A method for predicting rate of penetration based on CPLNet

Abstract

The application discloses a mechanical drilling speed prediction method based on a CPLNet, which comprises the following steps: collecting drilling parameters and preprocessing; the drilling parameters comprise drilling pressure, rotating speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density and drilling fluid density; spatial features and time features of the preprocessed drilling parameters are extracted; the spatial features and the time features are spliced in channel dimensions to obtain combined features; the combined features are fused and weighted to obtain fused features; and the mechanical drilling speed is predicted based on the fused features through a full connection network to obtain a drilling speed prediction value. The application combines a Peephole-LSTM and a CBAM attention mechanism, captures feature correlation and long memory, overcomes long-term dependence, and accurately predicts the drilling speed.

Description

Technical Field

[0001] This application relates to the field of intelligent monitoring technology in petroleum engineering, and more specifically, to a mechanical drilling rate prediction method based on CPLNet. Background Technology

[0002] In oil drilling engineering, the rate of penetration (ROP) refers to the speed at which the drill bit rotates during drilling. The ROP has a significant impact on drilling efficiency and quality. Existing data-driven time-series prediction methods for ROP have achieved good results. With the substantial improvement in computer preprocessing capabilities and the rapid advancement of artificial intelligence, ROP prediction methods based on machine learning are gradually emerging.

[0003] Traditional mechanical drilling rate prediction, often using a single LSTM or BiLSTM model, may fail to fully utilize the spatiotemporal information in the data, leading to reduced prediction accuracy. Furthermore, it fails to capture nonlinear relationships and interactions within the data, resulting in decreased accuracy and generalization. Moreover, when the input time series is too long, the model struggles to capture long-term dependencies, causing its learning performance to deteriorate with increasing time series length. Summary of the Invention

[0004] To address the aforementioned issues, this application provides a mechanical drilling rate prediction method based on CPLNet, aiming to improve the accuracy of drilling rate prediction by leveraging Peephole-LSTM long memory and CBAM dynamic weighting.

[0005] The first aspect of this invention provides a method for predicting mechanical drilling speed based on CPLNet, comprising: Collect and preprocess drilling parameters; drilling parameters include drilling pressure, rotation speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density; Extracting spatial and temporal characteristics of preprocessed drilling parameters; Spatial and temporal features are concatenated along the channel dimension to obtain combined features; the combined features are then fused and weighted to obtain fused features. Based on the fusion characteristics, the mechanical drilling rate is predicted through a fully connected network to obtain the drilling rate prediction value.

[0006] Furthermore, the preprocessing includes: data cleaning, wavelet transform denoising, feature engineering, Z-score normalization, and time window alignment.

[0007] Furthermore, in the spatial feature extraction process, the preprocessed drilling parameters are sequentially passed through a convolutional layer, a first batch normalization layer, and an adaptive pooling layer to obtain the spatial features of the preprocessed drilling parameters.

[0008] Furthermore, in the process of extracting time features, the preprocessed drilling parameters are sequentially passed through a multi-layer Peephole-LSTM network, a Dropout layer, and a second batch normalization layer to obtain the time features of the preprocessed drilling parameters. The multilayer Peephole-LSTM network includes memory units and gating units; the gating units include input gates, forget gates, and output gates; wherein, the forget gate, input gate, and output gate are all connected to the cell state in the memory unit through a peephole mechanism.

[0009] Furthermore, the calculation formula for the gate unit of the multilayer Peephole-LSTM network is as follows:

[0010]

[0011] in, For input gate, For the Gate of Oblivion For output gate, Let be the external input vector at time step t. , , , For the input weight matrix, , , , To hide the weight matrix, , , For additional peephole parameters, This represents the cell state at the previous moment. Let t represent the cell state at time t. , , , Let σ be the bias vector, and σ be the sigmoid activation function. Let be the hidden state vector at time step t, and tanh be the hyperbolic tangent activation function.

[0012] Furthermore, the fusion and weighting of combined features based on the CBAM attention mechanism includes: Channel attention weighting:

[0013] in, Here is the channel attention weight matrix. It is the sigmoid activation function. W' cFor channel attention weights, This is a global average pooling feature. This is a global max-pooling feature; Spatial attention weighting:

[0014] in, It is the sigmoid activation function. Spatial attention weights, This is a global average pooling feature. This represents the global max pooling feature.

[0015] Furthermore, it also includes residual connection and position encoding of the fused features; The residual connection satisfies:

[0016]

[0017] in, For element-wise multiplication, For input features, Features after channel attention The output is the fused feature.

