Semi-automatic drilling operation data mining method, device and equipment based on mode segmentation algorithm

Abstract

The invention discloses a semi-automatic drilling operation data mining method, device and equipment based on a mode segmentation algorithm, and the method comprises the steps: collecting drilling operation parameter data, and processing the drilling operation parameter data to obtain a multi-dimensional drilling parameter time sequence; mode mining is conducted on the multi-dimensional drilling parameter time sequence; and inputting the excavated mode into a drilling working condition classifier to obtain a drilling working condition classification result. Therefore, the method can be widely applied to the field of petroleum drilling and completion engineering.

Description

Technical Field

[0001] This invention relates to a semi-automatic drilling operation data mining method, apparatus, and equipment based on a pattern segmentation algorithm, and pertains to the field of oil drilling and completion technology. Background Technology

[0002] In the field of oil drilling and completion engineering, retrospective analysis of historical drilling processes is an important basis for optimizing engineering decisions and is of great significance for ensuring drilling operation safety and improving drilling efficiency.

[0003] Currently, the retrospective work on historical drilling operations mainly includes three aspects: (1) Timeliness analysis based on operation logs: Based on text processing technology, the data of manually filled drilling daily reports is read, and the timeliness of drilling operations is analyzed based on this. However, the data records the operation status relatively roughly and may contain errors and omissions, which cannot support more refined optimization of the drilling process. (2) Well log analysis: Real-time operation logging data at the well site records the entire drilling process in detail, but lacks a mapping relationship with the drilling operation status. Generally, depth or time curves are drawn using computer graphics technology, and professionals observe and analyze the graphs. Due to the large amount of data, this method is time-consuming and labor-intensive, and cannot fully utilize the statistical characteristics of the full amount of big data to mine effective information. (3) Data-driven intelligent working condition identification method: Some patented technologies and research papers propose to use machine learning methods to develop a working condition identification model based on real-time well site data, automatically construct a mapping dataset between logging data and drilling operation status, thereby realizing the retrospective analysis of historical operations. However, this type of method itself requires a large amount of logging data with correctly labeled operation status as training samples. Because data annotation is a labor-intensive task, only a small batch of manually annotated datasets are usually available for training, resulting in low model performance, poor generalization performance, and limited applicability. Summary of the Invention

[0004] This invention aims to at least solve one of the technical problems existing in the prior art. Therefore, in response to the above-mentioned problems, the purpose of this invention is to provide a semi-automatic method, apparatus, and equipment for mining historical drilling operation data based on an unsupervised pattern segmentation algorithm. This method can quickly mine potential patterns in massive historical operation data, assisting in the efficient acquisition of operational condition annotation data, thereby making the analysis and utilization of historical drilling operation data in the drilling engineering field easier.

[0005] To achieve the above-mentioned objectives, the technical solution adopted by this invention is as follows: In a first aspect, the present invention provides a semi-automatic drilling operation data mining method based on a pattern segmentation algorithm, comprising: Collect drilling operation parameter data and process the drilling operation parameter data to obtain a multidimensional drilling parameter time series; Pattern mining of multidimensional drilling parameter time series; Input the excavation pattern into the drilling condition classifier to obtain the drilling condition classification results.

[0006] Some possible implementations involve processing drilling operation parameter data, including: Several characteristic parameters related to drilling conditions were selected; Data cleaning is performed on the selected feature parameters; Standardize the cleaned data.

[0007] In some possible implementations, several characteristic parameters related to drilling conditions are selected, including: hook height, drill bit measurement depth, measurement well depth, pump pressure, cumulative pump strokes, inlet flow rate, torque, drilling pressure, rotational speed, mechanical drilling speed, suspended weight, and outlet flow rate.

[0008] In some possible implementations, the training process for the drilling condition classifier is as follows: Drilling operation parameter data of several wells were collected from historical drilling operation data. Several characteristic parameters related to drilling conditions were selected, and the selected characteristic parameters related to drilling conditions were processed to obtain a multidimensional drilling parameter time series. The AutoPlait algorithm is used to automatically segment and mine patterns in the time series of multidimensional drilling parameters. The labeled data of the actual drilling conditions corresponding to each mode constitute the training dataset; Based on the training dataset, a drilling condition classifier is trained using a convolutional neural network model or an ensemble learning model.

