Drill bit wear prediction method and device, electronic equipment and storage medium

Abstract

The invention relates to a drill bit wear prediction method and device, electronic equipment and a storage medium, and belongs to the technical field of drill bit wear prediction.The drill bit wear prediction method comprises the steps that drill bit wear data are obtained and preprocessed; an initial drill bit wear prediction model is constructed, model parameters of the initial drill bit wear prediction model are optimized by adopting two different types of optimization algorithms based on the preprocessed drill bit wear data, and a first drill bit wear prediction model and a second drill bit wear prediction model which are completely trained are correspondingly obtained; and respectively inputting to-be-predicted drill bit wear data into a first drill bit wear prediction model and a second drill bit wear prediction model which are completely trained, correspondingly obtaining a first prediction result and a second prediction result, and carrying out weighted fusion on the first prediction result and the second prediction result to obtain a drill bit wear prediction result, and the precision and stability of drill bit wear prediction are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of drill bit wear prediction, and in particular to a drill bit wear prediction method, device, electronic equipment and storage medium. Background Art

[0002] Drill bit wear is an inevitable problem in drilling projects, which will affect drilling efficiency and quality. Therefore, accurate prediction of drill bit wear is of great significance for optimizing drilling technology and improving drilling efficiency.

[0003] At present, the methods for predicting drill bit wear mainly include empirical formula method, model simulation method and artificial intelligence algorithm. The empirical formula method is usually based on a large amount of experimental data and experience summary, but its scope of application is limited. Different formulas may be required for different drilling conditions and drill bit types, and it is difficult to take into account all influencing factors; the finite element model can simulate the stress and deformation of the drill bit during the drilling process, thereby predicting the degree of drill bit wear. However, this method requires complex calculations and a large number of input parameters, and it may be difficult to accurately simulate some complex working conditions; the discrete element model simulates the interaction between the drill bit and the rock through the interaction of discrete particles, thereby predicting the wear of the drill bit, but this method is more difficult to set the particle parameters and determine the boundary conditions, and also has certain limitations; deep learning algorithms have been widely used in drill bit wear prediction and can automatically learn the features and patterns in the data. However, deep learning algorithms require a large amount of training data and computing resources, and have high requirements for data quality and preprocessing, making it difficult to achieve accurate prediction of drill bit wear.

[0004] However, the accuracy of drill bit wear prediction is crucial to the efficient progress of drilling projects. Empirical formula methods, model simulation methods and artificial intelligence algorithms often find it difficult to accurately predict drill bit wear when faced with complex drilling conditions and changeable drill bit wear mechanisms. Summary of the Invention

[0005] In view of this, it is necessary to provide a drill bit wear prediction method, device, electronic device and storage medium to solve the technical problem of low accuracy of drill bit wear prediction when facing complex drilling conditions and changeable drill bit wear mechanisms.

[0006] In order to solve the above problems, in a first aspect, the present invention provides a drill bit wear prediction method, comprising: Acquiring drill bit wear data and preprocessing the drill bit wear data; An initial drill bit wear prediction model is constructed, and based on the preprocessed drill bit wear data, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained accordingly; The drill bit wear data to be predicted is respectively input into the first drill bit wear prediction model and the second drill bit wear prediction model that have been fully trained, and the first prediction result and the second prediction result are obtained accordingly, and the first prediction result and the second prediction result are weightedly fused to obtain the drill bit wear prediction result.

[0007] In a possible implementation, the drill bit wear data includes drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, material properties, and drill bit wear amount.

[0008] In a possible implementation, preprocessing the drill bit wear data includes: performing outlier removal and noise filtering on the drill bit wear data, and normalizing the filtered drill bit wear data; An automatic encoder is used to extract features from the normalized drill wear data to extract low-dimensional key features and complete the preprocessing of the drill wear data.

