CN112664185A - Indicator diagram-based rod-pumped well working condition prediction method - Google Patents
Indicator diagram-based rod-pumped well working condition prediction method Download PDFInfo
- Publication number
- CN112664185A CN112664185A CN202011542249.7A CN202011542249A CN112664185A CN 112664185 A CN112664185 A CN 112664185A CN 202011542249 A CN202011542249 A CN 202011542249A CN 112664185 A CN112664185 A CN 112664185A
- Authority
- CN
- China
- Prior art keywords
- indicator diagram
- data
- working condition
- rod
- well
- 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
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method for predicting the working condition of a rod-pumped well based on an indicator diagram, wherein the method comprises the following steps: acquiring indicator diagram curve data of a plurality of pumping wells; converting indicator diagram curve data into a picture data set, sequencing the picture data set according to a time sequence to generate a time sequence picture training set, dividing the time sequence picture training set to obtain a training set and a test set, selecting a proper working condition prediction model according to the characteristics of the data set, and learning the indicator diagram working condition prediction model by combining a real indicator diagram working condition standard to obtain an oil pumping well working condition prediction model based on an indicator diagram; the indicator diagram working condition at the next moment is obtained by acquiring certain section of historical data of a certain pumping well and inputting the data into the indicator diagram working condition prediction model. The method and the system can generate the rod-pumped well working condition diagram based on the indicator diagram by utilizing a large amount of indicator diagram curve data, predict the rod-pumped well working condition at the next moment, and have high operation speed and high identification accuracy in the whole process. The method provides effective support for accurately grasping the pump detection cycle condition of the oil pumping unit well.
Description
Technical Field
The invention relates to a working condition prediction method, in particular to a working condition prediction method and a working condition prediction system based on an indicator diagram, and belongs to the technical field of oil well detection.
Background
In the mechanical oil extraction process, the sucker-rod pump oil extraction mode occupies an important position in the crude oil extraction in China. The current oil field widely adopts a maintenance strategy for the oil pumping well, namely, the oil pumping well is maintained when a certain fault causes the oil well not to be normally produced. Frequent pump detection not only causes yield loss, but also increases operation cost, and sometimes causes long operation waiting time of part of oil wells due to the reasons of service life, the limitation of the number of operation teams and the like, seriously affects the yield of the oil wells, and causes huge economic loss.
In recent years, a state-based maintenance mode is receiving more and more attention, wherein the development based on the indicator diagram analyzes the possible working condition of the oil pumping well in the next short time, and the further maintenance mode is a key link. In the traditional indicator diagram research, the current state of the indicator diagram is generally diagnosed based on a real-time indicator diagram image, and the prediction of the indicator diagram at the next time is not further researched, so that an oil pumping well working condition diagnosis model based on the indicator diagram is researched, the time sequence change of the indicator diagram is learned through a long-term and short-term memory neural network, the model parameters are continuously optimized to obtain an optimal model, and the method can greatly improve the overall development economic benefit of the oil field.
Disclosure of Invention
Aiming at the problems, the invention aims to avoid the great waste of human resources and financial resources caused by the uncertainty of the traditional oil pumping well pump inspection strategy method, thereby improving the quality and the overall economic benefit of an oil field operation team.
In order to achieve the purpose, the invention adopts the following technical scheme: the method for predicting the working condition of the rod-pumped well based on the indicator diagram comprises the following steps:
(1) and acquiring indicator diagram curve data of the oil field pumping well.
(2) And carrying out data preprocessing on the acquired data, and converting the acquired data into indicator diagram images.
(3) And (4) dividing the data into different data groups, and strengthening the time sequence strength of the data to reconstruct a data body.
(4) Performing algorithm modeling and searching optimal model parameters
(5) Construction of LSTM Algorithm model
And (1) acquiring indicator diagram curve data of the rod-pumped well, wherein the indicator diagram curve data comprises load tension borne by a polished rod suspension point and displacement of the suspension point to a wellhead.
And (2) mapping the data to a [0,1] interval through maximum-minimum standardization, drawing the processed curve data into an indicator diagram image, taking the indicator diagram image as a model independent variable, and taking the working condition of the pumping well at the next moment of the pumping well as a dependent variable.
