CN113987933A - Pumping unit well pump detection period prediction method based on BP neural network - Google Patents
Pumping unit well pump detection period prediction method based on BP neural network Download PDFInfo
- Publication number
- CN113987933A CN113987933A CN202111253670.0A CN202111253670A CN113987933A CN 113987933 A CN113987933 A CN 113987933A CN 202111253670 A CN202111253670 A CN 202111253670A CN 113987933 A CN113987933 A CN 113987933A
- Authority
- CN
- China
- Prior art keywords
- data
- neural network
- pump
- detection period
- pump detection
- 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
- 238000001514 detection method Methods 0.000 title claims abstract description 55
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 38
- 238000000034 method Methods 0.000 title claims abstract description 30
- 238000005086 pumping Methods 0.000 title claims abstract description 20
- 238000004519 manufacturing process Methods 0.000 claims abstract description 36
- 239000003129 oil well Substances 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims description 17
- 238000012549 training Methods 0.000 claims description 9
- 238000012545 processing Methods 0.000 claims description 5
- 238000006073 displacement reaction Methods 0.000 claims description 2
- 238000003062 neural network model Methods 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 2
- 239000012530 fluid Substances 0.000 claims 1
- 238000007689 inspection Methods 0.000 claims 1
- 238000010276 construction Methods 0.000 abstract description 3
- 230000008901 benefit Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000010606 normalization Methods 0.000 description 2
- 238000002790 cross-validation Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 239000007788 liquid Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B47/00—Survey of boreholes or wells
- E21B47/008—Monitoring of down-hole pump systems, e.g. for the detection of "pumped-off" conditions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- E—FIXED CONSTRUCTIONS
- E21—EARTH OR ROCK DRILLING; MINING
- E21B—EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS
- E21B2200/00—Special features related to earth drilling for obtaining oil, gas or water
- E21B2200/22—Fuzzy logic, artificial intelligence, neural networks or the like
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Geology (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Biomedical Technology (AREA)
- Economics (AREA)
- Health & Medical Sciences (AREA)
- Mining & Mineral Resources (AREA)
- Computer Hardware Design (AREA)
- Fluid Mechanics (AREA)
- Environmental & Geological Engineering (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Geochemistry & Mineralogy (AREA)
- Geophysics (AREA)
- Development Economics (AREA)
- Geometry (AREA)
- Game Theory and Decision Science (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
Abstract
The invention designs a method for predicting the pump detection period of an oil pumping unit well based on a BP neural network algorithm, and belongs to the technical field of oil well detection. The method includes obtaining a set of production-related data; preprocessing data; characteristic engineering; establishing a model according to a BP neural network algorithm, and finding out an optimal precision model for prediction; finally, the pump detection period is predicted through the model. By adopting the technical scheme, the invention can predict the future pump detection period on the premise of only analyzing the existing data. The invention can better discover the hidden relation among data by means of the BP neural network, can more accurately predict the pump detection period compared with other algorithms, and provides better decision service for the construction of the oil pumping well.
Description
Technical Field
The invention relates to a pump-testing period prediction method for a rod-pumped well, in particular to a pump-testing period prediction method for a rod-pumped well based on a BP neural network, and belongs to the technical field of oil well detection.
Background
In the oil production process, the determination of the pump detection period is very important for reasonably planning the production system. However, in actual production, a production system for regularly checking the pump is usually adopted, but a large amount of manpower and material resources are consumed, and even additional economic loss is caused due to untimely pump checking, so that the overall profit of the oil field is reduced. According to the method, from the perspective of historical data, a BP neural network algorithm is applied, the production state data and the law of the pump detection period are captured from the historical data, the pump detection period of an oil well is predicted through the real-time production state data, loss caused by unreasonable pump detection period is reduced, and the overall benefit is improved. Meanwhile, the method strengthens the application of a characteristic engineering theory, screens the characteristics with weak relevance and accelerates the operation speed of the algorithm. Finally, the relation between other characteristics and the pump detection period can be found out through a BP neural network algorithm, the pump detection period can be predicted by combining the algorithm with new production state data, reference can be provided for formulating a production system, the pump detection period can be well predicted based on a model of historical data, loss caused by unreasonable pump detection period is reduced, social resource investment can be reduced compared with regular pump detection, and the improvement of oil field development benefits is obviously influenced.
Disclosure of Invention
Aiming at the problems, the invention aims to provide a method for predicting the pump detection period of a pumping unit well based on a BP neural network algorithm, so that construction measures are arranged in advance, construction suggestions are guided, production strategies are optimized, and the overall profit of an oil field is improved.