[0018] A second aspect of this invention provides a mechanical drilling rate prediction device based on CPLNet, the device comprising: The data acquisition module is used to collect and preprocess drilling parameters, including drilling pressure, rotational speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density. The feature extraction module is used to extract the spatial and temporal features of the preprocessed drilling parameters; The splicing and fusion module is used to splice spatial and temporal features along the channel dimension to obtain combined features; the combined features are then fused and weighted to obtain fused features. The network output module is used to predict the mechanical drilling rate based on the fusion features through a fully connected network, and obtain the drilling rate prediction value.

[0019] A third aspect of the present invention provides an electronic device, characterized in that it includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement a CPLNet-based mechanical drilling rate prediction method.

[0020] A fourth aspect of the present invention provides a computer-readable storage medium, characterized in that a computer program is stored on the computer-readable storage medium, and the computer program, when executed by a processor, is a mechanical drilling rate prediction method based on CPLNet.

[0021] In this embodiment, the application employs a hybrid dual-path architecture. This architecture uses parallel convolutional neural networks (CNNs) and long short-term memory networks (LSTMs) to process spatial and temporal features respectively, enabling the model to simultaneously understand both instantaneous snapshots and dynamic evolution processes of the data. The model utilizes a Peephole LSTM with "peephole connections," a design that enhances the network's ability to capture long-term temporal dependencies and improves prediction accuracy. During the feature fusion stage, the model cleverly integrates the CBAM attention mechanism, which allows the model to adaptively adjust weights within the fused high-dimensional features, focusing on the most informative feature dimensions and effectively improving the signal-to-noise ratio and prediction accuracy. Attached Figure Description

[0022] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart of a mechanical drilling rate prediction method based on CPLNet proposed in an embodiment of this application; Figure 2 This is a flowchart of the Peephole-LSTM structure proposed in one embodiment of this application; Figure 3 This is a flowchart illustrating the CBAM attention mechanism structure proposed in one embodiment of this application; Figure 4 This is a correlation heatmap proposed in one embodiment of this application; Figure 5 This is a model loss curve proposed in one embodiment of this application; Figure 6 This is a density comparison chart of actual and predicted values ​​proposed in one embodiment of this application; Figure 7 This is a comparison chart of the actual ROP curve and the predicted curve proposed in one embodiment of this application; Figure 8 This is a structural diagram of a mechanical drilling rate prediction device based on CPLNet proposed in an embodiment of this application; Figure 9 This is a schematic diagram of an electronic device according to this application. Detailed Implementation

[0024] 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, 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.

[0025] Please refer to Figure 1 , Figure 1 This is a flowchart of a CPLNet-based mechanical drilling rate prediction method proposed in one embodiment of this application. Figure 1 As shown, a mechanical drilling rate prediction method based on CPLNet includes: Collect and preprocess drilling parameters; drilling parameters include drilling pressure, rotation speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density; Extracting spatial and temporal characteristics of preprocessed drilling parameters; Spatial and temporal features are concatenated along the channel dimension to obtain combined features; the combined features are then fused and weighted to obtain fused features. Based on the fusion characteristics, the mechanical drilling rate is predicted through a fully connected network to obtain the drilling rate prediction value.

[0026] In this embodiment, various data collected during the drilling process are compiled into a dataset, and correlation analysis is performed on it to create a heat map. The raw data is then cleaned and normalized, and the mechanical drilling rate is labeled in conjunction with the collected drilling data.

[0027] The dataset, as shown in Table 1, includes drilling parameters such as drilling pressure, rotational speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density. Table 1

[0028] The dataset is divided as follows: 80% training set and 20% validation set. The core parameters of the Peephole-LSTM and CBAM hybrid attention mechanism model are configured as follows: the input sequence time step seq_length is set to 10, the hidden layer dimension d_model is set to 256, the initial number of CNN channels is 64, the number of Peephole LSTM encoder layers n_encorder_layers is set to 2, the CBAM attention window size windows_size is set to 8, the dropout rate dropout_rate is set to 0.2, the number of units in the neural network layer is set to 64, the 1D convolutional layer kern_size is set to 3, the stride is 1, and the activation function is set to ReLU.

[0029] The concatenation of temporal and spatial features mainly involves processing the temporal features using Temporal Peephole LSTM, then processing the spatial features using Spatial CNN, and finally obtaining the concatenated features through normalization.

[0030] Furthermore, the preprocessing includes: data cleaning, wavelet transform denoising, feature engineering, Z-score normalization, and time window alignment.

[0031] In this embodiment, the normalization process uses the min-max normalization method to map features of different dimensions to a unified scale. Correlation analysis is then performed to obtain a heatmap of correlation analysis for each feature. Please refer to [reference needed]. Figure 4 , Figure 4 This is a correlation heatmap proposed in one embodiment of this application, to facilitate model training. Noise reduction is achieved using wavelet transform combined with a thresholding method.

[0032] Furthermore, in the spatial feature extraction process, the preprocessed drilling parameters are sequentially passed through a convolutional layer, a first batch normalization layer, and an adaptive pooling layer to obtain the spatial features of the preprocessed drilling parameters.

[0033] Furthermore, in the process of extracting time features, the preprocessed drilling parameters are sequentially passed through a multi-layer Peephole-LSTM network, a Dropout layer, and a second batch normalization layer to obtain the time features of the preprocessed drilling parameters. The multilayer Peephole-LSTM network includes memory units and gating units; the gating units include input gates, forget gates, and output gates; wherein, the forget gate, input gate, and output gate are all connected to the cell state in the memory unit through a peephole mechanism.

[0034] In this embodiment, please refer to Figure 2 , Figure 2 This is a flowchart of the Peephole-LSTM structure proposed in one embodiment of this application. Figure 2 As shown, Peephole LSTM consists of multiple memory units and three gated representations: forget gate, input gate, and output gate. Unlike the LSTM structure, Peephole LTTM innovatively adds a peephole mechanism to the gated units, enabling the model to extract the cell state in the memory units and improve long-term dependent capture.

[0035] Furthermore, the calculation formula for the gate unit of the multilayer Peephole-LSTM network is as follows:

[0036]

[0037] in, For input gate, For the Gate of Oblivion For output gate, Let be the external input vector at time step t. , , , For the input weight matrix, , , , To hide the weight matrix, , , For additional peephole parameters, This represents the cell state at the previous moment. Let t represent the cell state at time t. , , , Let σ be the bias vector, and σ be the sigmoid activation function. Let be the hidden state vector at time step t, and tanh be the hyperbolic tangent activation function.

[0038] In this embodiment, This represents the output of short-term memory at time t and the input at the next time step. Short-term memory is the hidden state. Output memory as One weighted component of long-term memory influences decision-making.

[0039] Furthermore, the fusion and weighting of combined features based on the CBAM attention mechanism includes: Channel attention weighting:

[0040] in, Here is the channel attention weight matrix. It is the sigmoid activation function. W' c For channel attention weights, This is a global average pooling feature. This is a global max-pooling feature; Spatial attention weighting:

[0041] in, It is the sigmoid activation function. Spatial attention weights, This is a global average pooling feature. This represents the global max pooling feature.

[0042] In this embodiment, please refer to Figure 3 , Figure 3 This is a flowchart illustrating the CBAM attention mechanism structure proposed in one embodiment of this application. Figure 3 As shown, the workflow of the CBAM attention mechanism is as follows: First, channel attention is applied to obtain channel information through max pooling and average pooling. Then, a shared multilayer perceptron (MLP) is used to process the pooling results. The results of the two branches are then added together and the channel attention weights are obtained through sigmoid. Finally, the input features are multiplied by the channel attention weights to obtain the channel attention weight matrix M. c : Channel attention weight matrix M c , can be represented as In order to reduce the computational parameters, in W c A dimensionality reduction coefficient r is used in the process.

[0043] The added spatial attention mechanism performs max pooling and average pooling separately along the channel dimension, concatenates the pooling results along the channel dimension, and generates spatial attention weights through convolutional layers and the sigmoid function: Spatial attention weight matrix M s , can be represented as Two pooling methods are used to generate feature maps along the channel dimension:

[0044] Multiplying the features by the spatial attention weights yields the spatial attention weight matrix M. s Finally, add residual connections in the add layer.

[0045] Furthermore, it also includes residual connection and position encoding of the fused features; The residual connection satisfies:

[0046]

[0047] in, For element-wise multiplication, For input features, Features after channel attention The output is the fused feature.

[0048] In this embodiment, the prediction results are evaluated using the mean squared error (MSE), mean absolute error (MAE), and correlation coefficient R. 2 As an evaluation metric, the mean squared error loss function is calculated using the following formula:

[0049] Where n is the number of samples, y i For sample labels, Predict the probability for the sample.

