Well logging curve abnormal value marking method, device, computer equipment and storage medium

Abstract

The present application relates to a method, device, computer equipment and storage medium for annotating outliers of well logging curves, wherein the method comprises: reading a preset set of trained machine learning models, selecting the best trained machine learning model under the current tuning standard from the set, inputting the well logging curve to be annotated into the selected trained machine learning model, and obtaining the anomaly annotating result of the well logging curve. In the whole process, the best trained machine learning model under the current tuning standard is selected, and the anomaly annotating of the well logging curve to be annotated is completed by the selected trained machine learning model, without relying on manual anomaly annotating, and the anomaly annotating of the well logging curve can be realized efficiently and accurately.

Description

Technical Field

[0001] The present application relates to the field of automation technology, and in particular to a method, device, computer equipment and storage medium for marking abnormal values ​​of a logging curve. Background Art

[0002] Well logging is a branch of geophysical exploration and is a general term for geophysical exploration methods used in drilling. According to the different physical properties of the rock used, it can be divided into electrical logging, radioactive logging, magnetic logging, acoustic logging, thermal logging and gravity logging. The parameters of the formation can be obtained through the logging curve, which is conducive to the exploration and development of oil. However, the logging curve is affected by various factors such as the environment during the measurement process, and some abnormal values ​​will be generated. These values ​​are not a true reflection of the formation properties. If these abnormal values ​​are not processed in the early preprocessing stage of the curve, the final logging interpretation results will be seriously affected. Therefore, the identification of abnormal values ​​in the logging curve is particularly important.

[0003] The traditional way to mark outliers on well logging curves is to run some well logging interpretation software on a computer to present the well logging curves in a visual form. Experienced interpreters check each curve from top to bottom to find outliers. The interpreters input the found outliers into the computer, and the computer integrates the input data with the original well logging curve to obtain the final well logging curve with outliers.

[0004] Although the above method can realize the anomaly annotation of logging curves, due to the great depth of logging, a large amount of data needs to be processed. It takes a lot of time to rely on manual processing and identification of this part of the data. In addition, there are deviations in the cognition and experience of different people during the processing process, which leads to inaccurate annotation of some outliers. Summary of the invention

[0005] Based on this, it is necessary to provide a method, device, computer equipment and storage medium for annotating outliers of well logging curves that are efficient and accurate in addressing the above technical problems.

[0006] A method for marking abnormal values ​​of a well logging curve, the method comprising:

[0007] Obtain the logging curves to be annotated and the tuning standards;

[0008] Reading a preset set of trained machine learning models, wherein the set of trained machine learning models includes a plurality of trained machine learning models, and the trained machine learning models are obtained by training sample logging curves carrying outlier annotations;

[0009] Selecting the best trained machine learning model under the tuning standard from the set of trained machine learning models;

[0010] The well logging curve to be annotated is input into the selected trained machine learning model to obtain the well logging curve outlier annotation result.

[0011] In one embodiment, before reading the preset trained machine learning model set, the method further includes:

[0012] Obtain sample logging curves with outlier annotations;

[0013] Randomly dividing the sample data into training sample data and testing sample data;

[0014] According to the training sample data, the multiple initial machine learning models are trained respectively to obtain multiple trained machine learning models;

[0015] Testing the multiple trained machine learning models respectively according to the test sample data to obtain model test results;

[0016] According to the model test results, the optimal trained machine learning model under different tuning standards is obtained, and a preset set of trained machine learning models is constructed.

[0017] In one embodiment, obtaining a sample well logging curve with an outlier annotation includes:

[0018] Get the original logging curve;

[0019] Performing depth correction on the original well logging curve to obtain the original well logging curve after depth correction;

[0020] The original well logging curve after the depth correction is obtained by a third party with outlier annotations to obtain a sample well logging curve with outlier annotations.

[0021] In one embodiment, performing depth correction on the original well logging curve to obtain the original well logging curve after depth correction includes:

[0022] Randomly select any one of the original logging curves as a reference curve;

[0023] Comparing other curves in the original logging curve with the reference curve to obtain depth errors between other curves in the original logging curve and the reference curve;

[0024] The depth of other curves in the original logging curve is calibrated according to the depth error to obtain the original logging curve after depth correction.