In order to achieve the purpose, the invention adopts the following technical scheme: the method for predicting the pump detection period of the oil pumping unit well based on the BP neural network algorithm comprises the following steps:
(1) collecting daily production data of the pumping well as a sample data set;
(2) sorting the preliminary data set, and adding a new characteristic parameter named as a pump detection period;
(3) carrying out data preprocessing on the data;
(4) screening data characteristic parameters by using a gray level correlation algorithm;
(5) modeling by using a BP neural network algorithm, and optimizing model parameters;
(6) constructing a pump detection period prediction model based on a BP neural network algorithm;
(7) and combining the pump detection period prediction model and the newly acquired data to obtain a pump detection period prediction value.
And (1) collecting daily production data of the pumping well as a model training data set.
And (2) adding a new characteristic parameter named as a pump detection period based on the production data acquired in the step (1). And then counting the days of each data in the production data from the next pump detection period as the value of the characteristic parameter of the pump detection period.
And (3) preprocessing data based on the data of the characteristic parameters of the pump detection period added in the step (2), including processing null values, setting data range constraints, deleting data which do not meet the range constraints, and normalizing the data.
And (4) analyzing the relation between each characteristic parameter and the characteristic parameter of the pump detection period by using a gray level correlation algorithm based on the data preprocessed in the step (3), and deleting the characteristic parameters with the correlation degree score smaller than 0.01.
And (5) based on the data set after the characteristic parameters are screened in the step (4), dividing a training set and a testing set according to the ratio of 3:1, putting the training set into a BP neural network model for fitting, optimizing parameters of the BP neural network algorithm by a grid search method, and keeping the model parameters of the BP neural network algorithm with the minimum error on the testing set.
And (6) bringing the model parameters into the BP neural network algorithm based on the model parameters obtained in the step (5) to obtain a pump detection period prediction model based on the BP neural network algorithm, and exporting an algorithm file to obtain an optimal algorithm model.
And (7) bringing newly acquired production data into the model based on the pump detection period prediction model obtained in the step (6) to obtain a pump detection period prediction result.
Due to the adoption of the technical scheme, the invention has the following advantages: 1. firstly, counting pump detection cycle days corresponding to historical production data, and adding new characteristic parameters serving as daily production data; 2. according to the method, abnormal data and error data are screened, missing values are filled, noise of a primary data set is reduced, and accuracy of a subsequent algorithm is well improved; 3. the min-max normalization method is adopted to normalize the data to the [0,1] interval, so that the dimension of each characteristic is removed, the influence of noise is reduced again, and the running speed and the accuracy of the algorithm are greatly improved; 4. by using a gray level correlation algorithm, characteristic parameters with weak correlation degree are screened, and the burden of the algorithm is reduced; 5. the model is optimized for the successfully trained algorithm model through a grid search method, so that the accuracy of the algorithm is improved; 6. the method can obtain the relation between the production data and the pump detection period by analyzing the existing data, and further predict the future pump detection period; 7. the method can provide reference suggestions for production decisions by predicting the number of pump detection cycle days in advance, and can reduce the investment of related production resources, save the cost and improve the benefits.
Drawings
FIG. 1 is a flow chart of a pump cycle prediction algorithm for a pumped well;
FIG. 2 is a schematic diagram of a BP neural network algorithm.
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.
A flow chart of a method for predicting a pumping cycle of a pumping unit based on a BP neural network algorithm in the embodiment is shown in fig. 1, and the method is specifically realized by the following steps:
1. daily production data of the pumping well is collected to serve as a sample data set. The acquired data characteristic parameters comprise casing pressure, oil pressure, water content, uplink current, downlink current, maximum load, minimum load, rated torque, stroke, pump diameter, displacement, daily liquid production, daily oil production and daily production time. And classifying the acquired data according to the wells, and respectively using the data as training sets of different oil well pump detection period models.
2. And (3) adding a new characteristic parameter to the production data set obtained in the step (1), and naming the new characteristic parameter as a pump detection period. And then counting the days of each data in the production data from the next pump detection period as the value of the characteristic parameter of the pump detection period.
