CN117495206A - Gas well drainage and production process well selection method based on artificial intelligence algorithm - Google Patents
Gas well drainage and production process well selection method based on artificial intelligence algorithm Download PDFInfo
- Publication number
- CN117495206A CN117495206A CN202311848890.7A CN202311848890A CN117495206A CN 117495206 A CN117495206 A CN 117495206A CN 202311848890 A CN202311848890 A CN 202311848890A CN 117495206 A CN117495206 A CN 117495206A
- Authority
- CN
- China
- Prior art keywords
- data
- gas
- well
- yield
- production
- 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.)
- Granted
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 121
- 238000010187 selection method Methods 0.000 title claims abstract description 20
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 19
- 238000000034 method Methods 0.000 claims abstract description 104
- 230000000694 effects Effects 0.000 claims abstract description 65
- 238000011156 evaluation Methods 0.000 claims abstract description 24
- 238000007726 management method Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000011112 process operation Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims abstract description 8
- 238000007405 data analysis Methods 0.000 claims abstract description 6
- 238000005457 optimization Methods 0.000 claims abstract description 6
- 238000012360 testing method Methods 0.000 claims abstract description 5
- 230000008569 process Effects 0.000 claims description 89
- 238000004140 cleaning Methods 0.000 claims description 27
- 230000008859 change Effects 0.000 claims description 18
- 230000001186 cumulative effect Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 13
- 238000003066 decision tree Methods 0.000 claims description 12
- 238000005065 mining Methods 0.000 claims description 12
- 230000007423 decrease Effects 0.000 claims description 10
- 230000009467 reduction Effects 0.000 claims description 6
- 230000005856 abnormality Effects 0.000 claims description 4
- 230000000638 stimulation Effects 0.000 claims description 4
- 238000010219 correlation analysis Methods 0.000 claims description 3
- 238000007689 inspection Methods 0.000 claims description 3
- 238000002360 preparation method Methods 0.000 claims description 3
- 238000012216 screening Methods 0.000 claims description 3
- 238000004806 packaging method and process Methods 0.000 claims description 2
- 238000005192 partition Methods 0.000 claims 1
- 239000007789 gas Substances 0.000 abstract description 179
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 abstract description 12
- 239000003345 natural gas Substances 0.000 abstract description 6
- 230000002354 daily effect Effects 0.000 description 31
- 238000004458 analytical method Methods 0.000 description 18
- 238000009826 distribution Methods 0.000 description 9
- 230000003068 static effect Effects 0.000 description 8
- 230000008901 benefit Effects 0.000 description 5
- 230000000630 rising effect Effects 0.000 description 4
- 230000007812 deficiency Effects 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000002343 natural gas well Substances 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 239000006260 foam Substances 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000005096 rolling process Methods 0.000 description 2
- -1 shale gas Chemical compound 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000033228 biological regulation Effects 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000003111 delayed effect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 230000008092 positive effect Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 238000010223 real-time analysis Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- 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
- G06N3/09—Supervised learning
-
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- 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
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Data Mining & Analysis (AREA)
- Strategic Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Educational Administration (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Development Economics (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Operations Research (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Quality & Reliability (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a gas well drainage and production process well selection method based on an artificial intelligence algorithm, which comprises the following steps: acquiring historical data; treating the historical data to obtain modeling data; establishing an operation effect evaluation index system, and evaluating the effect of a running drainage and production process well to obtain various process operation effects; defining positive and negative sample data sets through data analysis and determining a well selection model prediction target; obtaining a final sample data set; dividing the final sample data set into a training set and a testing set; substituting the data of the training set into various intelligent algorithm models according to various drainage and extraction processes, and performing parameter adjustment and algorithm optimization to obtain a plurality of well selection models; according to the well selection service characteristics and the types of the drainage and production processes selected by the current gas well, well selection model operation rules are established, and the situation of re-checking the well is avoided; establishing a well selection result output strategy according to the current information of the gas well; the method provides technical support for improving the digitization and intelligent management of the natural gas field.
Description
Technical Field
The invention relates to the technical field of gas well drainage and production process well selection, in particular to a gas well drainage and production process well selection method based on an artificial intelligence algorithm.
Background
In the production process of the shale gas well, along with the rapid decrease of the productivity and the gradual decrease of the bottom hole pressure, the self production capacity of the gas well with liquid is reduced, so that the bottom hole effusion is caused, the normal and stable production of the gas well is influenced, when the bottom hole effusion is serious, even the risk of flooding can occur, the gas well loses the gas production capacity, and the economic benefit is seriously influenced. Therefore, to ensure continuous and efficient production of a gas well, when or at the time of fluid accumulation in the gas well, it is necessary to identify in time, take drainage and production process measures, and determine which process is most suitable for the gas well. At present, the drainage and gas production process for shale gas wells mainly comprises a foam drainage process, a plunger lifting process, a gas lifting process, a spacing process and the like. For a certain gas well, the dimensions selected by the drainage process mainly consider the following directions: whether the gas well is at the optimal time of the upper process or not, otherwise, the investment benefit is low; the current state of the gas well is most suitable for the process, so that the optimal liquid discharge efficiency of the gas well is achieved.
Currently, for process well selection, common analysis methods include a production dynamic method, an inflow and outflow dynamic method, a plate method, a utility function method, and the like. However, these methods are all conventional mechanism methods, the accuracy of which does not have a certain universality, and which basically depend on manual calculation and analysis by gas well dynamic analysis engineers, and do not realize intelligent and automatic process optimization based on real-time dynamic data. Along with the gradual increase of the number of natural gas field development wells in China, the requirements for cost reduction and synergy are urgent, the improvement of the fine management capability of the gas wells is imperative, the single production area is almost the process requirement analysis of hundreds of wells every day, the conventional well selection analysis method has the problems of difficult well selection time determination, large well selection analysis workload, low efficiency, low accuracy, high labor cost and the like, and cannot meet the requirement of on-site large number of well selection analysis, and the fine management service capability of the gas wells cannot be effectively provided.