[0009] Some possible implementations involve using the AutoPlait algorithm to automatically segment and mine patterns in multidimensional drilling parameter time series, including: Use a cascaded Markov framework to model the data and obtain model parameters; Construct a cost function based on the minimum length theory; The model parameters are optimized based on the cost function and a three-level nested algorithm.

[0010] Some possible implementations involve using a cascaded Markov framework to model the data and obtain model parameters, including: Segment: Divide X into m consecutive segments The total number of segments is m, and X is a time series of multidimensional drilling operation parameters for a well: X={ }, data dimensions are d , length is n ; Pattern regime: Θ = { r HMMs ,…, }, each Include Each HMM is It is a potential regime of data, with a total number of patterns of r. Here is the state transition matrix. To observe the probability distribution, The initial state distribution; Model structure: C={m,r,S,Θ,F}, where F gives the region to which each segment belongs. .

[0011] In some possible implementations, the cost function is constructed based on the minimum length theory:

[0012] ; in, The encoding cost for each HMM, Let X be the encoding cost of data X under the conditions of a cascaded Markov model.

[0013] In some possible implementations, the model parameters are optimized based on a cost function using a three-layer nested algorithm. This three-layer nested algorithm consists of an innermost Cut Point Search algorithm, a middle Regime parameter estimation algorithm, and an outermost AutoPlait algorithm. The specific process is as follows: The Cut Point Search algorithm uses dynamic programming to find the optimal segmentation point on a time series. The input is the time series X and the current pattern set regime, and the output is the set of segments. The Regime parameter estimation algorithm includes the innermost Cut Point Search, which is used to estimate the parameters of the model. The input is the time series X, and the output is the parameters of the regime. The AutoPlait algorithm integrates two algorithms and performs iterative optimization. It starts by splitting the entire time series into one pattern. Each time, it calculates the cost function. If the cost decreases, it continues to split into new regions; otherwise, it maintains the current region.

[0014] Secondly, the present invention also provides a semi-automatic drilling operation data mining device based on a pattern segmentation algorithm, comprising: The data processing module is configured to collect drilling operation parameter data and process the drilling operation parameter data to obtain a multidimensional drilling parameter time series. The pattern mining module is configured to perform pattern mining on multidimensional drilling parameter time series; The working condition classification module is configured to input the excavation mode into the drilling working condition classifier to obtain the drilling working condition classification result.

[0015] Thirdly, the present invention also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to enable the processor to perform the method described thereon.

[0016] Because the present invention adopts the above technical solution, it has the following characteristics: 1. The present invention provides an automatic drilling condition identification method based on the AutoPlait algorithm, which can improve the accuracy of the condition identification model. By using the pre-segmentation of the AutoPlait algorithm, the integrity of the time series pattern caused by random segmentation during the training process of the drilling condition identification model is avoided, thereby improving the accuracy of drilling condition identification. At the same time, it enables more types of deep learning model architectures to be used for condition identification problems, thereby improving the level of intelligence in drilling operation management. 2. This invention uses the AutoPlait algorithm for automatic segmented calculation of drilling conditions as the core calculation of the historical drilling condition labeling and data analysis system, thereby making the collection of condition labeling data more convenient and the analysis and utilization of historical drilling operation data easier.

[0017] 3. This invention is an unsupervised machine learning method that requires no manual parameter tuning or prior training. In the task of labeling work condition classification data, it can automatically complete data processing and pattern recognition, greatly reducing the difficulty of manual labeling, improving work efficiency, and saving manual labeling time.

[0018] 4. This invention can capture state transition points more accurately. By automatically identifying state transition points, it can more accurately divide the different operation stages of the drilling process, making up for the shortcomings of data-driven models that have difficulty identifying condition transition points, and providing more accurate data support for drilling operation timeliness analysis.

[0019] 5. This invention learns the inherent patterns of local data in real time, without relying on human experience accumulation. It can adapt to different regions and drilling data quality environments, and has strong adaptability, robustness, and generalization ability.

