Auxiliary device for evaluating economic recoverability of deep coal bed gas resources

Abstract

The invention relates to the technical field of mineral resources, in particular to a deep coal bed gas resource economic recoverability evaluation auxiliary device which comprises a data input interface, a processor and an output interface which are in signal connection in sequence. The economic recoverability evaluation system comprises a productivity prediction module, a comprehensive evaluation module, an uncertainty analysis module, a scheme comparison module and a risk early warning module. The productivity prediction module is used for comparing the original data through a deep learning model and then outputting a productivity prediction sequence; and the scheme comparison module is used for outputting an optimal development scheme identifier to an output interface by adopting a multi-objective decision algorithm when the productivity prediction sequence is processed by the comprehensive evaluation module and the uncertainty analysis module to obtain index data. According to the invention, through a multi-module collaborative decision-making mechanism integrating productivity prediction, multi-scheme intelligent comparison and selection and dynamic risk early warning, automatic generation and output from original data to optimal development schemes and risk prevention and control suggestions are realized.

Description

Technical Field

[0001] This invention relates to the field of mineral resource technology, specifically to an auxiliary device for evaluating the economic recoverability of deep coalbed methane resources. Background Technology

[0002] Deep coalbed methane (usually referring to coalbed methane resources buried at depths exceeding 1500 meters) is an important alternative resource to conventional natural gas, and its development is of strategic significance for optimizing the energy structure and ensuring energy security. However, deep coalbed methane reservoirs are characterized by complex conditions, including high ground stress, high ground temperature, and low permeability, leading to significant technical challenges in exploration and development, high initial investment, long production cycles, and substantial geological and engineering uncertainties. Therefore, conducting a scientific and systematic resource economic recoverability assessment is not only a prerequisite for judging the feasibility of project investment but also a crucial step in optimizing development plans, controlling investment risks, and maximizing economic benefits. This assessment process requires integrating multi-dimensional factors such as geology, engineering, economics, and policy, going beyond the scope of traditional resource assessment to form decision support throughout the entire project lifecycle.

[0003] At the current technological level, although the literature "Design and Application of Economic Evaluation Auxiliary Decision-Making System for Coalbed Methane Projects" pointed out the limitations of early systems that only focused on financial evaluation and proposed an auxiliary decision-making framework covering capacity forecasting, financial evaluation, and uncertainty analysis, its functional output still has significant shortcomings. In this literature, the system remains limited to the calculation and listing of key economic indicators (such as net present value and internal rate of return) and sensitivity analysis of a single scheme, lacking mechanisms for automatic generation, parallel evaluation, and intelligent comparison of multiple development schemes. Furthermore, the system's risk considerations are mostly static analyses, making it difficult to achieve risk control based on real-time data-driven dynamic monitoring and early warning. Its final output is often scattered charts and data, failing to be integrated and transformed into decision-making recommendations that directly guide action. This forces decision-makers to rely on their own experience for comprehensive judgment, which, when facing high-risk, multi-variable, and complex projects like deep coalbed methane, restricts the scientific rigor, efficiency, and reliability of the decision-making process. Therefore, there is an urgent need for an intelligent evaluation device that can deeply integrate geological and engineering data, automatically optimize multiple schemes, and have dynamic risk early warning capabilities, so as to complete the leap from "data calculation" to "intelligent decision-making". Summary of the Invention

[0004] To address the aforementioned issues, this invention provides an auxiliary device for evaluating the economic recoverability of deep coalbed methane resources. By integrating a multi-module collaborative decision-making mechanism that combines production capacity forecasting, intelligent comparison of multiple schemes, and dynamic risk early warning, it achieves automated generation and output of raw data into optimized development schemes and risk control recommendations.

