CN117851928A - Method and system for predicting shale oil yield based on CNN-LSTM model - Google Patents
Method and system for predicting shale oil yield based on CNN-LSTM model Download PDFInfo
- Publication number
- CN117851928A CN117851928A CN202410258681.5A CN202410258681A CN117851928A CN 117851928 A CN117851928 A CN 117851928A CN 202410258681 A CN202410258681 A CN 202410258681A CN 117851928 A CN117851928 A CN 117851928A
- Authority
- CN
- China
- Prior art keywords
- cnn
- lstm model
- shale oil
- data
- lstm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 239000003079 shale oil Substances 0.000 title claims abstract description 122
- 238000000034 method Methods 0.000 title claims abstract description 64
- 238000004519 manufacturing process Methods 0.000 claims abstract description 95
- 238000012549 training Methods 0.000 claims abstract description 43
- 238000007781 pre-processing Methods 0.000 claims abstract description 31
- 230000015654 memory Effects 0.000 claims description 14
- 230000006870 function Effects 0.000 claims description 12
- 238000012360 testing method Methods 0.000 claims description 10
- 238000003860 storage Methods 0.000 claims description 9
- 238000004590 computer program Methods 0.000 claims description 7
- 238000011176 pooling Methods 0.000 claims description 7
- 238000011156 evaluation Methods 0.000 claims description 6
- 230000008569 process Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 5
- 238000010276 construction Methods 0.000 claims description 4
- 238000003745 diagnosis Methods 0.000 claims description 4
- 238000013527 convolutional neural network Methods 0.000 description 18
- 238000010586 diagram Methods 0.000 description 6
- 238000013528 artificial neural network Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 4
- 230000000306 recurrent effect Effects 0.000 description 4
- 230000003247 decreasing effect Effects 0.000 description 3
- 238000004880 explosion Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000003062 neural network model Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000007637 random forest analysis Methods 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 230000008033 biological extinction Effects 0.000 description 2
- 230000001186 cumulative effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000006403 short-term memory Effects 0.000 description 2
- 238000012731 temporal analysis Methods 0.000 description 2
- 230000002123 temporal effect Effects 0.000 description 2
- 238000000700 time series analysis Methods 0.000 description 2
- 238000013076 uncertainty analysis Methods 0.000 description 2
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 238000007636 ensemble learning method Methods 0.000 description 1
- 210000003608 fece Anatomy 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000035699 permeability Effects 0.000 description 1
- 230000001932 seasonal effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
- 238000010200 validation analysis Methods 0.000 description 1
Abstract
The embodiment of the invention discloses a method and a system for predicting shale oil yield based on a CNN-LSTM model, wherein the method for predicting shale oil yield based on the CNN-LSTM model comprises the following steps: acquiring shale oil production data in a certain period of time, and carrying out data preprocessing on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, and the Duong model is incorporated into a loss function of the CNN-LSTM model, so that the physical interpretability of the CNN-LSTM model is improved; inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model; inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted. The method for predicting shale oil yield based on the CNN-LSTM model solves the problem that the shale oil yield cannot be accurately predicted in the prior art.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method, a system, electronic equipment and a storage medium for predicting shale oil yield based on a CNN-LSTM model.
Background
Shale oil yield prediction is one of important tasks in oilfield development and production management, and an existing method for predicting shale oil yield comprises the following steps: simulating the dynamic behavior of a shale oil reservoir by using an oil reservoir numerical simulation technology to predict the shale oil yield, wherein the oil reservoir numerical simulation technology has high requirements on time, data and labor force;
predicting future shale oil production by analyzing past production data through a yield decremental model; but the yield-decreasing model has limited application in shale oil and requires further validation.
The shale oil production is predicted by a machine learning method that learns patterns in the historical data, which yields significant results in terms of the relationship between regression production and its influencing factors, but typically ignores dynamic changes in time series.
And predicting shale oil yield by constructing a neural network model. However, convolutional Neural Networks (CNNs) may ignore the timing of data when processing sequence data, recurrent Neural Networks (RNNs), while being able to take into account timing relationships, are prone to problems of gradient extinction or explosion when processing long sequences, while long and short term memory networks (LSTM), while alleviating the gradient problem, have high demands on large scale labeling data and high computational resources, and the complexity of the model makes it relatively poor in interpretation.