[0020] In summary, this invention can be widely applied in the field of oil drilling and completion engineering. Attached Figure Description

[0021] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Throughout the drawings, the same reference numerals denote the same parts. In the drawings: Figure 1 This is a training architecture diagram of the drilling condition classifier according to an embodiment of the present invention; Figure 2 This is an example diagram showing the segmented results of drilling operation parameters according to an embodiment of the present invention; Figure 3 This is a schematic diagram of the cascaded Markov model framework according to an embodiment of the present invention; Figure 4 This is a flowchart of the Cut Point Search algorithm according to an embodiment of the present invention; Figure 5 This is a flowchart of the Regime parameter estimation algorithm according to an embodiment of the present invention; Figure 6 This is a flowchart of the AutoPlait algorithm according to an embodiment of the present invention; Figure 7 This is the user interface for the manual annotation function module in this embodiment of the invention. Figure 8 This is a structural diagram of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0022] It should be understood that the terminology used herein is for the purpose of describing particular exemplary embodiments only and is not intended to be limiting. Unless the context clearly indicates otherwise, the singular forms “a,” “an,” and “described” as used herein may also include the plural forms. The terms “comprising,” “including,” “containing,” and “having” are inclusive and therefore indicate the presence of the stated features, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, steps, operations, elements, components, and / or combinations thereof. The method steps, processes, and operations described herein are not construed as requiring them to be performed in a particular order described or illustrated unless the order of performance is explicitly indicated. It should also be understood that additional or alternative steps may be used.

[0023] Although terms such as first, second, third, etc., may be used in this document to describe multiple elements, components, regions, layers, and / or segments, these elements, components, regions, layers, and / or segments should not be limited by these terms. These terms may be used only to distinguish one element, component, region, layer, or segment from another. Unless the context clearly indicates otherwise, terms such as "first," "second," and other numerical terms used herein do not imply order or sequence. Therefore, the first element, component, region, layer, or segment discussed below may be referred to as the second element, component, region, layer, or segment without departing from the teachings of the exemplary embodiments.

[0024] For ease of description, spatial relative terms may be used in the text to describe the relationship of one element or feature relative to another element or feature as shown in the figure. These relative terms include, for example, "inside," "outside," "middle," "outer," "below," "above," etc. Such spatial relative terms are intended to include different orientations of the device in use or operation, other than those depicted in the figure.

[0025] The present invention provides a semi-automatic drilling operation data mining method, apparatus, and equipment based on a pattern segmentation algorithm, comprising: collecting drilling operation parameter data and processing the drilling operation parameter data to obtain a multidimensional drilling parameter time series; performing pattern mining on the multidimensional drilling parameter time series; and inputting the mined patterns into a drilling condition classifier to obtain drilling condition classification results. Therefore, the present invention requires no manual parameter tuning or pre-training, and can automatically complete data processing and pattern recognition in the task of labeling drilling condition classification data, significantly reducing the difficulty of manual labeling, improving work efficiency, and saving manual labeling time.

[0026] Exemplary embodiments of the invention will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the invention and to fully convey the scope of the invention to those skilled in the art.

[0027] Example 1: The semi-automatic drilling operation data mining method based on pattern segmentation algorithm provided in this example includes: S1. Collect data and perform standardization processing to obtain multidimensional drilling parameter time series.

[0028] In this embodiment, the specific process of collecting and standardizing data includes: S11, Data Selection.

[0029] This embodiment collects drilling operation parameter data from several wells in historical drilling operation data. From these drilling operation parameters, 12 characteristic parameters related to drilling conditions are selected, as shown in Table 1. These data are recorded in chronological order, with one data point recorded every 5 seconds. The data for one well constitutes a multidimensional time series, containing approximately one million to several million data points. This is just one example, and is not limited to this.

[0030] Table 1 Characteristic parameters related to drilling conditions

[0031] S12, Data Cleaning.

[0032] In this embodiment, data cleaning refers to deleting abnormal data such as negative values ​​and infinity values ​​caused by sensor equipment problems. Parameter thresholds are set based on business experience to avoid the influence of random noise on data patterns. The filtering methods are shown in Table 2.

[0033] Table 2 Smoothing of Drilling Operation Parameter Thresholds

[0034] S13, Data Standardization.