[0009] In one possible implementation, the model parameters of the initial drill bit wear prediction model include a learning rate, a number of neurons, and a convolution kernel size; the two different types of optimization algorithms include an improved particle swarm optimization-grey wolf optimization algorithm; based on the preprocessed drill bit wear data, the model parameters of the initial drill bit wear prediction model are optimized using two different types of optimization algorithms, respectively, to obtain a first drill bit wear prediction model and a second drill bit wear prediction model that are trained completely, including: When the preprocessed drill bit wear data is used to train the initial drill bit wear prediction model, a particle swarm of an improved particle swarm optimization algorithm is constructed based on the model parameters of the initial drill bit wear prediction model, and the improved particle swarm optimization algorithm is used to perform a coarse search on the model parameters to obtain an optimal candidate solution, wherein the improved particle swarm optimization algorithm updates the particle position and particle velocity by dynamically adjusting the nonlinear inertia weight and the learning factor; Taking the optimal candidate solution as the initial population, fine-tuning the parameters using the grey wolf optimization algorithm to obtain optimal parameters, and adjusting the initial drill bit wear prediction model based on the optimal parameters to obtain a fully trained first drill bit wear prediction model; In one possible implementation, the model parameters of the initial drill bit wear prediction model further include self-attention weights and parameters of the LSTM layer; the two different types of optimization algorithms further include Bayesian optimization; and based on the preprocessed drill bit wear data, the model parameters of the initial drill bit wear prediction model are optimized using two different types of optimization algorithms, respectively, to obtain a first drill bit wear prediction model and a second drill bit wear prediction model that are trained completely, further comprising: Bayesian optimization is used to optimize the model parameters of the initial drill bit wear prediction model, wherein a probability model of the parameter space is constructed using Bayesian optimization, global optimal parameters are obtained based on the probability model, and the initial drill bit wear prediction model is adjusted based on the global optimal parameters to obtain a fully trained second drill bit wear prediction model.

[0010] In one possible implementation, the initial drill bit wear prediction model includes a CNN layer, an LSTM layer, a self-attention mechanism, and a fully connected layer; and training the initial drill bit wear prediction model using the preprocessed drill bit wear data includes: The drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, and material properties in the pre-processed drill bit wear data are used as input signals of the initial drill bit wear prediction model, and the drill bit wear amount is used as the output signal of the initial drill bit wear prediction model; Performing a convolution operation on the input signal using the CNN layer to obtain local features; Performing a pooling operation on the local features, and extracting features from the pooled local features through an LSTM layer to obtain a time series feature vector; The time series feature vector is weighted through a self-attention mechanism, and the weighted time series feature vector is input into the fully connected layer to obtain a prediction result of drill bit wear.

[0011] In a possible implementation, the training of the initial drill bit wear prediction model using the preprocessed drill bit wear data further includes: The performance of the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained are comprehensively evaluated using the mean square error, mean absolute error, determination coefficient and symmetric mean percentage error. The weights of the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained are determined based on the comprehensive evaluation results.

[0012] In a second aspect, the present invention further provides a drill bit wear prediction device, comprising: A data acquisition module, configured to acquire drill bit wear data and pre-process the drill bit wear data; a model optimization module for constructing an initial drill bit wear prediction model, and optimizing the model parameters of the initial drill bit wear prediction model using two different types of optimization algorithms based on the preprocessed drill bit wear data, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely; The prediction module is used to input the drill bit wear data to be predicted into the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained, respectively, to obtain the first prediction result and the second prediction result respectively, and to perform weighted fusion on the first prediction result and the second prediction result to obtain the drill bit wear prediction result.

[0013] In a third aspect, the present invention further provides an electronic device, comprising: a processor and a memory; The memory stores a computer-readable program executable by the processor; When the processor executes the computer-readable program, the steps in the drill bit wear prediction method described above are implemented.

[0014] In a fourth aspect, the present invention also provides a computer-readable storage medium for storing computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the drill wear prediction method described in any one of the above-mentioned method items.