Thus, there is a need for a method that accurately predicts shale oil production.
Disclosure of Invention
The embodiment of the invention aims to provide a method, a system, electronic equipment and a storage medium for predicting shale oil yield based on a CNN-LSTM model, which are used for solving the problem that the shale oil yield cannot be accurately predicted in the prior art.
To achieve the above objective, an embodiment of the present invention provides a method for predicting shale oil yield based on a CNN-LSTM model, the method specifically including:
acquiring shale oil production data in a certain period of time, and carrying out data preprocessing on the shale oil production data to obtain preprocessed data;
constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, and the Duong model is incorporated into a loss function of the CNN-LSTM model, so that the physical interpretability of the CNN-LSTM model is improved;
inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model;
inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted.
Based on the technical scheme, the invention can also be improved as follows:
further, the obtaining shale oil production data in a certain period of time, and performing data preprocessing on the shale oil production data to obtain preprocessed data, includes:
normalizing the shale oil production data, and converting the shale oil production data into normalized shale oil production data in [0,1 ].
Further, the constructing of the CNN-LSTM model comprises the following steps of;
the CNN comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN is used for extracting spatial characteristics of the preprocessed data;
extracting time characteristics of the preprocessed data through the LSTM;
and generating a predicted yield result through the full connection layer.
Further, the inputting the preprocessing data into the CNN-LSTM model for training, to obtain a trained CNN-LSTM model, includes:
dividing the preprocessing data into a training set and a testing set;
training the CNN-LSTM model based on the training set;
and evaluating the diagnosis result of the CNN-LSTM model meeting the performance condition based on the test set to obtain an evaluation index corresponding to the CNN-LSTM model.
Further, the training set is input into the CNN-LSTM model to perform training, so as to obtain a trained CNN-LSTM model, and the method further includes:
and evaluating the prediction precision of the CNN-LSTM model by determining coefficients, root mean square errors, average absolute errors and average absolute percentage errors.
Further, the training set is input into the CNN-LSTM model to perform training, so as to obtain a trained CNN-LSTM model, and the method further includes:
calculating a loss value of the CNN-LSTM model through a formula 1;
equation 1;
in the method, in the process of the invention,loss value of CNN-LSTM model, n is number of samples, +.>Is the predicted value of the CNN-LSTM model; />For actual birth value, +.>Loss value for Duong model, +.>Is->Is used for the scale factor of (a),size determination of->Effects on the CNN-LSTM model.
A system for predicting shale oil production based on a CNN-LSTM model, comprising:
the pretreatment module is used for acquiring shale oil production data in a certain period of time, and carrying out data pretreatment on the shale oil production data to obtain pretreated data;
the construction module is used for constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, the Duong model is incorporated into a loss function of the CNN-LSTM model, and the physical interpretability of the CNN-LSTM model is increased;
the training module is used for inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model;
inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted.
Further, the CNN comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN is used for extracting spatial characteristics of the preprocessed data;
the LSTM is used for extracting time characteristics of the preprocessed data;
the fully connected layer is used to generate a predicted yield result.
An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method when the computer program is executed.
A non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method.
The embodiment of the invention has the following advantages:
according to the method for predicting shale oil yield based on the CNN-LSTM model, shale oil production data in a certain period of time are obtained, and data preprocessing is carried out on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, and the Duong model is incorporated into a loss function of the CNN-LSTM model, so that the physical interpretability of the CNN-LSTM model is improved; inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model; inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted; solves the problem that the shale oil yield can not be accurately predicted in the prior art.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It will be apparent to those skilled in the art from this disclosure that the drawings described below are merely exemplary and that other embodiments may be derived from the drawings provided without undue effort.
The structures, proportions, sizes, etc. shown in the present specification are shown only for the purposes of illustration and description, and are not intended to limit the scope of the invention, which is defined by the claims, so that any structural modifications, changes in proportions, or adjustments of sizes, which do not affect the efficacy or the achievement of the present invention, should fall within the scope of the invention.
FIG. 1 is a flow chart of a method of predicting shale oil production based on a CNN-LSTM model of the present invention;
FIG. 2 is a first architecture diagram of a system for predicting shale oil production based on the CNN-LSTM model of the present invention;
FIG. 3 is a second architecture diagram of the system for predicting shale oil production based on the CNN-LSTM model of the present invention;
FIG. 4 is a schematic diagram of the structure of the RNN and LSTM of the present invention;
fig. 5 is a schematic diagram of an entity structure of an electronic device according to the present invention.
Wherein the reference numerals are as follows:
preprocessing module 10, construction module 20, training module 30, CNN-LSTM model 40, CNN401, LSTM402, full connectivity layer 403, electronic device 50, processor 501, memory 502, bus 503.
Detailed Description
Other advantages and advantages of the present invention will become apparent to those skilled in the art from the following detailed description, which, by way of illustration, is to be read in connection with certain specific embodiments, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Examples
Fig. 1 is a flowchart of an embodiment of a method for predicting shale oil yield based on a CNN-LSTM model, and as shown in fig. 1, the method for predicting shale oil yield based on the CNN-LSTM model according to the embodiment of the invention includes the following steps:
s101, shale oil production data in a certain period of time are obtained, and data preprocessing is carried out on the shale oil production data to obtain preprocessed data;
specifically, the shale oil production data is normalized, and the shale oil production data is converted into normalized shale oil production data in [0,1 ].
A preferred embodiment: according to the invention, shale oil production data of a certain block 2019, 5 months and 2022, 8 months are taken as experimental data, 40 months of production records are included, the shale oil production data are segmented according to the ratio of 7:3, the first 28 months are taken as training sets, and the subsequent 12 months are taken as test sets.
To facilitate training simulation of the CNN-LSTM model 40 and to counteract the effects of different dimensions of the production index, the input data and output data of the CNN-LSTM model 40 are preprocessed.
In the invention, shale oil production data is mapped to the [0,1] interval by adopting a normalization method.
The linear transformation formula is as follows:
；
wherein X is a normalized value, X is a characteristic value of the production index, xmin is a minimum value of the production index, and xmax is a maximum value of the production index.
S102, constructing a CNN-LSTM model 40;
specifically, the CNN-LSTM model 40 includes a CNN401, an LSTM402, and a full connectivity layer 403, and the Duong model is incorporated into the loss function of the CNN-LSTM model 40 to increase the physical interpretability of the CNN-LSTM model 40;
convolutional neural network (Convolutional Neural Network, CNN): the method is mainly used for processing data with a grid structure, such as images or time series data.
Long Short Term Memory neural network (LSTM): a variant of a recurrent neural network specifically designed to more effectively capture and address long-term dependencies.
The Duong model aims at the yield decreasing rule of the oil and gas field, and a large number of researchers propose one of various yield decreasing models;
the CNN comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN is used for extracting spatial characteristics of the preprocessed data;
extracting time characteristics of the preprocessed data through the LSTM;
and generating a predicted yield result through the full connection layer.
Recurrent neural network (Recurrent Neural Network, RNN): the cyclic connection enables information to be transferred in the network, which is particularly suitable for processing sequence data.
As a special RNN, LSTM enhances the memory module of the conventional RNN, as shown in FIG. 4. In fig. 4 (a), the memory module (composed of a single tanh layer or sigmoid layer) cannot effectively store history information for a long time, and this limitation may cause problems such as gradient extinction or gradient explosion. In fig. 4 (b), the introduction of the gate structure and memory cell states enables the LSTM to efficiently update and transmit critical information in the time series while retaining long-range information in its hidden layer. The loop network of the LSTM hidden layer consists of a forgetting gate, an input gate, an output gate and a hyperbolic tangent (tanh) layer. The processor state selectively retains the useful information of the previous step and extends it throughout the LSTM402. The gates in the interaction layer may add, delete or update information in the processor state based on the hidden state of the previous step and the input of the current step. The updated processor state and hidden state are transmitted back. The LSTM402 model can predict a single index using a single feature or multiple features.
In the present invention, based on the feature extraction of CNN401, LSTM402 is good at learning the inherent dependency of sequence data, and the function of both is fused to obtain CNN-LSTM model 40, so that the spatial and temporal features in time-series data can be recognized in a skilled manner.
S103, inputting the preprocessing data into the CNN-LSTM model 40 for training to obtain a trained CNN-LSTM model 40;
specifically, dividing the preprocessing data into a training set and a testing set;
training the CNN-LSTM model 40 based on the training set;
and evaluating the diagnosis result of the CNN-LSTM model 40 meeting the performance condition based on the test set to obtain an evaluation index corresponding to the CNN-LSTM model 40.
The prediction accuracy of the CNN-LSTM model 40 is evaluated by determining the coefficient (R2), root Mean Square Error (RMSE), mean Absolute Error (MAE), and Mean Absolute Percent Error (MAPE).
The evaluation index formula is as follows:
；
；
；
；
wherein:is the predicted value of the production prediction model; />Is the actual production value.
Calculating a loss value of the CNN-LSTM model 40 by formula 1;
equation 1;
in the method, in the process of the invention,loss value of CNN-LSTM model 40, n is number of samples, +.>Predicted values for the CNN-LSTM model 40; />For actual birth value, +.>Loss value for Duong model, +.>Is->Scale factor of>Size determination of->Effects on the CNN-LSTM model 40.
S104, inputting shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain a predicted yield result of the shale oil production data to be predicted;
the invention utilizes the same feature set to build a prediction model of the month output and the accumulated output. In particular, the CNN-LSTM model 40 based on physical constraints is compared with the single-time feature LSTM402 model (LSTM-2), the multi-feature LSTM model (LSTM-1) and the non-time series model RF;
random Forest (RF): an ensemble learning method improves the performance and generalization ability of a model by building and combining multiple decision trees together.
As shown in table 1. The results show that the monthly yield and cumulative yield curves of the CNN-LSTM model 40 based on physical constraints are more consistent with actual data than the other two LSTM402 models. From various indexes, the prediction performance of the model of the invention exceeds that of the LSTM402 model.
TABLE 1
At present, a method for predicting the shale oil yield in China almost has no decreasing curve, and the shale gas yield is mainly predicted. The invention applies a few shale gas yield prediction methods to shale oil, and aims to test the applicability of the shale gas yield prediction methods to shale oil. Wherein:
duong et al built a fracture controlled shale gas well yield decrementing model:
；
；
where q is the yield at a given time, q1 is the initial yield, t is the production time, a and m are the undetermined coefficients in the model, and Gp is the cumulative yield.
Wang et al modified the Duong model as follows:
；
where lambda is the empirical coefficient of the equation,
；
when lambda is less than or equal to 1/(2 lnt):
；
when λ > 1/(2 lnt):
；
niu et al propose a new descent curve expression based on the Duong model:
；
；
in the method, in the process of the invention,where A, B, C, m, n and p are constants,，/>，/>the method comprises the steps of carrying out a first treatment on the surface of the t is the production time in days; q is the yield in m3/d for a given time.
Evaluation indexes of the conventional shale oil yield prediction method include R2, RMSE, MAE and MAPE, as shown in Table 2. Among these metrics, lower RMSE, MAE and MAPE values reflect smaller errors and better stability of the model, and therefore lower values correspond to more superior model performance. R2 is an index for measuring the correlation between the predicted value and the actual value, and a higher R2 value indicates stronger correlation, so that the model performance is more excellent and the efficiency is higher. Compared with other machine learning models or traditional methods, the model provided by the invention has better performance on R2, RMSE, MAE and MAPE indexes, namely 0.9576, 3.1365, 2.3154 and 0.0402 respectively. This shows that the CNN-LSTM model 40 based on physical constraints achieves better performance advantages in shale oil yield predictions.
TABLE 2
According to the method for predicting shale oil yield based on the CNN-LSTM model, shale oil production data in a certain period of time are obtained, and data preprocessing is carried out on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model 40, wherein the CNN-LSTM model 40 comprises a CNN401, an LSTM402 and a full connection layer 403, and the Duong model is incorporated into a loss function of the CNN-LSTM model 40 to increase the physical interpretability of the CNN-LSTM model 40; inputting the preprocessing data into the CNN-LSTM model 40 for training to obtain a trained CNN-LSTM model 40; inputting the shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain the predicted yield result of the shale oil production data to be predicted. The method for predicting shale oil yield based on the CNN-LSTM model solves the problem that the shale oil yield cannot be accurately predicted in the prior art.
