Intelligent decision-making method for site selection of abandoned mine CAES storage cavern based on real-time self-adaptive correction

Abstract

The invention discloses a real-time self-adaptive correction abandoned mine CAES reservoir site selection intelligent decision-making method, and relates to the technical field of compressed air energy storage, and the method comprises the following steps: S1, collecting, fusing and preprocessing multi-source data; s2, constructing an intelligent site selection evaluation model of the AHP based on the improved ACO; and S3, constructing a decision system and a real-time feedback mechanism based on the intelligent evaluation model. According to the intelligent decision-making method for site selection of the abandoned mine CAES reservoir based on real-time self-adaptive correction, multi-source data fusion, an improved ant colony algorithm (ACO) based on the regret theory and an analytic hierarchy process (AHP) are combined, comprehensive intelligent evaluation can be carried out on abandoned mine reconstruction more scientifically and accurately, and the site selection accuracy of the abandoned mine CAES reservoir based on the real-time self-adaptive correction is improved. Basic principles of safety and reliability, technical feasibility and economic rationality are met.

Description

Technical Field

[0001] The invention relates to the technical field of compressed air energy storage, and in particular to a real-time adaptively corrected abandoned mine CAES reservoir site selection intelligent decision-making method. Background Art

[0002] Compressed air energy storage (CAES) technology, as a large-scale energy storage method, has the advantages of strong peak-shaving and frequency-regulating capabilities and is environmentally friendly. It can effectively store excess energy and release it during peak demand to achieve peak-shaving and valley-filling and balance the power load. It has gradually become one of the important technologies to solve the volatility problem of renewable energy. Abandoned mines have the potential to become CAES reservoirs due to their unique geological structure, closed space and good pressure-bearing capacity. At the same time, the transformation of abandoned mines into CAES reservoirs that can achieve long-term electricity storage is conducive to the recycling of space resources and the development of clean energy. However, the development of abandoned mine reservoirs involves complex geological conditions, such as the stability of the rock mass, the influence of groundwater, and the integrity of the surrounding rock structure. The coupling of these factors makes it extremely difficult to accurately evaluate the safety, long-term stability and energy storage efficiency of the reservoir.

[0003] In the existing technology, the site selection and evaluation of abandoned mine CAES reservoirs mainly rely on single geological or engineering data, such as preliminary judgments through simple analysis of parameters such as mine rock thickness, rock mass strength, and groundwater conditions. This method ignores the interaction between geological, engineering, and environmental factors, resulting in greater uncertainty in the evaluation results. In addition, traditional site selection methods are mostly based on manual experience and limited field data, which makes it difficult to cope with complex changes in geological conditions and lack the ability to systematically integrate and analyze multi-source data, thereby limiting the accuracy of reservoir site selection and the safety of engineering implementation. In addition, traditional evaluation methods often lack systematic decision support tools and cannot achieve dynamic updating and intelligent optimization of real-time data. With the development of artificial intelligence and big data technology, the comprehensive evaluation of abandoned mine CAES reservoirs using multi-source data fusion and intelligent site selection systems has become a feasible solution, but there is no mature technology specifically for abandoned mine geological evaluation. Summary of the invention

[0004] The purpose of the present invention is to provide a real-time adaptive and corrected abandoned mine CAES reservoir site selection intelligent decision-making method, which combines multi-source data fusion, an improved ant colony algorithm (ACO) based on regret theory and an analytic hierarchy process (AHP), and can more scientifically and accurately conduct a comprehensive intelligent evaluation of abandoned mine reconstruction, meeting the basic principles of safety, reliability, technical feasibility and economic rationality.

[0005] To achieve the above object, the present invention provides a real-time adaptive correction abandoned mine CAES reservoir site selection intelligent decision-making method, comprising the following steps: S1, collect, integrate and preprocess multi-source data; S2, constructing an intelligent evaluation model for site selection based on AHP with improved ACO; S3. Build a decision-making system and real-time feedback mechanism based on the intelligent evaluation model.

[0006] Preferably, S1 comprises the following steps: S11. Collect and integrate five types of data: geotechnical characteristics, hydrogeological characteristics, sealing characteristics of the overburden, mine characteristics, and safety and economic factors; S12. Data preprocessing: including data cleaning, missing value filling, noise filtering, and spatial consistency correction to ensure the consistency and accuracy of data processing.

[0007] Preferably, S2 comprises the following steps: S21. Construct a hierarchical analysis structure to decompose the abandoned mine site selection problem into four evaluation structures: target layer, criterion layer, criterion sub-layer and scheme layer; S22, construct a judgment matrix based on ACO-AHP; S23, preliminary comprehensive score, through the weighted summary of the criterion layer and the sub-criteria layer, to obtain the comprehensive score of the mine; S24. Based on the revision of regret theory, the uncertainty that may exist in the ACO path selection process is dealt with.