[0025] In one embodiment, obtaining the optimal trained machine learning model under different tuning criteria according to the model test results and constructing a preset set of trained machine learning models includes:

[0026] According to the model test results, the trained machine learning models are ranked according to different tuning standards, and multiple ranking sequences are obtained and the corresponding tuning standards are recorded;

[0027] The best trained machine learning model in the multiple sorting sequences is selected to match the corresponding tuning standard to construct a preset set of trained machine learning models.

[0028] In one embodiment, the step of inputting the well logging curve to be annotated into a selected trained machine learning model to obtain an anomaly annotation result of the well logging curve includes:

[0029] Inputting the well logging curve to be annotated into the selected trained machine learning model;

[0030] Get the selected trained machine learning model output data to get the outliers and the depth corresponding to the outliers;

[0031] According to the depth corresponding to the abnormal value, the abnormal value is marked in the logging curve to be marked, so as to obtain the logging curve abnormal value marking result.

[0032] In one embodiment, inputting the well logging curve to be annotated into the selected trained machine learning model includes:

[0033] Performing depth correction on the well logging curve to be annotated;

[0034] The depth-corrected well logging curve to be annotated is input into the selected trained machine learning model.

[0035] A logging curve abnormal value marking device, the device comprising:

[0036] Data acquisition module, used to obtain the logging curve to be annotated and the tuning standard;

[0037] A reading module, used to read a preset set of trained machine learning models, wherein the set of trained machine learning models includes a plurality of trained machine learning models, and the trained machine learning models are obtained by training sample logging curves carrying outlier annotations;

[0038] A model selection module, used to select the best trained machine learning model under the tuning standard from the set of trained machine learning models;

[0039] The outlier annotation module is used to input the well logging curve to be annotated into the selected trained machine learning model to obtain the outlier annotation result of the well logging curve.

[0040] In addition, the present application also provides an electronic device, which includes at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call the program instructions in the memory to execute the method for marking outliers of logging curves as described above.

[0041] In addition, the present application also provides a storage medium, which includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute to implement the above-mentioned logging curve outlier value marking method.

[0042] The above-mentioned method, device, computer equipment and storage medium for labeling outliers of well logging curves read a preset set of trained machine learning models, select the best trained machine learning model under the current tuning standard from the set, input the well logging curve to be labeled into the selected trained machine learning model, and obtain the labeling result of outliers of the well logging curve. In the whole process, the best trained machine learning model under the current tuning standard is selected, and the outliers of the well logging curve to be labeled are labeled by the selected trained machine learning model, without relying on manual outlier labeling, and the labeling of outliers of the well logging curve can be realized efficiently and accurately. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A diagram showing an application environment of a method for marking abnormal values ​​of a well logging curve in an embodiment;

[0044] Figure 2 A schematic diagram of a process of a method for marking abnormal values ​​of a well logging curve in one embodiment;

[0045] Figure 3 A schematic diagram of a flow chart of a method for marking abnormal values ​​of a well logging curve in another embodiment;

[0046] Figure 4 This is a schematic diagram of depth correction of well logging curves;

[0047] Figure 5 It is a structural block diagram of a well logging curve abnormal value marking device in one embodiment;

[0048] Figure 6 FIG. 4 is a diagram showing the internal structure of a device in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] The method for marking abnormal values ​​of well logging curves provided in this application can be applied to Figure 1 In the application environment shown. Among them, the terminal 102 communicates with the server 104 through the network. The server 104 pre-stores a set of trained machine learning models. The terminal 102 sends the logging curve to be labeled and the tuning standard under the current application scenario to the server 104. The server 104 obtains the logging curve to be labeled and the tuning standard, reads the preset set of trained machine learning models, selects the best trained machine learning model under the tuning standard in the trained machine learning model set, inputs the logging curve to be labeled into the selected trained machine learning model, obtains the logging curve anomaly value annotation result, and the server 104 can feed back the logging curve anomaly value annotation result to the terminal 102. Among them, the terminal 102 can be but not limited to various personal computers, laptops, smart phones, tablet computers and portable wearable devices, and the server 104 can be implemented with an independent server or a server cluster composed of multiple servers.

[0051] In one embodiment, Figure 2 As shown in FIG. 1 , a method for marking abnormal values ​​of a well logging curve is provided. The method is applied to Figure 1 Taking the server 104 in the example as an example, the following steps are included:

[0052] S200: Obtain the well logging curve to be annotated and the tuning standard.

[0053] The logging curve to be annotated refers to the object of the logging curve anomaly annotation. Specifically, the user can submit the logging curve to be annotated to the server on the terminal, and the server directly obtains the logging curve to be annotated uploaded by the terminal. The tuning standard can be understood as the evaluation scale of the model. In subsequent operations, different trained machine learning models can be selected for logging curve anomaly annotation under different evaluation scales. Specifically, the tuning standards include mean_absolute_error (mean absolute error), mean_squared_error (mean square error), median_absolute_error (median absolute error) and r2_score (goodness of fit). The above 4 tuning standards have their corresponding scale mathematical calculation formulas. In actual applications, professional operators operate at the terminal and upload the logging curve to be annotated to the server through the terminal. In addition, users will also select the corresponding tuning standard and upload it to the server according to the business needs of this time (the purpose or use of the logging curve anomaly annotation, etc.). For example, in the current application scenario, the porosity needs to be calculated, and the corresponding tuning standard is r2_score. Professional operators select the tuning standard - r2_score and upload it to the server. It is not necessary. In actual applications, the terminal can directly display the tuning standard for professional operators to click and select; or the terminal can only display the business demand options, and the terminal automatically finds the tuning indicators (evaluation scales) corresponding to the business demand options and uploads them to the server.

[0054] S400: Read a preset set of trained machine learning models, where the set of trained machine learning models includes multiple trained machine learning models, and the trained machine learning models are trained by sample logging curves carrying outlier annotations.

[0055] The preset set of trained machine learning models is a set of pre-set trained machine learning models, which includes multiple trained machine learning models. The multiple trained machine learning models can be obtained by training different machine learning models through sample data, and the tuning standard can also be taken into account in the set of trained machine learning models, that is, the model set can only include the optimal trained machine learning models under different tuning standards. The machine learning model can specifically include models based on 10 regression algorithms such as HuberRegressor, AdaBoostRegressor, Lasso, Ridge, SGDRegressor, LinearSVR, SVR, DecisionTreeRegressor, BaggingRegressor, GradientBoostRegressor, and RandomForestRegressor.

[0056] S600: Selecting the best trained machine learning model under the tuning criteria from the set of trained machine learning models.

[0057] The trained machine learning model set carries a variety of trained machine learning models, and there are tuning standards associated therewith, so the best trained machine learning model under the current tuning standard can be selected from them. Generally speaking, a tuning standard-optimal trained machine learning model correspondence table is stored in the trained machine learning model set, and based on the correspondence table and the tuning standard obtained in step S200, the best trained machine learning model under the tuning standard can be obtained.

[0058] S800: Input the well logging curve to be annotated into the selected trained machine learning model to obtain the anomaly annotation result of the well logging curve.

[0059] By inputting the annotated logging curve into the selected trained machine learning model, the logging curve anomaly annotation result corresponding to this logging curve annotation task can be obtained. Specifically, the outliers and the depths corresponding to the outliers can be directly obtained from the output data of the selected trained machine learning model. According to the depths corresponding to the outliers, the outliers are annotated to the logging curve to be annotated to obtain the logging curve anomaly annotation result.

[0060] The above-mentioned logging curve outlier annotation method reads a preset set of trained machine learning models, selects the best trained machine learning model under the current tuning standard from the set, inputs the logging curve to be annotated into the selected trained machine learning model, and obtains the logging curve outlier annotation result. In the whole process, the best trained machine learning model under the current tuning standard is selected, and the outlier annotation of the logging curve to be annotated is completed by the selected trained machine learning model, without relying on manual outlier annotation, and the logging curve outlier annotation can be realized efficiently and accurately.

[0061] like Figure 3 As shown, in one embodiment, before step S400, the process further includes:

[0062] S310: Obtaining a sample logging curve with outlier annotations.

[0063] The sample logging curve refers to a logging curve that has been marked with outliers, which can be a logging curve marked with outliers recently obtained from historical records, or a logging curve that is manually marked with outliers and uploaded by professional technicians in real time.

[0064] S320: Randomly divide the sample data into training sample data and test sample data.

