a processor, implements the drilling method based on top drive spindle stress field reconstruction as described in any one of claims 1 to 7.

Images