CN111783356A - Petroleum yield prediction method and device based on artificial intelligence - Google Patents

Petroleum yield prediction method and device based on artificial intelligence Download PDF

Info

Publication number
CN111783356A
CN111783356A CN202010604568.XA CN202010604568A CN111783356A CN 111783356 A CN111783356 A CN 111783356A CN 202010604568 A CN202010604568 A CN 202010604568A CN 111783356 A CN111783356 A CN 111783356A
Authority
CN
China
Prior art keywords
yield
time
petroleum
prediction
related information
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
Application number
CN202010604568.XA
Other languages
Chinese (zh)
Other versions
CN111783356B (en
Inventor
青鹏
张海峰
李轶
许卓群
杨鸣
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen International Graduate School of Tsinghua University
Original Assignee
Shenzhen International Graduate School of Tsinghua University
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen International Graduate School of Tsinghua University filed Critical Shenzhen International Graduate School of Tsinghua University
Priority to CN202010604568.XA priority Critical patent/CN111783356B/en
Publication of CN111783356A publication Critical patent/CN111783356A/en
Application granted granted Critical
Publication of CN111783356B publication Critical patent/CN111783356B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • 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
    • G06N3/045—Combinations of networks
    • 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
    • G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Biophysics (AREA)
  • Mathematical Physics (AREA)
  • Biomedical Technology (AREA)
  • Health & Medical Sciences (AREA)
  • Geology (AREA)
  • Mining & Mineral Resources (AREA)
  • Medical Informatics (AREA)
  • Computer Hardware Design (AREA)
  • Geometry (AREA)
  • Geophysics (AREA)
  • Environmental & Geological Engineering (AREA)
  • Fluid Mechanics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • Geochemistry & Mineralogy (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses an artificial intelligence-based petroleum yield prediction method. The method comprises the following steps: inputting the related information of the petroleum yield of a plurality of historical time points into a trained time series prediction model, and sequentially outputting the related information of the petroleum yield of the next time point so as to obtain a time series prediction value; judging abnormal events and corresponding time points for the time series predicted values, and calculating predicted values of subsequent petroleum yield related information of the abnormal event time points through a moving average model by taking reduction of loss caused by abnormal fluctuation as a target; and replacing the predicted value of the corresponding time point output by the time series model with the calculated predicted value of the related information of the petroleum yield to obtain a petroleum yield prediction result. The invention predicts the whole petroleum yield trend by training the time sequence prediction model, and predicts the yield value in a future period of time by using the moving average when abnormal points are met, thereby improving the prediction accuracy.

Description

Petroleum yield prediction method and device based on artificial intelligence
Technical Field
The invention relates to the technical field of petroleum yield prediction, in particular to a petroleum yield prediction method and device based on artificial intelligence.
Background
Petroleum is an important raw material in daily industrial production activities, the quantity of petroleum yield has important significance for guiding enterprise production and enterprise planning operation activities, and meanwhile, the petroleum yield can be accurately predicted to enable petroleum enterprises to make reasonable production tasks and avoid blind decisions. The yield change during oil exploitation is a dynamic process, and the oil production department needs to predict the yield of an oil well in the future for a period of time so as to adjust the production scheme of the oil well in a targeted manner, thereby realizing the stable production of the oil well. The main prediction methods for the single-factor time series variable problem of yield at present are regression method, grey system prediction, fuzzy system prediction and the like, for time series prediction, if linear relation is satisfied, traditional modeling and prediction methods such as autoregressive (CAR), sliding Mean (MA) and mixed autoregressive sliding mean (ARIMA) can be used, namely, if the linear relation is not satisfied, future values of a group of time series are assumed to be linearly related to historical values, and if the linear relation is not satisfied, the BP neural network can be used for prediction. These predictions are relatively inaccurate. In recent years, there are other prediction methods that can reflect the internal structure of data to some extent and play a certain role in solving the problem of nonlinear and abnormal time series, but there is a difficulty in realizing codes. In addition, the existing single-well production prediction does not consider the influence of abnormal events on the production prediction, and particularly the prediction difference is large in a period of time after the abnormal events occur in the oil well. When predicting production for a single well, it is relatively good to predict production if the time series of production is relatively smooth, but when an abnormal event is encountered, a situation occurs in which the predicted value and the actual value are significantly different.
Currently, the prediction of the whole oil field is generally carried out in the actual oil field production prediction stage, or the prediction of the single well production is carried out, for example, the single well production is predicted by using algorithms based on SSA and RBF, BP neural network, ARIMA, A RIMA-Kalman filter and the like. Due to the fact that the single-well crude oil production time sequence is aperiodic and nonlinear, a black box model is required to be used for modeling and predicting the single-well crude oil production time sequence, and the single-well crude oil production time sequence is subjected to multi-step prediction by combining singular spectrum analysis and a radial basis function neural network, the singular spectrum analysis method can effectively reduce error transfer in the multi-step prediction process, and therefore reliability of the multi-step prediction is improved. The BP neural network does not predict the yield in detail well, but only the overall trend. The ARIMA (autoregressive moving average) model is quite simple, requiring only endogenous variables and not other exogenous variables (so-called endogenous variables should mean that it depends only on the data itself, unlike regression which requires other variables).
In the prior art, the following problems exist in predicting production over a future period of time using time series:
1) various current models have the problem of low prediction accuracy, for example, a BP neural network can predict the trend of the whole yield, but for more areas, one average value is worth predicting; the ARIMA adopts an ARIMA model to predict time sequence data, which is required to be stable, if the data is unstable, the data cannot capture the regularity, and essentially only can capture the linear relation but not the nonlinear relation, and the accuracy is low when the data fluctuation is large.
2) The prediction under the ideal condition is considered in all the current models, the prediction after an abnormal event is not considered, particularly, after the abnormal condition, the trend of the actual value is inconsistent with that of the original predicted value, so that the predicted value is greatly biased to be greatly deviated from the actual value.
3) The calculation for the later period of the abnormal event is too fuzzy, and the abnormal event is not predicted according to the actual abnormal condition, but the whole is input into the model without considering the influence of the abnormal event on the whole prediction.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide an artificial intelligence-based oil yield prediction method and device, which improve the accuracy of the information related to the predicted oil yield by considering the influence of abnormal events on the prediction.
The invention provides an artificial intelligence-based petroleum yield prediction method. The method comprises the following steps:
inputting the related information of the petroleum yield of a plurality of historical time points into a trained time series prediction model, and sequentially outputting the related information of the petroleum yield of the next time point so as to obtain a time series prediction value;
judging abnormal events and corresponding time points for the time series predicted values, and calculating predicted values of subsequent petroleum yield related information of the abnormal event time points through a moving average model by taking reduction of loss caused by abnormal fluctuation as a target;
and replacing the predicted value of the corresponding time point output by the time series model with the calculated predicted value of the related information of the petroleum yield to obtain a petroleum yield prediction result.
In one embodiment, determining the abnormal event and the corresponding point in time comprises:
acquiring venturi data measured by a multiphase flowmeter;
acquiring the proportion of the change of the petroleum yield related data at the abnormal time point by using the Venturi data;
and if the current t moment yield and the t-1 moment yield change amplitude exceed the set threshold value and the Venturi data are changed, judging that an abnormal event occurs and determining a corresponding time point.
In one embodiment, the predicted value of oil production related information subsequent to the point in time of the anomaly event is calculated according to the following steps:
let K be XL+1/XL，XL+1For oil production-related information after the point in time of the abnormal event, XLThe related information of the petroleum yield before the abnormal event time point;
computing X using a moving average modelL+1The predicted value of the related information of the petroleum yield is expressed as
Figure BDA0002560503050000031
In one embodiment, the time series prediction model is a long-time memory network, and the training process includes: calculating the output value of the cell of the long-time and short-time memory network according to a forward calculation method; reversely calculating the error term of each long-time memory network cell; calculating a gradient for each weight based on the corresponding error term; a gradient-based optimization algorithm is applied to update the weights.
In one embodiment, the time series prediction model is a long-term memory network, and the prediction process includes:
inputting gas phase yield, liquid phase yield, water content and single well oil yield data measured by the multiphase flowmeter into a first LSTM neural network unit, outputting the data as a current cell state C and a hidden state H, and obtaining predicted values
Figure BDA0002560503050000032
Will predict the value
Figure BDA0002560503050000033
And inputting corresponding values of previous time points into LSTM cells and combining the previous cell state C and hidden state H to obtain predicted values
Figure BDA0002560503050000034
And further obtaining the time series prediction value.
In one embodiment, in training the time series prediction model, the sample data comprises: fluid operating pressure, pre-venturi differential pressure, post-venturi differential pressure, temperature, water cut, mass instantaneous flow, volume instantaneous flow, mass instantaneous flow, and cumulative gas phase production, liquid phase production, water cut, and single well production per minute.
In one embodiment, the petroleum-production-related information includes one or more of gas-phase production, liquid-phase production, water cut, and petroleum production.
Compared with the prior art, the method has the advantages that according to the output data measured by the multiphase flowmeter, the characteristics of the venturi original differential pressure measurement data are used for judging abnormal events, the time series output training time series prediction model is combined for predicting the overall petroleum output trend, and when abnormal points are met, the output value in a future period of time is predicted by using the sliding average, so that a higher accurate value is obtained.
Other features of the present invention and advantages thereof will become apparent from the following detailed description of exemplary embodiments thereof, which proceeds with reference to the accompanying drawings.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the invention and together with the description, serve to explain the principles of the invention.
FIG. 1 is a flow diagram of a method for artificial intelligence based oil production prediction, according to one embodiment of the present invention;
FIG. 2 is a schematic diagram of a training and prediction process for a long term memory network, according to one embodiment of the present invention;
FIG. 3 is an overall process schematic of an artificial intelligence based oil production prediction method according to one embodiment of the present invention;
FIG. 4 is a prior art oil production result predicted using a separate long and short term memory network;
FIG. 5 is a graphical illustration of the results of a prediction with long and short term memory networks and a running average, according to an embodiment of the invention.
Detailed Description
Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that: the relative arrangement of the components and steps, the numerical expressions and numerical values set forth in these embodiments do not limit the scope of the present invention unless specifically stated otherwise.
The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the invention, its application, or uses.
Techniques, methods, and apparatus known to those of ordinary skill in the relevant art may not be discussed in detail but are intended to be part of the specification where appropriate.
In all examples shown and discussed herein, any particular value should be construed as merely illustrative, and not limiting. Thus, other examples of the exemplary embodiments may have different values.
It should be noted that: like reference numbers and letters refer to like items in the following figures, and thus, once an item is defined in one figure, further discussion thereof is not required in subsequent figures.
The invention provides an artificial intelligence-based oil production prediction method, which uses a trained time series model to predict oil production related information, including but not limited to gas phase production, liquid phase production, water content, single well production and the like. In the following description, the prediction process will be described by taking a long-term memory network (LSTM) as an example.
The petroleum yield prediction method based on artificial intelligence provided by the invention needs to train a time sequence prediction model at first, as shown in figure 1, the model prediction process comprises the following steps:
step S110, constructing a model sample.
Specifically, the data sources of the model building sample process are fluid running pressure, venturi front differential pressure, venturi rear differential pressure, temperature, water content, mass instantaneous flow (oil), volume instantaneous flow (gas), mass instantaneous flow (water), and accumulated gas phase yield, liquid phase yield, water content and single well yield per minute obtained by the measurement of the three-phase flow meter device.
Further, the result data may be processed into average data per hour or average data per day.
Step S120, smoothing the data.
When a model sample is constructed, certain smoothing processing needs to be performed after all data are obtained so as to eliminate unnecessary abnormal points, thereby avoiding prediction errors caused by acquisition or calculation of the data.
And S130, training the time series model, obtaining a predicted value of the related information of the petroleum yield, and replacing the predicted value after the abnormal event by using the moving average.
The whole model construction adopts the current time point and N-1 single well yield values before the current time point to predict the next time. And then the predicted gas phase yield, liquid phase yield, water content and single well yield and the gas phase yield, liquid phase yield, water content and single well yield in the previous period are used as common input to predict the gas phase yield, liquid phase yield, water content and single well yield at the next time point.
For example, taking a long-short term memory network as an example, the input is the gas phase yield, the liquid phase yield, the water content and the oil yield of N time points, and the output is the gas phase yield, the liquid phase yield, the water content and the oil yield of the next time point. MA (moving average) is then used to replace the predicted values after the abnormal event.
Referring to fig. 2, the LSTM model includes an input gate, a hidden layer and an output gate, wherein the hidden layer includes a plurality of LSTM cell interiors (labeled as LSTM1, LSTM2, etc.), including an input gate i, a forgetting gate f, an output gate o, C, x is an input vector, and h is an output vector of the hidden layer. The calculation formulas of the input gate, the output gate and the forgetting gate in the whole cell are expressed as follows:
Figure BDA0002560503050000062
Figure BDA0002560503050000063
Figure BDA0002560503050000064
ct＝ft*ct-1+it*tanh(Wxcxt+Whcht-1+bc) (4)
ht＝ot*tanh(ct) (5)
Figure BDA0002560503050000061
wherein, for time t: i.e. itIs an input gate output, Wxi,Ehi,biCorresponding to the input gate weight, the hidden state weight and the offset respectively; f. oftIs forgotten gate output, Wxf,Whf,bfRespectively corresponding to a forgetting gate weight, a hidden state weight and an offset; otIs the output of an output gate, Wxo,Who,boRespectively corresponding to the weight of an output gate, the weight of a hidden state and the bias; c. CtIs the cell state of LSTM, htIs LSTM hidden state, WcbCIs the bias of the weights of the cell states.
The LSTM model training process can adopt a BPTT algorithm similar to the principle of a classical back propagation algorithm, and the whole process can be divided into four steps, namely calculating the output value of an LSTM cell according to a forward calculation method; calculating an error term of each LSTM cell reversely, wherein the error term comprises two reverse propagation directions according to time and a network level; calculating a gradient for each weight based on the corresponding error term; a gradient-based optimization algorithm is applied to update the weights. There are various types of gradient-based optimization algorithms, such as random gradient descent (SGD), AdaGrad, RMSProp, and others. The present invention preferably employs an adaptive momentum estimation algorithm (Adam). The Adam algorithm is an effective random optimization method based on gradient, combines the advantages of AdaGrad and RMSProp algorithms, can calculate adaptive learning rate for different parameters, and occupies less storage resources. Compared with other random optimization methods, the Adam method has better overall performance in practical application.
It should be noted that the prediction of the oil production-related information may be performed by using another time-series prediction model, for example, GRU.
To further understand the present invention, referring to fig. 3, the following describes the general process of the artificial intelligence based oil production prediction method of the present invention, including the following steps:
step S210, processing the venturi data gas phase yield, liquid phase yield, water content and single well oil yield data measured by the multiphase flowmeter, then performing standard processing on all the result data, and changing the data into a format required by the model, for example, the input format is:
(XN-6，XN-5，XN-4，XN-3，XN-2，XN-1，XN) The output is XN+1。
Step S220, inputting the formatted data into the LSTM model, outputting the data as the current cell state C and the hidden state H after passing through the first LSTM neural network unit, and obtaining the predicted cell state C and the hidden state H
Figure BDA0002560503050000071
Step S230, then
Figure BDA0002560503050000072
Input into the cell and combine with the last C and H, output as
Figure BDA0002560503050000073
And so on until output
Figure BDA0002560503050000074
Wherein
Figure BDA0002560503050000075
To
Figure BDA0002560503050000076
Is a predicted value.
Step S240, comparing the predicted value and the original value, and training the whole LSTM model according to the loss function.
And S250, predicting relevant information of petroleum yield by using the trained LSTM model, and obtaining time series predicted values of gas phase yield, liquid phase yield, water content, single well oil yield and the like at different moments.
Step S260, determining an abnormal event and an abnormal time point for obtaining the time series prediction value.
For example, according to abnormal changes of differential pressure, temperature and pressure signals obtained by multiphase flow measurement, when the deviation of an upstream measurement signal is large and the instantaneous production data exceeds 20% of the average range of the average production data, the abnormal production fluctuation point is determined, and the corresponding abnormal event occurrence time point is found. Namely, the abnormal event is judged to be that the change amplitude of the production exceeds 20% according to the current time t and the time t-1, and the data in the original Venturi data is changed to be considered as abnormal.
For simple calculation, the abnormal event is judged according to the current data, the type of the abnormal event and the change ratio K of the abnormal event relative to the yield under the normal condition are judged, and the K value represents the ratio of the yield at the t moment and the yield at the t-1 moment after the abnormal event occurs.
In step S270, predicted values of gas phase yield, liquid phase yield, water content and yield after the abnormal point are calculated using a moving average Method (MA), and the calculated predicted values are substituted for those of LSTM.
For example, if at XLIf an abnormal event occurs at the corresponding moment, the original LSTM predicted value is still used, and the method is inaccurate. The predicted value of LSTM after the abnormal time point is therefore replaced with the MA predicted value.
Specifically, let K ═ XL+1/XL(alternatively, X can be learned from eight columns of data measured in a multiphase flowmeterL+1Gas phase production, liquid phase production, water cut and single well production after abnormal time points, XLGas phase production, liquid phase production, water cut and single well production before the abnormal time point), and then calculating X by using MAL+1Subsequent prediction, e.g. as expressed by
Figure BDA0002560503050000081
X can be obtained according to the calculation formulaL+3、XL+4、XL+5…, X will be calculated using MAL+2And the subsequent values replace the original yield value predicted by the LSTM, and finally the required predicted values of gas phase yield, liquid phase yield, water content and single well yield are obtained.
The whole model prediction process is completed, wherein the gas phase yield, the liquid phase yield, the water content and the yield prediction is that an abnormal time point is judged based on venturi differential pressure, temperature and pressure data measured by a multiphase flowmeter, the gas phase yield, the liquid phase yield, the water content and the yield result data of each minute are combined, the whole data are input into a model for training after data processing, a basic LSTM prediction model is obtained, the gas phase yield, the liquid phase yield, the water content and the yield behind the abnormal point are calculated by using an MA algorithm and multiplied by a proportionality coefficient K, and the LSTM prediction gas phase yield, the liquid phase yield, the water content and the single well yield of the corresponding time point are replaced.
To further verify the technical effect of the present invention, comparative experiments were conducted, taking yield (yields) predictions as an example, wherein fig. 4 is the result of the prediction using a single LSTM in the prior art, and fig. 5 is the result of the prediction using the LSTM + MA combination of the present invention. It can be seen that the fitting effect of the prediction result (predict) and the True value (True) is better than that of the prior art.
In conclusion, the method can be used for long-term prediction of gas phase yield, liquid phase yield, water content and oil yield by combining Venturi data measured by a multiphase flowmeter with time series models such as LSTM and the like, has high prediction accuracy, and only needs corresponding yield result data in the prediction process without combining other data. In addition, the method utilizes the original Venturi data and the gas phase yield, the liquid phase yield, the water content and the yield result data to obtain the proportion of the gas phase yield, the liquid phase yield, the water content and the yield change at each abnormal time point, further uses a moving average model to predict the gas phase yield, the liquid phase yield, the water content and the single-well yield after the abnormal event, multiplies the proportion K after the abnormal event by the gas phase yield, the liquid phase yield, the water content and the single-well yield predicted by the corresponding time point LSTM, and accordingly enables the prediction result to be more accurate.
The present invention may be a system, method and/or computer program product. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied therewith for causing a processor to implement various aspects of the present invention.
The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic memory device, a magnetic memory device, an optical memory device, an electromagnetic memory device, a semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.
The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device via a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.
The computer program instructions for carrying out operations of the present invention may be assembler instructions, Instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, aspects of the present invention are implemented by personalizing an electronic circuit, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA), with state information of computer-readable program instructions, which can execute the computer-readable program instructions.
Aspects of the present invention are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.
These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.
The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions. It is well known to those skilled in the art that implementation by hardware, by software, and by a combination of software and hardware are equivalent.
Having described embodiments of the present invention, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen in order to best explain the principles of the embodiments, the practical application, or improvements made to the technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein. The scope of the invention is defined by the appended claims.

Claims (9)

1. An artificial intelligence-based petroleum yield prediction method comprises the following steps:
inputting the related information of the petroleum yield of a plurality of historical time points into a trained time series prediction model, and sequentially outputting the related information of the petroleum yield of the next time point so as to obtain a time series prediction value;
judging abnormal events and corresponding time points for the time series predicted values, and calculating predicted values of subsequent petroleum yield related information of the abnormal event time points through a moving average model by taking reduction of loss caused by abnormal fluctuation as a target;
and replacing the predicted value of the corresponding time point output by the time series model with the calculated predicted value of the related information of the petroleum yield to obtain a petroleum yield prediction result.
2. The method of claim 1, wherein determining an exception event and a corresponding point in time comprises:
acquiring venturi data measured by a multiphase flowmeter;
acquiring the proportion of the change of the petroleum yield related data at the abnormal time point by using the Venturi data;
and if the current t moment yield and the t-1 moment yield change amplitude exceed the set threshold value and the Venturi data are changed, judging that an abnormal event occurs and determining a corresponding time point.
3. The method according to claim 1, wherein the predicted value of oil production related information subsequent to the exceptional time point is calculated according to the following steps:
let K be XL+1/XL，XL+1For oil production-related information after the point in time of the abnormal event, XLThe related information of the petroleum yield before the abnormal event time point;
computing X using a moving average modelL+1The predicted value of the related information of the petroleum yield is expressed as
Figure FDA0002560503040000011
4. The method of claim 1, wherein the time series prediction model is an long and short term memory network, and the training process comprises: calculating the output value of the cell of the long-time and short-time memory network according to a forward calculation method; reversely calculating the error term of each long-time memory network cell; calculating a gradient for each weight based on the corresponding error term; a gradient-based optimization algorithm is applied to update the weights.
5. The method of claim 1, wherein the time series prediction model is a long-and-short memory network, and the prediction process comprises:
inputting gas phase yield, liquid phase yield, water content and single well oil yield data measured by the multiphase flowmeter into a first LSTM neural network unit, outputting the data as a current cell state C and a hidden state H, and obtaining predicted values
Figure FDA0002560503040000021
Will predict the value
Figure FDA0002560503040000022
And inputting corresponding values of previous time points into LSTM cells and combining the previous cell state C and hidden state H to obtain predicted values
Figure FDA0002560503040000023
And further obtaining the time series prediction value.
6. The method of claim 1, wherein in training the time series prediction model, sample data comprises: fluid operating pressure, pre-venturi differential pressure, post-venturi differential pressure, temperature, water cut, mass instantaneous flow, volume instantaneous flow, mass instantaneous flow, and cumulative gas phase production, liquid phase production, water cut, and single well production per minute.
7. The method of claim 1, wherein the petroleum production related information includes one or more of gas phase production, liquid phase production, water cut, and petroleum production.
8. A computer-readable storage medium, on which a computer program is stored, which program, when being executed by a processor, carries out the steps of the method according to any one of claims 1 to 7.
9. A computer device comprising a memory and a processor, on which memory a computer program is stored which is executable on the processor, characterized in that the steps of the method of any of claims 1 to 7 are implemented when the processor executes the program.
CN202010604568.XA 2020-06-29 2020-06-29 Oil yield prediction method and device based on artificial intelligence Active CN111783356B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010604568.XA CN111783356B (en) 2020-06-29 2020-06-29 Oil yield prediction method and device based on artificial intelligence

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010604568.XA CN111783356B (en) 2020-06-29 2020-06-29 Oil yield prediction method and device based on artificial intelligence

Publications (2)

Publication Number Publication Date
CN111783356A true CN111783356A (en) 2020-10-16
CN111783356B CN111783356B (en) 2024-03-29

Family

ID=72760081

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010604568.XA Active CN111783356B (en) 2020-06-29 2020-06-29 Oil yield prediction method and device based on artificial intelligence

Country Status (1)

Country Link
CN (1) CN111783356B (en)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112360399A (en) * 2020-11-30 2021-02-12 中国石油大学(北京) Method, device and equipment for predicting yield of coal bed gas
CN112560238A (en) * 2020-12-03 2021-03-26 中国石油天然气股份有限公司 Oil yield prediction method and device based on Starkeberg game model
CN113537592A (en) * 2021-07-15 2021-10-22 中国石油大学(北京) Oil and gas reservoir yield prediction method and device based on long-time and short-time memory network
CN115204533A (en) * 2022-09-16 2022-10-18 中国地质大学（北京） Oil-gas yield prediction method and system based on multivariable weighted combination model

Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104156473A (en) * 2014-08-25 2014-11-19 哈尔滨工业大学 LS-SVM-based method for detecting anomaly slot of sensor detection data
CN105205570A (en) * 2015-10-16 2015-12-30 国网重庆铜梁区供电有限责任公司 Power grid power sale quantity prediction method based on season time sequence analysis
CN106651001A (en) * 2016-11-08 2017-05-10 浙江理工大学 Needle mushroom yield prediction method based on improved neural network and implementation system
CN107346464A (en) * 2016-05-06 2017-11-14 腾讯科技（深圳）有限公司 Operational indicator Forecasting Methodology and device
CN107478280A (en) * 2017-08-17 2017-12-15 合肥工业大学 A kind of water-coal-slurry electromagnetic flowmeter signal processing method based on the analysis of excitation frequency higher hamonic wave
CN110119845A (en) * 2019-05-11 2019-08-13 北京京投亿雅捷交通科技有限公司 A kind of application method of track traffic for passenger flow prediction
CN110264352A (en) * 2019-05-16 2019-09-20 福建江夏学院 Stock index prediction method and device based on neural network model and time series
CN110284872A (en) * 2019-06-10 2019-09-27 中国石油大学(北京) The virtual flow rate calculation method and system of the underwater acquisition system of offshore gas field group
CN110516890A (en) * 2019-09-04 2019-11-29 重庆邮电大学 A kind of crop yield monitoring system based on Grey Combinatorial Model Method
CN110621846A (en) * 2017-05-04 2019-12-27 解决方案探寻有限公司 Recording data from a streaming network
CN111163092A (en) * 2019-12-30 2020-05-15 深信服科技股份有限公司 Flow abnormity detection method, device, equipment and storage medium
CN111279050A (en) * 2017-09-11 2020-06-12 吉奥奎斯特系统公司 Well planning system

Patent Citations (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104156473A (en) * 2014-08-25 2014-11-19 哈尔滨工业大学 LS-SVM-based method for detecting anomaly slot of sensor detection data
CN105205570A (en) * 2015-10-16 2015-12-30 国网重庆铜梁区供电有限责任公司 Power grid power sale quantity prediction method based on season time sequence analysis
CN107346464A (en) * 2016-05-06 2017-11-14 腾讯科技（深圳）有限公司 Operational indicator Forecasting Methodology and device
CN106651001A (en) * 2016-11-08 2017-05-10 浙江理工大学 Needle mushroom yield prediction method based on improved neural network and implementation system
CN110621846A (en) * 2017-05-04 2019-12-27 解决方案探寻有限公司 Recording data from a streaming network
CN107478280A (en) * 2017-08-17 2017-12-15 合肥工业大学 A kind of water-coal-slurry electromagnetic flowmeter signal processing method based on the analysis of excitation frequency higher hamonic wave
CN111279050A (en) * 2017-09-11 2020-06-12 吉奥奎斯特系统公司 Well planning system
CN110119845A (en) * 2019-05-11 2019-08-13 北京京投亿雅捷交通科技有限公司 A kind of application method of track traffic for passenger flow prediction
CN110264352A (en) * 2019-05-16 2019-09-20 福建江夏学院 Stock index prediction method and device based on neural network model and time series
CN110284872A (en) * 2019-06-10 2019-09-27 中国石油大学(北京) The virtual flow rate calculation method and system of the underwater acquisition system of offshore gas field group
CN110516890A (en) * 2019-09-04 2019-11-29 重庆邮电大学 A kind of crop yield monitoring system based on Grey Combinatorial Model Method
CN111163092A (en) * 2019-12-30 2020-05-15 深信服科技股份有限公司 Flow abnormity detection method, device, equipment and storage medium

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
刘巍 等: ""基于机器学习方法的油井日产油量预测"", 《石油钻采工艺》, vol. 42, no. 02, pages 70 - 75 *
石鹤群: ""面向产品生命周期的ARMA销售预测模型设计与实现"", 《硕士电子期刊》, vol. 2016, no. 03, pages 5 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112360399A (en) * 2020-11-30 2021-02-12 中国石油大学(北京) Method, device and equipment for predicting yield of coal bed gas
CN112560238A (en) * 2020-12-03 2021-03-26 中国石油天然气股份有限公司 Oil yield prediction method and device based on Starkeberg game model
CN113537592A (en) * 2021-07-15 2021-10-22 中国石油大学(北京) Oil and gas reservoir yield prediction method and device based on long-time and short-time memory network
CN113537592B (en) * 2021-07-15 2023-09-15 中国石油大学(北京) Oil and gas reservoir yield prediction method and device based on long-short-term memory network
CN115204533A (en) * 2022-09-16 2022-10-18 中国地质大学（北京） Oil-gas yield prediction method and system based on multivariable weighted combination model

Also Published As

Publication number Publication date
CN111783356B (en) 2024-03-29

Similar Documents

Publication Publication Date Title
CN111783356B (en) Oil yield prediction method and device based on artificial intelligence
Hu et al. Time series prediction method based on variant LSTM recurrent neural network
US10565516B2 (en) Updating prediction model
CN102568205B (en) Traffic parameter short-time prediction method based on empirical mode decomposition and classification combination prediction in abnormal state
JP7206874B2 (en) Control device, control method and program
US20190303755A1 (en) Water quality prediction
KR20190069582A (en) Reinforcement learning through secondary work
US11574550B2 (en) Learning of policy for selection of associative topic in dialog system
US20180314978A1 (en) Learning apparatus and method for learning a model corresponding to a function changing in time series
US11100388B2 (en) Learning apparatus and method for learning a model corresponding to real number time-series input data
US20200151545A1 (en) Update of attenuation coefficient for a model corresponding to time-series input data
CN111143989A (en) Frequency adjustment amount calculation method, module, system, storage medium, and device
CN116502774B (en) Time sequence prediction method based on time sequence decomposition and Legend projection
JP2019144779A (en) Causal estimation apparatus, causal estimation method, and program
CN114169645A (en) Short-term load prediction method for smart power grid
Beintema et al. Continuous-time identification of dynamic state-space models by deep subspace encoding
Lei et al. A novel time-delay neural grey model and its applications
US11755946B2 (en) Cumulative reward predictor training
CN117196665A (en) Method, equipment and device for predicting value parameters of object
CN116385059A (en) Method, device, equipment and storage medium for updating behavior data prediction model
KR102539223B1 (en) Deep learning method for accelerated processing of ami data stream in edge computing system
CN115630985A (en) Method and device for predicting consumption time
US20210248512A1 (en) Intelligent machine learning recommendation platform
US20220100254A1 (en) Configuring a power management system using reinforcement learning
Kashiwagi et al. Nonparametric nonlinear model predictive control

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