Carbonate gas reservoir reserve utilization effect evaluation method based on BP neural network

Abstract

The invention discloses a carbonate rock gas reservoir reserve utilization effect evaluation method based on a BP neural network, which belongs to the technical field of oil and gas field development, and is characterized by comprising the following steps: a, forming carbonate rock gas reservoir comprehensive classification; b, selecting characterization index parameters influencing the reserve utilization effect; c, determining influence factors of different types of gas reservoirs influencing the reserve utilization effect; d, determining weight values of different influence factors; e, establishing a gas reservoir reserve utilization quantification formula through nonlinear surface fitting; and f, by applying a BP neural network algorithm and combining with the reserve utilization effect evaluation method chart, establishing a reserve utilization effect evaluation model, and completing gas reservoir reserve utilization effect evaluation. According to the method, the change rule among the influence factors can be reflected more accurately and visually, and the established reserve utilization effect evaluation model can provide theoretical reference for further improvement of reserve utilization of the gas reservoir, deployment of the recovery ratio and optimization of the gas reservoir with excavation potential.

Description

Technical Field

[0001] The invention relates to the technical field of oil and gas field development, and in particular to a carbonate gas reservoir reserve producing effect evaluation method based on BP neural network. Background Art

[0002] Natural gas is the main resource to promote the green and low-carbon transformation of global energy, and plays an important bridging role in my country's low-carbon transformation and development. In my country's total natural gas production, carbonate gas reservoirs contribute about 50% of the production, but due to the influence of various geological factors such as complex structure, formation water distribution and water invasion intensity, low permeability, degree of fracture development, and fracture-cavity matching, the development and utilization of reserves of different types of gas reservoirs are different. The influence of low permeability and formation water has increased the difficulty of effective utilization of gas reservoir reserves, and the recovery rate of gas reservoirs has dropped significantly, seriously affecting the efficient development of gas reservoirs. Through in-depth research on the evaluation of the utilization effect of carbonate gas reservoir reserves, the purpose is to explore the potential for further utilization of reserves of different types of gas reservoirs, formulate a method for evaluating the utilization effect of reserves, and provide a theoretical basis and technical support for improving the utilization of reserves and the recovery rate of gas reservoirs. Carbonate gas reservoirs have complex reservoir spaces and various types. The reserve utilization is inconsistent. The reserve utilization of fractured gas reservoirs is almost 100%, but the reserve utilization of fractured-porous gas reservoirs varies greatly due to the influence of physical properties and formation water. On the one hand, the permeability of low-permeability gas reservoirs is low, the well control radius of gas wells is small, the propagation range is limited, the reserve control degree of single wells is low, and the reserve utilization degree is generally 20%-50%, which means the reserve utilization degree is low. On the other hand, formation water intrusion into gas reservoirs will form liquid accumulation at the bottom of the well. When the gas production of the gas well is not enough to carry out all the accumulated liquid, it will cause the liquid accumulation at the bottom of the well to increase, resulting in a decrease in gas well production or even waterlogging and shutdown. Therefore, the influence of formation water on the reserve utilization of gas reservoirs is more obvious, which greatly reduces the effective utilization of reserves. A large number of studies at home and abroad have shown that the factors affecting the utilization of gas reservoir reserves are many and complex, and there is a lack of systematic methods for evaluating the effect of reserve utilization. How to quantify and characterize the factors affecting the production effect of gas reservoir reserves plays an important guiding role in accurately evaluating the production effect of gas reservoir reserves, improving the degree of reserve production, and improving the recovery rate of gas reservoirs.

[0003] A Chinese patent document with announcement number CN106886046B and announcement date November 16, 2018 discloses a method and device for determining the available reserves of an unproduced block of a fracture-cavity gas reservoir, the method comprising: determining the reservoir type; obtaining the seismic carving volume and producing reserve parameters of the produced block; determining the prediction model parameters according to the relationship between the seismic carving volume and producing reserves of different reservoir types in the developed block; determining the available reserves of the undeveloped block according to the seismic carving volume and reservoir type of the unproduced block, combined with the fitting relationship and key parameters.

[0004] The method and device for determining the available reserves of unproduced blocks of fracture-cavity gas reservoirs disclosed in the patent document can realize the quantitative calculation of the available reserves of unproduced blocks of fracture-cavity carbonate gas reservoirs. However, the data of various influencing factors are highly discrete, and the changing laws between the influencing factors cannot be accurately and intuitively reflected, which affects the optimization of gas reservoirs. Summary of the invention

[0005] In order to overcome the defects of the above-mentioned prior art, the present invention provides a carbonate gas reservoir reserve production effect evaluation method based on BP neural network. The present invention characterizes the reserve production effect by comprehensively considering the reserve production degree and the reserve production quality, and uses the BP neural network to assign weights to indicators of different types of gas reservoirs, thereby making up for the discreteness between the influencing factor data, and being able to more accurately and intuitively reflect the changing laws between the influencing factors. On this basis, a reserve production effect evaluation model is established, which can provide a theoretical reference for the next step of improving the reserve production and recovery rate of gas reservoirs and optimizing the gas reservoirs with mining potential.

[0006] The present invention is achieved through the following technical solutions: The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network is characterized by comprising the following steps: a. Based on the geological characteristics of gas reservoirs, a comprehensive classification of carbonate gas reservoirs is formed according to reservoir type and physical properties; b. Selecting the characteristic index parameters that affect the reserve production effect; c. Use the Pearson coefficient method to determine the factors affecting the reserve production effect of different types of gas reservoirs; d. Carry out quantitative characterization analysis of factors affecting the production effect of gas reservoir reserves through BP neural network, and determine the weight values ​​of different influencing factors; e. Analyze the relationship between reserves production and various influencing factors through nonlinear surface fitting, and establish a quantitative formula for gas reservoir reserves production; f. Use the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, give the output value for model training, calculate the error and adjust the weight value, complete the iteration, combine with the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the gas reservoir reserve production effect evaluation.

[0007] In the step a, the comprehensive classification of carbonate gas reservoirs specifically refers to dividing carbonate gas reservoirs into fracture-pore type gas reservoirs, fracture-cavity type gas reservoirs and fracture-pore type gas reservoirs.

[0008] In the step e, the quantitative formula for gas reservoir reserve production includes a fitting calculation formula for reserve production degree and a fitting calculation formula for reserve production quality; The reserve utilization degree is calculated by fitting through formula 1; Formula 1 In the formula, is the reserve utilization degree, is the permeability, is the effective thickness, is π, is the porosity, is the gas production rate during the stable production period; The reserve production quality is calculated by fitting through formula 2; Formula 2 In the formula, is the reserve production quality, is the permeability, is the effective thickness, is π, is the porosity, It is the gas production rate during the stable production period.

[0009] In the step f, the output values ​​include the reserve production degree and the reserve production quality.

[0010] The reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization.

[0011] The reserve production quality refers to the dynamic reserve extraction degree under the unit pressure drop amplitude in the stage, which is used to characterize the quality of current reserve production.

[0012] In the step f, during the model training process, the transfer function from the input layer to the hidden layer is Formula 3, and the transfer function from the hidden layer to the output layer is Formula 4; Formula 3 In the formula, is the output of the hidden layer to the jth neuron, is the weight value from the input layer to the hidden layer, is the transfer function from the hidden layer to the output layer, is the input mode; Formula 4 In the formula, is the output of the output layer to the kth neuron, is the transfer function of the output layer, is the weight value from the hidden layer to the output layer, is the output of the hidden layer to the jth neuron.

[0013] The step f, calculating the error and adjusting the weight value, refers to calculating the error by using formula 5 and adjusting the weight value by using formula 6; Formula 5 In the formula, is the error between the network output and the target output, is the target output, Output for the network; Formula 6 In the formula, For each weight resize, is the learning rate, is the partial derivative of the error with respect to the weight.

[0014] In the step f, the reserve production effect evaluation method chart is obtained through the normal distribution law of the parameters.

[0015] In step f, the evaluation of gas reservoir reserve production effect includes three levels, specifically: When the reserve production degree is higher than 70% and the reserve production quality is higher than 0.8, the reserve production effect of the gas reservoir is evaluated as good; When the reserve production degree is greater than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is greater than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as poor.

[0016] The BP neural network described in the present invention refers to a multi-layer feedforward neural network trained according to an error back propagation algorithm.

[0017] The beneficial effects of the present invention are mainly manifested in the following aspects: 1. The present invention, a. Based on the geological characteristics of gas reservoirs, according to the reservoir type-physical properties, forms a comprehensive classification of carbonate gas reservoirs; b. Selects characterization index parameters that affect the reserve production effect; c. Uses the Pearson coefficient method to determine the influencing factors that affect the reserve production effect of different types of gas reservoirs; d. Carries out quantitative characterization analysis of the influencing factors of the gas reservoir reserve production effect through BP neural network, and determines the weight values ​​of different influencing factors; e. Through nonlinear surface fitting, analyzes the relationship between reserve production and the change of each influencing factor, and establishes a quantitative formula for gas reservoir reserve production; f. Uses the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, gives the output value for model training, calculates the error and Adjust the weight value, complete the iteration, combine the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the reserve production effect evaluation of gas reservoirs. As a complete technical solution, compared with the existing technology, the reserve production effect is characterized by comprehensively considering the reserve production degree and the reserve production quality, and the BP neural network is used to assign weights to indicators of different types of gas reservoirs, which makes up for the discreteness between the influencing factor data and can more accurately and intuitively reflect the changing laws between the influencing factors. On this basis, a reserve production effect evaluation model is established, which can provide a theoretical reference for the next step of improving the reserve production and recovery rate of gas reservoirs and the optimization of gas reservoirs with mining potential.

[0018] 2. In the present invention, the reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization and can reflect the spread of reserves that have been invested in development.

[0019] 3. In the present invention, the reserve production quality refers to the dynamic reserve recovery degree under the unit pressure drop amplitude of the stage, which is used to characterize the quality of current reserve production and can reflect the reservoir geological conditions and the development level of the corresponding period.

[0020] 4. The present invention adopts a nonlinear surface fitting method to analyze the relationship between reserve production and various influencing factors, establishes a reserve production effect evaluation model, and can achieve quantitative evaluation of the reserve production degree and production quality of developed gas reservoirs.

[0021] 5. The present invention obtains a reserve production effect evaluation model after BP neural network training, which can systematically, scientifically, clearly and accurately evaluate the reserve production effect of gas reservoirs. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments: Figure 1 It is a flowchart of the present invention; Figure 2 It is the fitting curve diagram of the reserve utilization degree of low permeability gas reservoir; Figure 3This is the fitting curve of the reserves production quality of low permeability gas reservoirs. DETAILED DESCRIPTION

[0023] Example 1 See also Figure 1 The carbonate gas reservoir reserve production effect evaluation method based on BP neural network includes the following steps: a. Based on the geological characteristics of gas reservoirs, a comprehensive classification of carbonate gas reservoirs is formed according to reservoir type and physical properties; b. Selecting the characteristic index parameters that affect the reserve production effect; c. Use the Pearson coefficient method to determine the factors affecting the reserve production effect of different types of gas reservoirs; d. Carry out quantitative characterization analysis of factors affecting the production effect of gas reservoir reserves through BP neural network, and determine the weight values ​​of different influencing factors; e. Analyze the relationship between reserves production and various influencing factors through nonlinear surface fitting, and establish a quantitative formula for gas reservoir reserves production; f. Use the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, give the output value for model training, calculate the error and adjust the weight value, complete the iteration, combine with the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the gas reservoir reserve production effect evaluation.

[0024] This embodiment is the most basic implementation method. As a complete technical solution, compared with the prior art, it characterizes the reserve production effect by comprehensively considering the reserve production degree and the reserve production quality, and uses the BP neural network to assign weights to indicators of different types of gas reservoirs, thereby compensating for the discreteness between the influencing factor data and being able to more accurately and intuitively reflect the changing laws between the influencing factors. On this basis, a reserve production effect evaluation model is established, which can provide a theoretical reference for the next step of improving the reserve production and recovery rate of gas reservoirs and the optimization of gas reservoirs with mining potential.

[0025] Example 2 See also Figure 1 The carbonate gas reservoir reserve production effect evaluation method based on BP neural network includes the following steps: a. Based on the geological characteristics of gas reservoirs, a comprehensive classification of carbonate gas reservoirs is formed according to reservoir type and physical properties; b. Selecting the characteristic index parameters that affect the reserve production effect; c. Use the Pearson coefficient method to determine the factors affecting the reserve production effect of different types of gas reservoirs; d. Carry out quantitative characterization analysis of factors affecting the production effect of gas reservoir reserves through BP neural network, and determine the weight values ​​of different influencing factors; e. Analyze the relationship between reserves production and various influencing factors through nonlinear surface fitting, and establish a quantitative formula for gas reservoir reserves production; f. Use the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, give the output value for model training, calculate the error and adjust the weight value, complete the iteration, combine with the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the gas reservoir reserve production effect evaluation.

[0026] In the step a, the comprehensive classification of carbonate gas reservoirs specifically refers to dividing carbonate gas reservoirs into fracture-pore type gas reservoirs, fracture-cavity type gas reservoirs and fracture-pore type gas reservoirs.

[0027] Furthermore, in the step e, the quantitative formula for gas reservoir reserve production includes a fitting calculation formula for reserve production degree and a fitting calculation formula for reserve production quality; The reserve utilization degree is calculated by fitting through formula 1; Formula 1 In the formula, is the reserve utilization degree, is the permeability, is the effective thickness, is π, is the porosity, is the gas production rate during the stable production period; The reserve production quality is calculated by fitting through formula 2; Formula 2 In the formula, is the reserve production quality, is the permeability, is the effective thickness, is π, is the porosity, It is the gas production rate during the stable production period.

[0028] In the step f, the output values ​​include the reserve production degree and the reserve production quality.

[0029] The reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization.

[0030] The reserve production quality refers to the dynamic reserve extraction degree under the unit pressure drop amplitude in the stage, which is used to characterize the quality of current reserve production.

[0031] In the step f, during the model training process, the transfer function from the input layer to the hidden layer is Formula 3, and the transfer function from the hidden layer to the output layer is Formula 4; Formula 3 In the formula, is the output of the hidden layer to the jth neuron, is the weight value from the input layer to the hidden layer, is the transfer function from the hidden layer to the output layer, is the input mode; Formula 4 In the formula, is the output of the output layer to the kth neuron, is the transfer function of the output layer, is the weight value from the hidden layer to the output layer, is the output of the hidden layer to the jth neuron.

[0032] The step f, calculating the error and adjusting the weight value, refers to calculating the error by using formula 5 and adjusting the weight value by using formula 6; Formula 5 In the formula, is the error between the network output and the target output, is the target output, Output for the network; Formula 6 In the formula, For each weight resize, is the learning rate, is the partial derivative of the error with respect to the weight.

[0033] This embodiment is a preferred implementation method. The reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization and can reflect the available amount of reserves that have been invested in development. The reserve utilization quality refers to the dynamic reserve recovery degree under the unit pressure drop amplitude of the stage, which is used to characterize the quality of the current reserve utilization and can reflect the reservoir geological conditions and the development level of the corresponding period.

[0034] Example 3 See also Figure 1 The carbonate gas reservoir reserve production effect evaluation method based on BP neural network includes the following steps: a. Based on the geological characteristics of gas reservoirs, a comprehensive classification of carbonate gas reservoirs is formed according to reservoir type and physical properties; b. Selecting the characteristic index parameters that affect the reserve production effect; c. Use the Pearson coefficient method to determine the factors affecting the reserve production effect of different types of gas reservoirs; d. Carry out quantitative characterization analysis of factors affecting the production effect of gas reservoir reserves through BP neural network, and determine the weight values ​​of different influencing factors; e. Analyze the relationship between reserves production and various influencing factors through nonlinear surface fitting, and establish a quantitative formula for gas reservoir reserves production; f. Use the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, give the output value for model training, calculate the error and adjust the weight value, complete the iteration, combine with the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the gas reservoir reserve production effect evaluation.

[0035] In the step a, the comprehensive classification of carbonate gas reservoirs specifically refers to dividing carbonate gas reservoirs into fracture-pore type gas reservoirs, fracture-cavity type gas reservoirs and fracture-pore type gas reservoirs.

[0036] In the step e, the quantitative formula for gas reservoir reserve production includes a fitting calculation formula for reserve production degree and a fitting calculation formula for reserve production quality; The reserve utilization degree is calculated by fitting through formula 1; Formula 1 In the formula, is the reserve utilization degree, is the permeability, is the effective thickness, is π, is the porosity, is the gas production rate during the stable production period; The reserve production quality is calculated by fitting through formula 2; Formula 2 In the formula, is the reserve production quality, is the permeability, is the effective thickness, is π, is the porosity, It is the gas production rate during the stable production period.

[0037] In the step f, the output values ​​include the reserve production degree and the reserve production quality.

[0038] The reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization.

[0039] The reserve production quality refers to the dynamic reserve extraction degree under the unit pressure drop amplitude in the stage, which is used to characterize the quality of current reserve production.

[0040] In the step f, during the model training process, the transfer function from the input layer to the hidden layer is Formula 3, and the transfer function from the hidden layer to the output layer is Formula 4; Formula 3 In the formula, is the output of the hidden layer to the jth neuron, is the weight value from the input layer to the hidden layer, is the transfer function from the hidden layer to the output layer, is the input mode; Formula 4 In the formula, is the output of the output layer to the kth neuron, is the transfer function of the output layer, is the weight value from the hidden layer to the output layer, is the output of the hidden layer to the jth neuron.

[0041] The step f, calculating the error and adjusting the weight value, refers to calculating the error by using formula 5 and adjusting the weight value by using formula 6; Formula 5 In the formula, is the error between the network output and the target output, is the target output, Output for the network; Formula 6 In the formula, For each weight resize, is the learning rate, is the partial derivative of the error with respect to the weight.

[0042] In the step f, the reserve production effect evaluation method chart is obtained through the normal distribution law of the parameters.

[0043] Furthermore, in step f, the evaluation of gas reservoir reserve production effect includes three levels, specifically: When the reserve production degree is higher than 70% and the reserve production quality is higher than 0.8, the reserve production effect of the gas reservoir is evaluated as good; When the reserve production degree is greater than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is greater than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as poor.

[0044] This embodiment is the best implementation method. It uses a nonlinear surface fitting method to analyze the relationship between reserve production and various influencing factors, establishes a reserve production effect evaluation model, and can achieve quantitative evaluation of the reserve production degree and production quality of developed gas reservoirs.

[0045] The reserve production effect evaluation model is obtained through BP neural network training, which can systematically, scientifically, clearly and accurately evaluate the reserve production effect of gas reservoirs.

[0046] The principle of the present invention is as follows: Firstly, the multi-layer perceptron neural network method is used to determine the indicator weight values ​​based on deep learning of big data neural network, which serves as the basis for model establishment. Secondly, the nonlinear surface fitting method is used to establish a quantitative evaluation model for relevant dynamic and static influencing factors, reserve production degree, and reserve production quality. Finally, the reserve production effect evaluation model is obtained after BP neural network training to realize the reserve production effect evaluation of developed gas reservoirs and undeveloped gas reservoirs.

[0047] The Longmen Feixianguan gas reservoir, Wushankan Carboniferous gas reservoir and Moxi Leikoupo gas reservoir were selected to verify the reserve production effect evaluation model established by the present invention. The error between the predicted value and the actual value is within 3%, and the fitting degree is above 96%, indicating that the established model is relatively reliable and can be used for the analysis of the evaluation system and optimization indicators, see Table 3, Figure 2 and Figure 3 .

[0048]

Claims

1. A method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network, characterized in that: The following steps are involved: a. Based on the geological characteristics of gas reservoirs, a comprehensive classification of carbonate gas reservoirs is formed according to reservoir type and physical properties; b. Selecting the characteristic index parameters that affect the reserve production effect; c. Use the Pearson coefficient method to determine the factors affecting the reserve production effect of different types of gas reservoirs; d. Carry out quantitative characterization analysis of factors affecting the production effect of gas reservoir reserves through BP neural network, and determine the weight values ​​of different influencing factors; e. Analyze the relationship between reserves production and various influencing factors through nonlinear surface fitting, and establish a quantitative formula for gas reservoir reserves production; f. Use the BP neural network algorithm to import the dynamic and static parameters of the gas reservoir, give the output value for model training, calculate the error and adjust the weight value, complete the iteration, combine the reserve production effect evaluation method chart, establish the reserve production effect evaluation model, and complete the gas reservoir reserve production effect evaluation.

2. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In the step a, the comprehensive classification of carbonate gas reservoirs specifically refers to dividing carbonate gas reservoirs into fracture-pore type gas reservoirs, fracture-cavity type gas reservoirs and fracture-pore type gas reservoirs.

3. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In the step e, the quantitative formula for gas reservoir reserve production includes a fitting calculation formula for reserve production degree and a fitting calculation formula for reserve production quality; The reserve utilization degree is calculated by fitting through formula 1; Formula 1 In the formula, is the reserve utilization degree, is the permeability, is the effective thickness, is π, is the porosity, is the gas production rate during the stable production period; The reserve production quality is calculated by fitting through formula 2; Formula 2 In the formula, is the reserve production quality, is the permeability, is the effective thickness, is π, is the porosity, It is the gas production rate during the stable production period.

4. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In the step f, the output values ​​include the reserve production degree and the reserve production quality.

5. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 4 is characterized in that: The reserve utilization degree refers to the ratio of dynamic reserves to geological reserves, which is used to characterize the current reserve utilization.

6. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 4, characterized in that: The reserve production quality refers to the dynamic reserve extraction degree under the unit pressure drop amplitude in the stage, which is used to characterize the quality of current reserve production.

7. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In the step f, during the model training process, the transfer function from the input layer to the hidden layer is Formula 3, and the transfer function from the hidden layer to the output layer is Formula 4; Formula 3 In the formula, is the output of the hidden layer to the jth neuron, is the weight value from the input layer to the hidden layer, is the transfer function from the hidden layer to the output layer, is the input mode; Formula 4 In the formula, is the output of the output layer to the kth neuron, is the transfer function of the output layer, is the weight value from the hidden layer to the output layer, is the output of the hidden layer to the jth neuron.

8. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: The step f, calculating the error and adjusting the weight value, refers to calculating the error by using formula 5 and adjusting the weight value by using formula 6; Formula 5 In the formula, is the error between the network output and the target output, is the target output, Output for the network; Formula 6 In the formula, For each weight resize, is the learning rate, is the partial derivative of the error with respect to the weight.

9. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In the step f, the reserve production effect evaluation method chart is obtained through the normal distribution law of the parameters.

10. The method for evaluating the production effect of carbonate gas reservoir reserves based on BP neural network according to claim 1, characterized in that: In step f, the evaluation of gas reservoir reserve production effect includes three levels, specifically: When the reserve production degree is higher than 70% and the reserve production quality is higher than 0.8, the reserve production effect of the gas reservoir is evaluated as good; When the reserve production degree is greater than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is greater than 0.8, the reserve production effect of the gas reservoir is evaluated as medium; When the reserve production degree is less than 70% and the reserve production quality is less than 0.8, the reserve production effect of the gas reservoir is evaluated as poor.

Images