[0005] To achieve the above objectives, the technical solution of the present invention is as follows: an auxiliary device for evaluating the economic recoverability of deep coalbed methane resources, comprising a data input interface, a processor, and an output interface connected in sequence by signals, wherein the data input interface is used to receive raw data transmitted by several sensors deployed in deep coalbed methane; the processor is equipped with an economic recoverability evaluation system, which includes a production capacity prediction module, a comprehensive evaluation module, an uncertainty analysis module, a scheme comparison module, and a risk warning module; The capacity forecasting module is used to compare the raw data with the deep learning model and output the capacity forecasting sequence to the comprehensive evaluation module and the uncertainty analysis module; The comprehensive evaluation module is used to receive the capacity forecast sequence and external historical data transmitted from the data input interface. It uses net present value, internal rate of return, investment payback period, economic net present value and social discount rate calculation models to obtain financial evaluation indicators and national economic evaluation indicators and transmit them to the scheme comparison module. The uncertainty analysis module receives capacity forecast sequences, financial evaluation indicators, national economic evaluation indicators, and cost data transmitted from the data input interface. It uses single-factor and multi-factor sensitivity analysis methods, nonlinear break-even models, Monte Carlo simulation, and risk probability distribution modeling to obtain sensitivity indicators, break-even point data, and risk indicator data. The sensitivity indicators and break-even point data are then transmitted to the scheme comparison module, and the risk indicator data is transmitted to the risk warning module. In the scheme comparison module, when the capacity forecast sequence is processed by the comprehensive evaluation module and the uncertainty analysis module to obtain the indicator data, a multi-objective decision-making algorithm is used to output the preferred development scheme identifier to the output interface. The risk warning module, when the capacity forecast sequence is processed by the uncertainty analysis module to obtain risk indicator data, uses a dynamic threshold judgment and risk level classification method to output a risk warning signal to the output interface. The output interface has a built-in decision output unit, which is used to comprehensively evaluate the preferred development scheme identification and risk warning signals, and output the economic feasibility evaluation results and decision recommendations to the user terminal for the user to view.

[0006] Furthermore, the data input interface is also used to receive external historical data, including financial and socioeconomic parameters, which are directly transmitted to the comprehensive evaluation module; it is also used to receive cost data corresponding to the collection of deep coalbed methane and transmit it to the uncertainty analysis module.

[0007] Furthermore, the raw data received by the production capacity prediction module includes historical gas production data and geological feature data. It adopts a time-series deep learning model based on the attention mechanism. The time-series deep learning model is a Transformer-LSTM hybrid model, which is used to extract the long-term dependence and local features of the gas production sequence. The production capacity prediction sequence is then output to the comprehensive evaluation module and the uncertainty analysis module.

[0008] Furthermore, the comprehensive evaluation module includes a financial evaluation unit and a national economic evaluation unit: The financial evaluation unit receives the capacity forecast sequence and financial parameters, uses the net present value, internal rate of return and investment payback period calculation model to output financial evaluation indicators to the scheme comparison module. The National Economic Evaluation Unit receives the production capacity forecast sequence and socio-economic parameters, uses the economic net present value and social discount rate calculation model to output the national economic evaluation indicators to the scheme comparison module.

[0009] Furthermore, the uncertainty analysis module includes a sensitivity analysis unit, a break-even analysis unit, and a risk probability analysis unit: The sensitivity analysis unit receives capacity forecast sequences, financial evaluation indicators, and national economic evaluation indicators. It uses single-factor and multi-factor sensitivity analysis methods to output sensitivity indicators to the risk warning module and the scheme comparison module. The break-even analysis unit receives capacity forecast sequences and cost data, uses a non-linear break-even model, and outputs break-even point data to the risk probability analysis unit and the scheme comparison module. The risk probability analysis unit receives sensitivity index and equilibrium point data, uses Monte Carlo simulation and risk probability distribution modeling, and outputs risk index data to the risk early warning module.

[0010] Furthermore, the scheme comparison module includes a scheme generation unit, an indicator normalization unit, and a multi-objective decision-making unit: The scheme generation unit is used to call the pre-set scheme parameter library to perform multi-dimensional combination when it receives financial evaluation indicators and national economic evaluation indicators from the comprehensive evaluation module, as well as sensitivity indicators and equilibrium point data from the uncertainty analysis module, and generates a scheme set containing at least three alternative development schemes, and transmits the scheme set to the indicator normalization unit. The index normalization unit receives the set of alternatives, uses the range standardization method to perform dimensionless processing on the heterogeneous evaluation indices of each alternative, and outputs the standardized alternative matrix to the multi-objective decision-making unit after all indices have been normalized. The multi-objective decision-making unit receives a standardized scheme matrix and uses the entropy weight method-TOPSIS coupling algorithm to calculate the relative closeness of each scheme to the ideal solution. When the relative closeness is higher than the preset feasibility threshold, the corresponding scheme identifier is output as the preferred development scheme identifier to the output interface.