3. And (4) carrying out data preprocessing on the data processed in the step (2). Firstly, processing null values, and deleting a certain row or a certain column when the row or the column is all null values; when an individual data item in a row or in a column is null, it is padded by a mean or 0. Secondly, processing abnormal values, when data in a certain column of data exceeds a theoretical range, the data are regarded as invalid data, and deleting the line, wherein the temperature is 1000 ℃, and the line cannot be deleted; when the relation between two columns of data does not accord with objective rules, deleting the data in the row, for example, the sleeve pressure is less than the oil pressure, which is unreasonable, and deleting the data. Then, the data type is converted, since the algorithm cannot process the character type data, the character type data is converted into Int integer type data, for example, three types of small layer numbers, namely "F1", "F2" and "F3", and then the types are converted into 1, 2 and 3 respectively, and then the training is carried out by turning into the algorithm. And finally, after all the data are converted into numbers, carrying out quantitative removal, toughening and normalization on the data, wherein min-max standardization is adopted to normalize the data, and dimensions are removed, and the min-max theoretical formula is as follows:
wherein
Is the normalized characteristic parameter, X is the original characteristic parameter, XminDenotes the minimum value of X, XmaxRepresents the maximum value in x. Through this process, a new data set is generated.
4. And (4) performing characteristic processing on the new data set generated in the step (3). And (4) preferably selecting the features with high association degree through a grey association degree algorithm, and simultaneously rejecting the features with low association degree. Wherein the relevance calculation weight formula is as follows:
the correlation coefficient calculation formula in the formula is as follows:
rho is a resolution coefficient, rho is more than 0 and less than 1, and the smaller rho is, the larger the difference between the correlation coefficients is, and the stronger the distinguishing capability is. Usually ρ is 0.5. x is the number of0(k)，xi(k) Respectively representing the kth number of the mother sequence and the kth number of the ith characteristic value of the subsequence. Zetai(k) The correlation coefficient of the kth value representing the ith feature with the kth value of the mother sequence. And finally, setting a threshold value of 0.01 according to the size of the association degree, namely deleting all the features smaller than the association degree and reserving the features larger than the threshold value.
5. And (4) dividing the new data set generated in the step (4) into a training set and a test set according to the ratio of 3:1, and carrying out BP neural network algorithm modeling. The BP neural network comprises an input layer, a hidden layer and an output layer, wherein the input layer is the characteristic attribute obtained through the step 3, the hidden layer is invisible by default, and the output layer is the pump detection period. Then optimizing BP neural network parameters through cross validation and a grid search method, measuring model accuracy by using MSE mean square error, wherein the MSE calculation formula is as follows:
wherein y'iIs the predicted value of the ith sample, yiIs the actual value of the ith sample, and n is the number of samples. After minimizing the MSE, the record of the algorithm parameters at this time is saved.
6. And (5) bringing the optimized parameters obtained in the step (5) into a BP neural network algorithm, storing and exporting an algorithm model file, and further obtaining an optimal pump detection period prediction model based on the BP neural network algorithm.
7. And (4) bringing the newly acquired oil well production data into the model obtained in the step (6) for prediction, and putting the newly acquired production data into the pump detection period prediction model to obtain a corresponding pump detection period prediction value.
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 scheme of the present invention should not be excluded from the protection scope of the present invention.
Claims (8)
1. The method for predicting the pump detection period of the oil pumping unit well based on the BP neural network algorithm comprises the following steps:
(1) collecting daily production data of the pumping well as a sample data set;
(2) sorting the preliminary data set, and adding a new characteristic parameter named as a pump detection period;
(3) preprocessing the data;
(4) screening data characteristic parameters by using a gray level correlation algorithm;
(5) modeling by using a BP neural network algorithm, and optimizing model parameters;
(6) constructing a pump detection period prediction model based on a BP neural network algorithm;
(7) and combining the pump detection period prediction model and the newly acquired data to obtain a pump detection period prediction value.
2. The method for predicting pump inspection cycle of a pumping unit based on BP neural network algorithm according to claim 1, wherein the characteristic parameters of daily production data obtained in the step (1) comprise casing pressure, oil pressure, water content, up current, down current, maximum load, minimum load, rated torque, stroke, pump diameter, displacement, daily fluid production, daily oil production, and daily production time. And classifying the acquired data according to the wells, and respectively using the data as training sets of different oil well pump detection period models.
3. The method for predicting pump-testing period of pumping unit based on BP neural network algorithm of claim 1, wherein in step (2), based on the production data set obtained in step (1), a new characteristic parameter is added, named "pump-testing period". And counting the days of each data in the production data from the next pump detection period as the value of the characteristic parameter corresponding to the pump detection period.
4. The method for predicting pump-testing period of pumping unit based on BP neural network algorithm of claim 1, wherein said step (3) of performing data preprocessing including processing null values, setting data range constraints and deleting data that do not conform to the range constraints, and normalizing the data based on the data added the "pump-testing period" characteristic parameter in step (2).
5. The method for predicting the pump-testing period of a pumping unit based on the BP neural network algorithm as claimed in claim 1, wherein in the step (4), based on the data preprocessed in the step (3), the relation between each characteristic parameter and the characteristic parameter of the pump-testing period is analyzed by using a gray-scale correlation algorithm, and the characteristic parameter with the correlation score of less than 0.01 is deleted.
6. The method for predicting the pumping unit well pump detection period based on the BP neural network algorithm as claimed in claim 1, wherein in the step (5), based on the data obtained in the step (4), as a data set, the data set is divided into a training set and a testing set according to the proportion of 3:1, the training set is put into the BP neural network model for fitting, then the parameters of the BP neural network algorithm are optimized through a grid search method, and the model parameters of the BP neural network algorithm with the smallest error on the testing set are reserved.
7. The method for predicting the pump detection period of a pumping unit well based on the BP neural network algorithm as claimed in claim 1, wherein the model parameters obtained in the step (5) are substituted into the BP neural network algorithm in the step (6) to obtain a pump detection period prediction model based on the BP neural network algorithm, and an algorithm file is stored and derived to obtain an optimal pump detection period prediction model based on the BP neural network.
8. The method for predicting pump-testing period of pumping unit well based on BP neural network algorithm of claim 1, wherein in step (7), based on the pump-testing period prediction model obtained in step (6), the newly acquired production data is put into the pump-testing period prediction model to obtain the corresponding pump-testing period prediction value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111253670.0A CN113987933A (en) | 2021-10-27 | 2021-10-27 | Pumping unit well pump detection period prediction method based on BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111253670.0A CN113987933A (en) | 2021-10-27 | 2021-10-27 | Pumping unit well pump detection period prediction method based on BP neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113987933A true CN113987933A (en) | 2022-01-28 |
Family
ID=79742343
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111253670.0A Pending CN113987933A (en) | 2021-10-27 | 2021-10-27 | Pumping unit well pump detection period prediction method based on BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113987933A (en) |
-
2021
- 2021-10-27 CN CN202111253670.0A patent/CN113987933A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108009716B (en) | Multi-level evaluation method for influence factors of horizontal well volume fracturing effect | |
| CN112529341B (en) | Drilling well leakage probability prediction method based on naive Bayesian algorithm | |
| CN107305601B (en) | Method for analyzing efficiency factors of oil pumping well system | |
| CN113807570B (en) | XGBoost-based reservoir dam risk level assessment method and system | |
| CN111210170B (en) | Environment-friendly management and control monitoring and evaluation method based on 90% electricity distribution characteristic index | |
| CN108875118B (en) | Method and device for evaluating accuracy of prediction model of silicon content of blast furnace molten iron | |
| CN114048436A (en) | Construction method and construction device for forecasting enterprise financial data model | |
| CN112664184A (en) | Pumping unit well pump detection period prediction method based on support vector regression algorithm | |
| CN115470962A (en) | LightGBM-based enterprise confidence loss risk prediction model construction method | |
| CN112664185A (en) | Indicator diagram-based rod-pumped well working condition prediction method | |
| CN109594967A (en) | A kind of lock of tool drilling detection method for early warning based on well logging big data | |
| CN108830405B (en) | Real-time power load prediction system and method based on multi-index dynamic matching | |
| CN113030633B (en) | GA-BP neural network-based power distribution network fault big data analysis method and system | |
| CN116629686A (en) | Method and device for evaluating enterprise energy consumption data | |
| CN113987933A (en) | Pumping unit well pump detection period prediction method based on BP neural network | |
| CN116383645A (en) | Intelligent system health degree monitoring and evaluating method based on anomaly detection | |
| CN115081485B (en) | AI-based magnetic flux leakage internal detection data automatic analysis method | |
| CN113298148B (en) | Ecological environment evaluation-oriented unbalanced data resampling method | |
| CN115456260A (en) | Customer service telephone traffic prediction method | |
| CN114004151A (en) | Oil pumping well fault prediction method based on LSTM-VGGNet | |
| CN115630571A (en) | Oil well indicator diagram automatic diagnosis method based on ensemble learning | |
| CN112765549B (en) | KNN-Matrix method for repairing abnormal data of building energy consumption monitoring | |
| CN113988410A (en) | Cross-region tight oil reservoir oil well productivity prediction method based on KNN algorithm and polynomial regression algorithm combination | |
| CN115883182A (en) | Method and system for improving network security situation element identification efficiency | |
| CN114575802A (en) | High water cut oil reservoir oil well yield prediction method based on machine learning |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication |