Bayesian stochastic volatility gas production forecasting method based on monte carlo simulation

Abstract

The application discloses a kind of based on Monte Carlo simulation's bayesian random fluctuation natural gas production prediction method, it is related to natural gas production prediction technical field.The application obtains natural gas historical collection data, carries out target function construction, calculates the initial value of yield probability;According to yield probability initial value, introduce Monte Carlo error judgment mechanism, carry out mean square error estimation calculation to natural gas production by natural gas collection historical factor data;Natural gas collection historical factor data is unfolded to multiple factors, and a multiple factor judgment weight model is formed, and the multiple factor judgment weight model is predicted to natural gas production by bayesian training formula.The application can better avoid natural gas prediction data overfitting phenomenon by using Monte Carlo simulation combined with bayesian calculation, and the robustness of data is enhanced, the approximate weight distribution calculation is carried out by the whole natural gas data multiple factor calculation model, and the accuracy of natural gas prediction judgment is improved.

Description

Technical Field

[0001] This invention relates to the field of natural gas production forecasting technology, and more specifically to a method for forecasting natural gas production based on Monte Carlo simulation and Bayesian stochastic fluctuations. Background Technology

[0002] Economic output of an oil and gas field refers to the amount of oil or gas produced that can achieve the operational goals under existing production and operation conditions. It represents the minimum oil or gas production threshold required to reach the enterprise's operational objectives. Traditionally, oil production calculations have been simplified by estimating oil or gas prices and drilling costs, treating these as fixed values ​​to obtain the current year's oil or gas production. However, in reality, during oil and gas field development, oil and gas production, crude oil prices, natural gas prices, and drilling costs are not constant but rather fluctuate according to a certain probability distribution.

[0003] Existing technologies use Monte Carlo simulation to simulate various factors affecting oil and gas field production throughout the development life of an oil and gas field. By establishing an economic production prediction model for oil and gas field development, the economic oil or gas production in the initial year is calculated. This is used to determine whether the operation is worthwhile or to make appropriate adjustments to the existing plan based on the calculation results to meet the company's profit requirements.

[0004] Monte Carlo simulation, also known as simulated sampling or statistical experimentation, is a simulation technique guided by mathematical statistics. Essentially, it simulates possible random phenomena by generating random numbers according to a certain probability distribution, and is a practical method in probability analysis. Currently, in the oil and gas industry, Monte Carlo simulation is mainly used for evaluating exploration risks and predicting reserves.

[0005] For predicting natural gas production, engineering models can be used to calculate rough forecasts, and probability assessment is a form of situation assessment. However, this method is inaccurate for estimating the linear state of natural gas production. Due to the uncertainty of actual natural gas production, although Monte Carlo simulations have some linear classification and are accompanied by normally distributed data to reflect the process of prediction, directly using Monte Carlo simulations will increase the error of the prediction conclusions. This necessitates that those skilled in the art solve the corresponding technical problems. Summary of the Invention

[0006] To overcome the defects and shortcomings of the existing technologies, this invention provides a Bayesian stochastic fluctuation natural gas production prediction method based on Monte Carlo simulation. The purpose of this invention is to address the problem that directly using Monte Carlo simulation in the existing technologies increases the error in the prediction conclusions. This invention utilizes Monte Carlo simulation combined with Bayesian computation to better avoid overfitting of natural gas prediction data, enhancing the robustness of the data. By using a multi-factor calculation model of all natural gas data to perform approximate weight distribution calculations, the accuracy of natural gas prediction judgments is improved.

[0007] To address the problems existing in the prior art, the present invention is achieved through the following technical solution.

[0008] This invention provides a method for predicting natural gas production based on Bayesian stochastic fluctuations using Monte Carlo simulation, characterized by the following steps:

[0009] S1. Obtain historical natural gas acquisition data, construct the objective function based on the historical natural gas acquisition data, and calculate the initial value of the production probability;

[0010] S2. Based on the initial production probability value calculated in step S1, a Monte Carlo error judgment mechanism is introduced, and the root mean square error of natural gas production is estimated by collecting historical factor data of natural gas.

[0011] S3. Perform multi-factor expansion on historical data of natural gas collection to form a multi-factor judgment weight model. Use the Bayesian training formula to predict natural gas production using this multi-factor judgment weight model.

[0012] In step S1, the objective function is constructed using historical natural gas acquisition data. Specifically, this involves analyzing the acquired historical natural gas acquisition data to obtain information on the geographical distribution of natural gas development, the drilling depth parameters of natural gas development strata, and machine failure data during the natural gas development process. The objective function is then constructed using these information.

[0013] The machine failure data in the natural gas development process includes machine failure occurrence mode parameters, failure mechanism thresholds, and failure impact analysis thresholds.

[0014] Before constructing the objective function, it is necessary to determine the relationship between the threshold for machine failure mechanism and the threshold for failure impact analysis during natural gas development.

[0015] In the process of obtaining the geographical distribution of natural gas development, several time periods of machine failure mechanism thresholds are determined and arranged in chronological order from far to near to form a time set of natural gas development geographical locations.

[0016] In the process of obtaining drilling depth parameters for natural gas development, instability is obtained by acquiring the threshold values ​​of each fault mechanism at sequential time nodes.

[0017] The objective function S is expressed as: ;

[0018] Where s1 is the formula for summing the gas production differences between natural gas wells during the development process and the predicted depth of natural gas development; i is the number of natural gas wells, j is the number of natural gas development cycles, and c is the number of development cycles. ij d represents the actual pressure value of natural gas well i during development j; ij Let be the predicted pressure value for natural gas well i during development phase j. Let z be the development cycle coefficient for natural gas well i. ij For natural gas well i, the drilling depth of the equipment during development j, y ij The cost required for drilling equipment to reach the depth required for natural gas well i during development j. The equipment loss coefficient is denoted by m and n, where m and n are positive integers.

[0019] Where s2 is the effective collection volume of natural gas developed, and r i Let Q be the actual gas output during the development of natural gas well i, α be the natural gas development regulation threshold, and Q be the actual gas output. j For natural gas development, the adaptability of gas wells to operation in the jth phase. R is the fitness adjustment factor. j To determine the failure rate of a natural gas well after j cycles of operation, This is a failure rate adjustment factor;

[0020] Where s3 is the dynamic forecast value of natural gas development quantity, The gas output per unit time of natural gas well development, where t is time. This represents the dynamic change in natural gas development volume per unit time. This represents the random disturbance component of the dynamic changes in natural gas development. This represents the average gas output per unit time during natural gas development.

[0021] In step S1, historical natural gas collection data is obtained through a cloud server.

[0022] In step S1, the initial value of the production probability is calculated, specifically... ,in, The random variables for natural gas collection are X1, X2, X3, ... X. n The mean, through the natural gas development decision function Perform logical calculations for natural gas development decisions and calculate probability functions. The product of the standard deviation of natural gas development (μ) The mean The natural gas collection unit vector, P(*) represents the initial value of the natural gas production probability.

[0023] In step S2, based on the initial production probability calculated in step S1, a Monte Carlo error judgment mechanism is introduced. Specifically, this means...

[0024] Based on the initial production probability calculated in step S1, a Monte Carlo error judgment mechanism is introduced. Perform natural gas data estimation, Let N be the root mean square error of natural gas development, and N be the incremental amount of natural gas development. The incremental adjustment coefficient for natural gas development is fixed. In this case, obtaining the incremental natural gas development N improves the accuracy of error judgment.

[0025] Historical data on natural gas development factors are corrected using an error judgment mechanism. The parameters u0 of the prediction model u(t) for the natural gas development depth U are extracted and encoded. The parameters v0 of the prediction model v(t) for the natural gas development stratum pressure value V are also encoded. p The parameters w(t) of the prediction model for the development time W of natural gas equipment are encoded. q Encode; set the value range of parameter u0 to [u min ,u max ], parameter v p The range of values ​​for is [v min ,v max ], parameter w q The range of values ​​for is [w min ,w max ], where the minimum and maximum values ​​are both within the normative threshold range for natural gas development.

[0026] Mean square deviation of natural gas production in historical data on natural gas development The value is obtained through estimation calculation. Where B is the logical threshold for incremental natural gas development per unit area. The adjustment factor is i, which represents the number of natural gas wells, and m is a positive integer.

[0027] In step S3, the historical data of natural gas collection is analyzed using multiple factors to form a multi-factor judgment weight model, specifically:

[0028] Weighting of natural gas development depth, natural gas development stratum pressure value, and natural gas equipment development time among historical factors of natural gas extraction is calculated.

[0029] Among them, the weight of natural gas development depth is , This refers to the burial depth in the middle of the gas reservoir's producing layer;

[0030] The weight of the rock pressure value for natural gas development is , This represents the dynamic change value of rock strata pressure;

[0031] Natural gas equipment development time weighting , Values ​​that change chronologically during development.

[0032] For a known dataset C containing predictions of natural gas development from all natural gas wells, obtain the weights of the natural gas development depth in dataset C. Distribution, weight of rock pressure values ​​in natural gas development Distribution and time weighting of natural gas equipment development Distribution, setting total parameters , Let θ be the Gaussian distribution variance of the predicted production threshold x for natural gas development, and let θ be the production distribution weight of the predicted production threshold x for natural gas development. To predict the production threshold for natural gas development, x is the expectation of a Gaussian distribution, and the posterior weight distribution is... ,in, To minimize the approximate weight distribution;

[0033] Using Bayesian training formula ,definition The sum of the natural gas development forecast dataset C, which represents the forecast of natural gas production. The likelihood distribution of the predicted production threshold x for natural gas development based on the current real-time natural gas development volume. The weighted distribution of the predicted production threshold x for the initial development production of natural gas.

[0034] In step S3, it is pre-defined that during the natural gas production forecasting process, due to the numerical instability of key time nodes in the natural gas production forecast target value at a certain moment, the mean of the natural gas production forecast target value is selected in the unstable state. By selecting the mean of the natural gas production forecast target value, the reliability curve of natural gas production is characterized. Several time nodes are selected to eliminate invalid natural gas production forecast target values. The floating-point solution of the natural gas production forecast target value is solved by the least squares method. Fuzzy search is performed, and after inverse transformation, the integer solution of fuzziness and its covariance matrix are obtained.

[0035] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0036] This invention utilizes Monte Carlo simulation combined with Bayesian computation to better avoid overfitting of natural gas forecast data and enhance the robustness of the data. By performing approximate weight distribution calculations using a multi-factor calculation model of all natural gas data, the accuracy of natural gas forecasting is improved. Attached Figure Description

[0037] Figure 1 This is the overall flowchart of the present invention;

[0038] Figure 2 This is a schematic diagram illustrating the predicted effect of the present invention. Detailed Implementation

[0039] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example 1

[0041] As a preferred embodiment of the present invention, please refer to the appendix to the specification. Figure 1 and attached Figure 2 As shown in the figure, this embodiment discloses a method for predicting Bayesian stochastic fluctuation natural gas production based on Monte Carlo simulation. The method includes the following steps:

[0042] S1. Obtain historical natural gas acquisition data, construct the objective function based on the historical natural gas acquisition data, and calculate the initial value of the production probability;

[0043] S2. Based on the initial production probability value calculated in step S1, a Monte Carlo error judgment mechanism is introduced, and the root mean square error of natural gas production is estimated by collecting historical factor data of natural gas.

[0044] S3. Perform multi-factor analysis on historical natural gas collection data to form a multi-factor judgment weight model. Use a Bayesian training formula to predict natural gas production using this multi-factor judgment weight model (see attached). Figure 2 (As shown).

[0045] In this embodiment, the use of Monte Carlo simulation combined with Bayesian computation can better avoid overfitting of natural gas forecast data and enhance the robustness of the data. By using a multi-factor calculation model of all natural gas data to perform approximate weight distribution calculation, the accuracy of natural gas forecast judgment is improved.

[0046] Example 2

[0047] As another preferred embodiment of the present invention, this embodiment is a description of the specific implementation of step S1 in embodiment 1 above.

[0048] As one implementation method of this embodiment, in step S1, historical natural gas collection data is obtained through a cloud server.

[0049] As an example, historical natural gas data can also be obtained through offline data transmission methods, such as data disk copying, online data transmission, and other methods, to retrieve historical natural gas data from cloud servers or physical servers.

[0050] As another implementation method of this embodiment, the geographical distribution of natural gas development, the drilling depth parameters of natural gas development strata, and the machine failure data during the natural gas development process are analyzed from the acquired historical natural gas data. An objective function is constructed using the geographical distribution of natural gas development, the drilling depth parameters of natural gas development strata, and the machine failure data during the natural gas development process to calculate the production probability handling.

[0051] Example 3

[0052] As another preferred embodiment of the present invention, this embodiment is a description of the specific implementation of Embodiments 1 and 2.

[0053] As one implementation of this embodiment, the machine fault data in the natural gas development process includes machine fault occurrence mode parameters, fault mechanism thresholds, and fault impact analysis thresholds.

[0054] Before constructing the objective function, it is necessary to determine the relationship between the threshold for machine failure mechanism and the threshold for failure impact analysis during natural gas development.

[0055] Furthermore, in the process of obtaining the geographical distribution of natural gas development, several time periods of machine failure mechanism thresholds are determined and arranged in chronological order from far to near to form a time set of natural gas development geographical locations.

[0056] Furthermore, during the process of obtaining drilling depth parameters for natural gas development strata, the instability obtained according to the fault mechanism thresholds at sequential time nodes is used to calculate the initial probability value of natural gas development production data through the calculation of the objective function S.

[0057] The objective function S is expressed as ;

[0058] Where s1 is the formula for summing the gas production differences between natural gas wells during the development process and the predicted depth of natural gas development; i is the number of natural gas wells, j is the number of natural gas development cycles, and c is the number of development cycles. ijd represents the actual pressure value of natural gas well i during development j; ij Let be the predicted pressure value for natural gas well i during development phase j. Let z be the development cycle coefficient for natural gas well i. ij For natural gas well i, the drilling depth of the equipment during development j, y ij The cost required for drilling equipment to reach the depth required for natural gas well i during development j. The equipment loss coefficient is denoted by m and n, where m and n are positive integers.

[0059] Where s2 is the effective collection volume of natural gas developed, and r i Let Q be the actual gas output during the development of natural gas well i, α be the natural gas development regulation threshold, and Q be the actual gas output. j For natural gas development, the adaptability of gas wells to operation in the jth phase. R is the fitness adjustment factor. j To determine the failure rate of a natural gas well after j cycles of operation, This is a failure rate adjustment factor;

[0060] Where s3 is the dynamic forecast value of natural gas development quantity, The gas output per unit time of natural gas well development, where t is time. This represents the dynamic change in natural gas development volume per unit time. This represents the random disturbance component of the dynamic changes in natural gas development. This represents the average gas output per unit time during natural gas development.

[0061] Example 4

[0062] As another preferred embodiment of the present invention, this embodiment is a description of the specific implementation of step S2 in embodiment 1 above.

[0063] As one implementation method of this embodiment, a probabilistic initial value calculation is performed on the natural gas production. ,in, The random variables for natural gas collection are X1, X2, X3, ... X. n The mean, through the natural gas development decision function Perform logical calculations for natural gas development decisions and calculate probability functions. The product of the standard deviation of natural gas development (μ) The mean The natural gas collection unit vector, P(*) represents the initial value of the natural gas production probability.

[0064] Based on the initial value of the production probability calculated above, a Monte Carlo error judgment mechanism is introduced. Perform natural gas data estimation, Let N be the root mean square error of natural gas development, and N be the incremental amount of natural gas development. The incremental adjustment coefficient for natural gas development is fixed. In this case, obtaining the incremental natural gas development N improves the accuracy of error judgment.

[0065] Historical data on natural gas development factors are corrected using an error judgment mechanism. The parameters u0 of the prediction model u(t) for the natural gas development depth U are extracted and encoded. The parameters v0 of the prediction model v(t) for the natural gas development stratum pressure value V are also encoded. p The parameters w(t) of the prediction model for the development time W of natural gas equipment are encoded. q Encode; set the value range of parameter u0 to [u min ,u max ], parameter v p The range of values ​​for is [v min ,v max ], parameter w q The range of values ​​for is [w min ,w max ], where the minimum and maximum values ​​are both within the normative threshold range for natural gas development.

[0066] Mean square deviation of natural gas production in historical data on natural gas development The value is obtained through estimation calculation. Where B is the logical threshold for incremental natural gas development per unit area. The adjustment factor is i, which represents the number of natural gas wells, and m is a positive integer.

[0067] Example 5

[0068] As another preferred embodiment of the present invention, this embodiment is a description of the specific implementation of step S3 in embodiment 1 above.

[0069] As one implementation method of this embodiment, in step S3, the historical data of natural gas collection is expanded using multiple factors to form a multi-factor judgment weight model, specifically referring to:

[0070] The weights of natural gas development depth, natural gas development stratum pressure value, and natural gas equipment development time are calculated for historical factors of natural gas extraction.

[0071] Among them, the weight of natural gas development depth is , This refers to the burial depth in the middle of the gas reservoir's producing layer;

[0072] The weight of the rock pressure value for natural gas development is , This represents the dynamic change value of rock strata pressure;

[0073] Natural gas equipment development time weighting , Values ​​that change chronologically during development.

[0074] For a known dataset C containing predictions of natural gas development from all natural gas wells, obtain the weights of the natural gas development depth in dataset C. Distribution, weight of rock pressure values ​​in natural gas development Distribution and time weighting of natural gas equipment development Distribution, setting total parameters , Let θ be the Gaussian distribution variance of the predicted production threshold x for natural gas development, and let θ be the production distribution weight of the predicted production threshold x for natural gas development. To predict the production threshold for natural gas development, x is the expectation of a Gaussian distribution, and the posterior weight distribution is... ,in, To minimize the approximate weight distribution.

[0075] As another preferred embodiment of this example, the Bayesian training formula is used. ,definition The sum of the natural gas development forecast dataset C, which represents the forecast of natural gas production. The likelihood distribution of the predicted production threshold x for natural gas development based on the current real-time natural gas development volume. The weighted distribution of the predicted production threshold x for the initial development production of natural gas.

[0076] In step S3, it is pre-defined that during the natural gas production forecasting process, due to the numerical instability of key time nodes in the natural gas production forecast target value at a certain moment, the mean of the natural gas production forecast target value is selected in the unstable state. By selecting the mean of the natural gas production forecast target value, the reliability curve of natural gas production is characterized. Several time nodes are selected to eliminate invalid natural gas production forecast target values. The floating-point solution of the natural gas production forecast target value is solved by the least squares method. Fuzzy search is performed, and after inverse transformation, the integer solution of fuzziness and its covariance matrix are obtained.

[0077] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for predicting natural gas production based on Bayesian stochastic fluctuations using Monte Carlo simulation, characterized in that, The method includes the following steps: S1. Acquire historical natural gas acquisition data. Analyze the acquired historical natural gas acquisition data to obtain the geographical distribution of natural gas development, drilling depth parameters of natural gas development strata, and machine failure data during the natural gas development process. The machine failure data during the natural gas development process includes machine failure occurrence mode parameters, failure mechanism thresholds, and failure impact analysis thresholds. Before constructing the objective function, it is necessary to determine the relationship between the machine failure mechanism thresholds and the failure impact analysis thresholds during the natural gas development process. In the process of acquiring the geographical distribution of natural gas development, determine several time periods for the machine failure mechanism thresholds and arrange them in chronological order from far to near to form a time set of natural gas development geographical locations. In the process of acquiring drilling depth parameters of natural gas development strata, determine the instability of acquiring each failure mechanism threshold according to the sequential time nodes. Correct the historical natural gas acquisition data through an error judgment mechanism, extract and encode the parameter u0 of the prediction model u(t) for the natural gas development depth U, and the parameter v of the prediction model v(t) for the natural gas development strata pressure value V. p The parameters w(t) of the prediction model for the development time W of natural gas equipment are encoded. q Encode; set the value range of parameter u0 to [u min ,u max ], parameter v p The range of values ​​for is [v min ,v max ], parameter w q The range of values ​​for is [w min ,w max The minimum and maximum values ​​are both within the normative threshold range for natural gas development; An objective function was constructed based on the geographical distribution of natural gas development, drilling depth parameters of natural gas development strata, and machine failure data during natural gas development, and the initial value of production probability was calculated. S2. Based on the initial production probability value calculated in step S1, a Monte Carlo error judgment mechanism is introduced, and the root mean square error of natural gas production is estimated by collecting historical factor data of natural gas. S3. Perform multi-factor expansion on historical natural gas collection data to form a multi-factor judgment weight model. Use the Bayesian training formula to predict natural gas production using this multi-factor judgment weight model. Pre-determine that during the natural gas production prediction process, due to the numerical instability of key time nodes in the natural gas production prediction target value at a certain moment, select the mean of the natural gas production prediction target value in the unstable state. By selecting the mean of the natural gas production prediction target value, characterize the reliability curve of natural gas production. Select several time nodes to remove invalid natural gas production prediction target values. Solve the floating-point solution of the natural gas production production prediction target value using the least squares method, perform fuzzy search, and after inverse transformation, obtain the integer solution of the fuzzyness and its covariance matrix.

2. The method for predicting Bayesian stochastic fluctuation natural gas production based on Monte Carlo simulation as described in claim 1, characterized in that, In step S1, historical natural gas collection data is obtained through a cloud server.

Images