And (3) dividing different indicator diagrams into different data groups according to the serial numbers of the pumping wells, and reordering the pumping well information of each group according to the time occurrence sequence. The data body is reconstructed by setting the size and the step length through a sliding window method, and the time sequence change among data is strengthened.
And (4) constructing an LSTM model by taking the reconstructed data volume selected in the step (3) as training data of the model and the working condition of the rod-pumped well as an output value of the algorithm, and searching for the optimal weight w and the optimal bias b by using gradient descent calculation.
And (5) searching an optimal value for the step (4) to set parameters of the LSTM, and inputting data into the LSTM model to obtain a model with optimal prediction accuracy.
Due to the adoption of the technical scheme, the invention has the following advantages: 1. dimension is removed through standardization, different features are made to be comparable, and the influence of the feature numerical level caused by the dimension on an analysis result is eliminated. 2. The sequence is sorted according to the time evolution sequence of the indicator diagram, and the connectivity among data is strengthened based on the setting of the sliding window, so that the accuracy of model prediction is improved. 3. An LSTM algorithm was chosen that is well suited for dealing with time series related problems.
Drawings
FIG. 1 is a flow chart of an algorithm for predicting the condition of a pumped well;
FIG. 2 is a schematic view of an LSTM;
FIG. 3 is a schematic view of a sliding window
Detailed Description
In order to more clearly illustrate the technical solution of the present invention, the following detailed description of the present invention is provided with reference to the accompanying drawings and examples.
The flow chart of the pumping well working condition prediction model method based on the indicator diagram is shown in figure 1, and the method is realized by the following steps:
1. the curve data of the indicator diagram of the oil pumping well in 10 continuous years of the oil field are collected. The load tension born by the polished rod suspension point and the displacement of the suspension point to the wellhead are obtained.
2. As for the step 2, because the indicator diagram curve data have different dimensions and magnitudes, if the screened data are directly used for modeling training, the algorithm is influenced greatly by the higher median value of the data, and the characteristic of lower mean value of the values is ignored. Therefore, min-max normalization is performed on the screened data, and each eigenvalue result is mapped to [0,1], and the specific conversion function is as follows:
wherein x, min and max are sample data of the characteristic value, the minimum value in the sample data of the characteristic value and the maximum value in the sample data of the characteristic value respectively.
And drawing the processed curve data into a indicator diagram image, taking the drawn indicator diagram as an independent variable required by the algorithm, and taking the corresponding indicator diagram marking information as a dependent variable required by the selected algorithm.
3. And 3, dividing the data processed in the step 2 into different data groups according to the number of the pumping unit well. And secondly, reordering the images of each data group according to the time sequence to construct a time-sequence indicator diagram data body. And finally, reading the data of the reconstructed data volume in a sliding window mode, as shown in a schematic diagram of fig. 3, selecting a window size of 6 and a step length of 1, and changing the data volume, thereby enhancing the time sequence strength of the data.
4. And 4, dividing the sample changed in the step 4 into a training set and a testing set, carrying out LSTM algorithm modeling, continuously optimizing the bias w and the bias b in the LSTM through a gradient descent algorithm, using the mean square error MSE of a predicted value and a true value as the performance index of the model, setting the final model parameters of the LSTM by setting the parameters set by the minimum mean square error, wherein the model is more optimal when the mean square error is smaller. The mean square error calculation formula is as follows:
whereinIs the predicted value of the ith sample, yiIs the actual value of the ith sample, and n is the number of samples. The gradient descent algorithm has the calculation formula as follows:
whereinThe value of the jth parameter is updated for the ith time shown, η is the step size of the gradient descent, and L is the loss function, i.e., MSE.
The above examples are only for illustrating the present invention, and the implementation steps of the methods and the like can be changed, and all equivalent changes and modifications based on the technical solution of the present invention should not be excluded from the protection scope of the present invention.