[0008] Preferably, S22 comprises the following steps: S221, constructing a first-level judgment matrix according to the criterion layer, and constructing a judgment matrix for the sub-criterion under each criterion layer; S222, optimizing the judgment matrix by using the ACO algorithm, making decisions according to the relative weights aij in the judgment matrix, and simulating multiple decision paths by using the ant colony algorithm; S223. Verify the consistency of the matrix using the consistency index CI and the consistency ratio CR. During the system iteration process, perform consistency checks at any time to ensure that the matrix remains consistent after each round of path selection.

[0009] Preferably, the regret value in S24 The calculation formula is as follows: ; in, is the potential optimal score of the unselected path, is the score of the current path; Regret Value , indicating that the current path is inferior to other paths and needs to be corrected; regret value , indicating that the score of the current path is consistent with the optimal path and does not need to be corrected.

[0010] Preferably, the modified state transition probability formula in S24 is: ; in, represents the information density, represents the heuristic function value; The revised pheromone update formula is: ; in, It represents the volatility coefficient of pheromone, which is used to describe the attenuation degree of pheromone concentration; Weight adjustment correction formula: ; The revised scoring formula is: ; in, is the correction coefficient, which is used to control the adjustment strength of the regret value, usually Take a value between.

[0011] Preferably, step S3 includes data flow control, model parameter adjustment, real-time feedback mechanism and algorithm iteration control. The user adjusts the model parameters according to the scoring results, and the system recalculates and adjusts the path selection according to the user's feedback, supporting dynamic adjustment of the decision.

[0012] Preferably, step S3 also includes an interface layout divided into data input, scoring result display and dynamic adjustment areas, and intuitively displays the scores and weight distribution of each criterion and sub-criteria through bar charts and radar charts. Users can quickly input data, view scores, and adjust weights and parameters through a concise operation interface.

[0013] Therefore, the present invention adopts the above-mentioned abandoned mine CAES reservoir site selection intelligent decision-making method with real-time adaptive correction, which has the following beneficial effects: (1) The present invention uses the ACO-AHP hierarchical analysis method to decompose the complex mine site selection problem into multiple levels, and through the construction of a reasonable judgment matrix, quantitative analysis and weight allocation of various influencing factors are carried out, which significantly improves the scientific nature of site selection decisions. At the same time, through the multi-source data fusion method, literature data, field survey data and historical engineering data are effectively integrated, avoiding the evaluation bias caused by a single data source in traditional methods, and ensuring the accuracy of decision-making results.

[0014] (2) The present invention introduces the regret theory to quantify the potential risks and regret values ​​after decision-making brought about by weights and scores during the evaluation process, effectively reducing the decision-making bias that may be caused by fluctuations in geological conditions or data errors, and improving the robustness and risk response capabilities of the site selection process.

[0015] (3) The present invention combines ACO-regret theory with AHP to enable the system to have full-process optimization capabilities from initial weight allocation (AHP) to dynamic path optimization (ACO) to future scenario simulation and correction (regret theory). Through multiple rounds of scoring and path selection, the system ensures that the final decision can converge to the global optimal solution, and at the same time has the ability of adaptive correction, realizing intelligent multi-source data fusion and decision optimization, improving the system's automation level, and reducing human intervention.

[0016] (4) The present invention dynamically processes and updates the real-time monitoring data through an intelligent decision-making system. The system automatically adjusts the weights of the evaluation indicators and the evaluation results according to the changes in the on-site geological conditions, ensuring the real-time and accuracy of the scoring decisions. In particular, it can be adjusted according to the latest data during the long-term project implementation process, providing highly flexible decision support.

[0017] (5) The decision-making evaluation method in this invention is not only applicable to the site selection decision of abandoned mine CAES storage, but also has wide applicability. Its multi-source data fusion and intelligent site selection evaluation system can be extended to other underground space utilization and risk assessment fields in complex geological environments, and has good versatility and extensibility.

[0018] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is the target hierarchical structure model diagram of the present invention; Figure 2 A flowchart of constructing a judgment matrix based on ACO-AHP for the present invention; Figure 3 This is the architecture design of the intelligent decision-making system of the present invention. DETAILED DESCRIPTION

[0020] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.

[0021] Unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0022] Example See also Figure 1-3 The present invention provides a real-time adaptive correction method for the intelligent decision-making method of abandoned mine CAES reservoir site selection. By combining multi-source data fusion technology with the risk assessment method of optimizing the ACO-AHP analytic hierarchy process, an intelligent decision-making evaluation system is systematically constructed to cope with complex geological conditions and engineering requirements, including the following steps: S1 Multi-source data fusion and preprocessing This method combines literature experience data, historical and field engineering data, and uses multi-source data fusion technology to systematically integrate multiple key factors such as mine geological structure, rock layer thickness, rock mass strength, groundwater impact, etc. Through quantitative analysis of various types of data, the present invention further introduces a dynamic weight allocation mechanism to allocate different weights based on the importance and influence of the data to ensure the rationality and scientificity of each evaluation indicator.

[0023] S11 Multi-source data collection and integration In view of the complex geological environment of abandoned mines, the data sources for site selection evaluation are diverse, including but not limited to: literature experience data, on-site geological exploration data and related engineering practice data. In the site selection process, the data is mainly divided into five categories, and the sources of each type of data and their acquisition methods are as follows: Geological structural characteristics Fault development, fold development, and earthquake history are analyzed through regional geological surveys, historical records, and remote sensing images, and GIS technology is used to form a model of fault distribution and activity history; the inclination of coal seams is determined through drilling data and geological profiles, and analyzed in combination with three-dimensional geological modeling.

[0024] Hydrogeological characteristics The distribution of surface water sources, groundwater system distribution, recharge and discharge conditions rely on hydrogeological surveys and remote sensing image data; the conditions of old goaf water in goafs need to be comprehensively assessed through groundwater level monitoring wells and historical mining hydrological data.

[0025] Sealing characteristics of upper cover The lithology and mechanical parameters of the caprock are obtained through field sampling and laboratory testing, including mechanical parameters such as uniaxial compressive strength and rock elastic modulus; the thickness of the caprock and the development of fractures can be obtained by combining seismic wave reflection technology and drilling data; the permeability of the rock mass is evaluated through field penetration tests and numerical simulations (such as COMSOL numerical simulation software).

[0026] Mine characteristics The mine depth and storage capacity are determined through historical mine data, engineering drawings and geological surveys; the mining method and roof rock properties are determined through historical engineering records and on-site investigations; the disposal of harmful gases needs to be evaluated based on the monitoring data of the mine ventilation system; the digital modeling situation is achieved by obtaining a digital three-dimensional model of the mine through laser scanning and BIM technology.

[0027] Safety and economic factors The distribution of surrounding photovoltaic or wind power energy is determined through satellite image data and related energy distribution reports; the transportation and power grid conditions in the mining area are evaluated based on regional infrastructure construction data; the surface building density and surface population density are based on regional planning maps and demographic data; and policy support and capital investment are analyzed in combination with relevant government planning documents and financial reports.

[0028] S12 Data Preprocessing Missing data handling The present invention adopts a missing data filling method based on adaptive linear interpolation to process missing data in geological structures, hydrological data and mine characteristics. By introducing a weight coefficient λ i , interpolate the values ​​of adjacent points. The calculation formula is: ; in, It is the weight obtained by optimizing the non-deterministic optimization algorithm. It can be optimized non-deterministically based on the trend of historical data or global data to find the optimal The optimization method is as follows: ; Through the prior probability distribution of known data points, the posterior probability distribution of missing data is derived to fill in the missing values. Perform a preliminary statistical analysis on the data set to determine the prior distribution type of the data (such as normal distribution or lognormal distribution), and use the derived posterior probability as the filling value of the missing data. It is mainly applicable to geological structural characteristics (such as fault development) and hydrological characteristics (such as groundwater system distribution). According to the distribution characteristics of known data, it can minimize the deviation of the filled data and ensure the rationality and accuracy of the data.

[0029] Spatial consistency correction In view of the differences in spatial resolution, coordinate system, projection method, etc. among data from different sources (such as geological maps, mine exploration data, remote sensing images, etc.), the present invention adopts a multi-source data spatial consistency correction algorithm to ensure that all data sets have a unified spatial coordinate system and resolution.

[0030] (1) Perform coordinate system conversion on all data sets, converting data in different coordinate systems into the target coordinate system (such as WGS-84) to ensure that all data share the same geographic reference.

[0031] (2) Multi-scale convolution kernels are introduced to resample data of different resolutions to ensure that the pixel size and accuracy of spatial data remain consistent while avoiding over-smoothing. The calculation formula is as follows: ; ; in, is the interpolated value. is an adaptive multi-scale convolution kernel, the scale of the convolution kernel r Determined by the local spatial frequency. are the original data points. It is the value after upsampling the low-resolution data. It is the value after downsampling of high-resolution data.

[0032] Time series completion For data with time series characteristics such as earthquake history and hydrological change data, the present invention adopts a time series completion method based on the ARIMA model to generate forecast data for the missing period by capturing the autocorrelation and trend characteristics in the time series.

[0033] Perform a stationarity test (such as a unit root test) on the time series data to determine whether the data is a stationary series. If the series is not stationary, perform a difference process to make the data series stationary. Model the time series data using the ARIMA model, use the model to predict the missing period, and generate an estimated value for the missing period. Use the prediction results as the complement of the time series data. The ARIMA model calculation formula is: ; in, Y t is the current value of the time series; c is a constant term; is the autoregression coefficient, indicating the The value of the moment is the current moment The intensity of the impact; is the sliding average coefficient, indicating the The error term at each moment For the current moment Y t The intensity of the impact; is the error (noise) term, which represents the unpredictable component or the part that cannot be explained by the model.

[0034] Noise Filtering In view of the noise in the collected experimental data and sensor data, the present invention adopts the Kalman filtering algorithm to filter the noise, which is suitable for processing the noise problem in the real-time data such as mine sensors and groundwater level monitoring, and can effectively filter out sudden outliers and sensor errors to improve data quality.

[0035] Kalman filtering is a recursive optimal estimation algorithm suitable for dealing with noise and uncertainty problems in dynamic systems. The state transition model and observation model calculation formula are as follows: ; ; in, is the real state of the system, is the observed value, F and H is the system matrix, and are system noise and observation noise, respectively.

[0036] 3D model reconstruction The present invention adopts a three-dimensional model reconstruction method based on BIM technology for the three-dimensional spatial data of the mine, which is used to accurately reconstruct the mine structure and evaluate the stability and capacity of the reservoir. According to the geological exploration data and mining records of the mine, a geometric model of the mine is established, the three-dimensional spatial structure of the mine is determined, and parametric modeling is performed using BIM software. Combined with the buried depth, storage capacity and mining method of the mine, the geometric characteristics and engineering information of the mine are integrated into a unified three-dimensional model. The important structures such as the mine roof and side walls are reconstructed in detail, and the stability of the structure is evaluated by Abaqus numerical simulation software. The stress distribution of the mine under different pressure conditions is simulated by establishing a finite element mesh model, defining the lithology and mechanical parameters of the mine, and applying boundary conditions. According to the analysis results of the three-dimensional model, a capacity assessment report of the mine reservoir is generated, and the stability of the mine is comprehensively analyzed.

[0037] S2 builds an intelligent evaluation model for site selection based on AHP with improved ACO ACO-based AHP is used to conduct hierarchical analysis of multi-dimensional evaluation indicators in the site selection of CAES reservoirs in abandoned mines. A judgment matrix is ​​constructed through AHP, and multiple influencing factors such as geological structure, hydrological conditions, and mine characteristics are hierarchically quantified. The weights of each criterion and sub-criteria are preliminarily allocated, and the preliminary comprehensive score of each site selection scheme is calculated. In order to optimize the subjectivity of AHP, ACO introduces dynamic path selection and pheromone update mechanism to adaptively adjust the weights of each level. Through multiple rounds of iterative optimization of the scores and weights of each option, the global optimal solution is gradually approached, thereby improving the accuracy of site selection evaluation and the robustness of decision-making. Regret theory is introduced into the optimization process of the ACO-AHP model. Regret theory quantifies the potential regret value of each site selection scheme in different scenarios, dynamically optimizes the score and weight distribution, and compensates for the deviation that may occur when ACO is in the local optimal solution.

[0038] S21 Construction of hierarchical analysis structure In the site selection process of converting abandoned mines into CAES reservoirs, a large number of complex and multi-dimensional geological, hydrological and economic factors are involved. In order to achieve scientific and accurate decision-making evaluation, the present invention introduces the decision-making process of ACO to optimize the AHP model, aiming to solve the comprehensive scoring problem of abandoned mine reconstruction through dynamic path selection and intelligent optimization. Through the application of ACO in the AHP model, by simulating the behavior of ant colonies, combining data-driven intelligent scoring and heuristic functions in the path selection process, the model can dynamically optimize weight distribution and path selection, and intelligently complete the comprehensive evaluation of multiple criteria and multiple factors.

[0039] The abandoned mine site selection problem is decomposed into four evaluation layers: target layer, criterion layer, criterion sub-layer and scheme layer. The target hierarchy model is shown in Figure 1 .

[0040] Define the target layer A :A comprehensive geological evaluation method for the reconstruction of abandoned mines into CAES reservoirs. Set as the final global target node of ACO, all ant paths eventually point to the target layer in order to find the global optimal site selection plan.

[0041] Defining Criteria Layers B : B 1 Geological structural features, B 2 Hydrogeological characteristics, B 3 Sealing characteristics of the upper covering layer, B 4 Mine characteristics, B 5 Safety and economic factor characteristics. Each criterion layer serves as a key node in path selection. When ants search for paths, they need to pass through these five key nodes and evaluate the importance of each criterion layer based on their pheromone concentration and heuristic function.