[0035] In this embodiment, data standardization refers to the standardization of data across all dimensions. The Min-Max normalization calculation formula is as follows:

[0036] In the formula, These represent the maximum and minimum values ​​of the data, respectively.

[0037] S2. Segment and pattern mining of multidimensional drilling parameter time series.

[0038] In this embodiment, the AutoPlait algorithm automatically segments and mines patterns in a multidimensional drilling parameter time series. The AutoPlait algorithm is an algorithm for automatically mining time series patterns. It features no parameter tuning or pre-training required, and can automatically find segmentation points in the time series and extract latent patterns. Its output can be used as a component of semi-automated software as a reference before manual annotation, or it can be combined with existing working condition identification models to compensate for their difficulty in identifying state transitions, making it suitable for drilling data analysis. The AutoPlait algorithm is a classic algorithm for handling pattern segmentation problems, renowned for its high accuracy and full automation. The application process of the AutoPlait algorithm includes: S21. Use a cascaded Markov framework to model the data.

[0039] In this embodiment, the multidimensional operation parameters can exhibit different distribution characteristics over time. Based on this, algorithms can be used to automatically capture these data characteristics and achieve automatic segmentation of the time series data. The meaning of segmenting drilling operation time series data is as follows: Figure 2 As shown. Modeling data using a cascaded Markov model (MLCM) can be achieved through... Figure 3 The implementation is as follows: Each independent Hidden Markov Model (HMM) (black box) is an independent pattern identified by the algorithm, which may correspond to a drilling operation state; the same pattern can also contain some sub-states, corresponding to some data distribution features with smaller granularity; transformations can occur between sub-states and between HMM models. However, it should be noted that the total number of patterns and the number of states in a pattern are not manually set but are automatically determined by the model to find the optimal solution. In fact, in drilling operations, there is not only one way to classify drilling conditions. This embodiment can identify the best segmentation that can be achieved by the information carried by the data obtained in step S1 without being affected by human prior factors, providing an independent reference for business personnel. After modeling the data obtained in step S1 using MLCM, the data structure is shown in Table 3.

[0040] Table 3 Data Structure After Modeling

[0041] S22. Construct a cost function based on the minimum length theory.

[0042] In this embodiment, the minimum length theory specifically states that when using a cascaded Markov framework to describe time-series data X of drilling operation parameters, the optimal model C={m,r,S,Θ,F} should minimize the total coding length required to describe the data. Based on the minimum length theory, the total coding length required to describe the data X of a well obtained in step S1 should be equal to the coding length required to describe the aforementioned cascaded hidden Markov model (C) plus the coding length required to describe the data X under the conditions of this model.

[0043] Based on the encoding principle, the total encoding length required to describe the data in this embodiment, i.e., the cost function, is:

[0044] ; (1) The coding cost required to describe r patterns in an MLCM:

[0045] Length of each HMM encoding: .

[0046] k is the number of states in a single HMM, and cF is a constant factor. n is the length of the time series, m is the total number of segments into which the MLCM model divides the time series, and r is the total number of patterns.

[0047] (2) The encoding cost of data under a given MLCM is the sum of the encoding lengths required for all segments: ) Among them, segmentation This indicates the observed data fragments given the HMM model parameters. The probability, This represents the length of the i-th segment; This indicates the mode to which it belongs. u The self-conversion probability.

[0048] S23. Based on the cost function, optimize the model parameters using a three-level nested algorithm.

[0049] In this embodiment, the AutoPlait optimization technique comprises a three-layer optimization algorithm. The innermost Cut Point Search (CPS) algorithm primarily uses dynamic programming to find the optimal segmentation point on the time series. The input is the time series X and the current pattern set (regime), and the output is the set of segments. The middle layer, Regime parameter estimation algorithm, includes the innermost Cut Point Search algorithm and is used to estimate the model parameters. The input is the time series X, and the output is the parameters of the regime. The AutoPlait algorithm integrates the two algorithms and performs iterative optimization, starting with splitting the entire time series as one segment per pattern. Each time, the cost function described is calculated, and splitting continues if the cost decreases. This nested approach allows the pattern regime and segments to be updated simultaneously, thus achieving fully automated pattern mining of time series data without any manual operation or parameter adjustment. This is highly advantageous for analyzing massive amounts of drilling operation parameter sensor data. The implementation steps of this three-layer nested optimization algorithm are as follows: Figures 4-6 As shown.

