Shale gas yield prediction method and system

Abstract

The invention provides a shale gas yield prediction method and system, and the method comprises the steps: carrying out the collection and data preprocessing of geological and engineering data, and obtaining the preprocessed geological and engineering data; acquiring the data association strength of the current geological and engineering data according to the preprocessed geological and engineering data; adjusting feature weights corresponding to the geological and engineering data according to the data association strength of the current geological and engineering data; fusing the feature data corresponding to the current geological and engineering data by using the adjusted feature weight to obtain fused feature data; and inputting the fused feature data into the shale gas yield prediction model which has completed construction and training verification, and obtaining a shale gas yield prediction result. The system comprises modules corresponding to the steps of the method.

Description

Technical Field

[0001] This invention proposes a method and system for predicting shale gas production, belonging to the field of shale gas production prediction technology. Background Technology

[0002] Shale gas, as an important unconventional natural gas resource, plays a crucial role in optimizing the energy supply structure through its efficient development and utilization. Accurate shale gas production forecasting provides key information for development planning and rational resource allocation, helping to improve the economics and recovery rate of shale gas development. However, current shale gas production forecasting faces numerous challenges. Shale gas reservoirs have complex geological conditions, with reservoir properties, gas content, and other geological factors intertwined with engineering factors such as fracturing parameters and horizontal well trajectories, resulting in complex nonlinear correlations between geological and engineering data. Existing technologies struggle to accurately characterize the dynamic differences in the impact of various geological and engineering factors on production when processing this data: on the one hand, the lack of effective quantification of the correlation strength between geological and engineering data leads to unreasonable feature weight allocation, failing to adapt to the data characteristics of different shale gas reservoirs or development stages; on the other hand, the simplistic data fusion methods fail to fully utilize the correlations between data to optimize feature expression, making the feature data input into the prediction model unable to accurately reflect the production impact mechanism, ultimately affecting the accuracy and reliability of the prediction model and making it difficult to meet the needs of refined production forecasting in shale gas development.

[0003] Therefore, there is an urgent need for a shale gas production prediction method that can accurately quantify data correlation, dynamically adjust feature weights, and optimize feature fusion, so as to improve the accuracy of production prediction and provide strong technical support for the efficient development of shale gas. Summary of the Invention

[0004] This invention provides a shale gas production prediction method and system to solve the aforementioned technical problems in the prior art. The technical solution adopted is as follows:

[0005] A method for predicting shale gas production, the method comprising:

[0006] Collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data;

[0007] The data correlation strength of the current geological and engineering data is obtained based on the preprocessed geological and engineering data;

[0008] The feature weights corresponding to the geological and engineering data are adjusted based on the data correlation strength of the current geological and engineering data;

[0009] The feature data corresponding to the current geological and engineering data are fused using the adjusted feature weights to obtain the fused feature data;

[0010] The fused feature data is input into the shale gas production prediction model that has been constructed and trained to obtain shale gas production prediction results.

[0011] Furthermore, geological and engineering data are collected and preprocessed to obtain preprocessed geological and engineering data, including:

[0012] Geological and engineering data are collected to obtain geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data;

[0013] The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient.

[0014] The geological and engineering data are subjected to noise reduction and data cleaning processes to obtain the denoised and cleaned geological and engineering data.

[0015] Furthermore, based on the preprocessed geological and engineering data, the data correlation strength of the current geological and engineering data is obtained, including:

[0016] Real-time access to daily gas and water production, and obtaining the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days;

[0017] When the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, the permeability, fracture density, temperature, and pressure gradient corresponding to the geological location at which the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold are retrieved as reference data.

[0018] The first data correlation strength between dynamic data, static data, and environmental data is obtained by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data.

[0019] The first data association strength is obtained by the following formula:

[0020]

[0021] Among them, S 01 Indicates the first data correlation strength; n represents the number of times the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold; T gi and P gi λ represents the normalized temperature and pressure gradients when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively; 01 and λ 02Ri and Rb represent the sensitivity coefficients of the influence of temperature and pressure gradients on dynamic data, respectively, with values ​​ranging from 0.5 to 1.4 and from 0.3 to 1.2. gi and H gi Q represents the normalized permeability and fracture density at the geological location when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold; qi and Q si These represent the rate of change of daily gas production and daily water production when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively.

[0022] By utilizing the permeability and fracture density, as well as temperature and pressure gradients at the geological location where the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold, the second data correlation strength between static data and environmental data is obtained.

[0023] The second data association strength is obtained using the following formula:

[0024]

[0025] Among them, S 02 Indicates the strength of the second data association; λ 03 and λ 04 The values ​​represent the influence intensity coefficients of environmental data on permeability and crack density, respectively, with values ​​ranging from 0.4 to 1.2 and from 0.7 to 1.3.

[0026] Furthermore, the feature weights corresponding to the geological and engineering data are adjusted based on the data correlation strength of the current geological and engineering data, including:

[0027] Retrieve the preset feature weights corresponding to each feature parameter contained in the static data;

[0028] The preset feature weights corresponding to each feature parameter contained in the static data are adjusted using the first data association strength and the second data association strength, so as to obtain the adjusted feature weights corresponding to each feature parameter contained in the static data.

[0029] The adjusted feature weights of the static data are obtained using the following formula:

[0030]

[0031] Where J represents the adjusted feature weight corresponding to each feature parameter contained in the static data; J0 represents the preset feature weight corresponding to each feature parameter contained in the static data; S 01 Indicates the first data association strength; S02 Indicates the strength of the second data association;

[0032] Retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data;

[0033] The preset feature weights corresponding to each feature parameter contained in the dynamic data are adjusted using the first data association strength, and the adjusted feature weights corresponding to each feature parameter contained in the dynamic data are obtained.

[0034] The adjusted feature weights of the dynamic data are obtained using the following formula:

[0035] D = D0 * (1 + S) 01 )

[0036] Where D represents the adjusted feature weight corresponding to each feature parameter contained in the dynamic data; D0 represents the preset feature weight corresponding to each feature parameter contained in the dynamic data.

[0037] Retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data;

[0038] The preset feature weights corresponding to each feature parameter contained in the environmental data are adjusted using the second data association strength to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

[0039] The adjusted feature weights of the environmental data are obtained using the following formula:

[0040]

[0041] Where G represents the adjusted feature weight corresponding to each feature parameter contained in the environmental data; G0 represents the preset feature weight corresponding to each feature parameter contained in the environmental data.

[0042] Furthermore, the adjusted feature weights are used to fuse the feature data corresponding to the current geological and engineering data to obtain the fused feature data, including:

[0043] Retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data;

[0044] The feature parameters contained in the static data, dynamic data, and environmental data are combined with their corresponding adjusted feature weights and weighted to generate fused feature data.

[0045] A shale gas production prediction system, the shale gas production prediction system comprising:

[0046] The data collection and preprocessing module is used to collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data.

[0047] The data association strength acquisition module is used to obtain the data association strength of the current geological and engineering data based on the preprocessed geological and engineering data;

[0048] The weight adjustment module is used to adjust the feature weights corresponding to the geological and engineering data according to the data correlation strength of the current geological and engineering data;

[0049] The fusion module is used to fuse the feature data corresponding to the current geological and engineering data using the adjusted feature weights, and obtain the fused feature data.

[0050] The prediction module is used to input the fused feature data into the shale gas production prediction model that has been constructed and trained to obtain the shale gas production prediction results.

[0051] Furthermore, the data collection and preprocessing module includes:

[0052] The collection module is used to collect geological and engineering data, and to acquire geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data;

[0053] The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient.

[0054] The preprocessing module is used to perform noise reduction and data cleaning on the geological and engineering data to obtain the noise-reduced and cleaned geological and engineering data.

[0055] Furthermore, the data association strength acquisition module includes:

[0056] Try using the retrieval module to retrieve daily gas and water production in real time, and obtain the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days.

[0057] The reference data acquisition module is used to retrieve the permeability, fracture density, temperature and pressure gradient corresponding to the geological location when the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold, as reference data.

[0058] The first data association strength acquisition module is used to acquire the first data association strength between dynamic data, static data, and environmental data by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data.

[0059] The first data association strength is obtained by the following formula:

[0060]

[0061] Among them, S 01 Indicates the first data correlation strength; n represents the number of times the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold; T gi and P gi λ represents the normalized temperature and pressure gradients when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively; 01 and λ 02 Ri and Rb represent the sensitivity coefficients of the influence of temperature and pressure gradients on dynamic data, respectively, with values ​​ranging from 0.5 to 1.4 and 0.3 to 1.2. gi and H gi Q represents the normalized permeability and fracture density at the geological location when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold; qi and Q si These represent the rate of change of daily gas production and daily water production when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively.

[0062] The second data association strength acquisition module is used to obtain the second data association strength between static data and environmental data by utilizing the permeability and fracture density, as well as temperature and pressure gradients, corresponding to the geological location when the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold.

[0063] The second data association strength is obtained using the following formula:

[0064]

[0065] Among them, S 02 Indicates the strength of the second data association; λ 03 and λ 04 The values ​​represent the influence intensity coefficients of environmental data on permeability and crack density, respectively, with values ​​ranging from 0.4 to 1.2 and from 0.7 to 1.3.

[0066] Furthermore, the weight adjustment module includes:

[0067] The first weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the static data.

[0068] The first feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the static data using the first data association strength and the second data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the static data.

[0069] The adjusted feature weights of the static data are obtained using the following formula:

[0070]

[0071] Where J represents the adjusted feature weight corresponding to each feature parameter contained in the static data; J0 represents the preset feature weight corresponding to each feature parameter contained in the static data; S 01 Indicates the first data association strength; S 02 Indicates the strength of the second data association;

[0072] The second weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data.

[0073] The second feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the dynamic data using the first data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the dynamic data.

[0074] The adjusted feature weights of the dynamic data are obtained using the following formula:

[0075] D = D0 * (1 + S) 01 )

[0076] Where D represents the adjusted feature weight corresponding to each feature parameter contained in the dynamic data; D0 represents the preset feature weight corresponding to each feature parameter contained in the dynamic data.

[0077] The third weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data.

[0078] The third feature weight adjustment module is used to adjust the preset feature weights corresponding to each feature parameter contained in the environmental data using the second data association strength, and to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

[0079] The adjusted feature weights of the environmental data are obtained using the following formula:

[0080]

[0081] Where G represents the adjusted feature weight corresponding to each feature parameter contained in the environmental data; G0 represents the preset feature weight corresponding to each feature parameter contained in the environmental data.

[0082] Furthermore, the fusion module includes:

[0083] The feature weight retrieval module is used to retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data.

[0084] The feature data fusion module is used to combine the feature parameters contained in the static data, dynamic data and environmental data with their corresponding adjusted feature weights to generate fused feature data.

[0085] Beneficial effects of this invention:

[0086] This invention proposes a shale gas production prediction method and system that dynamically adjusts weights and optimizes feature fusion through correlation strength analysis. This makes the features of the input model more closely match the logic of production impact, reducing errors caused by unreasonable weights and insufficient feature utilization in traditional methods, and improving prediction accuracy. It effectively captures the complex correlation between geological and engineering data, breaking through the limitations of single, static analysis, and adapting to the complex scenarios of intertwined geological and engineering factors in shale gas reservoirs, thus enhancing the method's universality. Through weight adjustment and feature fusion, it provides the prediction model with more representative and information-rich inputs, helping the model to perform better and providing reliable data support for production assessment and scheme optimization in shale gas development. Attached Figure Description

[0087] Figure 1 This is a flowchart of the method described in this invention;

[0088] Figure 2 This is a system block diagram of the system described in this invention. Detailed Implementation

[0089] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0090] This invention proposes a method for predicting shale gas production, such as... Figure 1 As shown, the shale gas production prediction method includes:

[0091] Collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data;

[0092] The data correlation strength of the current geological and engineering data is obtained based on the preprocessed geological and engineering data;

[0093] The feature weights corresponding to the geological and engineering data are adjusted based on the data correlation strength of the current geological and engineering data;

[0094] The feature data corresponding to the current geological and engineering data are fused using the adjusted feature weights to obtain the fused feature data;

[0095] The fused feature data is input into the shale gas production prediction model that has been constructed and trained to obtain shale gas production prediction results.

[0096] Specifically, the adjusted feature weights are used to fuse the feature data corresponding to the current geological and engineering data to obtain the fused feature data, including:

[0097] Retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data;

[0098] The feature parameters contained in the static data, dynamic data, and environmental data are combined with their corresponding adjusted feature weights and weighted to generate fused feature data.

[0099] Meanwhile, the shale gas production prediction model is a prediction model using a convolutional neural network, and its structure is as follows:

[0100] Input layer: Receives preprocessed geological, engineering and environmental data (such as permeability, daily gas production, temperature gradient, etc.);

[0101] 1D convolutional layer 1; extracting local time series features;

[0102] Pooling layer 1: Reduces data dimensionality, decreases computational load, and retains key features.

[0103] 1D convolutional layer 2; further extract deep features (such as the composite relationship between temperature gradient and yield);

[0104] Fully connected layer; integrates features extracted from convolutional layers and performs non-linear mapping;

[0105] Output layer; outputs predicted shale gas production values.

[0106] The working principle of the above technical solution is as follows: First, geological and engineering data are collected and preprocessed, including noise removal and missing data filling, to provide high-quality data for subsequent analysis. Based on the preprocessed data, the correlation between geological and engineering data is mined, the degree of correlation is quantified, and the potential relationship between different data on production is clarified. According to the correlation strength, the weights of corresponding features of geological and engineering data are optimized, with higher weights assigned to features with strong correlations to highlight key influencing factors. The adjusted weights are used to fuse feature data, integrating multi-dimensional information to form a fused feature that more accurately reflects the production impact mechanism. The fused feature is input into a trained prediction model, and using model algorithms (such as machine learning, mechanism-data fusion, etc.) to map relationships, the shale gas production prediction results are output.

[0107] The above technical solution achieves the following effects: By dynamically adjusting weights and optimizing feature fusion through correlation strength analysis, the features of the input model are made more closely aligned with the logic of production impact, reducing errors caused by unreasonable weights and insufficient feature utilization in traditional methods, thus improving prediction accuracy. It effectively captures the complex correlations between geological and engineering data, breaking through the limitations of single, static analysis, adapting to the complex scenarios of intertwined geological and engineering factors in shale gas reservoirs, and enhancing the method's universality. Through weight adjustment and feature fusion, it provides the prediction model with more representative and information-rich inputs, helping the model to perform better and providing reliable data support for production assessment and scheme optimization in shale gas development.

[0108] Meanwhile, existing technologies typically rely on manually set or fixed-weight machine learning models, which cannot adapt to dynamic changes in reservoir and production conditions. The above-mentioned technical solution dynamically adjusts feature weights through data correlation strength, enabling the model to respond in real-time to changes in the correlation between parameters such as permeability and fracture density and production. Existing technologies may simply stitch together multi-source data without considering the interactions between data. The above-mentioned technical solution achieves deep data fusion by calculating the correlation strength between static data (geological parameters), dynamic data (production parameters), and environmental data (temperature, pressure). Existing technologies perform poorly in complex reservoirs with high heterogeneity, high temperature, and high pressure. The above-mentioned technical solution enhances the model's adaptability to complex reservoir conditions through dynamic weight adjustment and multi-source data correlation. Existing technologies may lead to unclear decision-making basis due to model complexity (such as the black box nature of deep learning). The above-mentioned technical solution provides clear logical basis through correlation strength and weight adjustment formulas. Existing technologies may lead to invalid operations due to the inability to identify key parameters. The above-mentioned technical solution guides the prioritization of engineering measures through the identification of high-weight parameters.

[0109] One embodiment of the present invention involves collecting and preprocessing geological and engineering data to obtain preprocessed geological and engineering data, including:

[0110] Collect geological and engineering data to obtain geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data;

[0111] The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient.

[0112] The geological and engineering data are subjected to noise reduction and data cleaning processes to obtain the denoised and cleaned geological and engineering data.

[0113] The effects of the above technical solutions are as follows: By removing noise interference, data curves are smoother, more accurately reflecting reservoir and production dynamics. After filling missing values ​​and correcting outliers, data integrity is significantly improved, avoiding analytical biases caused by missing or incorrect data. Unifying data units makes model training more stable, avoiding weight imbalances caused by differences in units. For example, in production prediction models, standardized permeability, daily gas production, and temperature data can participate fairly in feature weight allocation. Noise reduction and cleaning reduce random fluctuations and errors in the data, lowering the model's sensitivity to noise and improving generalization ability. Preprocessed data more closely approximates the actual reservoir and production state, providing a reliable foundation for subsequent correlation strength calculations and production prediction. Through noise reduction and cleaning, the characteristics of static, dynamic, and environmental data are clearer, helping the model capture key parameters (such as the correlation between permeability and production) and improve prediction accuracy. Noise reduction and cleaning reduce the amount of data (e.g., removing outliers), lowering computational complexity and accelerating subsequent analysis processes. Standardized data is easier to interpret, allowing engineers to intuitively understand the contribution of each parameter to production.

[0114] One embodiment of the present invention involves obtaining the data correlation strength of current geological and engineering data based on preprocessed geological and engineering data, including:

[0115] Real-time access to daily gas and water production, and obtaining the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days;

[0116] When the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, the permeability, fracture density, temperature, and pressure gradient corresponding to the geological location at which the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold are retrieved as reference data.

[0117] The first data correlation strength between dynamic data, static data, and environmental data is obtained by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data.

[0118] The first data association strength is obtained by the following formula:

[0119]

[0120] Among them, S 01 Indicates the first data correlation strength; n represents the number of times the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold; T gi and P gi λ represents the normalized temperature and pressure gradients when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively; 01 and λ 02 Ri and Rb represent the sensitivity coefficients of the influence of temperature and pressure gradients on dynamic data, respectively, with values ​​ranging from 0.5 to 1.4 and 0.3 to 1.2. gi and H gi Q represents the normalized permeability and fracture density at the geological location when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold; qi and Q si These represent the rate of change of daily gas production and daily water production when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively.

[0121] By utilizing the permeability and fracture density, as well as temperature and pressure gradients at the geological location where the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold, the second data correlation strength between static data and environmental data is obtained.

[0122] The second data association strength is obtained using the following formula:

[0123]

[0124] Among them, S 02 Indicates the strength of the second data association; λ 03 and λ 04 The values ​​represent the influence intensity coefficients of environmental data on permeability and crack density, respectively, with values ​​ranging from 0.4 to 1.2 and from 0.7 to 1.3.

[0125] The working principle of the above technical solution is as follows: The system continuously monitors and acquires daily gas production and water production data in real time. For each consecutive two days, the system calculates the rate of change of daily gas production and water production to capture short-term fluctuations in production dynamics. Preset thresholds are set for the rate of change of daily gas production and water production. When the rate of change of either exceeds this threshold, the system determines it as an abnormal change and triggers subsequent data retrieval and analysis processes. When an abnormal change is detected, the system automatically retrieves geological and engineering data corresponding to the geological location where the abnormal change occurred, including permeability, fracture density, temperature, and pressure gradient. This data is used as reference data for subsequent correlation strength analysis.

[0126] The above technical solution achieves the following effect: by real-time monitoring and calculation of the rate of change in daily gas and water production, the system can accurately capture dynamic changes during the production process, providing timely and accurate data support for subsequent analysis. This is achieved by introducing a first data correlation strength S. 01 Second data correlation strength S 02 The system can quantify the correlation strength between dynamic and static data, environmental data, and the correlation strength between static and environmental data. This quantitative analysis helps to gain a deeper understanding of the complex heterogeneity and multi-field coupling effects of shale gas reservoirs. By incorporating correlation strength into the production prediction model, the system can more accurately reflect the combined impact of geological, engineering, and environmental factors on shale gas production. This helps improve the accuracy of production prediction and provides a reliable basis for the formulation and optimization of development plans. By quantifying data correlation strength, the system can reveal the intrinsic connections and interaction mechanisms between different parameters, thereby enhancing the interpretability of the production prediction model. This helps engineers better understand the model's prediction results and improve the scientific and rational nature of decision-making. This technical solution, through dynamically adjusting feature weights and fusing multi-source data, can adapt to shale gas reservoirs with different geological conditions and development stages. This helps improve the model's generalization ability and adaptability, enabling it to play a role in a wider range of application scenarios.

[0127] Traditional methods typically rely on static or periodic sampling data for analysis, making it difficult to capture dynamic changes during the production process in real time (such as sudden fluctuations in gas production or abrupt changes in formation parameters). The above-mentioned technical solution calculates the rate of change of daily gas and water production, enabling immediate detection of abnormal fluctuations (such as a sudden increase in production after fracturing or formation water breakthrough). Correlation analysis is triggered only when the rate of change exceeds a threshold, avoiding ineffective calculations and improving efficiency. If the rate of change of gas production in a well suddenly increases from 5% to 20% on a given day, the system automatically retrieves the geological parameters (permeability, fracture density) and environmental parameters (temperature, pressure gradient) at that moment to quickly pinpoint the cause of the anomaly (such as fracture propagation or changes in formation fluid properties). Existing methods often rely on empirical formulas or simple correlation analysis, failing to quantify the nonlinear interactions between geological, engineering, and environmental parameters. The advantages of the above-mentioned technical solution: First correlation strength S01 Used to quantify the correlation strength between dynamic data (gas production / water production change rate) and static data (permeability, fracture density) and environmental data (temperature, pressure gradient). Second correlation strength S 02 Quantify the correlation strength between static data (permeability, crack density) and environmental data. This is achieved through normalization and the sensitivity coefficient (λ). 01 -λ 04 This eliminates dimensional differences and achieves unified quantification of multi-source data. If the correlation strength S01 between the permeability and gas production change rate of a certain well is significantly higher than that of other wells, it can be inferred that the permeability of this well contributes more to the production, and the fracturing parameters should be optimized first. Traditional models ignore parameter correlation, resulting in a large deviation between the predicted results and actual production data. Feature weight adjustment driven by correlation strength: based on S... 01 and S 02 The weights of geological, engineering, and environmental parameters are dynamically adjusted to enhance the model's sensitivity to key parameters. Normalization and sensitivity intensity coefficients are used to ensure the model conforms to shale gas flow patterns (e.g., increased temperature may enhance the positive correlation between permeability and production). If the sensitivity intensity coefficient λ of a well's temperature gradient on its production... 01 With a higher λ, the model will focus more on the impact of temperature changes on yield, thereby improving prediction accuracy. Black-box models such as deep learning struggle to explain the relationship between parameters and yield. The above technical solution clarifies the contribution of each parameter to yield through S01 and S02, making it easier for engineers to understand the model's prediction results. 01 -λ 04 The value range (e.g., 0.5-1.4, 0.3-1.2) provides an intuitive reference for the degree of influence of the parameter. If the correlation strength S between fracture density and production in a well is... 01 The strength coefficient λ, which is relatively high and sensitive to crack density in terms of production, is also high. 02 Given the relatively large fracture density, engineers can infer that it is a key factor affecting production, necessitating priority optimization of the fracturing process. Traditional methods predict failure under extreme conditions such as high temperature and pressure, lacking a dynamic response to environmental factors. The aforementioned technical solution quantifies the asymmetric impact of environmental factors on parameter correlations through normalization of temperature and pressure gradients and the use of sensitivity intensity coefficients. When environmental parameters change (e.g., temperature increases), the model automatically adjusts parameter weights to maintain prediction accuracy. In high-temperature reservoirs, if the sensitivity intensity coefficient λ of the temperature gradient on production... 01 The higher the temperature, the more the model will focus on the impact of temperature changes on production, avoiding prediction failure due to environmental changes. Retraining the model with full data is computationally expensive and difficult to update in real time. The above-mentioned technical solution uses triggered data retrieval, only retrieving data when the rate of change exceeds a threshold, reducing unnecessary calculations. A dynamic adjustment mechanism based on correlation strength avoids retraining with full data, reducing computational costs. If the rate of change in gas production of a well does not exceed a threshold for several consecutive days, updates are triggered only during abnormal fluctuations, significantly improving efficiency.

[0128] One embodiment of the present invention adjusts the feature weights corresponding to the geological and engineering data based on the data association strength of the current geological and engineering data, including:

[0129] Retrieve the preset feature weights corresponding to each feature parameter contained in the static data;

[0130] The preset feature weights corresponding to each feature parameter contained in the static data are adjusted using the first data association strength and the second data association strength, so as to obtain the adjusted feature weights corresponding to each feature parameter contained in the static data.

[0131] The adjusted feature weights of the static data are obtained using the following formula:

[0132]

[0133] Where J represents the adjusted feature weight corresponding to each feature parameter contained in the static data; J0 represents the preset feature weight corresponding to each feature parameter contained in the static data; S 01 Indicates the first data association strength; S 02 Indicates the strength of the second data association;

[0134] Retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data;

[0135] The preset feature weights corresponding to each feature parameter contained in the dynamic data are adjusted using the first data association strength, and the adjusted feature weights corresponding to each feature parameter contained in the dynamic data are obtained.

[0136] The adjusted feature weights of the dynamic data are obtained using the following formula:

[0137] D = D0 * (1 + S) 01 )

[0138] Where D represents the adjusted feature weight corresponding to each feature parameter contained in the dynamic data; D0 represents the preset feature weight corresponding to each feature parameter contained in the dynamic data.

[0139] Retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data;

[0140] The preset feature weights corresponding to each feature parameter contained in the environmental data are adjusted using the second data association strength to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

[0141] The adjusted feature weights of the environmental data are obtained using the following formula:

[0142]

[0143] Where G represents the adjusted feature weight corresponding to each feature parameter contained in the environmental data; G0 represents the preset feature weight corresponding to each feature parameter contained in the environmental data.

[0144] The working principle of the above technical solution is as follows: The core of this solution is to achieve deep fusion of multi-source data correlation and adaptive model optimization by dynamically adjusting the feature weights of geological and engineering data. First, the system pre-assigns initial weights to each feature parameter in static data (such as permeability and fracture density), dynamic data (such as daily gas production and water production), and environmental data (such as temperature and pressure gradients). These weights are set based on historical experience or domain knowledge. Subsequently, the correlation strength (S) between the two types of data is calculated. 01 and S 02 Driving weight adjustment: S 01 Quantify the correlation between dynamic data and static / environmental data, S 02 Quantify the correlation between static data and environmental data. During the adjustment process, the weight of static data is simultaneously affected by S. 01 and S 02 The weights of dynamic data are only affected by S. 01 The influence, while the weight of environmental data is only affected by S 02 Influence.

[0145] The above technical solution achieves the following results: It significantly improves the accuracy, adaptability, and engineering value of shale gas production prediction through a dynamic weight adjustment mechanism. Firstly, it enhances prediction accuracy: weight adjustment, based on real-time calculated data correlation strength, enables the model to dynamically respond to changes in reservoir and production conditions. Secondly, by fusing the correlation strength of multi-source data, the model can adapt to complex reservoir conditions (such as high temperature and pressure, and high heterogeneity), reducing dependence on a single data source and avoiding prediction biases caused by changes in reservoir conditions. The mathematical formula and correlation strength of the weight adjustment provide a clear logical basis for the model, allowing engineers to intuitively understand the reasons for changes in the weights of each parameter. For example, high-weight parameters directly point to key factors affecting production, providing a scientific basis for engineering decisions. Finally, it optimizes resource allocation: the weight adjustment results can guide the prioritization of engineering measures. This technical solution achieves efficient utilization of multi-source data through dynamic weight allocation, providing a full-chain optimization capability from data to decision for the efficient development of unconventional oil and gas reservoirs.

[0146] The invention provides a shale gas production prediction system, such as... Figure 2 As shown, the shale gas production prediction system includes:

[0147] The data collection and preprocessing module is used to collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data.

[0148] The data association strength acquisition module is used to obtain the data association strength of the current geological and engineering data based on the preprocessed geological and engineering data;

[0149] The weight adjustment module is used to adjust the feature weights corresponding to the geological and engineering data according to the data correlation strength of the current geological and engineering data;

[0150] The fusion module is used to fuse the feature data corresponding to the current geological and engineering data using the adjusted feature weights, and obtain the fused feature data.

[0151] The prediction module is used to input the fused feature data into the shale gas production prediction model that has been constructed and trained to obtain the shale gas production prediction results.

[0152] The fusion module includes:

[0153] The feature weight retrieval module is used to retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data.

[0154] The feature data fusion module is used to combine the feature parameters contained in the static data, dynamic data and environmental data with their corresponding adjusted feature weights to generate fused feature data.

[0155] The working principle of the above technical solution is as follows: First, geological and engineering data are collected and preprocessed, including noise removal and missing data filling, to provide high-quality data for subsequent analysis. Based on the preprocessed data, the correlation between geological and engineering data is mined, the degree of correlation is quantified, and the potential relationship between different data on production is clarified. According to the correlation strength, the weights of corresponding features of geological and engineering data are optimized, with higher weights assigned to features with strong correlations to highlight key influencing factors. The adjusted weights are used to fuse feature data, integrating multi-dimensional information to form a fused feature that more accurately reflects the production impact mechanism. The fused feature is input into a trained prediction model, and using model algorithms (such as machine learning, mechanism-data fusion, etc.) to map relationships, the shale gas production prediction results are output.

[0156] The above technical solution achieves the following effects: By dynamically adjusting weights and optimizing feature fusion through correlation strength analysis, the features of the input model are made more closely aligned with the logic of production impact, reducing errors caused by unreasonable weights and insufficient feature utilization in traditional methods, thus improving prediction accuracy. It effectively captures the complex correlations between geological and engineering data, breaking through the limitations of single, static analysis, adapting to the complex scenarios of intertwined geological and engineering factors in shale gas reservoirs, and enhancing the method's universality. Through weight adjustment and feature fusion, it provides the prediction model with more representative and information-rich inputs, helping the model to perform better and providing reliable data support for production assessment and scheme optimization in shale gas development.

[0157] Meanwhile, existing technologies typically rely on manually set or fixed-weight machine learning models, which cannot adapt to dynamic changes in reservoir and production conditions. The above-mentioned technical solution dynamically adjusts feature weights through data correlation strength, enabling the model to respond in real-time to changes in the correlation between parameters such as permeability and fracture density and production. Existing technologies may simply stitch together multi-source data without considering the interactions between data. The above-mentioned technical solution achieves deep data fusion by calculating the correlation strength between static data (geological parameters), dynamic data (production parameters), and environmental data (temperature, pressure). Existing technologies perform poorly in complex reservoirs with high heterogeneity, high temperature, and high pressure. The above-mentioned technical solution enhances the model's adaptability to complex reservoir conditions through dynamic weight adjustment and multi-source data correlation. Existing technologies may lead to unclear decision-making basis due to model complexity (such as the black box nature of deep learning). The above-mentioned technical solution provides clear logical basis through correlation strength and weight adjustment formulas. Existing technologies may lead to invalid operations due to the inability to identify key parameters. The above-mentioned technical solution guides the prioritization of engineering measures through the identification of high-weight parameters.

[0158] In one embodiment of the present invention, the data collection and preprocessing module includes:

[0159] The collection module is used to collect geological and engineering data, and to acquire geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data;

[0160] The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient.

[0161] The preprocessing module is used to perform noise reduction and data cleaning on the geological and engineering data to obtain the denoised and cleaned geological and engineering data.

[0162] The effects of the above technical solutions are as follows: By removing noise interference, data curves are smoother, more accurately reflecting reservoir and production dynamics. After filling missing values ​​and correcting outliers, data integrity is significantly improved, avoiding analytical biases caused by missing or incorrect data. Unifying data units makes model training more stable, avoiding weight imbalances caused by differences in units. For example, in production prediction models, standardized permeability, daily gas production, and temperature data can participate fairly in feature weight allocation. Noise reduction and cleaning reduce random fluctuations and errors in the data, lowering the model's sensitivity to noise and improving generalization ability. Preprocessed data more closely approximates the actual reservoir and production state, providing a reliable foundation for subsequent correlation strength calculations and production prediction. Through noise reduction and cleaning, the characteristics of static, dynamic, and environmental data are clearer, helping the model capture key parameters (such as the correlation between permeability and production) and improve prediction accuracy. Noise reduction and cleaning reduce the amount of data (e.g., removing outliers), lowering computational complexity and accelerating subsequent analysis processes. Standardized data is easier to interpret, allowing engineers to intuitively understand the contribution of each parameter to production.

[0163] In one embodiment of the present invention, the data association strength acquisition module includes:

[0164] Try using the retrieval module to retrieve daily gas and water production in real time, and obtain the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days.

[0165] The reference data acquisition module is used to retrieve the permeability, fracture density, temperature and pressure gradient corresponding to the geological location when the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold, as reference data.

[0166] The first data association strength acquisition module is used to acquire the first data association strength between dynamic data, static data, and environmental data by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data.

[0167] The first data association strength is obtained by the following formula:

[0168]

[0169] Among them, S 01 Indicates the first data correlation strength; n represents the number of times the rate of change corresponding to either daily gas production or daily water production exceeds a preset rate of change threshold; T gi and P giλ represents the normalized temperature and pressure gradients when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively; 01 and λ 02 Ri and Rb represent the sensitivity coefficients of the influence of temperature and pressure gradients on dynamic data, respectively, with values ​​ranging from 0.5 to 1.4 and 0.3 to 1.2. gi and H gi Q represents the normalized permeability and fracture density at the geological location when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold; qi and Q si These represent the rate of change of daily gas production and daily water production when the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, respectively.

[0170] The second data association strength acquisition module is used to obtain the second data association strength between static data and environmental data by utilizing the permeability and fracture density, as well as temperature and pressure gradients, corresponding to the geological location when the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold.

[0171] The second data association strength is obtained using the following formula:

[0172]

[0173] Among them, S 02 Indicates the strength of the second data association; λ 03 and λ 04 The values ​​represent the influence intensity coefficients of environmental data on permeability and crack density, respectively, with values ​​ranging from 0.4 to 1.2 and from 0.7 to 1.3.

[0174] The working principle of the above technical solution is as follows: The system continuously monitors and acquires daily gas production and water production data in real time. For each consecutive two days, the system calculates the rate of change of daily gas production and water production to capture short-term fluctuations in production dynamics. Preset thresholds are set for the rate of change of daily gas production and water production. When the rate of change of either exceeds this threshold, the system determines it as an abnormal change and triggers subsequent data retrieval and analysis processes. When an abnormal change is detected, the system automatically retrieves geological and engineering data corresponding to the geological location where the abnormal change occurred, including permeability, fracture density, temperature, and pressure gradient. This data is used as reference data for subsequent correlation strength analysis.

[0175] The above technical solution achieves the following effect: by real-time monitoring and calculation of the rate of change in daily gas and water production, the system can accurately capture dynamic changes during the production process, providing timely and accurate data support for subsequent analysis. This is achieved by introducing a first data correlation strength S.01 Second data correlation strength S 02 The system can quantify the correlation strength between dynamic and static data, environmental data, and the correlation strength between static and environmental data. This quantitative analysis helps to gain a deeper understanding of the complex heterogeneity and multi-field coupling effects of shale gas reservoirs. By incorporating correlation strength into the production prediction model, the system can more accurately reflect the combined impact of geological, engineering, and environmental factors on shale gas production. This helps improve the accuracy of production prediction and provides a reliable basis for the formulation and optimization of development plans. By quantifying data correlation strength, the system can reveal the intrinsic connections and interaction mechanisms between different parameters, thereby enhancing the interpretability of the production prediction model. This helps engineers better understand the model's prediction results and improve the scientific and rational nature of decision-making. This technical solution, through dynamically adjusting feature weights and fusing multi-source data, can adapt to shale gas reservoirs with different geological conditions and development stages. This helps improve the model's generalization ability and adaptability, enabling it to play a role in a wider range of application scenarios.

[0176] Traditional methods typically rely on static or periodic sampling data for analysis, making it difficult to capture dynamic changes during the production process in real time (such as sudden fluctuations in gas production or abrupt changes in formation parameters). The above-mentioned technical solution calculates the rate of change of daily gas and water production, enabling immediate detection of abnormal fluctuations (such as a sudden increase in production after fracturing or formation water breakthrough). Correlation analysis is triggered only when the rate of change exceeds a threshold, avoiding ineffective calculations and improving efficiency. If the rate of change of gas production in a well suddenly increases from 5% to 20% on a given day, the system automatically retrieves the geological parameters (permeability, fracture density) and environmental parameters (temperature, pressure gradient) at that moment to quickly pinpoint the cause of the anomaly (such as fracture propagation or changes in formation fluid properties). Existing methods often rely on empirical formulas or simple correlation analysis, failing to quantify the nonlinear interactions between geological, engineering, and environmental parameters. The advantages of the above-mentioned technical solution: First correlation strength S 01 Used to quantify the correlation strength between dynamic data (gas production / water production change rate) and static data (permeability, fracture density) and environmental data (temperature, pressure gradient). Second correlation strength S 02 Quantify the correlation strength between static data (permeability, crack density) and environmental data. This is achieved through normalization and the sensitivity coefficient (λ). 01 -λ 04 This eliminates dimensional differences and achieves unified quantification of multi-source data. If the correlation strength S01 between the permeability and gas production change rate of a certain well is significantly higher than that of other wells, it can be inferred that the permeability of this well contributes more to the production, and the fracturing parameters should be optimized first. Traditional models ignore parameter correlation, resulting in a large deviation between the predicted results and actual production data. Feature weight adjustment driven by correlation strength: based on S... 01 and S 02The weights of geological, engineering, and environmental parameters are dynamically adjusted to enhance the model's sensitivity to key parameters. Normalization and sensitivity intensity coefficients are used to ensure the model conforms to shale gas flow patterns (e.g., increased temperature may enhance the positive correlation between permeability and production). If the sensitivity intensity coefficient λ of a well's temperature gradient on its production... 01 With a higher λ, the model will focus more on the impact of temperature changes on yield, thereby improving prediction accuracy. Black-box models such as deep learning struggle to explain the relationship between parameters and yield. The above technical solution clarifies the contribution of each parameter to yield through S01 and S02, making it easier for engineers to understand the model's prediction results. 01 -λ 04 The value range (e.g., 0.5-1.4, 0.3-1.2) provides an intuitive reference for the degree of influence of the parameter. If the correlation strength S between fracture density and production in a well... 01 The strength coefficient λ, which is relatively high and sensitive to crack density in terms of production, is also high. 02 Given the relatively large fracture density, engineers can infer that it is a key factor affecting production, necessitating priority optimization of the fracturing process. Traditional methods predict failure under extreme conditions such as high temperature and pressure, lacking a dynamic response to environmental factors. The aforementioned technical solution quantifies the asymmetric impact of environmental factors on parameter correlations through normalization of temperature and pressure gradients and the use of sensitivity intensity coefficients. When environmental parameters change (e.g., temperature increases), the model automatically adjusts parameter weights to maintain prediction accuracy. In high-temperature reservoirs, if the sensitivity intensity coefficient λ of the temperature gradient on production... 01 The higher the temperature, the more the model will focus on the impact of temperature changes on production, avoiding prediction failure due to environmental changes. Retraining the model with full data is computationally expensive and difficult to update in real time. The above-mentioned technical solution uses triggered data retrieval, only retrieving data when the rate of change exceeds a threshold, reducing unnecessary calculations. A dynamic adjustment mechanism based on correlation strength avoids retraining with full data, reducing computational costs. If the rate of change in gas production of a well does not exceed a threshold for several consecutive days, updates are triggered only during abnormal fluctuations, significantly improving efficiency.

[0177] In one embodiment of the present invention, the weight adjustment module includes:

[0178] The first weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the static data.

[0179] The first feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the static data using the first data association strength and the second data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the static data.

[0180] The adjusted feature weights of the static data are obtained using the following formula:

[0181]

[0182] Where J represents the adjusted feature weight corresponding to each feature parameter contained in the static data; J0 represents the preset feature weight corresponding to each feature parameter contained in the static data; S 01 Indicates the first data association strength; S 02 Indicates the strength of the second data association;

[0183] The second weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data.

[0184] The second feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the dynamic data using the first data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the dynamic data.

[0185] The adjusted feature weights of the dynamic data are obtained using the following formula:

[0186] D = D0 * (1 + S) 01 )

[0187] Where D represents the adjusted feature weight corresponding to each feature parameter contained in the dynamic data; D0 represents the preset feature weight corresponding to each feature parameter contained in the dynamic data.

[0188] The third weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data.

[0189] The third feature weight adjustment module is used to adjust the preset feature weights corresponding to each feature parameter contained in the environmental data using the second data association strength, and to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

[0190] The adjusted feature weights of the environmental data are obtained using the following formula:

[0191]

[0192] Where G represents the adjusted feature weight corresponding to each feature parameter contained in the environmental data; G0 represents the preset feature weight corresponding to each feature parameter contained in the environmental data.

[0193] The working principle of the above technical solution is as follows: The core of this solution is to achieve deep fusion of multi-source data correlation and adaptive model optimization by dynamically adjusting the feature weights of geological and engineering data. First, the system pre-assigns initial weights to each feature parameter in static data (such as permeability and fracture density), dynamic data (such as daily gas production and water production), and environmental data (such as temperature and pressure gradients). These weights are set based on historical experience or domain knowledge. Subsequently, the correlation strength (S) between the two types of data is calculated. 01 and S 02 Driving weight adjustment: S 01 Quantify the correlation between dynamic data and static / environmental data, S 02 Quantify the correlation between static data and environmental data. During the adjustment process, the weight of static data is simultaneously affected by S. 01 and S 02 The weights of dynamic data are only affected by S. 01 The influence, while the weight of environmental data is only affected by S 02 Influence.

[0194] The above technical solution achieves the following results: It significantly improves the accuracy, adaptability, and engineering value of shale gas production prediction through a dynamic weight adjustment mechanism. Firstly, it enhances prediction accuracy: weight adjustment, based on real-time calculated data correlation strength, enables the model to dynamically respond to changes in reservoir and production conditions. Secondly, by fusing the correlation strength of multi-source data, the model can adapt to complex reservoir conditions (such as high temperature and pressure, and high heterogeneity), reducing dependence on a single data source and avoiding prediction biases caused by changes in reservoir conditions. The mathematical formula and correlation strength of the weight adjustment provide a clear logical basis for the model, allowing engineers to intuitively understand the reasons for changes in the weights of each parameter. For example, high-weight parameters directly point to key factors affecting production, providing a scientific basis for engineering decisions. Finally, it optimizes resource allocation: the weight adjustment results can guide the prioritization of engineering measures. This technical solution achieves efficient utilization of multi-source data through dynamic weight allocation, providing a full-chain optimization capability from data to decision for the efficient development of unconventional oil and gas reservoirs.

[0195] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for predicting shale gas production, characterized in that, The shale gas production prediction method includes: Collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data; The data correlation strength of the current geological and engineering data is obtained based on the preprocessed geological and engineering data; The feature weights corresponding to the geological and engineering data are adjusted based on the data correlation strength of the current geological and engineering data; The feature data corresponding to the current geological and engineering data are fused using the adjusted feature weights to obtain the fused feature data; The fused feature data is input into the shale gas production prediction model that has been constructed and trained to obtain shale gas production prediction results.

2. The shale gas production prediction method according to claim 1, characterized in that, Geological and engineering data are collected and preprocessed to obtain preprocessed geological and engineering data, including: Collect geological and engineering data to obtain geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data; The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient. The geological and engineering data are subjected to noise reduction and data cleaning processes to obtain the denoised and cleaned geological and engineering data.

3. The shale gas production prediction method according to claim 1, characterized in that, Based on the preprocessed geological and engineering data, the data correlation strength of the current geological and engineering data is obtained, including: Real-time access to daily gas and water production, and obtaining the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days; When the rate of change of either the daily gas production or the daily water production exceeds a preset rate of change threshold, the permeability, fracture density, temperature, and pressure gradient corresponding to the geological location at which the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold are retrieved as reference data. The first data correlation strength between dynamic data, static data, and environmental data is obtained by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data. By utilizing the permeability, fracture density, temperature, and pressure gradients corresponding to the geological location when the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold, a second data correlation strength between static data and environmental data is obtained.

4. The shale gas production prediction method according to claim 1, characterized in that, The feature weights corresponding to the geological and engineering data are adjusted based on the data correlation strength of the current geological and engineering data, including: Retrieve the preset feature weights corresponding to each feature parameter contained in the static data; The preset feature weights corresponding to each feature parameter contained in the static data are adjusted using the first data association strength and the second data association strength, so as to obtain the adjusted feature weights corresponding to each feature parameter contained in the static data. Retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data; The preset feature weights corresponding to each feature parameter contained in the dynamic data are adjusted using the first data association strength, and the adjusted feature weights corresponding to each feature parameter contained in the dynamic data are obtained. Retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data; The preset feature weights corresponding to each feature parameter contained in the environmental data are adjusted using the second data association strength to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

5. The shale gas production prediction method according to claim 1, characterized in that, The adjusted feature weights are used to fuse the feature data corresponding to the current geological and engineering data to obtain the fused feature data, including: Retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data; The feature parameters contained in the static data, dynamic data, and environmental data are combined with their corresponding adjusted feature weights and weighted to generate fused feature data.

6. A shale gas production prediction system, characterized in that, The shale gas production prediction system includes: The data collection and preprocessing module is used to collect and preprocess geological and engineering data to obtain preprocessed geological and engineering data. The data association strength acquisition module is used to obtain the data association strength of the current geological and engineering data based on the preprocessed geological and engineering data; The weight adjustment module is used to adjust the feature weights corresponding to the geological and engineering data according to the data correlation strength of the current geological and engineering data; The fusion module is used to fuse the feature data corresponding to the current geological and engineering data using the adjusted feature weights, and obtain the fused feature data. The prediction module is used to input the fused feature data into the shale gas production prediction model that has been constructed and trained to obtain the shale gas production prediction results.

7. The shale gas production prediction system according to claim 6, characterized in that, The data collection and preprocessing module includes: The collection module is used to collect geological and engineering data, and to acquire geological and engineering data; wherein, the geological and engineering data includes static data, dynamic data and environmental data; The static data includes permeability and fracture density; the dynamic data includes daily gas production and daily water production; and the environmental data includes temperature and pressure gradient. The preprocessing module is used to perform noise reduction and data cleaning on the geological and engineering data to obtain the noise-reduced and cleaned geological and engineering data.

8. The shale gas production prediction system according to claim 6, characterized in that, The data association strength acquisition module includes: Try using the retrieval module to retrieve daily gas and water production in real time, and obtain the rate of change of daily gas and water production for each adjacent two days based on the daily gas and water production for each adjacent two days. The reference data acquisition module is used to retrieve the permeability, fracture density, temperature and pressure gradient corresponding to the geological location when the rate of change of either the daily gas production or the daily water production exceeds the preset rate of change threshold, as reference data. The first data association strength acquisition module is used to acquire the first data association strength between dynamic data, static data, and environmental data by using the reference data and the rate of change of daily gas production and daily water production corresponding to each reference data. The second data association strength acquisition module is used to obtain the second data association strength between static data and environmental data by utilizing the permeability and fracture density, as well as temperature and pressure gradients, corresponding to the geological location when the rate of change of either daily gas production or daily water production exceeds a preset rate of change threshold.

9. The shale gas production prediction system according to claim 6, characterized in that, The weight adjustment module includes: The first weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the static data. The first feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the static data using the first data association strength and the second data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the static data. The second weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the dynamic data. The second feature weight adjustment module is used to adjust the preset feature weight corresponding to each feature parameter contained in the dynamic data using the first data association strength, and to obtain the adjusted feature weight corresponding to each feature parameter contained in the dynamic data. The third weight retrieval module is used to retrieve the preset feature weights corresponding to each feature parameter contained in the environmental data. The third feature weight adjustment module is used to adjust the preset feature weights corresponding to each feature parameter contained in the environmental data using the second data association strength, and to obtain the adjusted feature weights corresponding to each feature parameter contained in the environmental data.

10. The shale gas production prediction system according to claim 6, characterized in that, The fusion module includes: The feature weight retrieval module is used to retrieve the adjusted feature weights corresponding to the feature parameters contained in static data, dynamic data, and environmental data. The feature data fusion module is used to combine the feature parameters contained in the static data, dynamic data and environmental data with their corresponding adjusted feature weights to generate fused feature data.

Images