[0065] The obtained sample logging curves are randomly divided into training sample data and test sample data. Generally speaking, the data volume of the training sample data is larger than that of the test sample data. For example, the sample logging curves can be evenly divided into 10 parts, 8 of which are used as training sample data and 2 of which are used as test sample data; for another example, 7 of them can be used as training sample data and 3 of them can be used as test sample data.

[0066] S330: Train multiple initial machine learning models respectively according to the training sample data to obtain multiple trained machine learning models.

[0067] The training sample data is used as training data to train multiple initial machine learning models to obtain multiple trained machine learning models. Specifically, the initial machine learning models include initial models based on 10 regression algorithms, including HuberRegressor, AdaBoostRegressor, Lasso, Ridge, SGDRegressor, LinearSVR, SVR, DecisionTreeRegressor, BaggingRegressor, GradientBoostRegressor, and RandomForestRegressor. The above-mentioned 10 initial machine learning models are trained separately using the above-mentioned training sample data, and finally multiple (10) trained machine learning models can be obtained.

[0068] S340: Test multiple trained machine learning models separately according to the test sample data to obtain model test results.

[0069] S350: According to the model test results, the optimal trained machine learning model under different tuning standards is obtained, and a preset set of trained machine learning models is constructed.

[0070] Select the above test sample data to test multiple trained machine learning models, calculate the error between the predicted value of each trained machine learning model and the true value in the test sample data, use different tuning standards to perform reference calculations on these error values, select the optimal catenary machine learning model under the current tuning standard, and finally obtain the optimal trained machine learning model under different tuning standards to build a preset set of trained machine learning models. Taking the tuning standard mean_absolute_error (mean absolute error) as an example, calculate the errors obtained by multiple trained machine learning models in processing each group of test sample data, calculate the average absolute error of these errors, and select the corresponding trained machine learning model with the smallest average absolute error as the optimal trained machine learning model under the tuning standard mean_absolute_error (mean absolute error).

[0071] In one embodiment, obtaining a sample well logging curve with an outlier annotation includes:

[0072] The original logging curve is obtained; the original logging curve is depth-corrected to obtain the original logging curve after depth correction; the original logging curve after depth correction is obtained by a third party for outlier annotation to obtain a sample logging curve with outlier annotation.

[0073] Since the original logging curve is affected by the weight and tension changes of the instrument itself during the field acquisition process, different instruments will have different depth displays for the same formation when acquiring data, and the original logging curve needs to be calibrated to ensure that each curve has a corresponding response at the same depth. The above-mentioned third party can be a terminal operated by a technician or a server that uses other methods to accurately mark the outliers of the logging curve. In actual applications, the server can send the original logging curve after depth correction to the terminal of a professional technician, who will mark the outliers on his terminal. The terminal will upload the sample logging curve with the outlier annotation to the server, and the server will obtain the sample logging curve with the outlier annotation.

[0074] In one embodiment, performing depth correction on the original well logging curve to obtain the depth-corrected original well logging curve includes:

[0075] Randomly select any curve in the original logging curve as the reference curve; compare other curves in the original logging curve with the reference curve to obtain the depth error between other curves in the original logging curve and the reference curve; calibrate the depth of other curves in the original logging curve according to the depth error to obtain the original logging curve after depth correction.

[0076] The process of logging curve depth calibration can be found in Figure 4 The example shown is Figure 4 For example, first select a curve as the benchmark ( Figure 4 The other curves are compared with the reference curve. If there is an error in depth, the depth of the curve with error is increased or decreased by a corresponding number to make the curves match in depth.

[0077] In one embodiment, according to the model test results, obtaining the optimal trained machine learning model under different tuning standards, and constructing a preset set of trained machine learning models includes:

[0078] According to the model test results, the trained machine learning models are ranked according to their quality under different tuning standards, and multiple ranking sequences are obtained and the corresponding tuning standards are recorded; the best trained machine learning model in the multiple ranking sequences is selected to match the corresponding tuning standard, and a preset set of trained machine learning models is constructed.

[0079] According to the model test results, different tuning standards are used to rank the quality of the trained machine learning models, that is, different evaluation scales are used to evaluate the trained machine learning models, and the quality ranking sequence of the trained machine learning models corresponding to different tuning standards is obtained. This ranking sequence can be arranged in order from best to worst under the current tuning standard, and the best trained machine learning model is selected, and the best trained machine learning model is matched with the corresponding tuning standard to construct a preset set of trained machine learning models. In other words, the preset set of trained machine learning models stores multiple optimal trained machine learning models under different tuning standards.

[0080] In one embodiment, inputting the well logging curve to be labeled into the selected trained machine learning model includes: performing depth correction on the well logging curve to be labeled; and inputting the depth-corrected well logging curve to be labeled into the selected trained machine learning model.

[0081] For the well logging curve to be annotated, there may also be depth errors, so it is necessary to perform depth correction on it in the same way as mentioned above, and then input the depth-corrected well logging curve to be annotated into the selected trained machine learning model to improve the accuracy of annotating outliers in the well logging curve.

[0082] It should be understood that although Figure 2-3 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 2-3 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0083] like Figure 5 As shown, a logging curve abnormal value marking device, the device comprises:

[0084] Data acquisition module 200, used to obtain the well logging curve to be annotated and the tuning standard;

[0085] A reading module 400 is used to read a preset set of trained machine learning models, where the set of trained machine learning models includes multiple trained machine learning models, and the trained machine learning models are trained by sample logging curves carrying outlier annotations;

[0086] A model selection module 600 is used to select the best trained machine learning model under the tuning criteria from the set of trained machine learning models;

[0087] The outlier annotation module 800 is used to input the well logging curve to be annotated into the selected trained machine learning model to obtain the outlier annotation result of the well logging curve.

[0088] The above-mentioned logging curve outlier annotation device reads a preset set of trained machine learning models, selects the best trained machine learning model under the current tuning standard from the set, inputs the logging curve to be annotated into the selected trained machine learning model, and obtains the logging curve outlier annotation result. In the whole process, the best trained machine learning model under the current tuning standard is selected, and the outlier annotation of the logging curve to be annotated is completed by the selected trained machine learning model, without relying on manual outlier annotation, and the logging curve outlier annotation can be realized efficiently and accurately.

[0089] In one of the embodiments, the above-mentioned logging curve outlier annotation device includes a model building module, which is used to obtain sample logging curves carrying outlier annotations; randomly divide the sample data into training sample data and test sample data; train multiple initial machine learning models according to the training sample data to obtain multiple trained machine learning models; test the multiple trained machine learning models according to the test sample data to obtain model test results; according to the model test results, obtain the optimal trained machine learning model under different tuning standards, and construct a preset set of trained machine learning models.

[0090] In one of the embodiments, the model building module is also used to obtain the original logging curve; perform depth correction on the original logging curve to obtain the original logging curve after depth correction; obtain the original logging curve after depth correction by a third party for outlier annotation to obtain a sample logging curve with outlier annotations.

[0091] In one of the embodiments, the model building module is also used to randomly select any one curve in the original logging curve as a baseline curve; compare other curves in the original logging curve with the baseline curve to obtain the depth error between the other curves in the original logging curve and the baseline curve; calibrate the depth of other curves in the original logging curve according to the depth error to obtain the original logging curve after depth correction.

[0092] In one of the embodiments, the model building module is also used to rank the trained machine learning models according to the model testing results under different tuning standards, obtain multiple ranking sequences and record the corresponding tuning standards; select the best trained machine learning model in the multiple ranking sequences to match the corresponding tuning standards, and build a preset set of trained machine learning models.

[0093] In one embodiment, the outlier annotation module 800 is also used to input the logging curve to be annotated into the selected trained machine learning model; obtain the output data of the selected trained machine learning model to obtain the outliers and the depths corresponding to the outliers; and annotate the outliers into the logging curve to be annotated according to the depths corresponding to the outliers to obtain the outlier annotation results of the logging curve.

[0094] In one embodiment, the outlier annotation module 800 performs depth correction on the well logging curve to be annotated; and inputs the well logging curve to be annotated after the depth correction into the selected trained machine learning model.

[0095] For the specific definition of the logging curve anomaly value marking device, please refer to the definition of the logging curve anomaly value marking method above, which will not be repeated here. Each module in the above-mentioned logging curve anomaly value marking device can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0096] The above-mentioned logging curve outlier value marking device includes a processor and a memory. The above-mentioned data acquisition module 200, reading module 400, model selection module 600, and outlier value marking module 800 are stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize corresponding functions. The processor includes a kernel, and the kernel retrieves the corresponding program unit from the memory. One or more kernels can be set, and the logging curve outlier value marking is realized by adjusting the kernel parameters.