According to the method for predicting shale oil yield based on the CNN-LSTM model, a convolutional neural network (CNN 401) and a long-short-term memory (LSTM 402) network are integrated, and physical constraint is carried out based on the Duong model. A convolutional neural network model (CNN 401) is introduced to enable the CNN-LSTM model 40 to have efficient capabilities in identifying hidden relationships between features of the dataset and in performing high-precision automatic feature extraction in noisy datasets. The long-short-term memory neural network model (LSTM 402) is introduced, so that the CNN-LSTM model 40 can better process time series data, and the problems of gradient disappearance and gradient explosion in long-time series training are solved. The empirical model Duong model constraints are added to the neural network model to enhance interpretability. The physically constrained CNN-LSTM model 40 exhibits excellent accuracy, efficiency, and cost effectiveness in predicting shale oil production compared to conventional methods.
The following are some other improvements to methods of predicting shale oil production based on the CNN-LSTM model:
(1) Physical modeling: models based on physical principles and geologic features may provide a deeper understanding of shale oil production. This includes prediction by building more complex physical models taking into account reservoir properties, permeability, fracture network, etc.
(2) Time series analysis: shale oil production data typically has time series characteristics. The trend and periodicity of the yield over time can be better captured using time series analysis methods such as ARIMA (autoregressive integrated moving average) or seasonal decomposition.
(3) Data driving method: more real-time data and sensor data are utilized to predict by applying a data-driven method. This may involve big data analysis, real-time monitoring and machine learning techniques to more flexibly accommodate changes in actual production conditions.
(4) Uncertainty analysis: uncertainty is an important factor in shale oil production predictions. The introduction of uncertainty analysis methods, such as monte carlo simulation, can provide more comprehensive prediction results and take into account various sources of uncertainty.
(5) Model interpretation: for some application scenarios, especially in decision support systems, the interpretation of the model is critical. The method develops a model with strong interpretation so that a decision maker can understand the principle behind the model to better make decisions.
(6) And (3) real-time optimization: and implementing a real-time optimization strategy by using the real-time data and the model so as to improve the shale oil yield to the greatest extent. This may involve techniques in terms of control theory, optimization algorithms, etc.
FIGS. 2-3 are architecture diagrams of embodiments of a system for predicting shale oil production based on a CNN-LSTM model in accordance with the present invention; as shown in fig. 2-3, the system for predicting shale oil yield based on the CNN-LSTM model provided by the embodiment of the invention includes the following steps:
the pretreatment module 10 is used for acquiring shale oil production data in a certain period of time, and carrying out data pretreatment on the shale oil production data to obtain pretreated data;
the preprocessing module 10 is further configured to:
normalizing the shale oil production data, and converting the shale oil production data into normalized shale oil production data in [0,1 ].
The building module 20 is configured to build a CNN-LSTM model 40, where the CNN-LSTM model 40 includes a CNN401, an LSTM402, and a full connection layer 403, and incorporate a dung model into a loss function of the CNN-LSTM model 40 to increase a physical interpretability of the CNN-LSTM model 40;
the building block 20 is also configured to:
the CNN401 comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN401 is used for extracting spatial characteristics of the preprocessed data;
extracting time features of the preprocessed data through the LSTM 402;
a predicted yield result is generated by the fully connected layer 403.
The training module 30 is configured to input the preprocessing data into the CNN-LSTM model 40 for training, to obtain a trained CNN-LSTM model 40;
the training module 30 is further configured to:
dividing the preprocessing data into a training set and a testing set;
training the CNN-LSTM model 40 based on the training set;
and evaluating the diagnosis result of the CNN-LSTM model 40 meeting the performance condition based on the test set to obtain an evaluation index corresponding to the CNN-LSTM model 40.
The prediction accuracy of the CNN-LSTM model 40 is evaluated by determining coefficients, root mean square error, mean absolute error, and mean absolute percentage error.
Calculating a loss value of the CNN-LSTM model 40 by formula 1;
equation 1;
in the method, in the process of the invention,loss value of CNN-LSTM model 40, n is number of samples, +.>Predicted values for the CNN-LSTM model 40; />For actual birth value, +.>Loss value for Duong model, +.>Is->Is used for the scale factor of (a),size determination of->Effects on the CNN-LSTM model 40.
Inputting the shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain the predicted yield result of the shale oil production data to be predicted.
The CNN401 comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN401 is used for extracting spatial characteristics of the preprocessed data;
the LSTM402 is configured to extract temporal features of the preprocessed data;
the fully connected layer 403 is used to generate predicted yield results.
