Time series logging data prediction method, system, medium and equipment based on fusion trend index PSO-LSTM

Abstract

The present invention relates to the field of oil exploration logging data processing, and discloses a time series logging data prediction method, system, medium and equipment based on the fusion trend indicator PSO-LSTM, which includes: integrating the trend indicator into the loss function of LSTM, guiding LSTM to approach the correct change trend of logging data during training, and ensuring the trend of predicted data; introducing a penalty function to improve the loss function, retaining the direction of the trend change of intermediate data; introducing a particle swarm optimization algorithm, combining it with LSTM, and using the PSO search capability to adjust the hyperparameters to obtain a better network structure. The present invention is suitable for real-time dynamic prediction schemes of oil industry processes, ensuring the correct change trend of predicted logging data and improving prediction accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of oil exploration logging data processing, and in particular to a method, system, medium and equipment for predicting time series logging data based on a fusion trend indicator PSO-LSTM. Background Art

[0002] In recent years, with the continuous deepening of oil exploration and development, the requirements for data detection in the drilling process have become increasingly higher. Mud logging data can reflect a variety of key information in the drilling process, such as geological characteristics, drilling parameters, etc. Traditional logging data acquisition and analysis have a certain lag. In addition, the oil industry attaches importance to data mining and data-driven algorithms, and the development of high-precision real-time drilling sensors has made it possible to conduct automated and intelligent research on real-time prediction and optimization of drilling speed in the drilling process. How to carry out real-time dynamic prediction schemes in the oil industry process to ensure the correct change trend of predicted logging data has become a technical problem that needs to be solved urgently. Summary of the invention

[0003] In view of the above problems, the purpose of the present invention is to provide a time series logging data prediction method, system, medium and equipment based on the fusion trend indicator PSO-LSTM, which can guide LSTM to approach the correct change trend of the data during the training process and automatically adjust the complex network structure of LSTM by machine optimization.

[0004] To achieve the above purpose, the present invention adopts the following technical scheme: a time series logging data prediction method based on PSO-LSTM with fusion trend index, which includes: integrating trend index into the loss function of LSTM to guide LSTM to approach the correct change trend of logging data during training to ensure the trend of predicted data; introducing penalty function to improve the loss function to retain the direction of trend change of intermediate data; introducing particle swarm optimization algorithm, combining it with LSTM, and using PSO search capability to adjust hyperparameters to obtain a better network structure.

[0005] Furthermore, trend indicators are incorporated into the LSTM loss function, including: The new loss function LOSS is obtained by combining the mean square error MSE of the predicted logging data value and the root mean square error RMSE of the predicted value difference; The error of differencing the predicted values By transforming and adding, we get:

[0006] If the positive and negative values ​​cancel each other out in the above formula, the trend information between the two points will be destroyed; in, is the value of the kth output result, is the true value of the kth data, and k is the number of output results.

[0007] Furthermore, the new loss function LOSS is:

[0008] In the formula, To balance the weights of the two indicators, For data The number of .

[0009] Furthermore, a penalty coefficient is introduced to eliminate the influence of the predicted value difference on the predicted data trend information, and the improved loss function is determined as:

[0010] In the formula, is the predicted difference between two points, is the true difference between the two points.

[0011] Furthermore, the particle swarm optimization algorithm is used to transform the prediction data from disorder to order in the optimization problem to obtain accurate prediction results.

[0012] Furthermore, the position formula and speed formula of the particle swarm optimization algorithm are updated, including:

[0013]

[0014] In the formula, is the position of the ith particle in the tth generation, is the speed of the ith particle when it moves from the tth generation to the t+1th generation; γ is the inertia weight of the speed; , They are the individual acceleration factor and social acceleration factor of the particle respectively; , is a random number on [0,1]; is the best historical position of the i-th particle up to the t-th generation; is the best position that all particles have passed up to the tth generation.