[0015] The beneficial effects of the present invention are as follows: based on the preprocessed drill bit wear data, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model, and a first drill bit wear prediction model and a second drill bit wear prediction model that are trained are obtained accordingly; the parameters of the initial drill bit wear prediction model are optimized by a three-step hybrid optimization of the IPSO-GWO optimization algorithm and the Bayesian optimization in the two different types of optimization algorithms, thereby reducing the complexity of hyperparameter tuning and greatly improving the adaptability of the model under complex drilling conditions; the drill bit wear data to be predicted is input into the first drill bit wear prediction model and the second drill bit wear prediction model that are trained accordingly, and a first prediction result and a second prediction result are obtained accordingly; the first prediction result and the second prediction result are weightedly fused to obtain a drill bit wear prediction result, thereby improving the accuracy and stability of drill bit wear prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For the technical personnel of the present invention, other drawings can be obtained based on these drawings without paying any creative work.

[0017] Figure 1A flow chart of an embodiment of a drill bit wear prediction method provided by the present invention; Figure 2 A schematic diagram of the IPSO-GWO hybrid optimization process of the drill bit wear prediction method provided by the present invention; Figure 3 A schematic diagram of the Bayesian optimization process of the drill bit wear prediction method provided by the present invention; Figure 4 A schematic structural diagram of an embodiment of a drill bit wear prediction device provided by the present invention; Figure 5 This is a schematic structural diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0018] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0019] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present invention. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute a separate or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0020] Before presenting the embodiments, the following terms are explained.

[0021] The Improved Particle Swarm Optimization (IPSO) algorithm is an improved version of the traditional Particle Swarm Optimization (PSO) algorithm. It aims to address the shortcomings of the standard PSO in terms of convergence speed, global search capability, and local extremum escape. IPSO improves the optimization performance of the algorithm by introducing new strategies, parameter adjustments, or mixing other algorithmic ideas.

[0022] The present invention discloses a drill bit wear prediction method, device, electronic device, and storage medium that can be used in a computer. The method, device, or computer-readable storage medium involved in the present invention can be integrated with the above-mentioned device or can be relatively independent.

[0023] A specific embodiment of the present invention discloses a drill bit wear prediction method, which can be executed by a computer, specifically by one or more processors of the computer. Figure 1 As shown in Figure 2, the drill bit wear prediction method includes: S101, obtaining drill bit wear data and preprocessing the drill bit wear data; It should be noted that the redundancy problem of the original data was improved by preprocessing the drill wear data.

[0024] S102, constructing an initial drill bit wear prediction model, and optimizing the model parameters of the initial drill bit wear prediction model using two different types of optimization algorithms based on the preprocessed drill bit wear data, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely; It should be noted that by optimizing the model parameters using two different types of optimization algorithms in the hybrid intelligent optimization strategy, namely the IPSO-GWO optimization algorithm and the Bayesian optimization algorithm, the parameter accuracy of the drill bit wear prediction model is improved, the complexity of hyperparameter tuning is reduced, and the accuracy and stability of drill bit wear prediction are improved. S103, inputting the drill bit wear data to be predicted into the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained, respectively, obtaining a first prediction result and a second prediction result respectively, and performing weighted fusion on the first prediction result and the second prediction result to obtain a drill bit wear prediction result.

[0025] It should be noted that the first drill bit wear prediction model and the second first drill bit wear prediction model are fused through a deep integration method to reduce the local defects of a single model and enhance the robustness of the system.

[0026] In some embodiments, in step S101, drill bit wear data is obtained and preprocessed. The drill bit wear data includes drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, material properties and drill bit wear amount. The drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, and material properties data are collected from the drilling process, and the actual drill bit wear amount is recorded as an output label; data cleaning technology is used to remove outliers and filter noise from the drill bit wear data, identify and remove outliers in the data, use a filtering algorithm to filter out noise, ensure data quality, normalize the filtered drill bit wear data, and use the Min-Max normalization formula to map the filtered drill bit wear data to Interval, eliminating the dimensional differences of different features to make the data comparable, the normalization formula is: , in, is the original data, namely the drill wear data, is the normalized data, 、 are the minimum and maximum values ​​of the characteristic data respectively; After normalizing the drill bit wear data, an automatic encoder is used to extract features from the normalized drill bit wear data to extract low-dimensional key features and complete the preprocessing of the drill bit wear data. An automatic encoder AE is used to reduce the dimension and denoise the normalized drill bit wear data, extract low-dimensional key features, enhance the representativeness of the data features, complete data preprocessing and deep feature extraction, and output the processed data for subsequent use.