[0011] Furthermore, the risk warning module includes a dynamic threshold setting unit, a risk level classification unit, and a warning signal generation unit: The dynamic threshold setting unit is used to continuously receive risk indicator data, and calculate the mean and standard deviation of the risk indicator data using a time series-based sliding window. When new risk indicator data is detected to exceed the range of the historical mean ± 2 times the standard deviation, a dynamically updated risk threshold is generated and transmitted to the risk level classification unit. The risk level classification unit is used to receive dynamically updated risk thresholds and real-time risk indicator data. It uses a fuzzy C-means clustering algorithm to divide the risk level into three levels: low, medium, and high. When the clustering results obtained by the clustering algorithm show that the proportion of high-risk level samples exceeds 15%, the warning signal generation unit is triggered. The early warning signal generation unit is used to generate a risk warning signal containing a risk description and recommended measures based on the mapping relationship between risk level and early warning level when an early warning signal is triggered or a high-risk level identifier is received, and then output it to the output interface.

[0012] Furthermore, the decision output unit in the output interface includes a scheme integration unit and a decision suggestion generation unit: The scheme integration unit is used to simultaneously receive the preferred development scheme identifier and risk warning signal. When the risk warning signal is medium or above, the risk-benefit coordination rule is activated to modify the feasibility of the preferred scheme and output the final recommended scheme after integration to the decision suggestion generation unit. The decision recommendation generation unit receives the final recommended solution, uses template-based natural language generation technology to combine key parameters of the solution with risk warnings, and automatically generates decision recommendations including conclusions such as "recommend adoption", "careful evaluation" or "postpone development". These recommendations are then output to the user terminal along with the economic feasibility evaluation results.

[0013] Furthermore, the output interface also includes a visualization output unit: The visualization output unit is used to receive the economic feasibility evaluation results and decision recommendations generated by the decision output unit. When the judgment result data is not a single scalar, the multi-layer rendering engine is started to generate a comprehensive visualization interface that includes the capacity forecast curve, break-even point, sensitivity spider chart and scheme comparison radar chart. The output format is adaptively adjusted according to the user terminal type before display.

[0014] Furthermore, it also includes a data storage and verification module, which is connected to the data input interface and the processor signal; The data storage and verification module is used to receive and store all raw data, intermediate processed data and final result data. When the capacity prediction module or the comprehensive evaluation module initiates a data call request, it first verifies the integrity and validity of the target data. If data is missing or there is a logical conflict, it requests data retransmission to the data input interface or sends a data abnormality alarm to the processor. The data is only output to the data input interface after the data verification is passed.

[0015] The above approach has the following beneficial effects: 1. This solution uses a multi-objective decision-making algorithm in the solution comparison module to automatically generate a preferred development solution identifier. Combined with the dynamic threshold judgment and risk level classification of the risk warning module, it outputs risk warning signals in real time. The decision output unit further comprehensively evaluates the solutions and risks, directly generating decision recommendations including conclusions such as "recommended," "careful evaluation," or "postponement of development." This achieves the direct transformation of economic evaluation results into actionable decisions, significantly improving decision-making efficiency and scientific rigor.

[0016] 2. This scheme employs a Transformer-LSTM hybrid model based on an attention mechanism in the production capacity prediction module. This model effectively captures the long-term dependencies and local characteristics of gas production sequences, overcoming the limitations of traditional ARMA or grey models under complex geological conditions. Simultaneously, the uncertainty analysis module integrates Monte Carlo simulation and nonlinear break-even analysis to achieve multi-dimensional probabilistic assessment of risk factors. This approach, combining deep learning and probabilistic analysis, significantly improves the accuracy and reliability of production capacity prediction and risk identification, providing technical support for the economic recoverability evaluation of deep coalbed methane resources.

[0017] 3. This solution, through its visualization output unit, can automatically generate a comprehensive visualization interface based on evaluation results, including capacity forecast curves, break-even points, sensitivity spider diagrams, and scheme comparison radar charts. The output format is adaptively adjusted based on the user's terminal type. Furthermore, the data storage and verification module enables real-time verification and anomaly alerts for data throughout the entire process, ensuring data quality and system stability. This integrated visualization and adaptive design not only allows decision-makers to intuitively and comprehensively grasp evaluation information but also lowers the system's operational threshold, making it suitable for decision-making needs across multiple scenarios and levels. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the system framework of an embodiment of the auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to the present invention. Detailed Implementation

[0019] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0021] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0022] The following detailed description illustrates the specific implementation method: Example 1:

[0023] Traditional economic evaluation systems for coalbed methane projects, while possessing functions such as production capacity forecasting, financial evaluation, and uncertainty analysis, primarily output indicators and charts, lacking support for intelligent selection of multiple options and dynamic risk warnings. Decision-makers must rely on manual experience to compare the economics and risks of different options, a cumbersome and highly subjective process. Especially for deep coalbed methane resources, with their complex geological conditions, long investment cycles, and volatile risk factors, existing systems struggle to automatically translate evaluation results into decision recommendations, leading to low decision-making efficiency and susceptibility to bias.

[0024] Based on the above problems, the inventors have proposed the following: Figure 1The deep coalbed methane resource economic feasibility evaluation auxiliary device shown includes a data input interface, a processor, and an output interface connected in sequence. The data input interface is used to receive raw data transmitted by several sensors deployed in the deep coalbed methane. The data input interface is also used to receive external historical data, including financial parameters and socioeconomic parameters, which are directly transmitted to the comprehensive evaluation module. It is also used to receive cost data corresponding to the collection of deep coalbed methane and transmit it to the uncertainty analysis module.

[0025] The processor is equipped with an economic feasibility evaluation system, which includes a capacity forecasting module, a comprehensive evaluation module, an uncertainty analysis module, a scheme comparison module, and a risk warning module. The production capacity prediction module is used to compare the raw data with a deep learning model and output the production capacity prediction sequence to the comprehensive evaluation module and the uncertainty analysis module. The raw data received by the production capacity prediction module includes historical gas production data and geological feature data. It adopts a time-series deep learning model based on the attention mechanism. The time-series deep learning model is a Transformer-LSTM hybrid model, which is used to extract the long-term dependence and local features of the gas production sequence and output the production capacity prediction sequence to the comprehensive evaluation module and the uncertainty analysis module.

[0026] The comprehensive evaluation module is used to receive the capacity forecast sequence and external historical data transmitted from the data input interface. It uses net present value, internal rate of return, investment payback period, economic net present value and social discount rate calculation models to obtain financial evaluation indicators and national economic evaluation indicators and transmit them to the scheme comparison module. The comprehensive evaluation module includes a financial evaluation unit and a national economic evaluation unit: The financial evaluation unit receives the capacity forecast sequence and financial parameters, uses the net present value, internal rate of return and investment payback period calculation model to output financial evaluation indicators to the scheme comparison module. The National Economic Evaluation Unit receives the production capacity forecast sequence and socio-economic parameters, uses the economic net present value and social discount rate calculation model to output the national economic evaluation indicators to the scheme comparison module.

[0027] The uncertainty analysis module receives capacity forecast sequences, financial evaluation indicators, national economic evaluation indicators, and cost data transmitted from the data input interface. It uses single-factor and multi-factor sensitivity analysis methods, nonlinear break-even models, Monte Carlo simulation, and risk probability distribution modeling to obtain sensitivity indicators, break-even point data, and risk indicator data. The sensitivity indicators and break-even point data are then transmitted to the scheme comparison module, and the risk indicator data is transmitted to the risk warning module. The uncertainty analysis module includes a sensitivity analysis unit, a break-even analysis unit, and a risk probability analysis unit. The sensitivity analysis unit receives capacity forecast sequences, financial evaluation indicators, and national economic evaluation indicators. It uses single-factor and multi-factor sensitivity analysis methods to output sensitivity indicators to the risk warning module and the scheme comparison module. The break-even analysis unit receives capacity forecast sequences and cost data, uses a non-linear break-even model, and outputs break-even point data to the risk probability analysis unit and the scheme comparison module. The risk probability analysis unit receives sensitivity index and equilibrium point data, uses Monte Carlo simulation and risk probability distribution modeling, and outputs risk index data to the risk early warning module.

[0028] In the scheme comparison module, when the capacity forecast sequence is processed by the comprehensive evaluation module and the uncertainty analysis module to obtain the indicator data, a multi-objective decision-making algorithm is used to output the preferred development scheme identifier to the output interface. The scheme comparison module includes a scheme generation unit, an indicator normalization unit, and a multi-objective decision-making unit: The scheme generation unit is used to call the pre-set scheme parameter library to perform multi-dimensional combination when it receives financial evaluation indicators and national economic evaluation indicators from the comprehensive evaluation module, as well as sensitivity indicators and equilibrium point data from the uncertainty analysis module, and generates a scheme set containing at least three alternative development schemes, and transmits the scheme set to the indicator normalization unit. The index normalization unit receives the set of alternatives, uses the range standardization method to perform dimensionless processing on the heterogeneous evaluation indices of each alternative, and outputs the standardized alternative matrix to the multi-objective decision-making unit after all indices have been normalized. The multi-objective decision-making unit receives a standardized scheme matrix and uses the entropy weight method-TOPSIS coupling algorithm to calculate the relative closeness of each scheme to the ideal solution. When the relative closeness is higher than the preset feasibility threshold, the corresponding scheme identifier is output as the preferred development scheme identifier to the output interface.

[0029] The risk warning module, when the capacity forecast sequence is processed by the uncertainty analysis module to obtain risk indicator data, uses a dynamic threshold judgment and risk level classification method to output a risk warning signal to the output interface. The risk warning module includes a dynamic threshold setting unit, a risk level classification unit, and a warning signal generation unit. The dynamic threshold setting unit is used to continuously receive risk indicator data, and calculate the mean and standard deviation of the risk indicator data using a time series-based sliding window. When new risk indicator data is detected to exceed the range of the historical mean ± 2 times the standard deviation, a dynamically updated risk threshold is generated and transmitted to the risk level classification unit. The risk level classification unit is used to receive dynamically updated risk thresholds and real-time risk indicator data. It uses a fuzzy C-means clustering algorithm to divide the risk level into three levels: low, medium, and high. When the clustering results obtained by the clustering algorithm show that the proportion of high-risk level samples exceeds 15%, the warning signal generation unit is triggered. The early warning signal generation unit is used to generate a risk warning signal containing a risk description and recommended measures based on the mapping relationship between risk level and early warning level when an early warning signal is triggered or a high-risk level identifier is received, and then output it to the output interface.

[0030] The output interface has a built-in decision output unit, which is used to comprehensively evaluate the preferred development scheme identification and risk warning signal, and output the economic feasibility evaluation results and decision recommendations to the user terminal for the user to view. The decision output unit in the output interface includes a scheme integration unit and a decision suggestion generation unit: The scheme integration unit is used to simultaneously receive the preferred development scheme identifier and risk warning signal. When the risk warning signal is medium or above, the risk-benefit coordination rule is activated to modify the feasibility of the preferred scheme and output the final recommended scheme after integration to the decision suggestion generation unit. The decision recommendation generation unit receives the final recommended solution, uses template-based natural language generation technology to combine key parameters of the solution with risk warnings, and automatically generates decision recommendations including conclusions such as "recommended to adopt", "careful evaluation" or "postpone development". These recommendations are then output to the user terminal along with the economic feasibility evaluation results. The output interface also includes a visualization output unit: The visualization output unit is used to receive the economic feasibility evaluation results and decision recommendations generated by the decision output unit. When the judgment result data is not a single scalar, the multi-layer rendering engine is started to generate a comprehensive visualization interface that includes the capacity forecast curve, break-even point, sensitivity spider chart and scheme comparison radar chart. The output format is adaptively adjusted according to the user terminal type before display.

[0031] Taking the evaluation of a deep coalbed methane block in Huaibei as an example, the specific implementation process is as follows: After the system is started, the data input interface first receives real-time data streams (such as daily gas production and bottom hole pressure) from downhole sensors in the Huaibei block, as well as manually entered external historical data (including gas production records, geological structure maps, drilling costs, etc. for the past three years) and financial parameters (such as total investment, natural gas price, and loan interest rate).

[0032] These data are categorized and directed to different modules. Historical gas production sequences and geological feature data are sent to the production capacity prediction module. This module does not use a simple statistical model, but rather a Transformer-LSTM hybrid deep learning model. Specifically, LSTM units excel at capturing local patterns and cyclical regularities (such as seasonal fluctuations) in gas production over time, while the Transformer's attention mechanism can identify long-distance dependencies in long-term sequences (e.g., the long-term impact of a key well's commissioning on the overall block's production trend). Through learning and training on historical data, the model outputs a production capacity prediction sequence for the block over the next 10 years. This sequence is then simultaneously sent to the comprehensive evaluation module and the uncertainty analysis module as the basis for their calculations.

[0033] In the comprehensive evaluation module, the financial evaluation unit combines the capacity forecast series with financial parameters. It doesn't simply calculate a static net present value (NPV), but rather converts the projected gas production for each future year (considering its probability distribution) into cash flows, incorporating investment, operating costs, taxes, etc., to dynamically calculate the project's financial NPV, internal rate of return (IRR), and payback period. Simultaneously, the national economic evaluation unit works in parallel, using the social discount rate to assess the project's net contribution to the national economy (economic NPV) and its environmental benefits (such as emission reductions).

[0034] The concurrently running uncertainty analysis module is dedicated to outlining the project's "risk profile." Its sensitivity analysis unit receives data from capacity forecasts and financial evaluations, automatically performing "if-then" analyses. For example, it simulates how the project's financial IRR and NPV would change if gas prices fell by 10% or drilling costs rose by 15%, thus precisely quantifying that "gas price" is a more sensitive factor than "cost." The break-even analysis unit uses capacity sequences and cost data to calculate the minimum average annual gas production required for the project to break even over its entire lifecycle. All these sensitivity indicators and break-even point data are ultimately fed into the risk probability analysis unit. This unit uses Monte Carlo simulations to randomly combine thousands of possible gas price, cost, and production scenarios, ultimately outputting a risk probability distribution map of the project, visually displaying the probability that the project's IRR will fall below the benchmark rate of return (i.e., the probability of failure).

[0035] Because the assessment and decision-making regarding the recoverability of deep coalbed methane often requires weighing different options, the option comparison module's option generation unit automatically generates multiple complete alternative development options by calling a pre-set option parameter library (e.g., Option A - high investment, using advanced fracturing technology, rapid recovery; Option B - medium investment, conservative development, long stable production period; Option C - phased investment, gradual expansion). This is combined with the economic and risk indicators calculated by the previous module.

[0036] Since indicators such as NPV (unit: RMB 10,000), IRR (unit: %), payback period (unit: years), and risk probability (unit: %) have different dimensions and directions of advantage and disadvantage, the normalization unit of the indicators uses the range standardization method to unify them into dimensionless values, so that all indicators can be compared on the same scale.

[0037] The multi-objective decision-making unit employs an entropy-weighted method coupled with the TOPSIS algorithm: First, the entropy-weighted method objectively assigns weights based on the volatility of each indicator data (for example, if a significant difference in risk probability is found between different solutions, a higher weight is automatically assigned to the risk probability); then, the TOPSIS method calculates the distance between each solution and the "ideal optimal solution" and the "ideal worst solution," ultimately deriving a relative proximity score representing the merits of each solution. The system presets a feasibility threshold (e.g., 0.7), and only solutions with a proximity score higher than this threshold are output as preferred development solutions.

[0038] Throughout the system's operation, the risk warning module remains continuously active. Its dynamic threshold setting unit continuously monitors real-time risk indicators (such as failure probability) from the risk probability analysis unit via a sliding time window. It constantly calculates the mean and standard deviation of recent risk indicators. For example, assuming the project's historical average failure probability is around 10% and the standard deviation is 2%, if a new calculation shows the failure probability suddenly jumps to 15% (exceeding the mean plus twice the standard deviation), the unit immediately updates the risk threshold dynamically, signaling a change in system status.

[0039] After receiving new thresholds and real-time data, the risk level classification unit uses a fuzzy C-means clustering algorithm to cluster the current risk level with historical data, classifying it into three risk levels: "low," "medium," and "high." If the clustering results show that the proportion of samples classified as "high risk" exceeds a preset warning line (e.g., 15%), the warning signal generation unit will be immediately triggered. This unit generates a structured risk warning signal based on a preset mapping relationship (e.g., "high risk" corresponds to "red warning"), which includes not only the risk level but also a risk description and preliminary recommended measures (e.g., "A significant increase in the risk of cost overruns has been detected; it is recommended to review supply chain contracts and activate cost control contingency plans").

[0040] Finally, all information is aggregated at the output interface. The decision output unit's solution integration unit simultaneously receives the "preferred solution identifier" from the solution comparison module and the "risk warning signal" from the risk warning module. If the warning signal is intermediate or high, it activates the risk-benefit coordination rules to modify the feasibility of the preferred solution. For example, the originally optimal solution B was selected due to its high return, but it comes with intermediate risk. The integration unit might suggest adding a risk reserve for this solution or adjusting its initial investment pace to form the final recommended solution.

[0041] The decision recommendation generation unit uses template-based natural language generation technology to integrate key parameters of the final solution (such as expected IRR and investment payback period) with risk warnings, and automatically generate easily understandable decision recommendations such as "It is recommended to adopt solution B, which is expected to have good economic returns, but cost control needs to be closely monitored. It is recommended to establish a cost monitoring team in the first phase of the project."

[0042] Simultaneously, the visualization output unit generates a comprehensive visualization interface from the entire evaluation process and results using a multi-layer rendering engine. On this interface, decision-makers can simultaneously view future capacity forecast curves, a production-cost relationship graph marking the break-even point, a spider chart showing the sensitivity of various factors, and a radar chart comparing the advantages and disadvantages of different options. The system adaptively adjusts the display format based on the user's terminal.

[0043] Example 2:

[0044] As attached Figure 1 As shown, the difference from Embodiment 1 is that it also includes a data storage and verification module, which is connected to the data input interface and the processor signal. The data storage and verification module is used to receive and store all raw data, intermediate processed data and final result data. When the capacity prediction module or the comprehensive evaluation module initiates a data call request, it first verifies the integrity and validity of the target data. If data is missing or there is a logical conflict, it requests data retransmission to the data input interface or sends a data abnormality alarm to the processor. The data is only output to the data input interface after the data verification is passed.