[0015] Furthermore, the particle swarm optimization algorithm is introduced and combined with LSTM. The PSO search capability is used to adjust the hyperparameters to obtain a better network structure, including: Create multiple LSTM models; Give LSTM an initial value and initial speed within the preset range; Input training data to train the LSTM model, make predictions for the trained LSTM model, and calculate its fitness function. for:

[0016] In the formula, β is the weight of the equilibrium indicator; , is the number of training set and validation set data; , Predict the results for the training set and validation set. , is the true value of the training set and the validation set; Count the optimal parameters of each LSTM, and count the optimal parameters and optimal fitness of all LSTMs; Update speed and position: Use the optimal parameters and optimal fitness to update the position and speed of each LSTM model parameter, and perform iterative training until the termination condition is met.

[0017] A time series logging data prediction system based on PSO-LSTM with fusion trend index comprises: a first improvement module, in which trend index is integrated into the loss function of LSTM to guide LSTM to approach the correct change trend of logging data during training, so as to ensure the trend of prediction data; a second improvement module, in which penalty function is introduced to improve the loss function, so as to retain the direction of trend change of intermediate data; a prediction module, in which particle swarm optimization algorithm is introduced and combined with LSTM, and hyperparameters are adjusted by using PSO search capability to obtain a better network structure.

[0018] A computer-readable storage medium storing one or more programs, wherein the one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any of the above methods.

[0019] A computing device comprises: one or more processors, a memory and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any of the above methods.

[0020] The present invention adopts the above technical solution, which has the following advantages: The present invention integrates the particle swarm optimization algorithm into the logging data prediction method based on the improved LSTM algorithm, combines the two indicators of the mean square error of the predicted value and the root mean square error of the predicted value difference, and then introduces a penalty function to eliminate the problem of the lack of trend direction in the intermediate data. The optimization ability of PSO is used to automatically find the optimal parameters of the optimal LSTM instead of the labor-intensive grid search method. It is not only suitable for the real-time dynamic prediction scheme of the oil industry process, but also ensures the correct change trend of the predicted logging data and improves the prediction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is an overall flow chart of the time series logging data prediction method based on the fusion trend indicator PSO-LSTM in an embodiment of the present invention; Figure 2 It is a flowchart of the optimization and improvement of the LSTM algorithm in an embodiment of the present invention; Figure 3 It is a detailed flow chart of the time series logging data prediction method based on the fusion trend indicator PSO-LSTM in an embodiment of the present invention. DETAILED DESCRIPTION

[0022] LSTM has strong ability to handle long-term dependency problems and can effectively capture long-term dependencies in long sequence data. Its unique gating mechanism can retain important information for a long time. At the same time, LSTM can automatically adjust its internal state according to different input data. Compared with some traditional neural networks, the training process of LSTM is more stable and is not prone to gradient disappearance or explosion problems. When dealing with high-dimensional problems, PSO can find better solutions in a shorter time than some traditional optimization algorithms. By making full use of the advantages of PSO and LSTM, a network structure with higher prediction accuracy can be found, effectively saving manpower and material costs. Therefore, the present invention proposes a time series logging data prediction method, system, medium and equipment based on the fusion trend indicator PSO-LSTM, including: integrating the trend indicator into the loss function of LSTM, guiding LSTM to approach the correct change trend of logging data during training, and ensuring the trend of predicted data; further introducing a penalty function to improve the loss function, retaining the direction of the trend change of intermediate data; then introducing an optimization method to automatically adjust the network structure of LSTM, combining LSTM with a particle swarm optimization algorithm, and using the inertia and experience of the particles themselves to update the characteristics of their own speed and position at the next moment, so as to obtain the optimal LSTM parameters, thereby obtaining a network structure with higher prediction accuracy, stronger generalization ability and stronger robustness.

[0023] In order to make the purpose, technical solution and advantages of the embodiment of the present invention clearer, the technical solution of the embodiment of the present invention will be clearly and completely described below in conjunction with the drawings of the embodiment of the present invention. Obviously, the described embodiment is a part of the embodiment of the present invention, not all of the embodiments. Based on the described embodiment of the present invention, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.

[0024] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0025] In one embodiment of the present invention, a time series logging data prediction method based on the fusion trend indicator PSO-LSTM is provided. In this embodiment, Figures 1 to 3 As shown, the method comprises the following steps: S1. Incorporate trend indicators into the loss function of LSTM to guide LSTM to approach the correct change trend of logging data during training, thus ensuring the trend of the predicted data; S2. Introduce a penalty function to improve the loss function and retain the direction of the trend change of the intermediate data; S3. Introduce the particle swarm optimization algorithm, combine it with LSTM, and use the PSO search capability to adjust the hyperparameters to obtain a better network structure.

[0026] In the above step S1, the logging data in the oil drilling industry is continuously updated during the operation. The logging parameters can reflect important information about the formation. The trend indicator is incorporated into the loss function of the LSTM to guide the LSTM to approach the correct change trend of the logging data during the training process.

[0027] like Figure 2 As shown in the figure, incorporating trend indicators into the LSTM loss function includes the following steps: S11, combining the mean square error MSE of the predicted logging data value and the root mean square error RMSE of the predicted value difference to obtain a new loss function LOSS; The new loss function LOSS is:

[0028] In the formula, To balance the weights of the two indicators, For data The number of .

[0029] S12, the error of differentiating the predicted values By transforming and adding, we get:

[0030]

[0031] If the positive and negative offset occurs in the above formula, the trend information between the two points will be destroyed; is the value of the kth output result, is the true value of the kth data, and k is the number of output results.

[0032] In the above step S2, a penalty coefficient is introduced to eliminate the influence of the predicted value difference on the predicted data trend information. The loss function is as follows:

[0033] Then the improved loss function is determined as:

[0034] In the formula, is the predicted difference between two points, is the true difference between two points, is the penalty coefficient ( >1), the larger the penalty coefficient, the higher the importance of the trend.

[0035] The penalty function is proposed to eliminate the trend destruction caused by the error in calculating the difference value, where >1, The larger it is, the more important the trend is.

[0036] In the above step S3, a particle swarm optimization algorithm is used to transform the prediction data from disorder to order in the optimization problem, so as to obtain an accurate prediction result.

[0037] Among them, the position formula and speed formula of the particle swarm optimization algorithm are updated, including:

[0038]

[0039] In the formula, is the position of the ith particle in the tth generation, is the speed of the ith particle when it moves from the tth generation to the t+1th generation; γ is the inertia weight of the speed; , They are the individual acceleration factor and social acceleration factor of the particle respectively; , is a random number on [0,1]; is the best historical position of the i-th particle up to the t-th generation; is the best position that all particles have passed up to the tth generation.

[0040] In the above step S3, the particle swarm optimization algorithm is introduced and combined with LSTM, and the hyperparameters are adjusted using the PSO search capability to obtain a better network structure, such as Figure 2 , Figure 3 As shown, the following steps are included: 3.1) Create multiple LSTM models; 3.2) Give LSTM an initial value and initial speed within a preset range; 3.3) Input training data to train the LSTM model, make predictions for the trained LSTM model, and calculate its fitness function. for:

[0041] In the formula, β is the weight of the equilibrium indicator; , is the number of training set and validation set data; , Predict the results for the training set and validation set. , is the true value of the training set and the validation set; 3.4) Count the optimal parameters of each LSTM, and count the optimal parameters and optimal fitness of all LSTMs; 3.5) Update speed and position: Use the optimal parameters and optimal fitness (the above formula) to update the position and speed of each LSTM model parameter; 3.6) Iterate the above steps 3.3) to 3.5) until the termination condition is met.

[0042] In summary, the present invention incorporates the particle swarm optimization algorithm into the logging data prediction method based on the improved LSTM algorithm, which can be used for real-time dynamic prediction solutions of petroleum industry processes, ensuring the correct change trend of the predicted logging data and improving the prediction accuracy.

[0043] The mean square error of the predicted value and the root mean square error of the predicted value difference are combined, and the penalty function is introduced to eliminate the problem of the lack of trend direction in the intermediate data. Using the optimization ability of PSO, instead of the labor-intensive grid search method, to automatically find the optimal parameters of the optimal LSTM, it can effectively solve the problem that the existing methods are not accurate and comprehensive enough in predicting drilling and logging data, and has strong practicality and promotion.

[0044] In one embodiment of the present invention, a time series logging data prediction system based on a fusion trend indicator PSO-LSTM is provided, which includes: The first improvement module incorporates trend indicators into the LSTM loss function to guide LSTM to approach the correct change trend of logging data during training, thus ensuring the trend of the predicted data. The second improvement module introduces a penalty function to improve the loss function and retain the direction of the trend change of the intermediate data; In the prediction module, the particle swarm optimization algorithm is introduced and combined with LSTM. The PSO search capability is used to adjust the hyperparameters to obtain a better network structure.

[0045] In the above embodiment, the trend indicator is incorporated into the loss function of LSTM, including: The new loss function LOSS is obtained by combining the mean square error MSE of the predicted logging data value and the root mean square error RMSE of the predicted value difference; The error of differencing the predicted values By transforming and adding, we get:

[0046] If the positive and negative values ​​cancel each other out in the above formula, the trend information between the two points will be destroyed; in, is the value of the kth output result, is the true value of the kth data, and k is the number of output results.

[0047] Among them, the new loss function LOSS is:

[0048] In the formula, To balance the weights of the two indicators, For data The number of .

[0049] In the above embodiment, a penalty coefficient is introduced to eliminate the influence of the predicted value difference on the predicted data trend information, and the improved loss function is determined as:

[0050] In the formula, is the predicted difference between two points, is the true difference between the two points.

[0051] In the above embodiment, the particle swarm optimization algorithm is used to transform the prediction data from disorder to order in the optimization problem to obtain accurate prediction results.

[0052] In this embodiment, the position formula and speed formula of the particle swarm optimization algorithm are updated, including:

[0053]

[0054] In the formula, is the position of the ith particle in the tth generation, is the speed of the ith particle when it moves from the tth generation to the t+1th generation; γ is the inertia weight of the speed; , They are the individual acceleration factor and social acceleration factor of the particle respectively; , is a random number on [0,1]; is the best historical position of the i-th particle up to the t-th generation; is the best position that all particles have passed up to the tth generation.

[0055] In the above embodiment, the particle swarm optimization algorithm is introduced and combined with LSTM, and the hyperparameters are adjusted using the PSO search capability to obtain a better network structure, including: Create multiple LSTM models; Give LSTM an initial value and initial speed within the preset range; Input training data to train the LSTM model, make predictions for the trained LSTM model, and calculate its fitness function. for:

[0056] In the formula, β is the weight of the equilibrium indicator; , is the number of training set and validation set data; , Predict the results for the training set and validation set. , is the true value of the training set and the validation set; Count the optimal parameters of each LSTM, and count the optimal parameters and optimal fitness of all LSTMs; Update speed and position: Use the optimal parameters and optimal fitness to update the position and speed of each LSTM model parameter, and perform iterative training until the termination condition is met.

[0057] The system provided in this embodiment is used to execute the above-mentioned method embodiments. Please refer to the above-mentioned embodiments for specific processes and detailed contents, which will not be repeated here.

[0058] In one embodiment of the present invention, a computing device is provided, which may be a terminal, and may include: a processor, a communication interface, a memory, a display screen, and an input device. Among them, the processor, the communication interface, and the memory communicate with each other through a communication bus. The processor is used to provide computing and control capabilities. The memory includes a non-volatile storage medium and an internal memory, and the non-volatile storage medium stores an operating system and a computer program. When the computer program is executed by the processor, the method in each of the above embodiments is implemented; the internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a management network, NFC (near field communication) or other technologies. The display screen may be a liquid crystal display screen or an electronic ink display screen, and the input device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computing device, or an external keyboard, touchpad or mouse, etc. The processor may call the logic instructions in the memory.

[0059] In addition, the logic instructions in the above-mentioned memory can be implemented in the form of software functional units and can be stored in a computer-readable storage medium when sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0060] In one embodiment of the present invention, a computer program product is provided, wherein the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, wherein the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the methods provided by the above-mentioned method embodiments.

[0061] In one embodiment of the present invention, a non-transitory computer-readable storage medium is provided, wherein the non-transitory computer-readable storage medium stores server instructions, and the computer instructions enable a computer to execute the methods provided in the above embodiments.

[0062] The above embodiment provides a computer-readable storage medium, whose implementation principle and technical effect are similar to those of the above method embodiment, and will not be repeated here.

[0063] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks 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 produce 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 1 A device that provides the functions specified in a block or multiple blocks.

[0064] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0065] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0066] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A time series logging data prediction method based on the fusion trend indicator PSO-LSTM, characterized in that: include: Incorporate trend indicators into the LSTM loss function to guide LSTM to approach the correct change trend of logging data during training, thus ensuring the trend of the predicted data; The penalty function is introduced to improve the loss function and retain the direction of the trend change of the intermediate data; The particle swarm optimization algorithm is introduced and combined with LSTM. The PSO search capability is used to adjust the hyperparameters to obtain a better network structure.

2. The method for predicting time series logging data based on the fusion trend indicator PSO-LSTM as claimed in claim 1, characterized in that: Incorporate trend indicators into the LSTM loss function, including: The new loss function LOSS is obtained by combining the mean square error MSE of the predicted logging data value and the root mean square error RMSE of the predicted value difference; The error of differencing the predicted values By transforming and adding, we get: If the positive and negative values ​​cancel each other out in the above formula, the trend information between the two points will be destroyed; in, is the value of the kth output result, is the true value of the kth data, and k is the number of output results.

3. The time series logging data prediction method based on the fusion trend indicator PSO-LSTM as claimed in claim 2 is characterized in that: The new loss function LOSS is: In the formula, To balance the weights of the two indicators, For data The number of .

4. The method for predicting time series logging data based on the fusion trend indicator PSO-LSTM as claimed in claim 1, characterized in that: The penalty coefficient is introduced to eliminate the influence of the predicted value difference on the predicted data trend information, and the improved loss function is determined as: In the formula, is the predicted difference between two points, is the true difference between the two points.

5. The method for predicting time series logging data based on the fusion trend indicator PSO-LSTM as claimed in claim 1, characterized in that: The particle swarm optimization algorithm is used to transform the prediction data from disorder to order in the optimization problem to obtain accurate prediction results.

6. The method for predicting time series logging data based on the fusion trend indicator PSO-LSTM as claimed in claim 5, characterized in that: The position formula and speed formula of the particle swarm optimization algorithm are updated, including: In the formula, is the position of the ith particle in the tth generation, is the speed of the ith particle when it moves from the tth generation to the t+1th generation; γ is the inertia weight of the speed; , They are the individual acceleration factor and social acceleration factor of the particle respectively; , is a random number on [0,1]; is the best historical position of the i-th particle up to the t-th generation; is the best position that all particles have passed up to the tth generation.

7. The method for predicting time series logging data based on the fusion trend indicator PSO-LSTM as claimed in claim 1, characterized in that: The particle swarm optimization algorithm is introduced and combined with LSTM. The PSO search capability is used to adjust the hyperparameters to obtain a better network structure, including: Create multiple LSTM models; Give LSTM an initial value and initial speed within the preset range; Input training data to train the LSTM model, make predictions for the trained LSTM model, and calculate its fitness function. for: In the formula, β is the weight of the equilibrium indicator; , is the number of training set and validation set data; , Predict the results for the training set and validation set. , is the true value of the training set and the validation set; Count the optimal parameters of each LSTM, and count the optimal parameters and optimal fitness of all LSTMs; Update speed and position: Use the optimal parameters and optimal fitness to update the position and speed of each LSTM model parameter, and perform iterative training until the termination condition is met.

8. A time series logging data prediction system based on the fusion trend indicator PSO-LSTM, characterized in that: include: The first improvement module incorporates trend indicators into the LSTM loss function to guide LSTM to approach the correct change trend of logging data during training, thus ensuring the trend of the predicted data. The second improvement module introduces a penalty function to improve the loss function and retain the direction of the trend change of the intermediate data; In the prediction module, the particle swarm optimization algorithm is introduced and combined with LSTM. The PSO search capability is used to adjust the hyperparameters to obtain a better network structure.

9. A computer-readable storage medium storing one or more programs, characterized in that: The one or more programs include instructions, which, when executed by a computing device, cause the computing device to perform any one of the methods of claims 1 to 7.

10. A computing device, characterized in that include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, and the one or more programs include instructions for executing any one of the methods described in claims 1 to 7.

Images