[0042] Defining Criteria Sublayers C : C 1 Fault development, C 2 Fold development, C 3 earthquake history, C 4 Coal seam inclination, C 5 Surface water source distribution, C 6 Groundwater distribution and recharge and discharge conditions, C 7 Conditions of old water in goaf, C 8 Caprock lithology and mechanical parameters, C 9 Thickness of cover layer, C 10 Development of caprock fissures C 11 Rock permeability, C 12 Mine depth, C 13 Mine storage capacity C 14 Mining methods, C 15 Hazardous gas disposal C 16 Roof lithology, C 17 Digital modeling, C18 Distribution of surrounding photovoltaic or wind power energy, C 19 Traffic and power grid conditions in the mining area, C 20 Density of surface structures, C 21 Surface population density, C 22 Policy support and capital investment. Each sub-criteria is a secondary node in the path search. Ants need to go through these secondary nodes to make decisions on the path and choose the next path based on the sub-criteria score and pheromone concentration. The score result is used as a heuristic function for path selection. The higher the value, the more inclined the ant is to choose this path.

[0043] Defining solution layers D : Is the abandoned mine suitable for conversion into a CAES reservoir?

[0044] Table 1 Criteria layer and criteria sub-layer indicators ;

[0045] S22 Construction of judgment matrix based on ACO-AHP According to the criteria B 1 Geological structural features, B 2 Hydrogeological characteristics, B 3 Sealing characteristics of the upper cover layer, B 4 Mine characteristics, B 5 Safety and economic factors, construct the first-level judgment matrix as follows: ; in Representation Guidelines B 1 to B 5 The relative importance between them is determined by the initial value preset by experts and will be automatically adjusted by the ACO algorithm in subsequent steps.

[0046] Construct a judgment matrix for the sub-criteria under each criterion layer, as follows: B 1 Sub-criteria for geological structural characteristics C 1 - C 4 :Fault development( C 1 ), fold development ( C2 ), earthquake history ( C 3 ), coal seam inclination ( C 4 ). Construct the following judgment matrix: ; B 2 Sub-criteria for hydrogeological characteristics C 5 - C 7 :Surface water source distribution( C 5 ), groundwater distribution and recharge and discharge conditions ( C 6 ), old water conditions in goaf ( C 7 ). Construct the following judgment matrix: ; Similarly, construct the remaining criteria layers ( B 3- B 5) Sub-criteria judgment matrix. Each matrix represents the relative importance of its sub-criteria.

[0047] The ACO algorithm optimizes the judgment matrix process as follows: Each ant will select a path in the criterion layer and the criterion sub-layer according to the relative weights in the judgment matrix. Make a decision. The decision is based on the current pheromone concentration of each criterion and sub-criteria and the heuristic function value The heuristic function calculation formula is as follows: ; in is the score value of the sub-criteria, which is normalized and used as the heuristic function for path selection.

[0048] The probability formula for an ant to choose a path is as follows: ; According to the selection probability, decide which path node to choose. After each round of iteration, the ants will collect scoring information during the path selection process. According to the path each ant passes through and its final score, the system will update the pheromone concentration on the path. The higher the score, the more pheromone increases, and the weight adjustment is more significant. The calculation formula is as follows: ; in, Is the path The pheromone increment depends on the score result when the ant passes through this path. As the pheromone concentration increases, the relative importance of the corresponding path (i.e., the weight in the judgment matrix) increases. For example, in the judgment matrix of the criterion layer, if the path B 1 → B 2 The pheromone concentration continues to increase, and the a in the judgment matrix 12 The value will also increase, reflecting B 1 and B 2 As the ant colony's path selection and pheromone concentration change, the weight of each criterion and sub-criteria in the judgment matrix will be automatically adjusted, and the scoring results of each round will be dynamically updated to ensure that the weight adjustment depends on the actual results of the ant's path selection, rather than fixed expert experience. Ultimately, the weight of each criterion and sub-criteria reflects the best choice of the ant colony. The dynamic adjustment formula of the weight is as follows: ; Consistency Verification In the process of constructing the judgment matrix, even after dynamically adjusting the weights using the ACO-regret theory, it is still necessary to verify the consistency of the judgment matrix, using the consistency index CI and the consistency ratio CR for testing.

[0049] ; in, is the maximum eigenvalue of the judgment matrix, is the dimension of the matrix.

[0050] According to the consistency index CI, calculate the consistency ratio CR: ; in, RI is a random consistency indicator. CR <0.1, the judgment matrix passes the consistency test; if CR >0.1, the matrix needs to be modified. The ant colony algorithm will modify the inconsistent weights according to the path selection results. For those problem paths that cause inconsistency, the system will attenuate their pheromone concentrations and adjust the heuristic function , the system guides the ants to choose other paths, thus gradually eliminating inconsistencies. The pheromone attenuation calculation formula is: ; in, Indicates the path To Path j The higher the pheromone concentration, the greater the probability that the ant will choose this path; It represents the volatility coefficient of pheromone and is used to describe the attenuation degree of pheromone concentration.