In view of the above, the invention provides a well selection method of a gas well drainage and production process based on an artificial intelligence algorithm, and by creating a set of process intelligent optimal intelligent well selection model based on dynamic data and static data of natural gas well production, the problems of delayed process opportunity, low efficiency, large artificial workload, low process fitness and the like in the traditional process optimization are solved, technical support is provided for improving the digital and intelligent management of natural gas fields, and the technology is also suitable for the gas well development management of unconventional natural gas such as shale gas, dense gas and the like.
Disclosure of Invention
The invention aims to provide a gas well drainage and production process well selection method based on an artificial intelligence algorithm, which comprises the following steps: s1: acquiring historical data related to gas well production; s2: treating the historical data by using a data cleaning tool to obtain modeling data; s3: establishing an operation effect evaluation index system based on a drainage and production process theory, and evaluating the effect of a put-into-operation drainage and production process well through the operation effect evaluation index system to obtain various process operation effects; s4: screening time period data with the operation effects of the various processes being greater than a preset effect threshold value, defining positive and negative sample data sets through data analysis, and determining a well selection model prediction target; s5: designing derivative parameters, enriching parameter dimensions, and obtaining a final sample data set; s6: dividing the final sample data set into a training set and a testing set according to a proportion, and respectively training a well selection model and evaluating the effect of the trained well selection model; s7: substituting the data of the training set into various intelligent algorithm models according to various drainage processes, and performing parameter adjustment and algorithm optimization to obtain a plurality of well selection models; s8: according to the well selection service characteristics and the types of the drainage and production processes selected by the current gas well, well selection model operation rules are established, and the situation of re-checking the well is avoided; s9: and establishing a well selection result output strategy according to the current information of the gas well.
Furthermore, the method also comprises the steps of packaging the developed data cleaning interface, the intelligent model and the output rule into a well selection model interface, integrating the well selection model interface with a production management platform and deploying the well selection model interface and the well selection model interface in a user side server.
Further, the developed data cleaning interface, intelligent model and output rule are packaged into a well selection model interface, integrated with a production management platform and deployed in a client server, comprising: the data cleaning interface takes minutes as a unit to call production data from a database; performing quality inspection on the acquired production data, and classifying the production data with the detected abnormality to obtain complete data; executing data multidimensional operation on the complete data, and simultaneously carrying out multidimensional operation on different types of data at the same time to one piece of data to finish data preparation and obtain input data; and respectively inputting the input data into the plurality of well selection models, respectively outputting the probability of the gas well selected by the well in a certain time period by the models, and determining the well selection result of the gas well in the drainage process based on the probability.
Further, the treatment of the historical data comprises data effective time determination, data cleaning and continuous time processing; the data valid time is determined as a first piece of valid data of the gas well, a plurality of reference fields are determined according to the importance degree of the gas well parameters, and the first second or minute data of which the reference fields are not 0 at the same time is used as the first piece of valid data of the gas well; the other fields are filtered backwards after the previous piece of data; the data cleaning comprises pressure data cleaning, temperature data cleaning and yield data cleaning; the pressure data are cleaned to be filled with the pressure data exceeding a preset pressure range; the temperature data are cleaned, and the temperature data exceeding a preset temperature range are filled; the output data is cleaned to acquire and clean accumulated gas yield, and daily gas yield and instantaneous gas yield are filled based on the cleaned accumulated gas yield; the continuous time processing is to fill in missing data to generate complete time series for the same point in time of different fields.
Further, acquiring and cleaning the accumulated gas yield, and filling the daily gas yield and the instantaneous gas yield based on the cleaned accumulated gas yield, comprising: when the accumulated gas yield exceeds the range or the value of the preset accumulated gas yield [0,100000] to be 0.9999', deleting the data; when the accumulated gas yield is lost, obtaining the current lost accumulated gas yield based on the last accumulated gas yield and the previous accumulated gas yield of the lost accumulated gas yield; filling data with the average gas production per minute of less than or equal to 0.000000001; for data with gas production rate of more than 100 per minute, the data is reduced by 100,0000 times; the instantaneous gas production and daily gas production are determined based on the cumulative gas production on the day and the cumulative gas production on the day before.
Further, the establishing of the gas well drainage process operation effect evaluation index system comprises the following steps: determining a baseline production of the gas well without resorting to a drainage process; acquiring the actual yield of the gas well after adopting a drainage and production process; determining the yield increase gas quantity and the casing pressure change condition of the gas well based on the reference yield and the actual yield; and determining an effect evaluation division based on the yield increasing gas quantity and the casing pressure change condition.
Further, the calculation formula of the reference yield is as follows:
wherein,is the standard yield of the month; />Average yield for the month of last month; />Is the rate of decrease of the last month;
the daily yield increase amount is calculated according to the following formula:
wherein,yield increase amount for the current day; />Actual yield on the present day;
the calculation formula of the daily average yield increase amount is as follows:
wherein,for the average daily yield increase in 30 days after taking the drainage process measures, the unit is +.>；/>For the average daily gas production in 30 days after taking the drainage and mining process measures, the unit is +.>；/>For the average daily gas production in 30 days before taking the drainage and mining process measures, the unit is +.>；
The calculation formula of the casing pressure drop value of the gas well is as follows:
wherein,the unit of the reduction value of the 30 balance sleeve is MPa after the drainage and mining process measures are taken; />The unit is MPa for the jacket pressure of a 30 balance before taking the drainage and mining technological measures; />In order to adopt the measure of the drainage and production process, the balance is uniformly sleeved with pressure, and the unit is MPa.
Further, the calculation formula of the reduction rate is as follows:
wherein,is the rate of decrease; />Is the average gas production rate in the last month; />The average gas production rate in the last month is obtained.
Further, determining an effect assessment score based on the stimulation gas volume and the casing pressure variation, comprising: when the yield increase gas amount is less than or equal to 0: the sleeve pressure drop value is less than or equal to zero, and the effect is poor; the effect is good when the sleeve pressure drop value is in the (0, 0.5) interval, the effect is good when the sleeve pressure drop value is greater than 0.5, the effect is good when the sleeve pressure drop value is less than 0 when the yield gas amount is in the (0, 0.5) interval, the effect is good when the sleeve pressure drop value is greater than 0.5, the effect is good when the yield gas amount is in the (0.5, 1) interval, the effect is good when the sleeve pressure drop value is less than or equal to 0, the effect is good when the sleeve pressure drop value is greater than 0, and the effect is good when the yield gas amount is greater than 1, the effect is evaluated without considering the sleeve pressure change condition.
Further, related derivative parameters are designed by means of correlation analysis, business experience or decision tree models, and parameter dimensions are enriched.
The technical scheme of the embodiment of the invention has at least the following advantages and beneficial effects:
the scheme of the invention realizes the real-time extraction, real-time analysis and real-time processing of the production data through the data quality real-time processing technology, can simultaneously contain all production gas wells, and has the processing time of minutes without manual intervention; for prediction scenes and complicating scenes, the intelligent model has the advantages that through learning experience samples, internally hidden relationship logic which cannot be structured can be rapidly analyzed, and early warning of well selection time is realized; meanwhile, the built intelligent model can analyze the whole well data in real time, output the result, and only needs a small amount of manual verification on the result without manual intervention, thereby greatly saving the cost of human resources; the static data used by the invention has more dimensions, so that the covered gas well has more scene types, has universality, is beneficial to popularization and application of different types of gas fields, and provides technical support for improving the digital and intelligent management of the natural gas field; meanwhile, the invention is also applicable to the development and refined management of gas wells of unconventional natural gas such as shale gas, dense gas and the like.
Drawings
FIG. 1 is an exemplary flow chart of a well selection method for a gas well drainage process based on an artificial intelligence algorithm provided by the invention;
FIG. 2 is a schematic diagram showing the distribution of the average daily yield increase amount for 30 days before and after taking the drainage process measures;
FIG. 3 is a graphical representation of a first type of well casing pressure drop distribution based on stimulation gas classification.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. The components of the embodiments of the present invention generally described and illustrated in the figures herein may be arranged and designed in a wide variety of different configurations.
FIG. 1 is an exemplary flow chart of a well selection method for a gas well drainage process based on an artificial intelligence algorithm provided by the invention.
As shown in fig. 1, the flow of the well selection method of the gas well drainage and production process based on the artificial intelligence algorithm comprises the following steps:
step S1: historical data relating to gas well production is obtained.
The historical data comprises process account data, dynamic data and static data; the static data comprise data of a well bore, well deviation, well logging, well completion and the like; the dynamic data refers to data generated during real-time production; the gas well is a natural gas well.
Step S2: and treating the historical data by using a data cleaning tool to obtain the modeling data. The processing of the historical data includes data valid time determination, data cleansing and continuous time processing. For example, a set of data quality control interface protocols aiming at real-time production dynamic data, process account data and static data is established, so that automatic processing of the data is realized.
Determining data valid time as first valid data of the gas well according to the importance degree of the gas well parameters, and determining a plurality of reference fields, wherein the first second or minute data of the plurality of reference fields are not 0 at the same time; the remaining fields are filtered backwards from the previous piece of data. The gas well parameters comprise oil jacket pressure, needle valve back pressure, accumulated output, instantaneous gas yield, daily gas yield, throttle valve front temperature, throttle valve back temperature and the like; the reference fields include casing pressure, oil pressure, and accumulated production.
The data cleansing includes pressure data cleansing, temperature data cleansing and throughput data cleansing.
The pressure data is cleaned to fill in the pressure data exceeding the preset pressure range. The preset pressure range is a pressure range in which preset normal pressure data are located; the pressure data comprise oil pressure, sleeve pressure and needle valve back pressure; for example, for pressure related data, for data outside the range of the preset pressure [0.0001,100 ]; for data having a value of '0.9999', the pre-use value is filled in and the threshold range is allowed to be modified. When pressure data exceeds a threshold or fails to be obtained from the Pspace database for a certain period of time (e.g., 10 minutes, 30 minutes, etc.), an error is reported and no more data cleansing is performed until the data is detected to be within the normal data range.
And the temperature data is cleaned to be filled with the temperature data exceeding a preset temperature range. The preset temperature range is a temperature range in which preset normal temperature data are located; the temperature data comprises a pre-throttle temperature and a post-throttle temperature; for example, with respect to the temperature-related data, since the pre-throttle temperature and the post-throttle temperature are substantially consistent in trend and value. And compared with the temperature after the throttle valve, the temperature before the throttle valve has higher temperature loss proportion, so the temperature data only acquire and clean the temperature after the throttle valve. For data outside the range of preset temperature [ -20,80 ]; for data with a value of '-30'; for data having a value of '0.9999', the pre-use value is filled in and the threshold range is allowed to be modified. When the data exceeds a threshold or fails to be obtained from the Pspace database for a certain period of time (e.g., 10 minutes, 30 minutes, etc.), an error is reported and no more data cleansing is performed until the data is detected to be within the normal data range.
And the output data is cleaned to acquire and clean the accumulated gas yield, and the daily gas yield and the instantaneous gas yield are filled based on the cleaned accumulated gas yield. For yield related data: a zero clearing point appears in each day of daily gas production, and the time of the zero clearing point is not fixed, so that related cleaning logic is difficult to formulate; since the data used subsequently is minute-scale data, the instantaneous yield will increase the yield error when diluted, so the yield only acquires and cleans the cumulative gas yield, and then daily gas yield and instantaneous yield are calculated and filled by the cleaned cumulative gas yield to reduce the data error.
The method comprises the steps of acquiring and cleaning accumulated gas yield, filling daily gas yield and instantaneous gas yield based on the cleaned accumulated gas yield, and comprising the following steps:
and deleting the data when the accumulated gas yield exceeds the range or the value of the preset accumulated gas yield [0,100000] to be 0.9999'. The preset accumulated gas yield range is a preset normal accumulated gas yield range, and the threshold range is allowed to be modified and is determined according to actual conditions.
When the accumulated gas yield is lost, the current accumulated gas yield is obtained based on the last accumulated gas yield and the previous accumulated gas yield of the lost accumulated gas yield. For example, the difference in cumulative gas production between the one after the deficiency and the one before the deficiency is calculated and divided by N (the number of deficiency) to give the average yield per minute. The cumulative gas yield of the current row = the cumulative gas yield of the last row + the average yield per minute is then calculated from the first strip missing to the latest time strip by strip.
Filling data with the average gas production per minute of less than or equal to 0.000000001; for data with gas production per minute greater than 100, it was scaled down 100,0000 times. Wherein data of 0.000000001 or less of gas production per minute is filled with the production per minute of the previous line, which means the production per minute of the corresponding time of the previous day.
The instantaneous gas production and daily gas production are determined based on the cumulative gas production on the day and the cumulative gas production on the day before. Wherein, yield per minute = cumulative gas yield of current line-cumulative gas yield of last line; the instantaneous production and daily gas production were calculated and filled into the database using the following two formulas: instantaneous yield= (cumulative gas yield of current line-cumulative gas yield of last line) ×60×24; daily gas production = 8 a day: cumulative gas yield of 00-8 a day before morning: cumulative gas yield of 00.
When the data exceeds the threshold value or fails to acquire the data from the Pspace database for one hour continuously, the error is reported, and the data cleaning is not performed until the data is detected to be in the normal data range.
The continuous time processing is to fill in missing data to generate complete time series for the same point in time of different fields. For example, since different fields do not necessarily have data present at the same point in time, a complete time series is first generated to ensure that every well should have a piece of data for all fields per minute. The logic for missing value handling in the section where the usage data does not correspond to actual traffic for the missing data is filled. And in order to facilitate the subsequent data inquiry, the first data of eight-point whole in the morning of each day is taken as the data of the same day and stored in the day data table.
Step S3: and establishing an operation effect evaluation index system of the gas well extraction process based on the extraction process theory, and evaluating the effect of the operational extraction process well through the operation effect evaluation index system to obtain various process operation effects.
The gas well drainage and production process comprises a foam drainage process, a plunger process and a gas lift process; the process operation effect is related to the yield increase gas quantity and the casing pressure drop value. The specific effect evaluation system comprises: and (3) according to the time recorded by the historical data for implementing the process for the first time and related production data of the gas well, combining the traditional theoretical knowledge to formulate an evaluation system for the well choosing effect of the bubble row, the plunger and the gas lift process of the gas well. The yield (defined as reference yield) of the gas well can be obtained under the production period without adopting the process mainly by calculating the progressive rate, the actual yield and the reference yield are compared to obtain the yield increase amount, the change condition of the casing pressure of the gas well before and after the process is calculated, and the actual application effect of the history process well is comprehensively judged. The construction of the operation effect evaluation index system comprises the following steps:
a baseline production of the gas well without resorting to a drainage process is determined. Wherein, the calculation formula of the reference yield is:
wherein,is the standard yield of the month; />Average yield for the month of last month; />Is the rate of decrease in the last month. The calculation formula of the decreasing rate is as follows:
wherein,is the rate of decrease; />Is the average gas production rate in the last month; />The average gas production rate in the last month is obtained.
And obtaining the actual yield of the gas well after the drainage and production process is adopted.
And determining the yield increase gas quantity and the casing pressure change condition of the gas well based on the reference yield and the actual yield. The daily yield increase amount is calculated according to the following formula:
wherein,yield increase amount for the current day; />Is the actual yield of the day.
The calculation formula of the daily average yield increase amount is as follows:
wherein,to average daily yield increase amount within 30 days after taking drainage and production technological measuresThe unit is->；/>For the average daily gas production in 30 days after taking the drainage and mining process measures, the unit is +.>；/>For the average daily gas production in 30 days before taking the drainage and mining process measures, the unit is +.>。
The calculation formula of the casing pressure drop value of the gas well is as follows:
wherein,the unit of the reduction value of the 30 balance sleeve is MPa after the drainage and mining process measures are taken; />The unit is MPa for the jacket pressure of a 30 balance before taking the drainage and mining technological measures; />In order to adopt the measure of the drainage and production process, the balance is uniformly sleeved with pressure, and the unit is MPa.
As shown in fig. 2 and 3, the analysis and classification limits of the data distribution are divided into: a total of 118 process wells were selected for analysis. When the well selection evaluation system is established, the difference value of the average gas production values (average daily gas production increase change) of 30 days before and after regulation of the system is taken as the main judgment basis. As shown in fig. 2, the threshold value of the histogram underflow box is set to 0, because the process is implemented for the purpose of increasing daily gas yield, when the average daily gas yield change is less than or equal to 0, this adjustment has no positive effect, and this part is regarded as the first type; observing data distribution, adjusting the critical value of the overflow box to be 1.0, and taking the part as a fourth class; the body portion is then divided into a second class and a third class with a 0.5 dividing line. On the basis, the difference value of the average value of the jacket pressure before and after the process is used as the jacket pressure drop value and is used as a second index of the evaluation system. Wherein, aiming at the first class (yield increasing gas quantity is less than or equal to 0) under the yield increasing gas quantity dividing condition, the yield increasing gas quantity is divided into a difference class and a medium class only by taking whether the sleeve pressure is reduced as a standard; for the second class, as shown in FIG. 3, the threshold of the histogram underflow bin is set to 0; the threshold value of the overflow box is set to 0.5; the middle part is divided into a whole body, and the middle part is respectively divided into middle part, good part and good part. For the third class, taking 0.5 as a dividing line, dividing the third class into two classes of good and excellent with relatively equal quantity, and keeping the quantity of the good to be slightly larger than the quantity of the excellent; for the fourth class, the influence of the change of the sleeve pressure is not considered any more, and the sleeve pressure is classified as excellent. Specifically, determining an effect evaluation division based on the stimulation gas amount and the casing pressure change condition includes:
when the yield increase gas amount is less than or equal to 0:
the sleeve pressure drop value is less than or equal to zero, and the effect is poor; the effect is middle if the sleeve pressure drop value is in the (0, 0.5) interval, and good if the sleeve pressure drop value is more than 0.5;
when the yield increase gas amount is in the (0, 0.5) interval:
the sleeve pressure drop value is below 0, and the effect is the middle; the effect is good when the sleeve pressure drop value is in the (0, 0.5) interval, and the effect is excellent when the sleeve pressure drop value is more than 0.5;
when the yield increase gas amount is in the (0.5, 1) interval:
the sleeve pressure drop value is less than or equal to 0, and the effect is good; the sleeve pressure drop value is more than 0, and the effect is excellent;
when the yield increase gas amount is more than 1:
the evaluation was superior regardless of the change in the casing pressure.
Step S4: and screening the time period data with the operation effects of the various processes being greater than a preset effect threshold, and defining positive and negative sample data sets and determining a well selection model prediction target through data analysis.
The preset effect threshold value is a preset minimum value of a process operation effect, when the value of the process operation effect in a certain time period is larger than the preset effect threshold value after the gas well is put into operation and the production process is carried out, the data of the gas well in the time period is used as a sample data set, and otherwise, the data is not selected; negative samples in the positive and negative sample data set are production time period data with the fact that the gas production rate is reduced to be normal, the oil jacket pressure difference is stable or tends to be reduced, and the oil jacket pressure fluctuation is small; positive samples in the positive and negative sample data set are production time period data with large output fluctuation, gradually expanding oil jacket pressure difference and large oil jacket pressure fluctuation; the well selection model predicts that the goal is to improve production stability (e.g., oil casing differential pressure changes, production changes).
Step S5: and designing derivative parameters, enriching parameter dimensions, and obtaining a final sample data set. The final sample data set includes a positive and negative sample data set and a derivative sample data set; derived parameters refer to parameters that have relevance to the well model predictive target. The related derivative parameters can be designed by means of correlation analysis, business experience or decision tree model and the like, and the parameter dimensions are enriched, including:
analysis hypothesis
(1) The evaluation system used is a process measure effect evaluation system established before, mainly takes the increase of gas production and the decrease of casing pressure as main evaluation dimensions, and evaluates the gas production and casing pressure to four grades of excellent, good, medium and bad.
(2) Since each well can be regarded as an independent system and the static parameters can be regarded as initial values of the system, the influence of static data is ignored in the analysis process. Static parameters are introduced in the subsequent modeling process, so that the prediction accuracy of the model can be effectively improved
(3) The time of first implementing the process is a time node, and two or more equal-length continuous time periods are taken as a sample group and a control group.
(3) The time to first perform the process for each well is assumed to be the optimal time.
(4) The existing system does not consider the time dimension, and the well with the excellent rating and the good rating can have the condition of too late implementation process, and the well with the medium rating and the poor rating can be the condition of timely implementation process. Essentially this is an egg phase problem. One of the purposes of modeling is to find the best time point for well selection and process adjustment, and evaluation of the existing data with respect to the time point affects the final result of the model. The problem can be partially solved by processing data during modeling, and then the model is continuously optimized through iteration of the model, so that virtuous circle is achieved, and the problem is solved.
Derived parameter analysis
(1) Determination of yield threshold
The design of the derived parameters is not to find a specific rule, but only to verify whether the designed derived parameters have a certain correlation with the target, and a method for determining whether the designed derived parameters have a correlation is generally a business method and a data analysis method. The service method generally refers to determining that a certain parameter and a target have an association relationship through experience of service personnel or a rule commonly accepted in industry, for example, in a well selection service scene, the oil jacket pressure difference is also a derivative parameter in nature, no further deliberate demonstration is needed, and the method is accepted in industry, and specific parameters such as the oil jacket pressure difference change rate, the yield decline rate and the like. While some parameters need to be determined by data analysis methods. The analysis method mainly adopted in the research is mainly decision tree.
In the decision tree algorithm, the data of the first thirty days of the first use process is marked as 1, and the data of the first sixty days to thirty days of the first use process is marked as 0.
The node with the optimal daily yield is determined by the coefficient of the decision tree and the critical value is established to be not more than 0.079.
The daily yield critical value obtained by the decision tree analysis method can be obtained, the probability of daily yield below the critical value is basically stabilized to about 15% from 150 days to 30 days before the first implementation of the process, but the probability of daily gas yield below the critical value is obviously increased to 39% within 30 days before the first implementation of the process. It is believed that the likelihood of gas wells requiring use of the process increases substantially when daily gas production is below a threshold.
(2) Average of three days of oil pressure
In the decision tree algorithm of the previous section, the same operation is repeated after the daily output and the parameters with high autocorrelation are removed. The node at which the rolling average of the oil pressure over three days is optimal is determined by the coefficient of the decision tree, and a critical value of more than 6.237 is established.
After the threshold value determination is made on the rolling average of the oil pressure for three days, the change of the distribution with time is not obvious, and the parameter is discarded because the importance of the oil pressure is not high in consideration of the logic of the service.
(3) Rate of rise in casing pressure
It is first assumed that the rate of rise of the casing pressure has a correlation with the model predictive target (well selection time point). And because the well selection process is a classification problem with only two results (suitable for the process or unsuitable for the process), a decision tree algorithm in machine learning is used to explore whether a more obvious characteristic value exists in the casing pressure rising rate when approaching to a well selection time node or not, and the characteristic value can be used as a judgment basis. And next, analyzing the distribution of the data about the characteristic values to determine whether the sleeve pressure rising rate has correlation with a model prediction target.
Taking data in 30 days before the first process as positive samples, taking data in 60-30 days before the first process as negative samples, and determining the critical value of the sleeve pressure rising rate to be-0.001 after establishing decision tree model analysis. Since 0 has a very strong statistical meaning in statistics and is often used as a critical point, here, distribution analysis of sample data is performed using-0.001 and 0 as critical points, respectively.
In the study of the distribution, 150 days before the first use process to 60 days before the first use process were taken and equally divided into three time periods as a control group. When the selected critical value is 0, most of data are smaller than 0 value, but the distribution trend change is not obvious; however, when the critical value is-0.001, the time node of the first use process gradually approaches to the time node, and the frequency of times less than the critical value range which is not more than five appears within thirty days has a remarkable increasing trend.
As time gets closer to the time node of the first use process, the probability of the sleeve pressure rise rate occurring in the (-0.001,0) interval increases significantly, so the parameter has a significant relationship with the model prediction targets (well selection time, yield increase amount).
(4) Oil pressure first order guide
The node where the first order of oil pressure is optimal is determined by the coefficient of the foundation of the decision tree, and a critical value greater than 0 is established.
The relation between the sleeve pressure rising rate and the critical value in different time periods before the measure is adopted, and the frequency of the critical value above the critical value less than or equal to five appears within thirty days, and the change along with time is obviously increased.
Step S6: and dividing the final sample data set into a training set and a testing set according to a proportion, and respectively training the well selection model and evaluating the effect of the trained well selection model.
The ratio of the training set to the testing set is 7:3; the well selection model is used for receiving the processing data of the current gas well and outputting a well selection result; the well selection result comprises the optimal time of what kind of drainage process and the upper drainage process are adopted for the current gas well.
Step S7: substituting the data of the training set into various intelligent algorithm models according to various drainage processes, and performing parameter adjustment and algorithm optimization to obtain a plurality of well selection models.
The number of the well selection models is the same as the number of the drainage and production processes; the multiple types of intelligent algorithm models include LGBM, catboost and LSTM.
Step S8: according to the well selection service characteristics and the types of the drainage and production processes selected by the current gas well, well selection model operation rules are established, and the situation of re-checking the well is avoided. Current gas wells refer to gas wells that require a drainage process to be performed.
Step S9: and establishing a well selection result output strategy according to the current information of the gas well.
The current information comprises the conditions of natural gas well production, on-site platform production, existing equipment, existing technology and the like; the well selection result output strategy refers to a strategy meeting the implementation conditions of the gas well site construction, so that the final well selection model can be recommended to direct production.
Further comprising step S10: the developed data cleaning interface, intelligent model and output rule are packaged into a well selection model interface, and integrated with a production management platform, and deployed in a client server, comprising:
the data cleaning interface takes minutes as a unit to call production data from a database; the data cleaning interface is used for calling the production data according to the set standard.
Performing quality inspection on the acquired production data, and classifying the production data with the detected abnormality to obtain complete data; the classification processing mode comprises repairing data and deleting data; the database is a real-time production database; the production data comprise data related to primary pressure (such as oil pressure, casing pressure and the like), temperature, yield and the like; for example, the first is repairable, such as noise, missing of a certain minute, abnormal value of a certain minute, etc., and the type of abnormality is directly processed by means of weighted average, front value filling, etc.; the second is irreparable, such as a large number of misses, a large number of outliers, etc., this type of anomaly, in a way that the system records and throws out, the period of time not being as model input data.
Executing data multidimensional operation on the complete data, and simultaneously carrying out multidimensional operation on different types of data at the same time to one piece of data to finish data preparation and obtain input data; data support is provided for the subsequent model entry.
And respectively inputting the input data into the plurality of well selection models, respectively outputting the probability of the gas well selected by the well in a certain time period by the models, and determining the well selection result of the gas well in the drainage process based on the probability. For example, in a fixed time period, the multiple types of well selection models are used for acquiring data according to the type of the current drainage process of the gas well to carry out probability analysis on the current processes of the gas well, and after the analysis is completed, the result output is carried out on the optimal process of the well through the set model output rules.
The above is only a preferred embodiment of the present invention, and is not intended to limit the present invention, but various modifications and variations can be made to the present invention by those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (10)
1. A gas well drainage and production process well selection method based on an artificial intelligence algorithm is characterized by comprising the following steps:
s1: acquiring historical data related to gas well production;
s2: treating the historical data by using a data cleaning tool to obtain modeling data;
s3: establishing an operation effect evaluation index system of the gas well extraction process based on an extraction process theory, and evaluating the effect of the operational extraction process well through the operation effect evaluation index system to obtain various process operation effects;
s4: screening time period data with the operation effects of the various processes being greater than a preset effect threshold value, defining positive and negative sample data sets through data analysis, and determining a well selection model prediction target;
s5: designing derivative parameters, enriching parameter dimensions, and obtaining a final sample data set;
s6: dividing the final sample data set into a training set and a testing set according to a proportion, and respectively training a well selection model and evaluating the effect of the trained well selection model;
s7: substituting the data of the training set into various intelligent algorithm models according to various drainage processes, and performing parameter adjustment and algorithm optimization to obtain a plurality of well selection models;
s8: according to the well selection service characteristics and the types of the drainage and production processes selected by the current gas well, well selection model operation rules are established, and the situation of re-checking the well is avoided;
s9: and establishing a well selection result output strategy according to the current information of the gas well.
2. The well selection method of the gas well drainage process based on the artificial intelligence algorithm according to claim 1, further comprising packaging the developed data cleaning interface, the intelligent model and the output rule into a well selection model interface, integrating the well selection model interface with a production management platform, and deploying the well selection model interface in a client server.
3. The well selection method of the gas well drainage process based on the artificial intelligence algorithm according to claim 2, wherein the well selection model interface is packaged by the developed data cleaning interface, the intelligent model and the output rule, and is integrated with a production management platform, and is deployed in a client server, and the well selection method comprises the following steps:
the data cleaning interface takes minutes as a unit to call production data from a database;
performing quality inspection on the acquired production data, and classifying the production data with the detected abnormality to obtain complete data;
executing data multidimensional operation on the complete data, and simultaneously carrying out multidimensional operation on different types of data at the same time to one piece of data to finish data preparation and obtain input data;
and respectively inputting the input data into the plurality of well selection models, respectively outputting the probability of the gas well selected by the well in a certain time period by the models, and determining the well selection result of the gas well in the drainage process based on the probability.
4. The well selection method of the gas well drainage process based on the artificial intelligence algorithm according to claim 1, wherein the treatment of the historical data comprises data valid time determination, data cleaning and continuous time processing;
the data valid time is determined as a first piece of valid data of the gas well, a plurality of reference fields are determined according to the importance degree of the gas well parameters, and the first second or minute data of which the reference fields are not 0 at the same time is used as the first piece of valid data of the gas well; the other fields are filtered backwards after the previous piece of data;
the data cleaning comprises pressure data cleaning, temperature data cleaning and yield data cleaning;
the pressure data are cleaned to be filled with the pressure data exceeding a preset pressure range;
the temperature data are cleaned, and the temperature data exceeding a preset temperature range are filled;
the output data is cleaned to acquire and clean accumulated gas yield, and daily gas yield and instantaneous gas yield are filled based on the cleaned accumulated gas yield;
the continuous time processing is to fill in missing data to generate complete time series for the same point in time of different fields.
5. The artificial intelligence algorithm based well drainage process well selection method of claim 4, wherein the acquiring and cleaning of the accumulated gas production and filling of the daily gas production and the instantaneous gas production based on the cleaned accumulated gas production comprises:
when the accumulated gas yield exceeds the range or the value of the preset accumulated gas yield [0,100000] to be 0.9999', deleting the data;
when the accumulated gas yield is lost, obtaining the current lost accumulated gas yield based on the last accumulated gas yield and the previous accumulated gas yield of the lost accumulated gas yield;
filling data with the average gas production per minute of less than or equal to 0.000000001; for data with gas production rate of more than 100 per minute, the data is reduced by 100,0000 times;
the instantaneous gas production and daily gas production are determined based on the cumulative gas production on the day and the cumulative gas production on the day before.
6. The well selection method for a gas well drainage process based on an artificial intelligence algorithm according to claim 1, wherein the establishing a gas well drainage process operation effect evaluation index system comprises:
determining a baseline production of the gas well without resorting to a drainage process;
acquiring the actual yield of the gas well after adopting a drainage and production process;
determining the yield increase gas quantity and the casing pressure change condition of the gas well based on the reference yield and the actual yield;
and determining an effect evaluation division based on the yield increasing gas quantity and the casing pressure change condition.
7. The well selection method based on the artificial intelligence algorithm for the gas well drainage process according to claim 6, wherein the calculation formula of the reference yield is:
wherein,is the standard yield of the month; />Average yield for the month of last month; />Is the rate of decrease of the last month;
the daily yield increase amount is calculated according to the following formula:
wherein,yield increase amount for the current day; />Actual yield on the present day;
the calculation formula of the daily average yield increase amount is as follows:
wherein,for the average daily yield increase in 30 days after taking the drainage process measures, the unit is +.>；/>For the average daily gas production in 30 days after taking the drainage and mining process measures, the unit is +.>；/>For the average daily gas production in 30 days before taking the drainage and mining process measures, the unit is +.>；
The calculation formula of the casing pressure drop value of the gas well is as follows:
wherein,the unit of the reduction value of the 30 balance sleeve is MPa after the drainage and mining process measures are taken; />The unit is MPa for the jacket pressure of a 30 balance before taking the drainage and mining technological measures; />In order to adopt the measure of the drainage and production process, the balance is uniformly sleeved with pressure, and the unit is MPa.
8. The well selection method based on the artificial intelligence algorithm for the gas well drainage process of claim 7, wherein the calculation formula of the reduction rate is:
wherein,is the rate of decrease; />Is the average gas production rate in the last month; />The average gas production rate in the last month is obtained.
9. The artificial intelligence algorithm based well drainage process well selection method of claim 6, wherein determining an effect evaluation partition based on the stimulation gas volume and the casing pressure change condition comprises:
when the yield increase gas amount is less than or equal to 0:
the sleeve pressure drop value is less than or equal to zero, and the effect is poor; the effect is middle if the sleeve pressure drop value is in the (0, 0.5) interval, and good if the sleeve pressure drop value is more than 0.5;
when the yield increase gas amount is in the (0, 0.5) interval:
the sleeve pressure drop value is below 0, and the effect is the middle; the effect is good when the sleeve pressure drop value is in the (0, 0.5) interval, and the effect is excellent when the sleeve pressure drop value is more than 0.5;
when the yield increase gas amount is in the (0.5, 1) interval:
the sleeve pressure drop value is less than or equal to 0, and the effect is good; the sleeve pressure drop value is more than 0, and the effect is excellent;
when the yield increase gas amount is more than 1:
the evaluation was superior regardless of the change in the casing pressure.
10. The well selection method for the gas well drainage process based on the artificial intelligence algorithm according to claim 1, wherein the related derivative parameters and the abundant parameter dimensions are designed by means of correlation analysis, service experience or decision tree models.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311848890.7A CN117495206B (en) | 2023-12-29 | 2023-12-29 | Gas well drainage and production process well selection method based on artificial intelligence algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311848890.7A CN117495206B (en) | 2023-12-29 | 2023-12-29 | Gas well drainage and production process well selection method based on artificial intelligence algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117495206A true CN117495206A (en) | 2024-02-02 |
| CN117495206B CN117495206B (en) | 2024-03-22 |
Family
ID=89669375
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311848890.7A Active CN117495206B (en) | 2023-12-29 | 2023-12-29 | Gas well drainage and production process well selection method based on artificial intelligence algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117495206B (en) |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108843298A (en) * | 2018-06-26 | 2018-11-20 | 西南石油大学 | The quick well choosing method of coal bed gas well refracturing and device based on mining data |
| CN112360446A (en) * | 2020-11-20 | 2021-02-12 | 中国石油天然气股份有限公司 | Method for judging gas well type based on gas well production dynamic index system |
| US20220145746A1 (en) * | 2020-11-06 | 2022-05-12 | Halliburton Energy Services, Inc. | Dynamic Multi-Well Operation Integration in Wellbore Drilling |
| CA3123281A1 (en) * | 2021-06-25 | 2022-12-25 | Cenovus Energy Inc. | Apparatus and method for oil production forecasting |
| WO2023059701A1 (en) * | 2021-10-05 | 2023-04-13 | Schlumberger Technology Corporation | Systems and methods for predicting hydraulic fracturing design parmaters based on injection test data and machine learning |
| US20230186069A1 (en) * | 2021-12-09 | 2023-06-15 | Saudi Arabian Oil Company | Managing training wells for target wells in machine learning |
-
2023
- 2023-12-29 CN CN202311848890.7A patent/CN117495206B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108843298A (en) * | 2018-06-26 | 2018-11-20 | 西南石油大学 | The quick well choosing method of coal bed gas well refracturing and device based on mining data |
| US20220145746A1 (en) * | 2020-11-06 | 2022-05-12 | Halliburton Energy Services, Inc. | Dynamic Multi-Well Operation Integration in Wellbore Drilling |
| CN112360446A (en) * | 2020-11-20 | 2021-02-12 | 中国石油天然气股份有限公司 | Method for judging gas well type based on gas well production dynamic index system |
| CA3123281A1 (en) * | 2021-06-25 | 2022-12-25 | Cenovus Energy Inc. | Apparatus and method for oil production forecasting |
| WO2023059701A1 (en) * | 2021-10-05 | 2023-04-13 | Schlumberger Technology Corporation | Systems and methods for predicting hydraulic fracturing design parmaters based on injection test data and machine learning |
| US20230186069A1 (en) * | 2021-12-09 | 2023-06-15 | Saudi Arabian Oil Company | Managing training wells for target wells in machine learning |
Non-Patent Citations (8)
| Title |
|---|
| QIAN SHIHAO;DONG ZHENZHEN;SHI QIANQIAN;GUO WEI;ZHANG XIAOWEI;LIU ZHAOXIA;WANG LINGJUN;WU LEI;ZHANG TIANYANG;LI WEIRONG；: "Optimization of shale gas fracturing parameters based on artificial intelligence algorithm", 《ARTIFICIAL INTELLIGENCE IN GEOSCIENCES》, 22 August 2023 (2023-08-22), pages 95 - 110 * |
| 刘佳;柳迎红;冯汝勇;廖夏;李娜;: "煤层气井降压增产措施及合理选井", 煤炭科学技术, no. 1, 15 June 2018 (2018-06-15), pages 231 - 234 * |
| 孔鹏;李忠城;杨晨曦;张云鹏;汪若璇;: "基于PCA-NB方法的煤层气井压后产量预测研究", 特种油气藏, no. 06, 18 October 2019 (2019-10-18), pages 108 - 112 * |
| 张天翔;孙斌;朱卫平;刘川庆;甄怀宾;王建新;: "探究煤层气选井选层的界限及标准", 《化学工程与装备》, no. 02, 15 February 2017 (2017-02-15), pages 97 - 99 * |
| 杨兆中等: "基于灰色关联的逼近理想解排序法的煤层气井 重复压裂选井———以沁水盆地柿庄南区块为例", 《科学技术与工程》, vol. 20, no. 12, 28 April 2020 (2020-04-28), pages 4681 - 4685 * |
| 闫林林;李勇斌;刘冀;荆志强;李亚蓉;朱彤;: "苏里格气田排水采气工艺经济适用性研究", 内蒙古石油化工, no. 12, 30 December 2019 (2019-12-30), pages 1 - 7 * |
| 陈新浩;: "机器学习理论在压裂分析中的应用研究", 内蒙古石油化工, no. 02, 30 January 2010 (2010-01-30), pages 17 - 20 * |
| 霍雅迪;江厚顺;: "一种基于BP神经网络的气井重复压裂井优选方法", 天然气地球科学, no. 04, 10 April 2020 (2020-04-10), pages 552 - 558 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117495206B (en) | 2024-03-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108417033B (en) | Expressway traffic accident analysis and prediction method based on multi-dimensional factors | |
| CN112987666B (en) | Power plant unit operation optimization regulation and control method and system | |
| CN109255134B (en) | Method for acquiring fault condition of pumping well | |
| CN106886915B (en) | Advertisement click estimation method based on time attenuation sampling | |
| KR101904278B1 (en) | Method for decline curve analysis according to cumulative production incline rate in unconventional gas field | |
| CN105654229A (en) | Power grid automation system and equipment running state risk assessment algorithm | |
| CN108011367A (en) | A kind of Characteristics of Electric Load method for digging based on depth decision Tree algorithms | |
| CN111210170A (en) | Environment-friendly management and control monitoring and evaluation method based on 90% electricity distribution characteristic index | |
| CN109446184A (en) | Power generation big data preprocess method and system based on big data analysis platform | |
| CN106644983A (en) | Spectrum wavelength selection method based on PLS-VIP-ACO algorithm | |
| CN106528850A (en) | Door access data abnormal detection method based on machine learning clustering algorithm | |
| CN110223156A (en) | Automation model evolutionary algorithm based on gradually optimal feature selection | |
| CN117495206B (en) | Gas well drainage and production process well selection method based on artificial intelligence algorithm | |
| CN115618987A (en) | Production well production data prediction method, device, equipment and storage medium | |
| Tsui | The family formation process among US marriage cohorts | |
| CN105183612B (en) | The appraisal procedure of server free memory abnormal growth and operation conditions | |
| CN113298148B (en) | Ecological environment evaluation-oriented unbalanced data resampling method | |
| CN108493933A (en) | A kind of Characteristics of Electric Load method for digging based on depth decision Tree algorithms | |
| CN110196797A (en) | Automatic optimization method and system suitable for credit scoring card system | |
| CN115630571A (en) | Oil well indicator diagram automatic diagnosis method based on ensemble learning | |
| CN112801388B (en) | Power load prediction method and system based on nonlinear time series algorithm | |
| CN112360446B (en) | Method for judging gas well type based on gas well production dynamic index system | |
| CN112508267A (en) | Distribution transformer area load prediction method and system based on artificial intelligence | |
| CN111563647A (en) | Power information system detection method and system based on association rule and DEA | |
| CN113379093B (en) | Energy consumption analysis and optimization method for oil-gas gathering and transportation system |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |