Yield planning model construction method based on multivariate Gaussian mixture model

Abstract

The invention discloses a yield planning model construction method based on a multivariate Gaussian mixture model, and belongs to the field of gas reservoir yield target planning. Comprising the steps that a yield planning model is researched based on a gamma function model, and it is determined that final recoverable reserves have stability under model parameter changes; the constraint condition is changed, and it is determined that the influence on the gamma curve in the dynamic adjustment process conforms to the Gaussian model; selecting a plurality of URR influence factors, and establishing a yield planning model based on a multivariate Gaussian mixture model; and analyzing the output development trend under different constraint conditions and the influence of the main control factors on the planning model by using the established output planning model. According to the method, on the basis of the multivariate Gaussian mixture model, fuzzy analysis, weight analysis and other methods are adopted, the influence of all influence factors on the planned yield in different periods and the influence of main control factors on the planning model can be effectively analyzed, and reference is provided for formulating natural gas exploration and development planning.

Description

Technical Field

[0001] The present invention relates to the field of gas reservoir production target planning, and particularly to a method for constructing a production planning model based on a multivariate Gaussian mixture model. Background Art

[0002] Natural gas development requires a large amount of manpower and material resources. In natural gas development planning, peak production prediction and risk quantification analysis are required. Through peak production prediction and risk quantification analysis, a judgment mechanism that coordinates accuracy and rationality is formed, so as to achieve the purpose of optimizing development benefits.

[0003] However, the current prediction models in the process of natural gas development and production do not consider the influence of multiple factors such as the reserve-production ratio and recovery degree, resulting in low credibility of the production prediction results and unable to represent the actual production situation. Moreover, in terms of scientific and technological literature, the research on natural gas production planning models at home and abroad is still in its infancy. At the same time, through consulting relevant domestic and foreign literature and investigating the existing methods for peak production prediction and risk quantification analysis, it is found that the current production predictions in the industry do not consider internal and external influencing factors and do not carry out production planning for influencing factor analysis, resulting in low credibility of the production prediction results and unable to represent the actual production situation. In terms of existing patents, there are no patents related to natural gas production planning models.

[0004] Generally speaking, the current natural gas production planning often has problems such as great prediction difficulty and risk factors not being considered, which are important technical problems that need to be solved urgently by those skilled in the art. Therefore, it is necessary to establish a production planning model with multi-factor constraints, study the influence of main control factors on the model, and guide the development and production planning of gas fields. Summary of the Invention

[0005] The present invention aims to solve the problems of great prediction difficulty and low credibility of prediction results in the existing natural gas production planning model, and proposes a method for constructing a production planning model based on a multivariate Gaussian mixture model. This method is based on the multivariate Gaussian mixture model and uses methods such as fuzzy analysis and weight analysis to effectively analyze the influence of various influencing factors on the planned production at different times, as well as the influence of main control factors on the planning model, providing a reference for the formulation of natural gas exploration and development planning.

[0006] In order to achieve the above invention purpose, the technical solution of the present invention is as follows:

[0007] A method for constructing a production planning model based on a multivariate Gaussian mixture model, characterized by comprising the following steps:

[0008] Step a, study the production planning model based on the gamma function model, and determine the stability of the ultimately recoverable reserves under the change of model parameters;

[0009] Step b: Change the constraint conditions to determine that the influence during the dynamic adjustment of the gamma curve conforms to the Gaussian model;

[0010] Step c: Select multiple URR influencing factors and establish a production planning model based on the multivariate Gaussian mixture model;

[0011] Step d: Use the established production planning model to analyze the production development trend under different constraint conditions, the influence mechanism of the main control factors on different planning periods, and the influence of the main control factors on the planned production.

[0012] Preferably, in step a, the gamma distribution transformed from the gamma function model has the following general form:

[0013]

[0014] Among them, a is the shape parameter, β is the rate parameter, and the case of a > 1 is used to study the stability of the ultimate recoverable reserves.

[0015] Preferably, the establishment of the production planning model based on the multivariate Gaussian mixture model includes:

[0016] Select the reserve-production ratio, recovery degree, conversion degree, and decline rate as the influencing factors of the ultimate recoverable reserves;

[0017] Construct a membership function for multi-factor fuzzy analysis using the logistic equation;

[0018] Determine the value range of multi-factor fuzzy analysis using the logistic equation;

[0019] Predict the initial values of each influencing factor and calculate the normalized weights of each factor according to the membership function;

[0020] Use the normalized weights of each factor to establish a production planning model based on the multivariate Gaussian mixture model.

[0021] Preferably, the establishment of the production planning model based on the multivariate Gaussian mixture model includes:

[0022] Select the reserve-production ratio, recovery degree, conversion degree, and decline rate as the influencing factors of the ultimate recoverable reserves;

[0023] Combined with the weight fuzzy analysis of each influencing factor, preliminarily analyze the characteristic parameters of the four influencing factors;

[0024] Obtain the weight values of each characteristic parameter;

[0025] Select the average vector of each influencing factor to obtain the mixed event probability value;

[0026] Re - estimate the weight values of each characteristic parameter to obtain the estimated initial parameter values of the model.

[0027] Preferably, the weight values of each characteristic parameter need to satisfy the condition ω i is the mixed weight value.

[0028] Preferably, the average vector of each influencing factor is: Among them, is the characteristic vector of each influencing factor; T is the number of all influencing factor values among all influencing factors.

[0029] Preferably, the mixed event probability value is: represents the basic density of each Gaussian model, and M is the number of Gaussian models.

[0030] Preferably, the estimated initial parameter values of the model include:

[0031]

[0032]

[0033] In the formula, is the re - estimated weight value of each influencing factor, is the vectorized mean value.

[0034] Preferably, use the established production planning model to analyze the production development trend under different constraint conditions, including: draw the prediction curve of the production planning model based on the established two - dimensional multivariate Gaussian mixture model, study the change trend of the production planning model under different growth rates when the ultimate recoverable reserve is fixed; and study the change trend of the production planning model under different ultimate recoverable reserves when the growth rate is fixed.

[0035] Preferably, use the established production planning model to analyze the influence mechanism of the main control factors on different planning periods, including: draw the prediction curve of the production planning model based on the established two - dimensional multivariate Gaussian mixture model, and combine the individual production change curves under different URR influencing factors to determine the influence of each influencing factor on the planned production in different planning periods.

[0036] To sum up, the present invention has the following advantages:

[0037] 1. The present invention studies the influence of various factors on the gas reservoir production planning model, determines that the dynamic change of production under constraints conforms to the Gaussian process, then considers influencing factors such as the reserve-production ratio, and establishes a production planning model based on the multivariate Gaussian mixture model. Finally, according to the production planning models under different constraints, the influence of the main controlling factors is studied. Using this method can clarify the influence mechanism of the change of constraints on the natural gas production planning model and the influence of the main controlling factors on the planning model, and provide guidance for the formulation of natural gas exploration and development plans.

[0038] 2. Based on the multivariate Gaussian mixture model, the present invention uses methods such as fuzzy analysis and weight analysis to effectively analyze the influence of each factor on the planned production in different periods and the influence of the main controlling factors on the planning model, provides a reference for the formulation of natural gas exploration and development plans, explores a new direction for the research of natural gas development strategic goals, and promotes the new development of domestic oil and gas enterprises in the field of production planning.

[0039] 3. In the process of establishing the model, the present invention uses the Gaussian mixture model. First, the weight value is obtained through characteristic parameters, and then the weight and mean are re-estimated using the calculated value of the mixed event probability, so as to combine the Gaussian models of four different factors into a multivariate Gaussian mixture model, and the influence of each factor is reflected by the image of the mixed model. This model construction method can consider the influence of any number of factors, combine multiple single-factor Gaussian models into a mixed Gaussian model under the influence of multiple factors, and more intuitively reflect the influence degree and law of the factors. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic diagram of the gamma distribution.

[0041] Figure 2 It is a schematic diagram of the result of the Gaussian mixture model.

[0042] Figure 3 It is the production planning model curve of different growth rates in the central Sichuan paleo-uplift gas area.

[0043] Figure 4 It is the production planning model curve under different conditions.

[0044] Figure 5 It is the Gaussian process and model confidence interval of production prediction in the central Sichuan paleo-uplift gas area.

[0045] Figure 6 It is the production planning model diagram under different growth rates k.

[0046] Figure 7 It is the production planning model under the influence of different URRs.

[0047] Figure 8 It is the influence mechanism diagram of the reserve-production ratio on the production planning model of the upper production period;

[0048] Figure 9 It is a diagram of the influence mechanism of the decline rate on the production planning model for the upper production period;

[0049] Figure 10 It is a diagram of the influence mechanism of the recovery degree on the production planning model for the upper production period;

[0050] Figure 11 It is a schematic diagram of the change of the production volume K with time during the stable production period planned under the coupling of reservoir and production;

[0051] Figure 12 and 13 It is a schematic diagram of the influence of the reservoir - production ratio on the production planning model. Specific implementation manners

[0052] To illustrate the present invention more clearly, the present invention will be further described below in conjunction with preferred embodiments and the accompanying drawings. Those skilled in the art should understand that the content specifically described below is illustrative rather than restrictive, and should not be used to limit the protection scope of the present invention.

[0053] Taking the gas reservoir in the central ancient uplift of a certain basin as an example, the method for constructing a production planning model based on a multivariate Gaussian mixture model of the present invention will be described below in conjunction with data and graphs.

[0054] Step 1: Determination of the theoretical basis for gas reservoir production planning research

[0055] <1>Based on the gamma function model, study the production planning model, dynamically adjust the growth rate during the production increase period and the stable production period, and at the same time adjust the model parameters of the function to study whether the ultimate recoverable reserve (URR for short) changes, and determine the stability of the ultimate recoverable reserve URR.

[0056] The ultimate recoverable reserve URR can be regarded as the area between the production curve and the horizontal axis numerically. Continuously calculating or integrating the production curve can obtain the value of the ultimate recoverable reserve URR. And using the gamma function can effectively and simply solve the integral operation. Therefore, in this embodiment, the gamma function model is selected to study the production planning model.

[0057] The gamma function is a class of functions that extend the factorial function to real and complex numbers, and has important applications in analysis, probability theory, partial differential equations, and combinatorics.

[0058] The general expression of the gamma function is shown in formula (1).

[0059]

[0060] The gamma function is frequently used in probability and statistics, and it appears in the Poisson distribution and the beta distribution. The probability distribution that has the most direct connection is the gamma distribution obtained by transforming the gamma function. To better understand the gamma distribution, first modify the variable in the gamma function:

[0061]

[0062] Divide both sides of the equation by Γ(a) to get:

[0063]

[0064] This exactly conforms to the definition of a continuous probability distribution, and the integrand is the density function of the gamma distribution as follows:

[0065]

[0066] If we let u = βx, then d(βx) = βdx. Therefore, formula (4) can be transformed into:

[0067]

[0068] Thus, the general form of the gamma distribution is obtained:

[0069]

[0070] The mean of the gamma distribution is: The variance of the gamma distribution is: Where a is called the shape parameter, which determines the shape of the distribution curve; β is called the rate parameter or inverse scale parameter, which determines the steepness of the curve. The gamma distribution is as Figure 1 shown. In the figure, β is the inverse scale parameter. The larger 1 / β is, the steeper the curve. The graph of the gamma distribution depends on the value of the shape parameter a. When a ≤ 1, f(X = x; a, β) is a decreasing function. When a > 1, f(X = x; a, β) is a unimodal function. To better study the changing trend of natural gas production, in this invention, the case of a > 1 is taken, and the gamma function model is used to study the stability of URR under parameter changes.

[0071] In the application example, the ultimate recoverable reserve in the central Sichuan paleo-uplift gas area is about 600 billion cubic meters. Therefore, the integral of the gamma function is URR (about 600 billion cubic meters). Adjust the model parameters of the gamma function and calculate the natural gas production in combination with the following formula:

[0072]

[0073] Finally, the production planning model curve of the central Sichuan paleo-uplift gas area as shown in Figure 3 is obtained. Each curve corresponds to a different growth rate k.

[0074] As can be seen from Figure 3 it, different model parameters will lead to different growth rates. However, the area between the curve and the horizontal axis, that is, the value of URR, remains almost unchanged, all around 600 billion cubic meters. This indicates that during the process of the production planning model curve dynamically adjusting the growth rate and the stable production period, URR remains basically stable. Adjusting the model parameters of the adjustment function will not affect the change of URR, but the stable production period can be quickly reached by adjusting the growth rate in the production stage. This is the theoretical basis for subsequent changing the constraint conditions and studying the production planning model.

[0075] <2>Study the influence mechanism of changes in constraint conditions on peak production and the planning model, and determine that the influence during the process of dynamic adjustment of the curve conforms to the Gaussian model.

[0076] The Gaussian process thinking of the constraint conditions is as follows:[[]]

[0077] For each influencing factor variable x of the production planning model, there exists a Gaussian variable y, and an n - dimensional Gaussian distribution can be obtained. There are infinitely many process points conforming to the Gaussian distribution between x and y, and its process is a Gaussian process, with a parameterized mean function and covariance function.

[0078] The probability density function of the one - dimensional Gaussian distribution is:[[]]

[0079]

[0080] The probability density function of the multi - dimensional Gaussian distribution is:[[]]

[0081]

[0082] In the formula, u is the mean, σ is the standard deviation, and n is the number of variables.

[0083] By adjusting the growth rate and the stable production period, the production planning is predicted, and the obtained model curve is as Figure 3 shown. As can be seen from the figure, although different growth rates, decline rates, and stable production period times have an impact on the shape and characteristics of the model, the overall trend of the production curve remains the same, and it is very close to the Figure 2 Gaussian model curve shown. Therefore, the influence of the constraint conditions on the production planning model curve conforms to the Gaussian process, and the Gaussian model can also be used for production prediction in subsequent research.

[0084] Step 2: Establish a production planning model based on the multi - dimensional Gaussian mixture model

[0085] <1>There are two methods to establish the production planning model in this step. The first one is:[[]]

[0086] Considering four influencing factors, namely the reserve-production ratio, recovery factor, conversion degree, and decline rate, a membership function for multi-factor fuzzy analysis is constructed using the logistic equation. The normalized weights of each influencing factor are calculated using weighted fuzzy analysis, and then the Gaussian mixture model is used to study the controlled relationship between the planning model and the four factors under different URR conditions.

[0087] The Gaussian mixture model can be regarded as a model composed of K single Gaussian models, and these K sub-models are the latent variables of the mixture model. Generally speaking, a mixture model can use any probability distribution. In this method, the Gaussian mixture model is used because the Gaussian distribution has good mathematical properties and good computational performance, which can better analyze the change trend of data.

[0088] Combined with formula (8), the Gaussian mixture density is the sum of the weights of M Gaussian densities, and its expression is:

[0089]

[0090] Among them, is the feature vector, is the Gaussian probability density value, M is the number of Gaussian models, that is, the Gaussian models established for the four influencing factors of the reserve-production ratio, recovery factor, conversion degree, and decline rate. ω i is the mixing weight value, and the value range comes from the weights obtained by the weighted fuzzy analysis of the URR influencing factors.

[0091] The mixing weights must meet the conditions:

[0092] In formula (9), the basic density is a D-dimensional Gaussian function, and its expression is:

[0093]

[0094] In the formula, among them is the mean vector, ∑ is the covariance matrix, D is the dimension of the feature vector, and T is the transpose symbol.

[0095] In this application example, the total resource volume of the central Sichuan paleo-uplift is about 300 billion cubic meters. Considering the URR range of influencing factors from 172.8 to 621 billion cubic meters. During the research on gas reservoir production planning, the reserve-production ratio, recovery factor, conversion degree, and decline rate are used as URR influencing factors, and an index set U = (x1, x2, x3, x4) of the weights of the influencing factors of the paleo-uplift gas reservoir production law is constructed. Among them, x1 represents the size of the reserve-production ratio, x2 represents the level of recovery factor, x3 represents the level of conversion degree, and x4 represents the size of the decline rate. A membership function for multi-factor fuzzy analysis is constructed using the logistic equation, and the expression is:

[0096]

[0097] Among them, a represents the lower limit of the membership function, b represents the lower limit of the membership function, and m is the median of point a and point b.

[0098] Then, the logistic equation is used to determine the value range of multi-factor fuzzy analysis. The value ranges of each influencing factor are as follows: 15 ≤ reserve-production ratio ≤ 60; 40% ≤ recovery rate ≤ 70%; 30% ≤ conversion degree ≤ 80%; 5% ≤ decline rate ≤ 15%.

[0099] Predict the initial values of each influencing factor to obtain: {reserve-production ratio 55, recovery rate 45%, conversion degree 60%, decline rate 10%}. Calculate the normalized weights of each factor according to the membership function: {0.39, 0.30, 0.20, 0.11}. When performing subsequent calculations, round to the nearest integer. The weights of each factor are: Calculate the normalized weights of each factor according to the membership function: {0.4, 0.3, 0.2, 0.1}.

[0100] The production prediction curve obtained based on each influencing factor and the curve under multi-factor control are as Figure 5 shown. It can be seen from the figure that under a single constraint condition, different factors have different effects on the production change trend, but there is a large gap from the production planning curve. This shows that considering only a single factor is far from enough. In contrast, within the comprehensive control results of the four factors and their 95% confidence interval, the overall production prediction result is very close to the planning curve. This also indicates that after comprehensively considering multiple factors, the accuracy of the production planning model is higher.

[0101] <2>The establishment method of the second production planning model is as follows:

[0102] Combined with the weight fuzzy analysis of influencing factors, preliminarily analyze the characteristic parameters of the four factors affecting natural gas production, and establish a multivariate Gaussian mixture model based on the four influencing factors of reserve-production ratio, recovery degree, conversion degree, and decline rate.

[0103] Based on the Gaussian mixture model, establish a production planning model and estimate the initial parameter values. The specific process is as follows:

[0104] First, preliminarily analyze the characteristic parameters of the four factors affecting natural gas production.

[0105] And find the weight values of each parameter for the characteristic parameters. The conditions that the parameter values need to meet are:

[0106]

[0107] After having the weight values of each parameter, select the average vector:

[0108] Obtain the mixed event probability value:

[0109] Re - estimate each parameter, and the results are shown in formulas (11) and (12).

[0110]

[0111]

[0112] Among them, M is the number of Gaussian models, ω i is the mixing weight value. is the feature vector of each influencing factor, T is the number of all influencing factor values in the influencing factors. For example, 50 production values in historical data correspond to 50 reserve - production ratios, and for the reserve - production ratio, T = 50. represents the basic density of each Gaussian model, is the weight value after re - estimating each influencing factor, is the vectorized mean value. Because the variable is a vector, and its mean value is also a vector.

[0113] Substitute the re - estimated parameters into formula (8). The finally established production planning model prediction graph is similar to Figure 2 and use this model to predict the production development trend of the gas reservoir under different constraint conditions. The specific results are shown in the content of step three.

[0114] Step three: Use the established production planning model to predict the production development trend under different constraint conditions

[0115] <1>Based on the established two - dimensional multivariate Gaussian mixture model, study the change trends of the production planning model curves under different growth rates k and different ultimate recoverable reserves URR respectively.

[0116] The production planning model under different growth rates k is as Figure 6 shown. It can be seen from the figure that the larger the growth rate k, the greater the slope of the production planning curve, and it will reach the peak production earlier, and the production peak will also be larger. Therefore, it will enter the decline period earlier. However, the curve area in the figure remains the same (maintaining URR = 500 billion cubic meters). Therefore, for the planning curve with a large growth rate, the cumulative production before entering the decline period is larger, and the cumulative production after the decline period is smaller.

[0117] The production planning model under different URR is as Figure 7As shown, it can be seen from the figure that since the growth rate remains unchanged, the slope of the production plan curve is also constant. The growth amplitudes of the three curves are exactly the same. However, the larger the URR, the later the time to reach the peak production, and the larger the peak production. Therefore, it will enter the decline period later. Moreover, the URR represents the area of the curve. Therefore, the larger the URR, the higher the production curve will be.

[0118] <2>Study the influence of the main influencing factors on the planned production in different planning periods.

[0119] The influence mechanism of the reserve-production ratio on the production planning model in the upper production period is as Figure 8 shown. The shaded part in the figure is the production change curve under the influence of the reserve-production ratio. Comparing this shaded part with the predicted curve of the production planning model at the top, it can be seen that the reserve-production ratio is mainly concentrated in the growth period of production. This indicates that the reserve-production ratio mainly controls the growth rate of the production planning model in the upper production period, and the K value should be adjusted starting from the reserve-production ratio.

[0120] The influence mechanism of the decline rate on the production planning model in the upper production period is as Figure 9 shown. The shaded part in the figure is the production change curve under the influence of the decline rate. Comparing the shaded part with the predicted curve of the production planning model at the top, it can be seen that the decline rate is mainly concentrated in the stable production period of production. This indicates that the decline rate controls the stable state of the stable production period.

[0121] The influence mechanism of the recovery factor on the production planning model in the upper production period is as Figure 10 shown. The shaded part in the figure is the production change curve under the influence of the recovery factor. Comparing the shaded part with the predicted curve of the production planning model at the top, it can be seen that the recovery factor is mainly concentrated in the decline period of production. This indicates that the recovery factor has the greatest impact on the production in the decline period, and the decline period can be extended by adjusting the recovery factor.

[0122] <3>Study the influence of the main control factors on the planning model.

[0123] Taking the reserve-production ratio as an example, the reserve-production ratio is the main control factor of the production planning model in the upper production period. The variation of the production K in the stable production period under the coupling of reserves and production with time is as Figure 11 shown. It can be seen from the figure that the coupling of reserves and production conforms to the logictics growth model, and different growth curves and peak times can be obtained by changing the reserve-production conversion rate.

[0124] At the same time, the influence of the reserve-production ratio on the production planning model in different years is as Figure 12 、 Figure 13 shown by each peak curve in. By superimposing all the curves, the final result is shown by the coupling curve at the end of the figure. This curve is the production planning model diagram under the coupling of reserves and production, its contour is the production prediction curve, and the area is the ultimate recoverable reserve URR of the gas reservoir.

[0125] The present invention takes into account influencing factors such as the reserve-production ratio, growth rate, decline rate, etc. Based on the multivariate Gaussian mixture model, methods such as fuzzy analysis and weight analysis are adopted to effectively analyze the influence of each factor on the planned production in different periods, as well as the influence of the main control factors on the planning model, providing a reference for the formulation of natural gas exploration and development plans, exploring a new direction for the research of natural gas development strategic goals, and promoting the new development of domestic oil and gas enterprises in the field of production planning.

[0126] The above are only preferred embodiments of the present invention, and do not impose any form of limitation on the present invention. Any simple modification or equivalent change made to the above embodiments based on the technical essence of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for constructing a yield planning model based on a multivariate Gaussian mixture model, characterized in that, It includes the following steps: Step a: Based on the gamma function model, study the production planning model and determine that the ultimate recoverable reserves are stable under changes in model parameters; Step b: Change the constraint conditions and determine that the influence during the dynamic adjustment of the gamma curve conforms to the Gaussian model; Step c: Select multiple URR influencing factors and establish a production planning model based on the multivariate Gaussian mixture model; Step d: Use the established production planning model to analyze the production development trend under different constraint conditions, the influence mechanism of the main control factors on different planning periods, and the influence of the main control factors on the planned production.

2. The method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 1, wherein In step a, the gamma distribution transformed from the gamma function model has the following general form: where a is the shape parameter and β is the rate parameter. The case of a > 1 is used to study the stability of the ultimate recoverable reserves.

3. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 1, characterized in that, The establishment of the production planning model based on the multivariate Gaussian mixture model includes: Select the reserve-production ratio, recovery degree, conversion degree, and decline rate as the influencing factors of the ultimate recoverable reserves; Use the logistic equation to construct the membership function for multi-factor fuzzy analysis; Use the logistic equation to determine the value range of multi-factor fuzzy analysis; Predict the initial values of each influencing factor and calculate the normalized weight of each factor according to the membership function; Use the normalized weights of each factor to establish a production planning model based on the multivariate Gaussian mixture model.

4. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 1, characterized in that, The establishment of the production planning model based on the multivariate Gaussian mixture model includes: Select the reserve-production ratio, recovery degree, conversion degree, and decline rate as the influencing factors of the ultimate recoverable reserves; Combined with the weight fuzzy analysis of each influencing factor, preliminarily analyze the characteristic parameters of the four influencing factors; Obtain the weight values of each characteristic parameter; Select the average vector of each influencing factor and obtain the mixed event probability value; Re-estimate the weight values of each characteristic parameter to obtain the estimated initial parameter values of the model.

5. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 4, characterized in that The weight values of each characteristic parameter need to satisfy the conditions ω i is the mixed weight value.

6. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 4, characterized in that The average vector of each influencing factor is as follows: Among them, is the eigenvector of each influencing factor; T is the number of all influencing factor values among each influencing factor.

7. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 4, characterized in that The mixed event probability value is: represents the basic density of each Gaussian model, and M is the number of Gaussian models.

8. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 4, characterized in that The estimated initial parameter values of the model include: In the formula, is the re - estimated weight value for each influencing factor, is the mean value after vectorization.

9. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 1, characterized in that, Using the established production planning model to analyze the production development trend under different constraint conditions, including: drawing the prediction curve of the production planning model based on the established two-dimensional multivariate Gaussian mixture model, and studying the change trend of the production planning model under different growth rates when the ultimate recoverable reserves are fixed; and studying the change trend of the production planning model under different ultimate recoverable reserves when the growth rate is fixed.

10. A method for constructing a yield planning model based on a multivariate Gaussian mixture model according to claim 1, characterized in that Using the established production planning model to analyze the influence mechanism of the main control factors on different planning periods, including: drawing the prediction curve of the production planning model based on the established two-dimensional multivariate Gaussian mixture model, and combining the individual production change curves under different URR influencing factors to determine the influence of each influencing factor on the planned production in different planning periods.

Images