[0027] In some embodiments, in step S102, an initial drill bit wear prediction model is constructed, and the initial drill bit wear prediction model includes a CNN layer, an LSTM layer, a self-attention mechanism, and a fully connected layer. The model parameters of the initial drill bit wear prediction model include a learning rate, a number of neurons, a convolution kernel size, and a self-attention weight; based on the preprocessed drill bit wear data, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model, and the two different types of optimization algorithms include an improved particle swarm optimization-grey wolf optimization algorithm and a Bayesian optimization, that is, the parameters of the initial drill bit wear prediction model are optimized using the IPSO-GWO optimization algorithm and the Bayesian optimization, respectively, to obtain a first drill bit wear prediction model and a second drill bit wear prediction model that are fully trained; the preprocessed drill bit wear data is used to train the initial drill bit wear prediction model, and the drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, and material properties in the preprocessed drill bit wear data are used as input signals of the initial drill bit wear prediction model, and the drill bit wear amount is used as the output signal of the initial drill bit wear prediction model. No.; the CNN layer is used to perform convolution operation on the input signal to obtain local features; its CNN layer includes a 1D convolution layer and a maximum pooling layer. The CNN layer performs a convolution operation on the input signal through the convolution kernel to extract the local spatial features of the data, performs a pooling operation on the local features, and performs a maximum pooling operation on the convolved features to reduce the feature dimension and reduce the amount of calculation. The local features after pooling are extracted through the LSTM layer to obtain a time series feature vector; the processed features are input to the LSTM layer. The LSTM layer uses a gating mechanism to capture long-term dependency information, learns the dynamic change law of drill bit wear, and obtains a time series feature vector. The time series feature vector is weighted by the self-attention mechanism. The self-attention mechanism weights the output of the LSTM layer, calculates the importance weights of different time steps, and highlights key information; the weighted time series feature vector is input to the fully connected layer to obtain the prediction result of drill bit wear. The features processed by the self-attention mechanism are input to the fully connected layer. The fully connected layer maps the features processed by the self-attention mechanism to the final prediction dimension and outputs the prediction result of drill bit wear; When the preprocessed drill bit wear data is used to train the initial drill bit wear prediction model, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model. The two different types of optimization algorithms use hybrid intelligent optimization strategies. The hybrid intelligent optimization strategy is divided into (improved particle swarm optimization algorithm) IPSO coarse search stage, (grey wolf optimization algorithm) GWO fine adjustment stage and Bayesian optimization stage, which makes the hyperparameter selection of the initial drill bit wear prediction model more accurate. Among them, IPSO balances exploration and development through dynamic adjustment strategy to avoid the problem of premature convergence of traditional PSO. GWO performs local tuning based on the global results of IPSO to improve parameter accuracy. The model parameters of the initial drill bit wear prediction model include learning rate, number of neurons and convolution kernel size. The learning rate, number of neurons and convolution kernel size of the initial drill bit wear prediction model are optimized by IPSO-GWO. The convolution kernel size is optimized. The process is as follows: based on the model parameters of the initial drill bit wear prediction model, a particle swarm of the improved particle swarm optimization algorithm is constructed, and the model parameters are roughly searched using the improved particle swarm optimization algorithm to obtain the optimal candidate solution. The improved particle swarm optimization algorithm updates the particle position and particle velocity by dynamically adjusting the nonlinear inertia weight and learning factor. With the optimal candidate solution as the initial population, the gray wolf optimization algorithm is used to fine-tune the model parameters to obtain the optimal parameters. The model parameters of the initial drill bit wear prediction model also include the self-attention weight and the parameters of the LSTM layer. The self-attention weight and the parameters of the LSTM layer of the initial drill bit wear prediction model are optimized through Bayesian optimization. The Bayesian optimization method is used to construct a probability model of the parameter space, and the global optimal parameters are obtained based on the probability model. For a flow chart of the IPSO-GWO hybrid optimization, please refer to Figure 2 In the IPSO rough search stage, the model parameters of the initial drill bit wear prediction model are initialized, that is, the particle swarm is initialized, and the nonlinear inertia weight and dynamic learning factor are defined by the particle swarm. The nonlinear inertia weight is adjusted to: , in, is the currently adjusted inertia weight value, is the maximum value of the inertia weight, is the minimum value of the inertia weight, The maximum number of adjustments, The number of times the current inertia weight is adjusted; Its dynamic learning factor is updated as: , , in, is the first dynamic learning factor, is the second dynamic learning factor, is the maximum value of the dynamic learning factor, is the minimum value of the dynamic learning factor, is the maximum number of iterations for dynamic learning factor updates, The current iteration number; The particle position and velocity are updated by dynamically adjusting the nonlinear inertia weight and the learning factor. The update rules of the particle position and velocity are as follows: , , in, is the velocity of the particle at the next moment, is the particle's velocity at the current moment, 、 is a random number, is the particle’s own best historical position, is the best historical position of the particle swarm, The position of the particle at the next moment, is the particle's current position; Calculate the fitness of each particle and evaluate the particle performance; check whether the preset termination conditions are met, such as reaching the maximum number of iterations or the fitness meets the requirements. If so, output the coarse search optimal candidate solution and enter the GWO fine adjustment stage; if not, continue to adjust the particle speed and position and proceed to the next round of search.