[0045] The specific implementation process is as follows: During the evaluation of the Huaibei deep coalbed methane block, when various data from the Huaibei block enter the system through the data input interface, the data storage and verification module first stores the received real-time sensor data (such as downhole pressure, temperature, instantaneous gas production), externally imported historical production data, geological parameters, and financial indicators in the time-series database according to the preset data structure and classification standards, and adds a timestamp and data source identifier to each data.

[0046] During subsequent evaluation, when the production capacity prediction module needs to access historical gas production data for model training, it will send a data request to the data storage and verification module. At this point, the data storage and verification module will first check the integrity of the target dataset; for example, it might find that the requested "daily gas production data for the first quarter of 2025" contains missing records for several consecutive days. Simultaneously, the data storage and verification module will also perform logical consistency checks, such as finding that the "cumulative gas production" field value for a certain day is less than the previous day.

[0047] Once such data quality issues are identified, the data storage and verification module will not directly send the defective data. Instead, it will send a command to the data input interface to request the retransmission of the original data for a specified time period, attempting to automatically repair any missing or abnormal data. If the data problem persists after retransmission, the data storage and verification module will send a structured data anomaly alarm to the processor. The alarm message will clearly indicate the type of abnormal data (missing / conflicting), its location (specific data table and time point), and its severity level.

[0048] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. An auxiliary device for evaluating the economic recoverability of deep coalbed methane resources, characterized in that, It includes a data input interface, a processor, and an output interface connected in sequence. The data input interface is used to receive raw data transmitted by several sensors deployed in deep coalbed methane. The processor is equipped with an economic feasibility evaluation system, which includes a production capacity prediction module, a comprehensive evaluation module, an uncertainty analysis module, a scheme comparison module, and a risk warning module. The capacity forecasting module is used to compare the raw data with the deep learning model and output the capacity forecasting sequence to the comprehensive evaluation module and the uncertainty analysis module; The comprehensive evaluation module is used to receive the capacity forecast sequence and external historical data transmitted from the data input interface. It uses net present value, internal rate of return, investment payback period, economic net present value and social discount rate calculation models to obtain financial evaluation indicators and national economic evaluation indicators and transmit them to the scheme comparison module. The uncertainty analysis module receives capacity forecast sequences, financial evaluation indicators, national economic evaluation indicators, and cost data transmitted from the data input interface. It uses single-factor and multi-factor sensitivity analysis methods, nonlinear break-even models, Monte Carlo simulation, and risk probability distribution modeling to obtain sensitivity indicators, break-even point data, and risk indicator data. The sensitivity indicators and break-even point data are then transmitted to the scheme comparison module, and the risk indicator data is transmitted to the risk warning module. In the scheme comparison module, when the capacity forecast sequence is processed by the comprehensive evaluation module and the uncertainty analysis module to obtain the indicator data, a multi-objective decision-making algorithm is used to output the preferred development scheme identifier to the output interface. The risk warning module, when the capacity forecast sequence is processed by the uncertainty analysis module to obtain risk indicator data, uses a dynamic threshold judgment and risk level classification method to output a risk warning signal to the output interface. The output interface has a built-in decision output unit, which is used to comprehensively evaluate the preferred development scheme identification and risk warning signals, and output the economic feasibility evaluation results and decision recommendations to the user terminal for the user to view.

2. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 1, characterized in that, The data input interface is also used to receive external historical data, including financial and socioeconomic parameters, which are directly transmitted to the comprehensive evaluation module. It is also used to receive cost data corresponding to the collection of deep coalbed methane and transmit it to the uncertainty analysis module.

3. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 2, characterized in that, The raw data received by the production capacity prediction module includes historical gas production data and geological feature data. It adopts a time-series deep learning model based on the attention mechanism. The time-series deep learning model is a Transformer-LSTM hybrid model, which is used to extract the long-term dependence and local features of the gas production sequence. The production capacity prediction sequence is output to the comprehensive evaluation module and the uncertainty analysis module.

4. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 3, characterized in that, The comprehensive evaluation module includes a financial evaluation unit and a national economic evaluation unit: The financial evaluation unit receives the capacity forecast sequence and financial parameters, uses the net present value, internal rate of return and investment payback period calculation model to output financial evaluation indicators to the scheme comparison module. The National Economic Evaluation Unit receives the production capacity forecast sequence and socio-economic parameters, uses the economic net present value and social discount rate calculation model to output the national economic evaluation indicators to the scheme comparison module.

5. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 4, characterized in that, The uncertainty analysis module includes a sensitivity analysis unit, a break-even analysis unit, and a risk probability analysis unit. The sensitivity analysis unit receives capacity forecast sequences, financial evaluation indicators, and national economic evaluation indicators. It uses single-factor and multi-factor sensitivity analysis methods to output sensitivity indicators to the risk warning module and the scheme comparison module. The break-even analysis unit receives capacity forecast sequences and cost data, uses a non-linear break-even model, and outputs break-even point data to the risk probability analysis unit and the scheme comparison module. The risk probability analysis unit receives sensitivity index and equilibrium point data, uses Monte Carlo simulation and risk probability distribution modeling, and outputs risk index data to the risk early warning module.

6. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 5, characterized in that, The scheme comparison module includes a scheme generation unit, an indicator normalization unit, and a multi-objective decision-making unit: The scheme generation unit is used to call the pre-set scheme parameter library to perform multi-dimensional combination when it receives financial evaluation indicators and national economic evaluation indicators from the comprehensive evaluation module, as well as sensitivity indicators and equilibrium point data from the uncertainty analysis module, and generates a scheme set containing at least three alternative development schemes, and transmits the scheme set to the indicator normalization unit. The index normalization unit receives the set of alternatives, uses the range standardization method to perform dimensionless processing on the heterogeneous evaluation indices of each alternative, and outputs the standardized alternative matrix to the multi-objective decision-making unit after all indices have been normalized. The multi-objective decision-making unit receives a standardized scheme matrix and uses the entropy weight method-TOPSIS coupling algorithm to calculate the relative closeness of each scheme to the ideal solution. When the relative closeness is higher than the preset feasibility threshold, the corresponding scheme identifier is output as the preferred development scheme identifier to the output interface.

7. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 6, characterized in that, The risk warning module includes a dynamic threshold setting unit, a risk level classification unit, and a warning signal generation unit. The dynamic threshold setting unit is used to continuously receive risk indicator data, and calculate the mean and standard deviation of the risk indicator data using a time series-based sliding window. When new risk indicator data is detected to exceed the range of the historical mean ± 2 times the standard deviation, a dynamically updated risk threshold is generated and transmitted to the risk level classification unit. The risk level classification unit is used to receive dynamically updated risk thresholds and real-time risk indicator data. It uses a fuzzy C-means clustering algorithm to divide the risk level into three levels: low, medium, and high. When the clustering results obtained by the clustering algorithm show that the proportion of high-risk level samples exceeds 15%, the warning signal generation unit is triggered. The early warning signal generation unit is used to generate a risk warning signal containing a risk description and recommended measures based on the mapping relationship between risk level and early warning level when an early warning signal is triggered or a high-risk level identifier is received, and then output it to the output interface.

8. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 7, characterized in that, The decision output unit in the output interface includes a scheme integration unit and a decision suggestion generation unit: The scheme integration unit is used to simultaneously receive the preferred development scheme identifier and risk warning signal. When the risk warning signal is medium or above, the risk-benefit coordination rule is activated to modify the feasibility of the preferred scheme and output the final recommended scheme after integration to the decision suggestion generation unit. The decision recommendation generation unit receives the final recommended solution, uses template-based natural language generation technology to combine key parameters of the solution with risk warnings, and automatically generates decision recommendations including conclusions such as "recommend adoption", "careful evaluation" or "postpone development". These recommendations are then output to the user terminal along with the economic feasibility evaluation results.

9. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 8, characterized in that, The output interface also includes a visualization output unit: The visualization output unit is used to receive the economic feasibility evaluation results and decision recommendations generated by the decision output unit. When the judgment result data is not a single scalar, the multi-layer rendering engine is started to generate a comprehensive visualization interface that includes the capacity forecast curve, break-even point, sensitivity spider chart and scheme comparison radar chart. The output format is adaptively adjusted according to the user terminal type before being displayed.

10. The auxiliary device for evaluating the economic recoverability of deep coalbed methane resources according to claim 9, characterized in that, It also includes a data storage and verification module, which is connected to the data input interface and the processor signal; The data storage and verification module is used to receive and store all raw data, intermediate processed data and final result data. When the capacity prediction module or the comprehensive evaluation module initiates a data call request, it first verifies the integrity and validity of the target data. If data is missing or there is a logical conflict, it requests data retransmission to the data input interface or sends a data abnormality alarm to the processor. The data is only output to the data input interface after the data verification is passed.

Images