Claims (4)
1. A method for predicting the working condition of a rod-pumped well based on an indicator diagram comprises the following steps:
(1) adopting indicator diagram curve data of the pumping well, wherein the indicator diagram curve data comprises load tension borne by a polished rod suspension point and displacement of the suspension point to a well head;
(2) converting the indicator diagram curve data into a picture data set, dividing the picture data set into different data set groups according to the number of the pumping well, removing singular data, modifying error mark data, and selecting a pumping well time sequence variation diagram as a pumping well working condition diagnosis data body;
(3) the data body is changed in a sliding window mode to strengthen the time sequence;
(4) and (4) carrying out training set and test set segmentation on the data body in the step (3) to establish a long-short term memory neural network-based prediction model.
(5) And continuously updating model parameters until the model parameters are optimal through a gradient descent algorithm, and using the Mean Square Error (MSE) as an evaluation standard of the model.
2. The indicator diagram-based method for predicting the conditions of a rod-pumped well as recited in claim 1, wherein: and (2) aiming at data processing, performing maximum-minimum standardized preprocessing on indicator diagram curve data to enable coordinate points of the indicator diagram curve data to be distributed in [0,1], converting the processed data into 36 × 36 pixel pictures, and dividing the pictures into different rod-pumped well working condition diagnosis data bodies according to the rod-pumped well number.
3. The indicator diagram-based method for predicting the working condition of the rod-pumped well according to claim 1 or 2, wherein: and (3) reading data in a sliding window mode, increasing the number of training data samples, setting the window size to be 6, and setting the sliding step length to be 1.
4. The indicator diagram-based method for predicting the working condition of the rod-pumped well according to claim 1 or 3, wherein: and (4) finding out the optimal parameters by combining the gradient descent algorithm and the mean square error evaluation method, wherein the input is a indicator image pixel matrix, and the output result is 11 typical pumping well working conditions.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011542249.7A CN112664185A (en) | 2020-12-23 | 2020-12-23 | Indicator diagram-based rod-pumped well working condition prediction method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202011542249.7A CN112664185A (en) | 2020-12-23 | 2020-12-23 | Indicator diagram-based rod-pumped well working condition prediction method |
Publications (1)
Publication Number | Publication Date |
---|---|
CN112664185A true CN112664185A (en) | 2021-04-16 |
Family
ID=75409404
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202011542249.7A Pending CN112664185A (en) | 2020-12-23 | 2020-12-23 | Indicator diagram-based rod-pumped well working condition prediction method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112664185A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113756786A (en) * | 2021-09-01 | 2021-12-07 | 中国石油大学(北京) | Method for predicting time sequence indicator diagram of rod-pumped well based on deep learning |
CN114517673A (en) * | 2022-02-28 | 2022-05-20 | 沈阳中科奥维科技股份有限公司 | Indicator diagram prediction method based on electric parameters |
CN116877051A (en) * | 2023-07-24 | 2023-10-13 | 西南石油大学 | Method and equipment for diagnosing production flow and pressure abnormality of oil extraction engineering |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106294131A (en) * | 2016-07-26 | 2017-01-04 | 南京航空航天大学 | A kind of big data creation method of emulation streaming containing the test of correlative character system-oriented |
CN110318731A (en) * | 2019-07-04 | 2019-10-11 | 东北大学 | A kind of oil well fault diagnostic method based on GAN |
CN111061151A (en) * | 2019-11-21 | 2020-04-24 | 东北大学 | Distributed energy state monitoring method based on multivariate convolutional neural network |
CN111240268A (en) * | 2020-01-14 | 2020-06-05 | 重庆大学 | Axle system thermal error modeling method and thermal error compensation system based on SLSTM neural network |
CN111810124A (en) * | 2020-06-24 | 2020-10-23 | 中国石油大学(华东) | Oil pumping well fault diagnosis method based on characteristic re-calibration residual convolution neural network model |
CN112112631A (en) * | 2020-08-24 | 2020-12-22 | 黄山学院 | Intelligent identification method for indicator diagram of oil pumping unit |
-
2020
- 2020-12-23 CN CN202011542249.7A patent/CN112664185A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106294131A (en) * | 2016-07-26 | 2017-01-04 | 南京航空航天大学 | A kind of big data creation method of emulation streaming containing the test of correlative character system-oriented |
CN110318731A (en) * | 2019-07-04 | 2019-10-11 | 东北大学 | A kind of oil well fault diagnostic method based on GAN |
CN111061151A (en) * | 2019-11-21 | 2020-04-24 | 东北大学 | Distributed energy state monitoring method based on multivariate convolutional neural network |
CN111240268A (en) * | 2020-01-14 | 2020-06-05 | 重庆大学 | Axle system thermal error modeling method and thermal error compensation system based on SLSTM neural network |
CN111810124A (en) * | 2020-06-24 | 2020-10-23 | 中国石油大学(华东) | Oil pumping well fault diagnosis method based on characteristic re-calibration residual convolution neural network model |
CN112112631A (en) * | 2020-08-24 | 2020-12-22 | 黄山学院 | Intelligent identification method for indicator diagram of oil pumping unit |
Non-Patent Citations (1)
Title |
---|
檀朝东等: "时序示功图驱动的抽油机井结蜡预测及清蜡效果评价", 《石油钻采工艺》 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113756786A (en) * | 2021-09-01 | 2021-12-07 | 中国石油大学(北京) | Method for predicting time sequence indicator diagram of rod-pumped well based on deep learning |
CN114517673A (en) * | 2022-02-28 | 2022-05-20 | 沈阳中科奥维科技股份有限公司 | Indicator diagram prediction method based on electric parameters |
CN116877051A (en) * | 2023-07-24 | 2023-10-13 | 西南石油大学 | Method and equipment for diagnosing production flow and pressure abnormality of oil extraction engineering |
CN116877051B (en) * | 2023-07-24 | 2024-05-14 | 西南石油大学 | Method and equipment for diagnosing production flow and pressure abnormality of oil extraction engineering |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112664185A (en) | Indicator diagram-based rod-pumped well working condition prediction method | |
CN109272123B (en) | Sucker-rod pump working condition early warning method based on convolution-circulation neural network | |
CN111460728B (en) | Method and device for predicting residual life of industrial equipment, storage medium and equipment | |
CN113807432B (en) | Air temperature forecast data correction method based on deep learning | |
CN112651167A (en) | Semi-supervised rolling bearing fault diagnosis method based on graph neural network | |
CN110132626A (en) | A kind of Fault Diagnoses of Oil Pump method based on multiple dimensioned convolutional neural networks | |
CN112664184A (en) | Pumping unit well pump detection period prediction method based on support vector regression algorithm | |
CN112861912A (en) | Deep learning-based method and system for identifying indicator diagram of complex working condition of pumping well | |
CN109635879B (en) | Coal mining machine fault diagnosis system with optimal parameters | |
CN113137211B (en) | Oil well production parameter self-adaptive control method based on fuzzy comprehensive decision | |
CN113095414A (en) | Indicator diagram identification method based on convolutional neural network and support vector machine | |
CN113627091A (en) | Device and method for predicting energy load | |
CN110288257A (en) | A kind of depth transfinites indicator card learning method | |
CN106022352A (en) | Submersible piston pump fault diagnosis method based on support vector machine | |
CN113283180A (en) | K-means and SVR combination-based tight reservoir horizontal well fracturing productivity prediction method and application | |
CN112507479A (en) | Oil drilling machine health state assessment method based on manifold learning and softmax | |
CN110598326A (en) | Well testing interpretation method based on artificial intelligence | |
CN114718861A (en) | Intelligent diagnosis method for working condition of screw pump well based on deep learning | |
CN114004151A (en) | Oil pumping well fault prediction method based on LSTM-VGGNet | |
CN116821828A (en) | Multi-dimensional time sequence prediction method based on industrial data | |
CN114941890B (en) | Central air conditioner fault diagnosis method and system based on image and depth blur | |
CN116244596A (en) | Industrial time sequence data anomaly detection method based on TCN and attention mechanism | |
CN115597879A (en) | Method for positioning and identifying fault information in data-driven diesel engine vibration signal | |
CN115292381A (en) | Virtual currency mining behavior identification method based on extreme gradient lifting algorithm | |
CN114648095A (en) | Air quality concentration inversion method based on deep learning |
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 | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20210416 |