[0028] In the GWO fine-tuning phase, the optimal candidate solution of IPSO is used as the initial population to simulate the hunting behavior of gray wolves. The Gray Wolf Optimization Algorithm (GWO) is used to conduct a detailed search in the local area. By dynamically adjusting the positions of α, β, and δ wolves, the hyperparameter selection is further refined to avoid falling into the local optimum. The search range is dynamically adjusted and the hyperparameters, such as the number of neurons in the LSTM hidden layer, are refined. Specifically, the optimal solution obtained in the IPSO phase is used to initialize the α wolf: the optimal candidate solution obtained in the IPSO phase is used as the initial α wolf in the GWO phase; a group of gray wolves is randomly generated, each gray wolf represents a model Candidate solutions for parameters, calculate fitness and determine α, β, and δ wolves: Evaluate the fitness of each gray wolf and sort them according to fitness to determine α wolf (optimal solution), β wolf (second-optimal solution), and δ wolf (third-optimal solution). Update the position of the gray wolf. Based on the positions of α wolf, β wolf, and δ wolf, update the position of the gray wolf population, that is, update the candidate solutions for the model parameters. Dynamically shrink the search range: According to the change in fitness, dynamically adjust the search range to search for the optimal solution more finely; determine whether the termination conditions are met. If the maximum number of iterations is reached or the fitness improvement is no longer significant, end the GWO stage and output the final optimal parameters.

[0029] For a flow chart of Bayesian optimization, please refer to Figure 3 In the Bayesian optimization stage, the Bayesian optimization method is used to construct a probability model of the parameter space, and the global optimal parameters are obtained based on the probability model. First, the parameters to be optimized are defined, and the parameters are divided into continuous parameters (such as learning rate, L2 regularization coefficient) and discrete parameters (such as the number of convolution kernels, the number of attention heads). Through the definition of parameter space, a search space is provided for Bayesian optimization. A proxy model is initialized, and a Gaussian process is used to model the nonlinear relationship between hyperparameters and validation set errors, which is used to approximate the true objective function to accelerate the optimization process; data partitioning, the drill wear data is divided into a training set and a validation set. The training set is used for model training, and the validation set is used to evaluate model performance; generate new hyperparameters: use the proxy model to generate new hyperparameter combinations for training the CNN-LSTM initial drill wear prediction Test the model, call the model and train it, use the generated hyperparameter combination to initialize the CNN-LSTM initial drill wear prediction model, and use the training set data for training, verify the error calculation, and after each training cycle, use the verification set to calculate the verification error of the model to evaluate the model performance; update the proxy model, and update the proxy model based on the new verification error results to make it more accurately approximate the true objective function. Determine the termination condition: check whether the termination condition is met, such as whether the verification error no longer decreases significantly, reaches the optimal level, or reaches the maximum number of iterations. If the condition is met, proceed to the next step; otherwise, return to generate new hyperparameters and train the model; output the optimal hyperparameters: when the termination condition is met, output the optimal hyperparameter combination found by Bayesian optimization, that is, obtain the global optimal parameters.

[0030] The initial drill bit wear prediction model is adjusted based on the optimal parameters to obtain the first drill bit wear prediction model, that is, the initial drill bit wear prediction model is adjusted based on the parameters optimized by IPSO-GWO to obtain the first drill bit wear prediction model; the initial drill bit wear prediction model is adjusted based on the global optimal parameters to obtain a fully trained second drill bit wear prediction model, that is, the initial drill bit wear prediction model is adjusted by the parameters optimized by the Bayesian optimization method to obtain a fully trained second drill bit wear prediction model; the preprocessed drill bit wear data is divided into a training set, a test set, and a validation set. When the loss of the validation set does not decrease for 10 consecutive rounds, the training of the initial drill bit wear prediction model is terminated; during the training During the process, the mean square error (MSE), mean absolute error (MAE), coefficient of determination (R²) and symmetric mean percentage error (SMAPE) are used to comprehensively evaluate the performance of the first drill bit wear prediction model and the second drill bit wear prediction model. Based on the calculated evaluation indicators, it is judged whether the model performance meets the requirements, such as whether the prediction error is within an acceptable range. If the performance of the first drill bit wear prediction model and the second drill bit wear prediction model does not meet the requirements, the model optimization and prediction are performed again; if the requirements are met, the process is terminated and the prediction results are output. Through multi-model fusion, the local defects of the single model are reduced and the system robustness is enhanced.

[0031] In some embodiments, in step S103, the drill bit wear data to be predicted is input into a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained, respectively, to obtain a first prediction result and a second prediction result respectively, and the first prediction result and the second prediction result are weightedly fused to obtain a drill bit wear prediction result, the drill bit wear data to be predicted is predicted using the first drill bit wear prediction model that has been trained, to obtain a first prediction result, and the drill bit wear data to be predicted is predicted using the second drill bit wear prediction model that has been trained, to obtain a second prediction result, and the weights of the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained are determined based on a comprehensive evaluation result of the model, and based on the weights of the first drill bit wear prediction model and the second drill bit wear prediction model, an ensemble learning method is used to perform weighted averaging on the first prediction result and the second prediction result to obtain a final prediction result, that is, the weights of the first drill bit wear prediction model and the second drill bit wear prediction model are applied to the corresponding first prediction result and the second prediction result, and weighted averaging is performed to obtain the drill bit wear prediction result, and the final prediction result is a drill bit wear prediction value, and the final prediction result outputted can be used to guide drilling engineering.

[0032] In summary, the drill bit wear prediction method provided by the present invention obtains drill bit wear data and preprocesses the drill bit wear data; constructs an initial drill bit wear prediction model, and based on the preprocessed drill bit wear data, uses two different types of optimization algorithms to optimize the model parameters of the initial drill bit wear prediction model, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely; the drill bit wear data to be predicted is input into the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained completely, thereby obtaining a first prediction result and a second prediction result, and performing weighted fusion on the first prediction result and the second prediction result to obtain a drill bit wear prediction result, thereby improving the accuracy and stability of drill bit wear prediction.

[0033] In order to better implement the drill bit wear prediction method in the embodiment of the present invention, based on the drill bit wear prediction method, correspondingly, Figure 4 As shown, an embodiment of the present invention further provides a drill bit wear prediction device for identifying sepiolite in rocks based on a well-trained hybrid neural network model. The hybrid neural network model includes a feature extraction layer, an early fusion layer, a mid-term fusion layer, and a late fusion layer. The drill bit wear prediction device 400 includes: The data acquisition module 401 is used to acquire drill bit wear data and pre-process the drill bit wear data; A model optimization module 402 is configured to construct an initial drill bit wear prediction model. Based on the preprocessed drill bit wear data, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely. The prediction module 403 is used to input the drill wear data to be predicted into the first drill wear prediction model and the second drill wear prediction model that have been trained, respectively, to obtain the first prediction result and the second prediction result respectively, and to perform weighted fusion on the first prediction result and the second prediction result to obtain the drill wear prediction result.

[0034] like Figure 5 As shown, the present invention also provides an electronic device 500 , which can be a computing device such as a mobile terminal, a desktop computer, a notebook, a palmtop computer, or a server. The electronic device 500 includes a processor 501 , a memory 502 , and a display 503 . Figure 5 Only some of the components of the electronic device 500 are shown, but it should be understood that implementation of all of the shown components is not required, and more or fewer components may be implemented instead.

[0035] In some embodiments, the memory 502 may be an internal storage unit of the electronic device 500, such as the hard drive or memory of the electronic device 500. In other embodiments, the memory 502 may also be an external storage device of the electronic device 500, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. Furthermore, the memory 502 may include both the internal storage unit of the electronic device 500 and an external storage device. The memory 502 is used to store application software installed on the electronic device 500 and various data, such as program code installed on the electronic device 500. The memory 502 may also be used to temporarily store data that has been output or is about to be output. In one embodiment, the memory 502 stores a drill wear prediction program, which can be executed by the processor 501 to implement the drill wear prediction method of various embodiments of the present invention.

[0036] In some embodiments, the processor 501 may be a central processing unit (CPU), a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 502 , such as a drill wear prediction method.

[0037] In some embodiments, display 503 can be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 503 is used to display identification information of the drill wear prediction program and to display a visual user interface. Components 501-503 of electronic device 500 communicate with each other via a system bus.

[0038] In some embodiments, when the processor 501 executes the drill wear prediction program in the memory 502, the various steps in the drill wear prediction method described in the above embodiments are implemented. Since the drill wear prediction method has been described in detail above, it will not be repeated here.

[0039] Accordingly, the present invention also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, the steps or functions of the drill wear prediction method provided in the above-mentioned method embodiments can be implemented.

[0040] Those skilled in the art will appreciate that all or part of the process steps of the above-described embodiments can be implemented by instructing related hardware through a computer program, and the program can be stored in a computer-readable storage medium, such as a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0041] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or replacements that can be easily conceived by any technician familiar with the technical neighbors within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A drill bit wear prediction method, characterized in that: include: Acquiring drill bit wear data and preprocessing the drill bit wear data; An initial drill bit wear prediction model is constructed, and based on the preprocessed drill bit wear data, two different types of optimization algorithms are used to optimize the model parameters of the initial drill bit wear prediction model, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely; The drill bit wear data to be predicted is respectively input into the first drill bit wear prediction model and the second drill bit wear prediction model that have been fully trained, and the first prediction result and the second prediction result are correspondingly obtained, and the first prediction result and the second prediction result are weightedly fused to obtain the drill bit wear prediction result.

2. The drill bit wear prediction method according to claim 1, characterized in that: The drill bit wear data includes drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, material properties and drill bit wear amount.

3. The drill bit wear prediction method according to claim 2, characterized in that: The pre-processing of the drill bit wear data includes: performing outlier removal and noise filtering on the drill bit wear data, and normalizing the filtered drill bit wear data; An automatic encoder is used to extract features from the normalized drill wear data to extract low-dimensional key features and complete the preprocessing of the drill wear data.

4. The drill bit wear prediction method according to claim 3, characterized in that: The model parameters of the initial drill bit wear prediction model include a learning rate, a number of neurons, and a convolution kernel size. The two different types of optimization algorithms include an improved particle swarm optimization-grey wolf optimization algorithm. Based on the preprocessed drill bit wear data, the model parameters of the initial drill bit wear prediction model are optimized using two different types of optimization algorithms, respectively, to obtain a first drill bit wear prediction model and a second drill bit wear prediction model that are fully trained, including: When the preprocessed drill bit wear data is used to train the initial drill bit wear prediction model, a particle swarm of an improved particle swarm optimization algorithm is constructed based on the model parameters of the initial drill bit wear prediction model, and the improved particle swarm optimization algorithm is used to perform a coarse search on the model parameters to obtain an optimal candidate solution, wherein the improved particle swarm optimization algorithm updates the particle position and particle velocity by dynamically adjusting the nonlinear inertia weight and the learning factor; The optimal candidate solution is used as the initial population, and the gray wolf optimization algorithm is used to fine-tune the model parameters to obtain the optimal parameters. The initial drill bit wear prediction model is adjusted based on the optimal parameters to obtain a fully trained first drill bit wear prediction model.

5. The drill bit wear prediction method according to claim 4, characterized in that: The model parameters of the initial drill bit wear prediction model also include self-attention weights and parameters of the LSTM layer, and the two different types of optimization algorithms also include Bayesian optimization; based on the preprocessed drill bit wear data, the model parameters of the initial drill bit wear prediction model are optimized using two different types of optimization algorithms, respectively, to obtain a first drill bit wear prediction model and a second drill bit wear prediction model that are trained completely, and further include: Bayesian optimization is used to optimize the model parameters of the initial drill bit wear prediction model, wherein a probability model of the parameter space is constructed using Bayesian optimization, global optimal parameters are obtained based on the probability model, and the initial drill bit wear prediction model is adjusted based on the global optimal parameters to obtain a fully trained second drill bit wear prediction model.

6. The drill bit wear prediction method according to claim 4, characterized in that: The initial drill bit wear prediction model includes a CNN layer, an LSTM layer, a self-attention mechanism, and a fully connected layer; the initial drill bit wear prediction model is trained using the preprocessed drill bit wear data, including: The drilling force, torque, vibration, acoustic emission signal, drill bit geometric parameters, and material properties in the pre-processed drill bit wear data are used as input signals of the initial drill bit wear prediction model, and the drill bit wear amount is used as the output signal of the initial drill bit wear prediction model; Performing a convolution operation on the input signal using the CNN layer to obtain local features; Performing a pooling operation on the local features, and extracting features from the pooled local features through an LSTM layer to obtain a time series feature vector; The time series feature vector is weighted through a self-attention mechanism, and the weighted time series feature vector is input into the fully connected layer to obtain a prediction result of drill bit wear.

7. The drill bit wear prediction method according to claim 6, characterized in that: The training of the initial drill bit wear prediction model using the pre-processed drill bit wear data further includes: The performance of the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained are comprehensively evaluated using the mean square error, mean absolute error, determination coefficient and symmetric mean percentage error. The weights of the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained are determined based on the comprehensive evaluation results.

8. A drill wear prediction device, characterized in that: include: A data acquisition module, configured to acquire drill bit wear data and pre-process the drill bit wear data; a model optimization module for constructing an initial drill bit wear prediction model, and optimizing the model parameters of the initial drill bit wear prediction model using two different types of optimization algorithms based on the preprocessed drill bit wear data, thereby obtaining a first drill bit wear prediction model and a second drill bit wear prediction model that have been trained completely; The prediction module is used to input the drill bit wear data to be predicted into the first drill bit wear prediction model and the second drill bit wear prediction model that have been trained, respectively, to obtain the first prediction result and the second prediction result respectively, and to perform weighted fusion on the first prediction result and the second prediction result to obtain the drill bit wear prediction result.

9. An electronic device, characterized in that: including memory and processor; The memory stores a computer-readable program executable by the processor; When the processor executes the computer-readable program, the steps of the drill bit wear prediction method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps of the drill wear prediction method according to any one of claims 1 to 7.

Images