[0097] An embodiment of the present application provides a storage medium on which a program is stored. When the program is executed by a processor, the method for marking abnormal values ​​of well logging curves is implemented.

[0098] An embodiment of the present application provides a processor, which is used to run a program, wherein the method for marking abnormal values ​​of well logging curves is executed when the program is run.

[0099] like Figure 6As shown, the embodiment of the present application provides a device 70, which includes at least one processor 701, and at least one memory 702 and a bus 703 connected to the processor 701; wherein the processor 701 and the memory 702 communicate with each other through the bus 703; the processor 701 is used to call the program instructions in the memory 702 to execute the above-mentioned logging curve abnormal value marking method. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0100] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing the following method steps:

[0101] Obtain the logging curves to be annotated and the tuning standards;

[0102] Reading a preset set of trained machine learning models, where the set of trained machine learning models includes multiple trained machine learning models, and the trained machine learning models are obtained by training sample logging curves carrying outlier annotations;

[0103] Select the best trained machine learning model under the tuning criteria from the set of trained machine learning models;

[0104] The well logging curve to be annotated is input into the selected trained machine learning model to obtain the anomaly annotation results of the well logging curve.

[0105] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0106] Obtain sample logging curves with outlier annotations; randomly divide sample data into training sample data and test sample data; train multiple initial machine learning models according to the training sample data to obtain multiple trained machine learning models; test multiple trained machine learning models according to the test sample data to obtain model test results; based on the model test results, obtain the optimal trained machine learning model under different tuning standards and build a preset set of trained machine learning models.

[0107] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0108] The original logging curve is obtained; the original logging curve is depth-corrected to obtain the original logging curve after depth correction; the original logging curve after depth correction is obtained by a third party for outlier annotation to obtain a sample logging curve with outlier annotation.

[0109] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0110] Randomly select any curve in the original logging curve as the reference curve; compare other curves in the original logging curve with the reference curve to obtain the depth error between other curves in the original logging curve and the reference curve; calibrate the depth of other curves in the original logging curve according to the depth error to obtain the original logging curve after depth correction.

[0111] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0112] According to the model test results, the trained machine learning models are ranked according to their quality under different tuning standards, and multiple ranking sequences are obtained and the corresponding tuning standards are recorded; the best trained machine learning model in the multiple ranking sequences is selected to match the corresponding tuning standard, and a preset set of trained machine learning models is constructed.

[0113] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0114] Input the logging curve to be annotated into the selected trained machine learning model; obtain the output data of the selected trained machine learning model to obtain the outliers and the depths corresponding to the outliers; annotate the outliers into the logging curve to be annotated according to the depths corresponding to the outliers, and obtain the outlier annotation results of the logging curve.

[0115] In one embodiment, the computer program product, when executed on a data processing device, is adapted to execute a program for initializing the following method steps:

[0116] Perform depth correction on the well logging curve to be labeled; input the well logging curve to be labeled after depth correction into the selected trained machine learning model.

[0117] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0118] In a typical configuration, the device includes one or more processors (CPU), memory and bus. The device may also include input / output interface, network interface and the like.

[0119] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0120] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0121] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0122] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0123] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for marking abnormal values ​​of a well logging curve, the method comprising: Obtaining a well logging curve to be annotated and an optimization standard, wherein the optimization standard includes a mean absolute error, a mean square error, a median absolute error, or a goodness of fit; Reading a preset set of trained machine learning models, wherein the set of trained machine learning models includes a plurality of trained machine learning models, and the trained machine learning models are obtained by training sample logging curves carrying outlier annotations; Selecting the best trained machine learning model under the tuning standard from the set of trained machine learning models; Inputting the well logging curve to be annotated into the selected trained machine learning model to obtain an outlier annotation result of the well logging curve; The process of obtaining the sample logging curve with outlier annotations includes: obtaining the original logging curve; performing depth correction on the original logging curve to obtain the original logging curve after depth correction; obtaining the sample logging curve with outlier annotations by a third party for the original logging curve after depth correction; The depth correction of the original well logging curve to obtain the depth-corrected original well logging curve comprises: randomly selecting any one of the original well logging curves as a reference curve; comparing other curves in the original well logging curve with the reference curve to obtain depth errors between the other curves in the original well logging curve and the reference curve; and depth correction of other curves in the original well logging curve according to the depth error to obtain the depth-corrected original well logging curve; Before reading the preset set of trained machine learning models, it also includes: obtaining a sample logging curve carrying an outlier annotation; randomly dividing the sample logging curve into training sample data and test sample data; training multiple initial machine learning models according to the training sample data to obtain multiple trained machine learning models; testing the multiple trained machine learning models according to the test sample data to obtain model test results; according to the model test results, obtaining the optimal trained machine learning model under different tuning standards, and constructing a preset set of trained machine learning models.

2. The method according to claim 1, characterized in that The step of obtaining the optimal trained machine learning model under different tuning criteria according to the model test results and constructing a preset set of trained machine learning models includes: According to the model test results, the trained machine learning models are ranked according to different tuning standards, and multiple ranking sequences are obtained and the corresponding tuning standards are recorded; The best trained machine learning model in the multiple sorting sequences is selected to match the corresponding tuning standard to construct a preset set of trained machine learning models.

3. The method according to claim 1, characterized in that The step of inputting the well logging curve to be annotated into the selected trained machine learning model to obtain the well logging curve outlier annotation result includes: Inputting the well logging curve to be annotated into the selected trained machine learning model; Get the selected trained machine learning model output data to get the outliers and the depth corresponding to the outliers; According to the depth corresponding to the abnormal value, the abnormal value is marked in the logging curve to be marked, so as to obtain the logging curve abnormal value marking result.

4. The method according to claim 1, characterized in that The step of inputting the well logging curve to be annotated into the selected trained machine learning model comprises: Performing depth correction on the well logging curve to be annotated; The depth-corrected well logging curve to be annotated is input into the selected trained machine learning model.

5. A logging curve abnormal value marking device, characterized in that: The device comprises: A data acquisition module, used to acquire the well logging curve to be annotated and the tuning standard, wherein the tuning standard includes the mean absolute error, the mean square error, the median absolute error, or the goodness of fit; A reading module, used to read a preset set of trained machine learning models, wherein the set of trained machine learning models includes a plurality of trained machine learning models, and the trained machine learning models are obtained by training sample logging curves carrying outlier annotations; A model selection module, used to select the best trained machine learning model under the tuning standard from the set of trained machine learning models; An outlier annotation module is used to input the well logging curve to be annotated into the selected trained machine learning model to obtain an outlier annotation result of the well logging curve; The process of obtaining the sample logging curve with outlier annotations includes: obtaining the original logging curve; performing depth correction on the original logging curve to obtain the original logging curve after depth correction; obtaining the sample logging curve with outlier annotations by a third party for the original logging curve after depth correction; The depth correction of the original well logging curve to obtain the depth-corrected original well logging curve comprises: randomly selecting any one of the original well logging curves as a reference curve; comparing other curves in the original well logging curve with the reference curve to obtain depth errors between the other curves in the original well logging curve and the reference curve; and depth correction of other curves in the original well logging curve according to the depth error to obtain the depth-corrected original well logging curve; The model building module is used to obtain sample logging curves with outlier annotations; randomly divide the sample logging curves into training sample data and test sample data; train multiple initial machine learning models according to the training sample data to obtain multiple trained machine learning models; test the multiple trained machine learning models according to the test sample data to obtain model test results; based on the model test results, obtain the optimal trained machine learning model under different tuning standards and build a preset set of trained machine learning models.

6. The device according to claim 5, characterized in that The model building module is also used to rank the trained machine learning models according to the model testing results under different tuning standards, obtain multiple sorting sequences and record the corresponding tuning standards; select the best trained machine learning model in the multiple sorting sequences to match the corresponding tuning standards, and build a preset set of trained machine learning models.

7. An electronic device, comprising at least one processor, and at least one memory and bus connected to the processor; wherein: The processor and the memory communicate with each other via the bus; The processor is used to call the program instructions in the memory to execute the method for marking abnormal values ​​of well logging curves as described in any one of claims 1 to 4.

8. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is run, the device where the storage medium is located is controlled to execute to implement the method for marking abnormal values ​​of well logging curves described in any one of claims 1 to 4.

Images