[0050] Furthermore, such as Figure 4 As shown, the Cut Point Search algorithm mainly uses dynamic programming combined with a hidden Markov model to find the optimal segmentation point. Its steps are as follows: Input the multidimensional drilling operation parameters time series X and remige parameter; initialization; Calculate the state probabilities p1,i(t) and p2,u(t), where, This represents the probability of being in state i of regime 1 at time t; This represents the probability of being in state u of regulation 2 at time t; Update path records L1, i(t) and L2, u(t), where, This represents the optimal path record for state i in period 1 at time t. This represents the optimal path record for state u in regime 2 at time t.

[0051] Select the optimal set of splitting points; Allocate segments; Returns the segment information set {m1,m2,S1,S2}.

[0052] Furthermore, such as Figure 5 As shown, the Regime parameter estimation algorithm is used to estimate the model's parameters. After obtaining the segmentation information parameters returned by the Cut Point Search algorithm, the Regime parameter estimation algorithm then calls the Baum-Welch algorithm to update the parameters. , Piecewise HMM model parameters , Update the transformation matrix Output The results are returned to the AutoPlait algorithm, and the specific steps are as follows: Input the time series of multidimensional drilling operation parameters X; Initialize model parameters , ; Iterative optimization; The CPS algorithm described above is used for initial time series segmentation; The Baum-Welch algorithm is applied to each segment to estimate the parameters of the HMM model; Update the transformation matrix △; To determine if the cost has improved, calculate the total cost function value. Terminate the iteration when the total cost function no longer decreases, and output the optimal set of parameters. .

[0053] Furthermore, such as Figure 6 As shown, the AutoPlait algorithm will receive The information is used to calculate the costs before and after the split. Used to calculate the new total number of segments m Used to calculate the new segment length , Used to calculate the encoding cost of model parameters and data encoding costs , The cost associated with the regime transition probability is used to calculate the cost before and after the split. If the cost before the split is greater than the cost after the split, it indicates that the split was effective, and the new [split] will be [transferred / implemented]. and Push Q onto the stack to continue optimization iterations; if the cost before splitting is less than the cost after splitting, the split is invalid, so add the current regime to the final result, and output the final total number of segments m, total number of regimes r, and segment set S. mode parameter set The steps of the AutoPlait algorithm are as follows: Input the time series of multidimensional drilling operation parameters X; Initialize a single regime parameter; Push the initial data onto stack Q; Check if stack Q is empty. If it is, output the final segment structure. If not, retrieve the current regime from Q and attempt to split it into segments. , ; Determine if the split reduces the total cost. If not, hold the current regime, update the final result set, and return to the step of determining if stack Q is empty. If so, accept the split, push the new regime onto stack Q, update the final result set, and return to the step of determining if stack Q is empty.

[0054] Furthermore, the output includes the total number of segments and the total number of regimes, r; the set of segments. This includes the start and end times of each segment; and the regime type to which each segment belongs. The final cost function value Model parameter sets and transformation matrices for each region It also displays the segmentation results of the time series data and the distribution of different regions.

[0055] S3. The data segments and patterns calculated in step S2 are automatically rendered to the front-end interface and displayed in the form of color blocks. At the same time, the drilling operation parameters of the input time series are displayed as curves.

[0056] In this embodiment, operational personnel use their experience to determine the actual drilling conditions corresponding to each mode, and input the completed legend to complete the data annotation. For segments in the S2 process that are inaccurately judged, manual correction can be performed by capturing the time period on the screen. The mapping between drilling operation parameters and operation status formed by saving the data, together with the historically accumulated annotated data, constitutes the second dataset.

[0057] S4. Based on the second dataset, a drilling condition classifier with accuracy meeting the requirements of general business applications can be trained using a convolutional neural network model or an ensemble learning model.

[0058] In this embodiment, after pattern mining and calibration of 15 wells, the predictive performance of the trained working condition classifier on 5 test wells is shown in Table 4.

[0059] Table 4 Performance of the drilling condition classifier trained on the second dataset

[0060] Furthermore, the convolutional neural network used in this embodiment comprises four modules, containing 3, 4, 6, and 3 residual blocks respectively. Each residual block contains two convolutional layers, with a 3×3 convolutional layer as input and a fully connected layer as output. The ensemble learning models used, XGBoost and LightGBM, both achieve performance levels comparable to those shown in Table 4, demonstrating the feasibility of the method described in this embodiment in the field of drilling and completion engineering. The efficiency of using the method described in this embodiment to complete the labeling task of five wells is 10 times that of traditional manual labeling methods, and this efficiency advantage will increase exponentially as the data scale expands.

[0061] Example 2: Following the pattern segmentation algorithm-based semi-automatic drilling operation data mining method provided in Example 1, this example provides a pattern segmentation algorithm-based semi-automatic drilling operation data mining device. The device provided in this example can implement the pattern segmentation algorithm-based semi-automatic drilling operation data mining method of Example 1. This device can be implemented through software, hardware, or a combination of both. For ease of description, this example is described by dividing the functionality into various units. Of course, in implementation, the functions of each unit can be implemented in one or more software and / or hardware components. For example, the device may include integrated or separate functional modules or units to execute the corresponding steps in the methods of Example 1. Since the device in this example is basically similar to the method example, the description process of this example is relatively simple. Relevant details can be found in the description of Example 1. The example of the pattern segmentation algorithm-based semi-automatic drilling operation data mining device provided by this invention is merely illustrative.

[0062] Specifically, the semi-automatic drilling operation data mining device based on pattern segmentation algorithm provided in this embodiment includes: The data processing module is configured to collect drilling operation parameter data and process the drilling operation parameter data to obtain a multidimensional drilling parameter time series. The pattern mining module is configured to perform pattern mining on multidimensional drilling parameter time series; The working condition classification module is configured to input the excavation mode into the drilling working condition classifier to obtain the drilling working condition classification result.

[0063] Furthermore, the training of the drilling condition classifier includes a data processing module, a main algorithm module, a manual verification module, and a condition classification module; The data processing module is used to select, clean, and transform the first sample set composed of drilling operation parameter data, making it easier to achieve automated data mining through pattern segmentation algorithms.

[0064] The main algorithm module uses a cascaded Markov model to model the processed data into different states that can be transformed into each other with a certain probability, i.e., different segments. The segments can be clustered into several patterns, and a fully automatic pattern segmentation algorithm is used to determine all model parameters, including the number of segments, segmentation points, and the number of optimal patterns, thereby completing the segmentation and pattern recognition of the data.

[0065] The manual verification module is a visual page built using Vue3 front-end technology. Professionals can use this page to map the automatically mined patterns to the actual working conditions in the drilling business field, and modify inappropriate automatic identifications to form a high-quality labeled second dataset.

[0066] The working condition classification module trains a drilling working condition classifier based on the second dataset to automatically identify working conditions in newly added unlabeled data.

[0067] Example 3: This example provides an electronic device that implements the semi-automated historical working condition data mining method based on the pattern segmentation algorithm provided in Example 2. The electronic device can be an electronic device for the client, such as a mobile phone, laptop, tablet computer, desktop computer, etc., to execute the method of Example 2.

[0068] like Figure 8 As shown, the electronic device includes a processor, a memory, a communication interface, and a bus. The processor, memory, and communication interface are connected via the bus to enable communication between them. The memory stores a computer program that can run on the processor. When the processor runs the computer program, it executes the method of Embodiment 1. The implementation principle and technical effects are similar to those of Embodiment 1, and will not be repeated here. Those skilled in the art will understand that... Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computing device on which the present application is applied. The specific computing device may include more or fewer components than shown in the figure, or combine certain components, or have different component arrangements.

[0069] In a preferred embodiment, the logical instructions in the aforementioned memory can be implemented as software functional units and sold or used as independent products, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), and optical discs.

[0070] In a preferred embodiment, the processor can be any type of general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), and is not limited thereto.

[0071] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In the description of this specification, the terms "a preferred embodiment," "furthermore," "specifically," "in this embodiment," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the embodiments in this specification. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described can be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0072] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A semi-automatic drilling operation data mining method based on pattern segmentation algorithm, characterized in that, include: Collect drilling operation parameter data and process the drilling operation parameter data to obtain a multidimensional drilling parameter time series; Pattern mining of multidimensional drilling parameter time series; Input the excavation pattern into the drilling condition classifier to obtain the drilling condition classification results.

2. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 1, characterized in that, Processing drilling operation parameter data, including: Several characteristic parameters related to drilling conditions were selected; Data cleaning is performed on the selected feature parameters; Standardize the cleaned data.

3. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 2, characterized in that, Several characteristic parameters related to drilling conditions were selected, including: hook height, drill bit measurement depth, measurement well depth, pump pressure, cumulative pump strokes, inlet flow rate, torque, drilling pressure, rotational speed, mechanical drilling speed, suspended weight, and outlet flow rate.

4. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 3, characterized in that, The training process for the drilling condition classifier is as follows: Drilling operation parameter data of several wells were collected from historical drilling operation data. Several characteristic parameters related to drilling conditions were selected, and the selected characteristic parameters related to drilling conditions were processed to obtain a multidimensional drilling parameter time series. The AutoPlait algorithm is used to automatically segment and mine patterns in the time series of multidimensional drilling parameters. The labeled data of the actual drilling conditions corresponding to each mode constitute the training dataset; Based on the training dataset, a drilling condition classifier is trained using a convolutional neural network model or an ensemble learning model.

5. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 4, characterized in that, The AutoPlait algorithm is used to automatically segment and mine patterns in multidimensional drilling parameter time series, including: Use a cascaded Markov framework to model the data and obtain model parameters; Construct a cost function based on the minimum length theory; The model parameters are optimized based on the cost function and a three-level nested algorithm.

6. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 5, characterized in that, The data is modeled using a cascaded Markov framework to obtain model parameters, including: Segment: Divide X into m consecutive segments The total number of segments is m, and X is a time series of multidimensional drilling operation parameters for a well: X={ }, data dimensions are d , length is n ; Pattern regime: Θ = { r HMMs ,…, }, each Include Each HMM is It is a potential regime of data, with a total number of patterns of r. Here is the state transition matrix. To observe the probability distribution, The initial state distribution; Model structure: C={m,r,S,Θ,F}, where F gives the region to which each segment belongs. .

7. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 5, characterized in that, Constructing a cost function based on minimum length theory: ； in, The encoding cost for each HMM, Let X be the encoding cost of data X under the conditions of a cascaded Markov model.

8. The semi-automatic drilling operation data mining method based on pattern segmentation algorithm according to claim 7, characterized in that, Based on the cost function, a three-layer nested algorithm is used to optimize the model parameters. The three-layer nested algorithm includes the innermost CutPoint Search algorithm, the middle Regime parameter estimation algorithm, and the outermost AutoPlait algorithm. The specific process is as follows: The Cut Point Search algorithm uses dynamic programming to find the optimal segmentation point on a time series. The input is the time series X and the current pattern set regime, and the output is the set of segments. The Regime parameter estimation algorithm includes the innermost Cut Point Search, which is used to estimate the parameters of the model. The input is the time series X, and the output is the parameters of the regime. The AutoPlait algorithm integrates two algorithms and performs iterative optimization. It starts by splitting the entire time series into one pattern. Each time, it calculates the cost function. If the cost decreases, it continues to split into new regions; otherwise, it maintains the current region.

9. A semi-automatic drilling operation data mining device based on a pattern segmentation algorithm, characterized in that, include: The data processing module is configured to collect drilling operation parameter data and process the drilling operation parameter data to obtain a multidimensional drilling parameter time series. The pattern mining module is configured to perform pattern mining on multidimensional drilling parameter time series; The working condition classification module is configured to input the excavation mode into the drilling working condition classifier to obtain the drilling working condition classification result.

10. An electronic device, characterized in that, include: At least one processor; And a memory communicatively connected to the processor; wherein the memory stores instructions executable by the processor, the instructions being executed by the processor to enable the processor to perform the method according to any one of claims 1-7.

Images