[0051] S23 Preliminary Comprehensive Rating After the consistency test of the judgment matrix is ​​passed, the weights of each criterion and sub-criteria can be extracted according to the eigenvector of the judgment matrix. The comprehensive score of the mine is obtained through weighted aggregation of the criterion layer and the sub-criteria layer.

[0052] For the criteria layer ( B 1 - B 5 ), extract the judgment matrix weight vector: ,in: is the weight of the geological structure feature, is the weight of the hydrogeological feature, and so on, until The sub-criteria under each criterion layer also extract weights from the corresponding judgment matrix. For example, in the geological structure characteristics ( B 1 The weights of the sub-criteria under They correspond to fault development, fold development, earthquake history, and coal seam inclination respectively.

[0053] Criteria Sub-layer Initial Scoring Determined by experts, and then adjusted based on the feedback and regret value of the ant colony. and rating After adjustment, the weighted score for each criterion layer is calculated , the calculation formula is: ; in, and It is the dynamically updated weight and score.

[0054] Overall Rating S It is the weighted sum of each criterion layer, reflecting the contribution of each criterion and sub-criterion to the overall score after ant colony optimization. The calculation formula is as follows: ; in, S Provide an overall score for the mine; is the optimized criterion layer weight.

[0055] S24 Decision Optimization Based on Regret Theory In order to deal with the uncertainty that may exist in the ACO path selection process and avoid local optimal solutions, the introduction of regret theory can prevent some low-scoring paths from being ignored due to poor short-term performance. By calculating the potential gap between each option and the optimal option, the weights and scores in the ACO-AHP model are dynamically corrected. When a path performs poorly but has potential advantages, the calculation of the regret value will increase the weight of the path, ensuring a more comprehensive scoring process and avoiding the trap of local optimal solutions. By introducing the regret value correction, the system can adaptively adjust the scores of each factor and ultimately output a more reasonable weight distribution and scoring results. At present, the scores of the ACO-based criteria and criterion sub-layers have been calculated. The actual score of the currently completed path is expressed as , by comparing the current path score with the potential optimal score, the regret value is calculated , the calculation formula is as follows: ; in, is the potential optimal score of the unselected path, is the score of the current path.

[0056] If regret is worth it , indicating that the current path is inferior to other paths and needs to be corrected. The larger the regret value, the greater the correction required. , it means that the score of the current path is consistent with the optimal path and does not need to be corrected.

[0057] The revised state transition probability formula is as follows: ; The revised pheromone update formula is as follows: ; The weight adjustment correction formula is as follows: ; The revised scoring formula is as follows: ; in, is the correction coefficient, which is used to control the adjustment strength of the regret value, usually Take a value between.

[0058] The architectural design of the intelligent decision-making system is based on modularization, high coupling, and parallel computing to ensure that the system can maintain high efficiency and stability when processing complex multi-source data and performing dynamic optimization. The overall design goal of the system architecture is to achieve intelligent, multi-dimensional, and global optimization of decisions on the reconstruction of abandoned mines into CAES. The architecture takes multi-source data input, judgment matrix generation and optimization, and comprehensive scoring and decision support as the three main lines as the core processes, and ensures the achievement of the global optimal solution through dynamic adjustment of ACO and regret theory. Its design concept combines the intelligent decision-making system with the real-time feedback mechanism, so that the system not only relies on static data, but can also be optimized in real time based on the latest input data. The overall architecture of the system is divided into the following key levels: Data layer: responsible for processing data input from multiple sources, and performing data preprocessing, cleaning and normalization to ensure the consistency and accuracy of subsequent model calculations. This layer also includes data dynamic update functions to cope with data changes or uncertainties.

[0059] Calculation and optimization layer: core algorithm layer, including AHP judgment matrix generation, ACO path optimization, regret value correction and dynamic score calculation. The task of this layer is to adjust the score and weight distribution through multiple iterations to finally obtain the optimal score result.

[0060] User interaction and visualization layer: The scoring results are presented through intuitive interfaces and charts, supporting users to compare and analyze multiple site selection options. This layer also includes a data input interface, a result display module, and a decision support module, which can generate a detailed site selection report.

[0061] System control and feedback layer: This layer implements the control logic of the system, including data flow control, model parameter adjustment, real-time feedback mechanism, and algorithm iteration control. The innovation of the system control layer is the introduction of a dynamic adjustment strategy based on an adaptive feedback mechanism, which can make real-time score corrections based on user preferences or changes in external conditions.

[0062] In order to improve the maintainability and scalability of the system, the system is divided into the following modules. Each module runs independently and interacts with data and tasks through API and middleware.

[0063] (1) Data input and management module: ① Multi-source data input: supports the collection of multi-dimensional data such as geology, hydrology, and engineering parameters. Data sources include field surveys, historical data, remote sensing data, etc.

[0064] ②Data preprocessing: includes functions such as data cleaning, missing value filling, noise filtering, and spatial consistency correction to ensure the consistency and accuracy of data processing.

[0065] ③ Dynamic data update: supports real-time data input and update. When the mine environment or engineering conditions change, the system can automatically adjust the input data through the data dynamic update module, and perform real-time recalculation and decision optimization.

[0066] (2) Judgment matrix generation and consistency verification module: ① Based on the AHP hierarchical analysis method, a judgment matrix is ​​constructed. The system automatically generates the matrix and performs consistency checks to ensure the logical consistency of the input data.

[0067] ②ACO optimization: The system introduces a dynamic path selection mechanism combined with ACO, and simulates multiple decision paths through the ant colony algorithm, which greatly improves the accuracy of weight allocation and avoids human weight allocation errors.

[0068] ③ Real-time consistency adjustment: During the system iteration process, consistency checks are performed at any time to ensure that the matrix after each round of path selection remains consistent.

[0069] (3) Rating and regret theory correction module: ① Scoring mechanism: The system scores each sub-criteria based on actual data. After the scoring is standardized, the score of each path is automatically calculated.

[0070] ② Regret theory correction: The system dynamically corrects the score through regret theory, especially for paths with low scores but greater potential value, calculates the regret value and adjusts its weight to prevent the system from falling into the dilemma of local optimal solution.

[0071] ③Pheromone concentration update: After each round of iteration, the ant colony algorithm updates the pheromone concentration on the path through score feedback and regret value correction, which affects the tendency of path selection in the next round.

[0072] (4) Comprehensive scoring and decision-making module: The system uses a comprehensive score of each criterion and sub-criteria, performs weighting, and ultimately outputs the optimal mine site selection plan.

[0073] ① Multiple iterations convergence: The system uses an iterative algorithm. Each round of iteration is based on the results of the previous scoring and is dynamically adjusted through ACO until all weights and scores converge to the global optimal solution.

[0074] ② Intelligent decision support: The system automatically generates a detailed decision report, which not only contains the scoring results of the mine, but also includes the analysis logic behind the scoring, regret value correction details and other information to help users better understand the basis behind the scoring.

[0075] (5) User interaction and visualization module: ① Visual modular interactive interface: The interface layout is divided into data input, scoring result display and dynamic adjustment areas, and the scoring and weight distribution of each criterion and sub-criteria are intuitively displayed through bar charts, radar charts and other charts. Users can quickly input data, view scores, and adjust weights and parameters through a simple operation interface.

[0076] ② Real-time feedback: Users can adjust model parameters based on scoring results. The system recalculates and adjusts path selection based on user feedback to support dynamic adjustment of decisions.

[0077] Therefore, the present invention adopts the above-mentioned real-time adaptive correction of abandoned mine CAES reservoir site selection intelligent decision-making method. The present invention uses the ACO-AHP hierarchical analysis method to decompose the complex mine site selection problem into multiple levels, and through the construction of a reasonable judgment matrix, quantitative analysis and weight distribution of each influencing factor are carried out, which significantly improves the scientific nature of the site selection decision. At the same time, through the multi-source data fusion method, literature data, field survey data and historical engineering data are effectively integrated, avoiding the evaluation bias caused by a single data source in the traditional method, and ensuring the accuracy of the decision-making results.

[0078] The present invention introduces regret theory to quantify the potential risks and regret values ​​after decision-making caused by weights and scores during the evaluation process, effectively reducing decision-making deviations that may be caused by fluctuations in geological conditions or data errors, and improving the robustness and risk response capabilities of the site selection process.

[0079] The present invention combines ACO-regret theory with AHP to enable the system to have full-process optimization capabilities from initial weight allocation (AHP), to dynamic path optimization (ACO), and then to future scenario simulation and correction (regret theory). Through multiple rounds of scoring and path selection, the system ensures that the final decision can converge to the global optimal solution, and at the same time has the ability of adaptive correction, realizing intelligent multi-source data fusion and decision optimization, improving the automation level of the system, and reducing human intervention.

[0080] The present invention dynamically processes and updates the real-time monitoring data through an intelligent decision-making system. The system automatically adjusts the weights of the evaluation indicators and the evaluation results according to the changes in the on-site geological conditions, ensuring the real-time and accuracy of the scoring decision, especially in the long-term project implementation process, it can be adjusted according to the latest data, providing highly flexible decision support.

[0081] By combining multi-source data with an intelligent decision-making system, the present invention can comprehensively analyze the geological conditions of the mine during the decision-making process, accurately identify potential geological risk points, and significantly improve the safety of project implementation. At the same time, the system visualization module can intuitively display the geological structure of the mine and the distribution of groundwater, providing key reference information for project planning and construction, and ensuring the long-term stability of reservoir operation.

[0082] The present invention reduces manual intervention and significantly shortens the time required for the decision-making process through automated data fusion and intelligent algorithm optimization. At the same time, systematic data processing and decision support significantly reduce the additional cost investment caused by improper decision-making, especially in reservoir development under complex geological conditions, which has obvious economic benefits.

[0083] The decision-making evaluation method in the present invention is not only applicable to the site selection decision of abandoned mine CAES storage, but also has wide applicability. Its multi-source data fusion and intelligent site selection evaluation system can be extended to other underground space utilization and risk assessment fields in complex geological environments, and has good versatility and expansibility.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.

Claims

1. A real-time adaptive correction method for intelligent decision-making on the site selection of abandoned mine CAES reservoirs, characterized in that: The following steps are involved: S1, collect, integrate and preprocess multi-source data; S2, constructing an intelligent evaluation model for site selection based on AHP with improved ACO; S3. Build a decision-making system and real-time feedback mechanism based on the intelligent evaluation model.

2. According to claim 1, a real-time adaptive correction abandoned mine CAES reservoir site selection intelligent decision-making method is characterized by: S1 includes the following steps: S11. Collect and integrate five types of data: geotechnical characteristics, hydrogeological characteristics, sealing characteristics of the overburden, mine characteristics, and safety and economic factors; S12. Data preprocessing: including data cleaning, missing value filling, noise filtering, and spatial consistency correction to ensure the consistency and accuracy of data processing.

3. According to claim 2, a real-time adaptive correction abandoned mine CAES reservoir site selection intelligent decision-making method is characterized by: S2 includes the following steps: S21. Construct a hierarchical analysis structure to decompose the abandoned mine site selection problem into four evaluation structures: target layer, criterion layer, criterion sub-layer and scheme layer; S22, construct a judgment matrix based on ACO-AHP; S23, preliminary comprehensive score, through the weighted summary of the criterion layer and the sub-criteria layer, to obtain the comprehensive score of the mine; S24. Based on the correction of regret theory, the uncertainty that may exist in the ACO path selection process is handled to avoid local optimal solutions.

4. The method for intelligent decision-making of abandoned mine CAES storage site selection with real-time adaptive correction according to claim 3 is characterized in that: S22 includes the following steps: S221, constructing a first-level judgment matrix according to the criterion layer, and constructing a judgment matrix for the sub-criterion under each criterion layer; S222, optimizing the judgment matrix by using the ACO algorithm, making decisions according to the relative weights aij in the judgment matrix, and simulating multiple decision paths by using the ant colony algorithm; S223. Verify the consistency of the matrix using the consistency index CI and the consistency ratio CR. During the system iteration process, perform consistency checks at any time to ensure that the matrix remains consistent after each round of path selection.

5. The method for intelligent decision-making of abandoned mine CAES storage site selection with real-time adaptive correction according to claim 4 is characterized in that: S24 Regret Value The calculation formula is as follows: ； in, is the potential optimal score of the unselected path, is the score of the current path; Regret Value , indicating that the current path is inferior to other paths and needs to be corrected; regret value , indicating that the score of the current path is consistent with the optimal path and does not need to be corrected.

6. The method for intelligent decision-making of abandoned mine CAES storage site selection with real-time adaptive correction according to claim 5 is characterized by: The revised state transition probability formula in S24 is: ； in, represents the information density, represents the heuristic function value; The revised pheromone update formula is: ； in, It represents the volatility coefficient of pheromone, which is used to describe the attenuation degree of pheromone concentration; Weight adjustment correction formula: ； The revised scoring formula is: ； in, is the correction coefficient, which is used to control the adjustment strength of the regret value, usually Take a value between.

7. The method for intelligent decision-making of abandoned mine CAES storage site selection with real-time adaptive correction according to claim 6 is characterized by: Step S3 includes data flow control, model parameter adjustment, real-time feedback mechanism and algorithm iteration control. The user adjusts the model parameters according to the scoring results, and the system recalculates and adjusts the path selection according to the user's feedback to support dynamic adjustment of decisions.

8. The method for intelligent decision-making of abandoned mine CAES storage site selection with real-time adaptive correction according to claim 7 is characterized by: Step S3 also includes an interface layout divided into data input, scoring result display and dynamic adjustment areas, which intuitively displays the scores and weight distribution of each criterion and sub-criteria through bar charts and radar charts. Users can quickly input data, view scores, and adjust weights and parameters through a simple operation interface.

Images