[0050] Generate loss functions for the training and test sets, test the results to determine if overfitting is occurring, and output a comparative graph for visualization. Please refer to [link / reference]. Figure 5 , Figure 5 This is a model loss curve proposed in one embodiment of this application. Figure 5 As shown, MAE, MSE, and R are generated simultaneously. 2 Among them, the smaller the MAE and MSE, the better the prediction effect. 2 The closer the value is to 1, the better the fitted regression model is.

[0051] The performance metrics of the CPLNet-based mechanical drilling rate prediction method are shown in Table 2, which contains the model performance parameters. Table 2

[0052] The following optimization techniques are applied to the model: the AdamW optimizer with weight decay and gradient pruning is used, and the learning rate is scheduled using the Bayesian Optimizer strategy. The learning rate is reduced when the validation loss stops improving. The Bayesian Optimizer strategy can be implemented using the Bayesian Optimizer library in Keras. After all metrics are satisfactory, the trained mechanical drilling rate prediction model can be used to predict the mechanical drilling rate. Please refer to [link / reference]. Figure 6 , Figure 7 , Figure 6 This is a density comparison chart of actual and predicted values ​​proposed in one embodiment of this application; Figure 7This is a comparison chart of the actual and predicted ROP curves according to an embodiment of this application. Figure 6 As shown in the density plot, the more high-density points a straight line passes through, the more accurate the prediction. Figure 7 As shown, the greater the overlap between the predicted value curve and the actual value curve, the more accurate the prediction.

[0053] Please refer to Figure 8 , Figure 8 This is a structural diagram of a CPLNet-based mechanical drilling rate prediction device proposed in one embodiment of this application. Figure 8 As shown in the figure, this disclosure also provides a mechanical drilling rate prediction device based on CPLNet, the device including: a data acquisition module 801, a feature extraction module 802, a splicing and fusion module 803, and a network output module 804; The data acquisition module 801 is used to acquire drilling parameters and perform preprocessing; the drilling parameters include drilling pressure, rotation speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density and drilling fluid density. Feature extraction module 802 is used to extract the spatial and temporal features of the preprocessed drilling parameters; The splicing and fusion module 803 is used to splice spatial features and temporal features along the channel dimension to obtain combined features; and to perform fusion and weighting on the combined features to obtain fused features. The network output module 804 is used to predict the mechanical drilling rate based on the fusion features through a fully connected network, and obtain the drilling rate prediction value.

[0054] This disclosure also provides an electronic device, please refer to... Figure 9 , Figure 9 This is a schematic diagram of an electronic device illustrated in an embodiment of this disclosure. For example... Figure 9 As shown, the electronic device 100 includes a memory 110 and a processor 120. The memory 110 and the processor 120 are connected via a bus. The memory 110 stores a computer program that can run on the processor 120 to implement the steps in the CPLNet-based mechanical drilling rate prediction method disclosed in this embodiment.

[0055] The disclosed embodiments also provide a computer-readable storage medium that, when the instructions in the computer-readable storage medium are executed by a processor of a computer device, enables the computer device to perform the steps in the CPLNet-based mechanical drilling rate prediction method of the present disclosure embodiments.

[0056] In summary, this invention fully leverages the advantages of CNN and Peephole LSTM to achieve deep feature extraction from drilling time-series data. The peephole structure, by adding directly monitorable and controllable cell states to the gating units, acquires more feature information, enabling the model to more accurately control the memory flow and enhancing its ability to retain long-term dependent information. The CBAM module, through channel and spatial attention and self-learning and self-modifying weight adjustments, dynamically acquires the importance of each feature, thereby further strengthening feature representation. Overall, the current model outperforms traditional LSTM and BiLSTM models in terms of long-sequence modeling ability, prediction accuracy, and stability. Furthermore, this invention employs a more reliable evaluation metric method, making the model's predictions more standardized and solving the drilling speed prediction problem.

[0057] This application describes embodiments with reference to flowchart illustrations and / or block diagrams of methods, apparatuses, electronic devices, and computer program products according to embodiments of this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0058] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0059] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal equipment, causing a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0060] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0061] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0062] Although preferred embodiments of the present application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of the present application.

[0063] 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 terminal device 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 terminal device. 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 terminal device that includes said element.

[0064] The above provides a detailed description of a CPLNet-based mechanical drilling rate prediction method provided in this application. 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 method and its core ideas. 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 method for predicting mechanical drilling rate based on CPLNet, characterized in that, include: Collect and preprocess drilling parameters; drilling parameters include drilling pressure, rotation speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density; Extracting spatial and temporal characteristics of preprocessed drilling parameters; Spatial and temporal features are concatenated along the channel dimension to obtain combined features; the combined features are then fused and weighted to obtain fused features. Based on the fusion features, the mechanical drilling rate is predicted through a fully connected network to obtain the drilling rate prediction value; During the time feature extraction process, the preprocessed drilling parameters are sequentially passed through a multi-layer Peephole-LSTM network, a Dropout layer, and a second batch normalization layer to obtain the time features of the preprocessed drilling parameters. The multilayer Peephole-LSTM network includes memory units and gating units; the gating units include input gates, forget gates, and output gates; wherein, the forget gate, input gate, and output gate are all connected to the cell state in the memory unit through a peephole mechanism; The fusion and weighting of combined features based on the CBAM attention mechanism includes: Channel attention weighting: in, Here is the channel attention weight matrix. It is the sigmoid activation function. for Channel attention weights W' c for Channel attention weights This is a global average pooling feature. This is a global max-pooling feature; Spatial attention weighting: in, It is the sigmoid activation function. Here is the spatial attention weight matrix. Spatial attention weights, This is a global average pooling feature. This represents the global max pooling feature.

2. The mechanical drilling rate prediction method based on CPLNet according to claim 1, characterized in that, The preprocessing includes: data cleaning, wavelet transform denoising, feature engineering, Z-score normalization, and time window alignment.

3. The mechanical drilling rate prediction method based on CPLNet according to claim 1, characterized in that, In the spatial feature extraction process, the preprocessed drilling parameters are sequentially passed through a convolutional layer, a first batch normalization layer, and an adaptive pooling layer to obtain the spatial features of the preprocessed drilling parameters.

4. The mechanical drilling rate prediction method based on CPLNet according to claim 1, characterized in that, The calculation formula for the gated unit of the multi-layer Peephole-LSTM network is as follows: in, For input gate, For the Gate of Oblivion For output gate, Let be the external input vector at time step t. , , , For the input weight matrix, , , , To hide the weight matrix, , , For additional peephole parameters, This represents the cell state at the previous moment. Let t represent the cell state at time t. , , , Let σ be the bias vector, and σ be the sigmoid activation function. Let be the hidden state vector at time step t, and tanh be the hyperbolic tangent activation function.

5. The mechanical drilling rate prediction method based on CPLNet according to claim 1, characterized in that, It also includes residual connection and position encoding of the fused features; The residual connection satisfies: in, For element-wise multiplication, As input features, Features after channel attention The output is the fused feature.

6. A mechanical drilling rate prediction device based on CPLNet, characterized in that, The device includes: The data acquisition module is used to collect and preprocess drilling parameters, including drilling pressure, rotational speed, torque, well depth, natural gamma, resistivity, pump pressure, displacement, rock density, and drilling fluid density. The feature extraction module is used to extract the spatial and temporal features of the preprocessed drilling parameters; The splicing and fusion module is used to splice spatial and temporal features along the channel dimension to obtain combined features; the combined features are then fused and weighted to obtain fused features. The network output module is used to predict the mechanical drilling rate based on the fusion features through a fully connected network, and obtain the drilling rate prediction value. The feature extraction module is specifically configured to, during the temporal feature extraction process, pass the preprocessed drilling parameters sequentially through a multi-layer Peephole-LSTM network, a Dropout layer, and a second batch normalization layer to obtain the temporal features of the preprocessed drilling parameters. The multilayer Peephole-LSTM network includes memory units and gating units; the gating units include input gates, forget gates, and output gates; wherein, the forget gate, input gate, and output gate are all connected to the cell state in the memory unit through a peephole mechanism; The splicing and fusion module is specifically configured to perform fusion and weighting of combined features based on the CBAM attention mechanism: Channel attention weighting: in, Here is the channel attention weight matrix. It is the sigmoid activation function. for Channel attention weights W' c for Channel attention weights This is a global average pooling feature. This is a global max-pooling feature; Spatial attention weighting: in, It is the sigmoid activation function. Here is the spatial attention weight matrix. Spatial attention weights, This is a global average pooling feature. This represents the global max pooling feature.

7. An electronic device, characterized in that, include: A processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the CPLNet-based mechanical drilling rate prediction method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the CPLNet-based mechanical drilling rate prediction method as described in any one of claims 1 to 5.

Images