According to the system for predicting shale oil yield based on the CNN-LSTM model, shale oil production data in a certain period of time is obtained through the preprocessing module 10, and data preprocessing is carried out on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model 40 by a construction module 20, wherein the CNN-LSTM model 40 comprises a CNN401, an LSTM402 and a full connection layer 403, and the Duong model is incorporated into a loss function of the CNN-LSTM model 40 to increase the physical interpretability of the CNN-LSTM model 40; inputting the preprocessing data into the CNN-LSTM model 40 through a training module 30 for training to obtain a trained CNN-LSTM model 40; inputting shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain a predicted yield result of the shale oil production data to be predicted; the system for predicting shale oil yield based on the CNN-LSTM model solves the problem that the shale oil yield cannot be accurately predicted in the prior art.
Fig. 5 is a schematic diagram of an entity structure of an electronic device according to an embodiment of the present invention, as shown in fig. 5, an electronic device 50 includes: a processor 501 (processor), a memory 502 (memory), and a bus 503;
wherein, the processor 501 and the memory 502 complete the communication with each other through the bus 503;
the processor 501 is configured to invoke program instructions in the memory 502 to perform the methods provided by the above-described method embodiments, for example, including: acquiring shale oil production data in a certain period of time, and carrying out data preprocessing on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model 40, wherein the CNN-LSTM model 40 comprises a CNN401, an LSTM402 and a full connection layer 403, and the Duong model is incorporated into a loss function of the CNN-LSTM model 40 to increase the physical interpretability of the CNN-LSTM model 40; inputting the preprocessing data into the CNN-LSTM model 40 for training to obtain a trained CNN-LSTM model 40; inputting the shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain the predicted yield result of the shale oil production data to be predicted.
The present embodiment provides a non-transitory computer readable storage medium storing computer instructions that cause a computer to perform the methods provided by the above-described method embodiments, for example, including: acquiring shale oil production data in a certain period of time, and carrying out data preprocessing on the shale oil production data to obtain preprocessed data; constructing a CNN-LSTM model 40, wherein the CNN-LSTM model 40 comprises a CNN401, an LSTM402 and a full connection layer 403, and the Duong model is incorporated into a loss function of the CNN-LSTM model 40 to increase the physical interpretability of the CNN-LSTM model 40; inputting the preprocessing data into the CNN-LSTM model 40 for training to obtain a trained CNN-LSTM model 40; inputting the shale oil production data to be predicted into the trained CNN-LSTM model 40 to obtain the predicted yield result of the shale oil production data to be predicted.
Those of ordinary skill in the art will appreciate that: all or part of the steps for implementing the above method embodiments may be implemented by hardware associated with program instructions, where the foregoing program may be stored in a computer readable storage medium, and when executed, the program performs steps including the above method embodiments; and the aforementioned storage medium includes: various storage media such as ROM, RAM, magnetic or optical disks may store program code.
The apparatus embodiments described above are merely illustrative, wherein elements illustrated as separate elements may or may not be physically separate, and elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on such understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the embodiments or the methods of some parts of the embodiments.
While the invention has been described in detail in the foregoing general description and specific examples, it will be apparent to those skilled in the art that modifications and improvements can be made thereto. Accordingly, such modifications or improvements may be made without departing from the spirit of the invention and are intended to be within the scope of the invention as claimed.
Claims (10)
1. A method for predicting shale oil yield based on a CNN-LSTM model, the method comprising:
acquiring shale oil production data in a certain period of time, and carrying out data preprocessing on the shale oil production data to obtain preprocessed data;
constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, and the Duong model is incorporated into a loss function of the CNN-LSTM model, so that the physical interpretability of the CNN-LSTM model is improved;
inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model;
inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted.
2. The method for predicting shale oil yield based on a CNN-LSTM model according to claim 1, wherein the obtaining shale oil production data for a certain period of time, performing data preprocessing on the shale oil production data to obtain preprocessed data, includes:
normalizing the shale oil production data, and converting the shale oil production data into normalized shale oil production data in [0,1 ].
3. The method for predicting shale oil yield based on a CNN-LSTM model of claim 1, wherein the constructing a CNN-LSTM model comprises;
the CNN comprises a convolution layer and a maximum pooling layer which are sequentially connected, and the CNN is used for extracting spatial characteristics of the preprocessed data;
extracting time characteristics of the preprocessed data through the LSTM;
and generating a predicted yield result through the full connection layer.
4. The method for predicting shale oil yield based on a CNN-LSTM model of claim 1, wherein inputting the pre-processed data into the CNN-LSTM model for training, obtaining a trained CNN-LSTM model, comprises:
dividing the preprocessing data into a training set and a testing set;
training the CNN-LSTM model based on the training set;
and evaluating the diagnosis result of the CNN-LSTM model meeting the performance condition based on the test set to obtain an evaluation index corresponding to the CNN-LSTM model.
5. The method for predicting shale oil yield based on a CNN-LSTM model of claim 1, wherein the inputting the training set into the CNN-LSTM model for training, to obtain a trained CNN-LSTM model, further comprises:
and evaluating the prediction precision of the CNN-LSTM model by determining coefficients, root mean square errors, average absolute errors and average absolute percentage errors.
6. The method for predicting shale oil yield based on a CNN-LSTM model of claim 1, wherein the inputting the training set into the CNN-LSTM model for training, to obtain a trained CNN-LSTM model, further comprises:
calculating a loss value of the CNN-LSTM model through a formula 1;
equation 1;
in the method, in the process of the invention,loss value of CNN-LSTM model, n is number of samples, +.>Is the predicted value of the CNN-LSTM model;for actual birth value, +.>Loss value for Duong model, +.>Is->Scale factor of>Size determination of->Effects on the CNN-LSTM model.
7. A system for predicting shale oil production based on a CNN-LSTM model, comprising:
the pretreatment module is used for acquiring shale oil production data in a certain period of time, and carrying out data pretreatment on the shale oil production data to obtain pretreated data;
the construction module is used for constructing a CNN-LSTM model, wherein the CNN-LSTM model comprises CNN, LSTM and a full connection layer, the Duong model is incorporated into a loss function of the CNN-LSTM model, and the physical interpretability of the CNN-LSTM model is increased;
the training module is used for inputting the preprocessing data into the CNN-LSTM model for training to obtain a trained CNN-LSTM model;
inputting shale oil production data to be predicted into the trained CNN-LSTM model to obtain a predicted yield result of the shale oil production data to be predicted.
8. The system for predicting shale oil production based on a CNN-LSTM model of claim 7, wherein the CNN comprises a layer of convolution layer and a layer of maximum pooling layer connected in sequence, the CNN being configured to extract spatial features of the preprocessed data;
the LSTM is used for extracting time characteristics of the preprocessed data;
the fully connected layer is used to generate a predicted yield result.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method according to any one of claims 1 to 6 when the computer program is executed.
10. A non-transitory computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410258681.5A CN117851928A (en) | 2024-03-07 | 2024-03-07 | Method and system for predicting shale oil yield based on CNN-LSTM model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410258681.5A CN117851928A (en) | 2024-03-07 | 2024-03-07 | Method and system for predicting shale oil yield based on CNN-LSTM model |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117851928A true CN117851928A (en) | 2024-04-09 |
Family
ID=90542062
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410258681.5A Pending CN117851928A (en) | 2024-03-07 | 2024-03-07 | Method and system for predicting shale oil yield based on CNN-LSTM model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117851928A (en) |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200056910A (en) * | 2019-09-26 | 2020-05-25 | 한국지질자원연구원 | Method for creating a shale gas production forecasting model using deep learning |
| CN112819240A (en) * | 2021-02-19 | 2021-05-18 | 北京科技大学 | Method for predicting shale oil yield based on physical constraint LSTM model |
| CN113627076A (en) * | 2021-07-20 | 2021-11-09 | 首都师范大学 | Lithium ion battery RUL prediction method based on HI and ANN |
| CN115375031A (en) * | 2022-08-31 | 2022-11-22 | 中国石油化工股份有限公司石油工程技术研究院 | Oil production prediction model establishing method, capacity prediction method and storage medium |
| CN115640888A (en) * | 2022-10-18 | 2023-01-24 | 中国石油大学(北京) | Yield prediction method of decreasing function embedded threshold sequence network |
| CN116523086A (en) * | 2022-01-18 | 2023-08-01 | 中国石油化工股份有限公司 | Single well production dynamic prediction method based on long-short-term memory depth neural network |
| CN117272841A (en) * | 2023-11-21 | 2023-12-22 | 西南石油大学 | Shale gas dessert prediction method based on hybrid neural network |
-
2024
- 2024-03-07 CN CN202410258681.5A patent/CN117851928A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20200056910A (en) * | 2019-09-26 | 2020-05-25 | 한국지질자원연구원 | Method for creating a shale gas production forecasting model using deep learning |
| CN112819240A (en) * | 2021-02-19 | 2021-05-18 | 北京科技大学 | Method for predicting shale oil yield based on physical constraint LSTM model |
| CN113627076A (en) * | 2021-07-20 | 2021-11-09 | 首都师范大学 | Lithium ion battery RUL prediction method based on HI and ANN |
| CN116523086A (en) * | 2022-01-18 | 2023-08-01 | 中国石油化工股份有限公司 | Single well production dynamic prediction method based on long-short-term memory depth neural network |
| CN115375031A (en) * | 2022-08-31 | 2022-11-22 | 中国石油化工股份有限公司石油工程技术研究院 | Oil production prediction model establishing method, capacity prediction method and storage medium |
| CN115640888A (en) * | 2022-10-18 | 2023-01-24 | 中国石油大学(北京) | Yield prediction method of decreasing function embedded threshold sequence network |
| CN117272841A (en) * | 2023-11-21 | 2023-12-22 | 西南石油大学 | Shale gas dessert prediction method based on hybrid neural network |
Non-Patent Citations (3)
| Title |
|---|
| LUO, SH等: "Study on the Production Decline Characteristics of Shale Oil: Case Study of Jimusar Field", 《FRONTIERS IN ENERGY RESEARCH》, 27 May 2022 (2022-05-27) * |
| 刘巍;刘威;谷建伟;: "基于机器学习方法的油井日产油量预测", 石油钻采工艺, no. 01, 20 January 2020 (2020-01-20) * |
| 王洪亮;穆龙新;时付更;窦宏恩;: "基于循环神经网络的油田特高含水期产量预测方法", 石油勘探与开发, no. 05, 13 July 2020 (2020-07-13) * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sangiorgio et al. | Robustness of LSTM neural networks for multi-step forecasting of chaotic time series | |
| Müller et al. | Surrogate optimization of deep neural networks for groundwater predictions | |
| Ma et al. | DeePr-ESN: A deep projection-encoding echo-state network | |
| CN111144542B (en) | Oil well productivity prediction method, device and equipment | |
| CN110909926A (en) | TCN-LSTM-based solar photovoltaic power generation prediction method | |
| CN112819136A (en) | Time sequence prediction method and system based on CNN-LSTM neural network model and ARIMA model | |
| CN109409561B (en) | Construction method of multi-time scale time sequence collaborative prediction model | |
| CN112508265A (en) | Time and activity multi-task prediction method and system for business process management | |
| CN111461463A (en) | Short-term load prediction method, system and equipment based on TCN-BP | |
| CN110956309A (en) | Flow activity prediction method based on CRF and LSTM | |
| CN114548591A (en) | Time sequence data prediction method and system based on hybrid deep learning model and Stacking | |
| CN111507505A (en) | Method for constructing reservoir daily input prediction model | |
| CN113484882B (en) | GNSS sequence prediction method and system of multi-scale sliding window LSTM | |
| Yeh et al. | Application of LSTM based on the BAT-MCS for binary-state network approximated time-dependent reliability problems | |
| Almqvist | A comparative study between algorithms for time series forecasting on customer prediction: An investigation into the performance of ARIMA, RNN, LSTM, TCN and HMM | |
| Wen et al. | MapReduce-based BP neural network classification of aquaculture water quality | |
| CN117175588A (en) | Space-time correlation-based electricity load prediction method and device | |
| CN117851928A (en) | Method and system for predicting shale oil yield based on CNN-LSTM model | |
| CN113569479B (en) | Long-term multi-step control method, device and storage medium for rock mass crack development of stone cave temple | |
| Egele et al. | Asynchronous distributed bayesian optimization at hpc scale | |
| CN111027680B (en) | Monitoring quantity uncertainty prediction method and system based on variational self-encoder | |
| Yakushin et al. | Neural network model for forecasting statistics of communities of social networks | |
| Lei et al. | A novel time-delay neural grey model and its applications | |
| Alam et al. | Remaining useful life estimation using event data | |
| Lizhe et al. | The prediction and optimization of Hydraulic fracturing by integrating the numerical simulation and the machine learning methods |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |