A geological warm pressure analysis method based on multi-agent collaborative decision-making

Abstract

The application provides a geological temperature and pressure analysis method based on multi-agent collaborative decision-making, and relates to the technical field of geological analysis. The method comprises the following steps: by establishing sample identification and performing quality diagnosis on the temperature and pressure analysis data, first distinguishing the balanced and unbalanced sample states, then combining the tectonic background to perform constrained screening and executability analysis on the temperature and pressure meter, performing parallel joint solving of multiple temperature and pressure meters under unified algorithm configuration, introducing a correction mechanism for the unbalanced sample, finally performing weighted fusion on the consistency, compatibility and confidence of multiple results, and outputting the temperature and pressure conclusion with interval significance and reliability annotation. The application can improve the accuracy of geological temperature and pressure analysis in complex geological environment.

Description

Technical Field

[0001] This invention relates to the technical field of geological analysis, and specifically to a geological temperature and pressure analysis method based on multi-agent collaborative decision-making. Background Technology

[0002] Current geothermal-barotropic analysis techniques are typically based on mineral coexistence and thermodynamic equilibrium assumptions. They involve selecting specific mineral pairs and substituting them into corresponding thermobarometry formulas or thermodynamic models to invert the temperature and pressure conditions during rock formation or evolution. These methods largely rely on human experience in selecting thermobarometers and assume that mineral composition represents a single, stable equilibrium state. In practical applications, they often follow a predetermined process of composition testing, endmember calculation, activity correction, and thermobaric solution, yielding single or limited numerical results. With the development of micro-area analysis techniques, existing technologies have gradually introduced calculation methods involving multiple measurement points or multiple mineral combinations. However, overall, they still primarily rely on single-model or few-model sequential calculations, and the interpretation of results heavily depends on researchers' subjective judgment of the model's applicability.

[0003] However, under complex geological conditions, especially in high-strain shear zones, mylonite systems, or multi-stage metamorphic environments, mineral composition is often significantly affected by compositional redistribution, diffusion restriction, and non-equilibrium kinetic processes, leading to contradictory or even significantly distorted calculation results from different thermobarometers. Current technologies lack an analytical mechanism that can systematically characterize the sample equilibrium state at a holistic level, uniformly constrain the applicable conditions of thermobarometers, and conduct consistency assessments of multi-model calculation results. Conclusions are typically drawn by manually eliminating outliers or empirically selecting a particular thermobarometer, making it difficult to quantify the reliability and sources of uncertainty of different model results. This can easily lead to misleading thermobaric judgments in complex tectonic settings. Summary of the Invention

[0004] This invention provides a geological temperature and pressure analysis method based on multi-agent collaborative decision-making, which can improve the accuracy of geological temperature and pressure analysis for complex geological environments.

[0005] In a first aspect, the present invention provides a geological temperature and pressure analysis method based on multi-agent collaborative decision-making, the method comprising: Acquire the temperature and pressure analysis datasets bound to the target sample, merge them to establish sample identifiers, and form a sample data package; Data quality diagnosis is performed based on the sample data packets to generate a quality profile set, and the sample data packets are divided into balanced subsets and unbalanced subsets according to the quality profile set and bound with balance labels. A task constraint set is constructed by combining the balance label with the sample structure background data, and a candidate set of thermobarometers matching the mineral pair identifiers is generated under the constraints of the task constraint set. An executability analysis is performed on the candidate thermometer / barometer set to form an executable thermometer / barometer set, and an algorithm configuration set is generated for the executable thermometer / barometer set; Based on the set of executable thermobarometers and the set of algorithm configurations, the temperature and pressure are solved in parallel to generate temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate. Non-equilibrium correction is introduced on the non-equilibrium subset to output the set of temperature and pressure calculation results. A consistency review is performed on the set of temperature and pressure calculation results. A recommended value range and a comprehensive confidence score are generated by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

[0006] Based on the above technical solutions, preferably, the step of performing data quality diagnosis based on the sample data packets to generate a quality profile set, and dividing the sample data packets into balanced subsets and unbalanced subsets according to the quality profile set and binding balance labels, specifically includes: The mineral pair identifiers and mineral chemical composition data in the sample data package are judged to be compatible in terms of elemental composition, stoichiometry, and endmember constraints, and incompatible records are marked as abnormal records. After removing the abnormal records, the remaining valid records are compared with the key element requirement template corresponding to the mineral pair identifier to calculate the missing ratio, substitution ratio and reasonableness of the key elements, thereby generating a component integrity index. By establishing charge conservation constraints for variable valence elements and combining them with the analysis of mass metadata, the consistency of valence state inference results at different measurement points in the same mineral spatial partition data is judged to generate a valence state consistency index. An endmember solvability assessment is performed on the valid records. The mineral chemical composition data is solved by endmember decomposition within the feasible region defined by the composition integrity index and the valence state consistency index. An endmember solvability index is generated based on the proportion of feasible solutions, the constraint conflict situation and the rationality of the endmember mole fraction. Spatial partition stability assessment is performed on the valid records. By aligning the endmember mole fraction sets of the same mineral in different mineral spatial partitions, a cross-partition composition gradient characterization is constructed. The rebalancing characteristics and kinetic interpretability of the composition gradient are determined by combining the sample construction background data, thereby generating a spatial partition stability index. The quality profile set is constructed based on the component integrity index, the valence consistency index, the endmember solvability index, and the spatial partition stability index. Based on the quality profile set, the sample data packets are subjected to balance discrimination. Records that simultaneously meet the end-member solvability threshold, spatial partition stability threshold, and valence consistency threshold are included in the balanced subset and bound with balance labels. Meanwhile, records that do not meet any of the threshold conditions are included in the unbalanced subset and bound with unbalanced source labels.

[0007] Based on the above technical solutions, preferably, the step of constructing a task constraint set by combining the balance label with the sample tectonic background data, and generating a candidate set of thermobarometers matching the mineral pair identifiers under the constraints of the task constraint set, specifically includes: The balance label and the sample tectonic background data are jointly aligned to form a tectonic balance context, which is simultaneously bound to mineral pair identifiers, balance labels, non-equilibrium source labels, tectonic unit identifiers, shear band location identifiers, deformation strength grades, and deformation indication features under the sample identifier. Based on the constructed equilibrium context, applicable mineral combination constraints are generated. The mineral combination rules in the expert knowledge base are mapped by mineral pair identifiers and combined with the equilibrium label to exclude thermometer and barometer types that do not have non-equilibrium fault tolerance. Deformation strength constraints and strain rate sensitive constraints are generated based on the deformation strength level and the deformation indication features; Based on the aforementioned unbalanced source labels, rebalancing risk constraints are generated, and spatial partition usage constraints are generated simultaneously. The pressure range constraint and temperature range constraint are generated by combining the structural unit identifier, the shear band location identifier, the deformation strength grade, and the balance label; The applicable mineral combination constraint, the deformation strength constraint, the strain rate sensitivity constraint, the rebalancing risk constraint, the spatial partitioning usage constraint, the pressure range constraint, and the temperature range constraint are combined to form the task constraint set; Based on the task constraint set, the initial set of thermometers is retrieved in the expert knowledge base using mineral pair identifiers as the main search key. The initial set of thermometers is filtered and sorted according to the task constraint set to generate the candidate set of thermometers.

[0008] Based on the above technical solutions, preferably, the step of performing executability analysis on the candidate thermometer / barometer set to form an executable thermometer / barometer set specifically includes: While maintaining the correspondence between the mineral pair identifiers and the thermobarometer candidate set, a candidate description package set is read for each thermobarometer candidate to form a candidate execution context; Based on the candidate execution context, the completeness of the candidate execution data of the thermometer and barometer is judged, and the thermometer and barometer candidates that do not meet the data requirement template are marked as data unexecutable candidates; After eliminating the data non-executable candidates, the thermobarometer candidates that do not have non-balance fault tolerance and rely on the assumption of complete balance are marked as premise conflict candidates. After eliminating the aforementioned conflict candidates, the theoretical boundary consistency judgment is performed on the remaining thermobarometer candidates, and the thermobarometer candidates that meet the conditions within the feasible domain are marked as boundary risk candidates. After eliminating the boundary risk candidates, a construction sensitivity consistency judgment is performed on the remaining thermobarometer candidates, and thermobarometer candidates that are incompatible with the construction environment are marked as construction incompatible candidates. After eliminating the incompatible candidates, a computational reproducibility evaluation is performed on the remaining thermobarometer candidates, and the thermobarometer candidates that are sensitive to disturbances are marked as numerically unstable candidates. Based on the data completeness judgment results, premise consistency judgment results, theoretical boundary consistency judgment results, construction sensitivity consistency judgment results, and reproducibility index, the thermobarometer candidates in the thermobarometer candidate set that are not marked as any unexecutable type are included in the executable thermobarometer set.

[0009] Based on the above technical solutions, preferably, the configuration set for the executable thermobarometer set generation algorithm specifically includes: Based on the sample identifier, a data mapping configuration is generated for each candidate read data requirement template of the executable thermobarometer set; The endmember decomposition strategy is determined based on the solid solution endmember set, endmember constraint conditions, and task constraint set, and the endmember decomposition constraints are selected according to the balance label to output the endmember solution configuration. The activity model configuration is generated by selecting a matching activity model based on the set of applicable premises and the set of theoretical boundaries. Based on the price state consistency index and the non-equilibrium source label, a fixed price state strategy is selected to generate a price state processing configuration; The unbalanced correction channel is enabled based on the balance label, and an unbalanced correction configuration is generated. The calibration data is extracted from the theoretical boundary set and combined with the task constraint set to form the calibration feasible domain, generating the experimental anchoring configuration. The selection of solution variables, initial value generation, search strategy, convergence criterion, abnormal termination conditions and restart strategy are uniformly and solidified to generate joint solution control configuration; The data mapping configuration, the endmember solving configuration, the activity model configuration, the valence state processing configuration, the non-equilibrium correction configuration, the experimental anchoring configuration, and the joint solving control configuration are aggregated and a consistency check is performed to form the algorithm configuration set.

[0010] Based on the above technical solutions, preferably, the parallel execution of the joint temperature and pressure solution based on the executable thermobarometer set and the algorithm configuration set, generating temperature results, pressure results, and uncertainty sets corresponding to each thermobarometer candidate while maintaining the result traceability chain, and introducing unbalance correction on the unbalance subset, specifically includes: Based on the data mapping configuration, variable assembly is performed on the sample data package to generate a candidate variable package; Perform endmember solving according to the endmember solving configuration to generate a set of endmember mole fractions; Generate an activity variable package based on the activity model configuration; Based on the valence state processing configuration, valence state processing is performed on the activity variable package to generate a modified activity variable package; Based on the joint solution control configuration, the joint solution main process is started, and a joint search is performed on the candidate values ​​of temperature and pressure within the solution domain constraint. At each candidate point, the temperature result and pressure result are determined according to the modified activity variable package. When the balance label indicates an unbalanced subset, an unbalanced correction is introduced in the joint solution master process. The unbalanced correction is based on the diffusion length parameter, composition gradient parameter and rebalancing weight parameter in the unbalanced correction configuration to weight and fuse the endmember contributions of the kernel partition and the edge partition. After obtaining the temperature and pressure results, an uncertainty set bound to the configuration identifier is generated; After the temperature results, pressure results, uncertainty set, and solution status are solved in parallel, the temperature and pressure calculation result set is formed under the sample identifier.

[0011] Based on the above technical solutions, preferably, the consistency review of the temperature and pressure calculation result set is performed by generating a recommended value range and a comprehensive confidence score through a weighted fusion of result dispersion, theoretical compatibility, and confidence level, so as to output a structured evaluation result bound to the sample identifier, specifically including: The set of temperature and pressure calculation results is subjected to reviewability screening to form a review result pool; For each thermometer / barometer candidate in the review result pool, a standardized result representation is constructed; Based on the standardized results, the dispersion of the calculation results is represented, and temperature dispersion index and pressure dispersion index are generated respectively. After obtaining the temperature dispersion index and the pressure dispersion index, a theoretical compatibility judgment is performed. Based on the set of applicable premises, the set of theoretical boundaries, the set of task constraints, the balance label and the non-balance source label, the theoretical assumption compatibility, the overlap of applicable intervals and the consistency of construction sensitivity among the candidate thermometers and barometers in the review result pool are jointly judged, and a compatibility matrix and compatibility index are generated. A candidate weight set is generated based on the compatibility matrix and empirical confidence parameters. The candidate weight set is then corrected by combining risk level labels, price state processing strategy stability, unbalanced correction activation status, and calibration feasible region constraint status to form fused weights. The temperature dispersion index, the pressure dispersion index, the compatibility index, and the candidate weight set are then weighted and fused to generate weighted center values ​​in the temperature and pressure dimensions, respectively. These values ​​are then combined with the uncertainty set to expand the range of recommended temperature and pressure values.

[0012] In a second aspect of the invention, a geothermal-barotropic analysis device based on multi-agent collaborative decision-making is provided. The device is used to execute a geothermal-barotropic analysis method based on multi-agent collaborative decision-making as described above. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to acquire and merge the temperature and pressure analysis dataset bound to the target sample to establish a sample identifier, thereby forming a sample data package; The processing module is used to perform data quality diagnosis based on the sample data package to generate a quality profile set, and to divide the sample data package into a balanced subset and an unbalanced subset according to the quality profile set and bind a balance label. The processing module is used to construct a task constraint set by combining the balance label and the sample structure background data, and generate a set of thermobarometer candidates that match the mineral pair identifiers under the constraints of the task constraint set. The processing module is used to perform executability analysis on the candidate set of thermobarometers to form an executable set of thermobarometers, and generate an algorithm configuration set for the executable set of thermobarometers; The processing module is used to perform joint temperature and pressure solving in parallel based on the set of executable thermobarometers and the set of algorithm configurations, generate temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate, introduce non-equilibrium correction on the non-equilibrium subset, and output a set of temperature and pressure calculation results. The output module is used to perform a consistency review on the set of temperature and pressure calculation results, and generate a recommended value range and a comprehensive confidence score by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

[0013] In a third aspect of the invention, an electronic device is provided, including a processor, a memory, a user interface, and a network interface, wherein the memory is used to store instructions, the user interface and the network interface are both used to communicate with other devices, and the processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any of the preceding embodiments.

[0014] In a fourth aspect of the invention, a non-transitory computer-readable storage medium is provided, the computer-readable storage medium storing instructions that, when executed, perform the method as described in any of the preceding claims.

[0015] In summary, one or more technical solutions provided in the embodiments of the present invention have at least the following technical effects or advantages: This invention improves the accuracy of geothermal and barometric analysis in complex geological environments by upgrading the traditional analysis process, which relies on a single equilibrium assumption and human experience to select a thermobarometer, through a multi-agent collaborative decision-making mechanism. This process involves a systematic discrimination and joint evaluation of sample state, tectonic background, and model applicability. Data quality diagnosis and quality profile sets explicitly distinguish between equilibrium and non-equilibrium subsets, avoiding the blind application of equilibrium thermobarometers under conditions of strong deformation or dynamic non-equilibrium. Simultaneously, it incorporates tectonic background data to uniformly constrain the applicability of thermobarometers and introduces non-equilibrium correction and uncertainty quantification in parallel joint solutions. Finally, a weighted fusion of result dispersion, theoretical compatibility, and confidence level outputs a recommended value range and a comprehensive confidence score. This overall approach suppresses misleading results caused by component redistribution and non-equilibrium effects under complex tectonic conditions, ensuring that thermobaric conclusions possess both numerical stability and geological interpretation consistency, thereby significantly improving the reliability of thermobaric inversion in complex geological environments. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a geological temperature and pressure analysis method based on multi-agent collaborative decision-making disclosed in an embodiment of the present invention. Figure 2 This is a schematic diagram of a geological temperature and pressure analysis device based on multi-agent collaborative decision-making disclosed in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of an electronic device disclosed in an embodiment of the present invention.

[0017] Explanation of reference numerals in the attached drawings: 201, acquisition module; 202, processing module; 203, output module; 301, processor; 302, communication bus; 303, user interface; 304, network interface; 305, memory. Detailed Implementation

[0018] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0019] In the description of the embodiments of the present invention, words such as "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design described as "for example" or "for instance" in the embodiments of the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Rather, the use of words such as "for example" or "for instance" is intended to present the relevant concepts in a specific manner.

[0020] In the description of the embodiments of the present invention, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0021] Existing geothermal-barometry analysis methods are mainly based on the assumption of mineral thermodynamic equilibrium and the selection of thermobarometers based on human experience. They usually use a single or a few models to calculate mineral composition data and give deterministic thermobaric results. However, in complex geological environments such as high-strain shear zones, mylonite systems, and multi-stage metamorphic alteration, mineral composition is easily affected by composition redistribution and non-equilibrium kinetic processes, resulting in discrete and inconsistent results from different thermobarometers. Existing technologies lack a systematic discrimination of sample equilibrium state, unified constraints on the applicable premises of thermobarometers, and a comprehensive evaluation mechanism for the reliability and uncertainty of multi-model results. They often rely on subjective selection results, which can easily lead to unstable or even misleading thermobaric analysis conclusions in complex tectonic contexts.

[0022] This embodiment discloses a geological temperature and pressure analysis method based on multi-agent collaborative decision-making, referring to... Figure 1 This includes the following steps S110-S160: S110: Acquire the temperature and pressure analysis dataset bound to the target sample, merge it to establish a sample identifier, and form a sample data package.

[0023] This invention discloses a geological temperature and pressure analysis method based on multi-agent collaborative decision-making, which is applied to a server. The server includes, but is not limited to, electronic devices such as mobile phones, tablets, wearable devices, and PCs (Personal Computers), and can also be a backend server running a geological temperature and pressure analysis method based on multi-agent collaborative decision-making. The server can be implemented using a standalone server or a server cluster composed of multiple servers.

[0024] The decision-making agent initiates sample initialization processing. The decision-making agent receives the analysis request of the target sample and generates a unique sample identifier. The sample identifier serves as a global index for all subsequent processing flows and is used to transfer status and results among multiple agents. At the same time, the decision-making agent determines the mineral pair identifier, analysis scale type, and expected temperature and pressure inversion target corresponding to the target sample based on the analysis request, and establishes a sample processing context under the sample identifier to constrain the scope and data type of subsequent data acquisition.

[0025] After the sample processing context is established, the decision-making agent coordinates the quality assessment agent to perform data scheduling on the original data source associated with the target sample. The original data source includes at least mineral chemical composition data, mineral spatial partitioning data, sample structural background data, and analytical quality metadata. Among them, mineral chemical composition data is used to characterize the elemental composition information of the mineral pair to identify the corresponding mineral; mineral spatial partitioning data is used to distinguish the compositional differences of different spatial locations such as the core, edge, fracture neighborhood, and fine-grained zone; sample structural background data is used to describe the structural unit, shear zone location, and deformation strength level of the target sample; and analytical quality metadata is used to describe the analytical method, calibration status, detection limit consistency, and outlier marking.

[0026] After the raw data scheduling is completed, the quality assessment agent performs format unification and semantic verification on various types of raw data. Format unification is used to convert mineral chemical composition data and mineral spatial partition data from different sources into a consistent data structure expression. Semantic verification is used to ensure that there is no ambiguous mapping between mineral pair identifiers, measurement point identifiers, spatial partition identifiers and element identifiers, and to generate corresponding data status markers when data missing, semantic conflict or abnormal label inconsistency is found, so as to ensure that subsequent processing is only carried out on the basis of traceable data.

[0027] After format unification and semantic verification are completed, the quality assessment agent performs preliminary association and binding processing on the data that has passed the verification. The association and binding processing associates the mineral chemical composition data with the corresponding mineral spatial partition data under the sample identifier. At the same time, the sample structural background data and analytical quality metadata are added as context attributes to the association results, thereby forming a composite data record with the measurement point as the basic unit, and maintaining the traceability relationship between each composite data record and the original data source.

[0028] After the composite data record is constructed, the decision-making agent performs aggregation processing on the composite data record. The aggregation processing grouped the composite data records belonging to the same mineral combination according to the mineral pair identifier, and generated a structured data container under the sample identifier to contain the mineral pair identifier, the composite data record set, the construction background attribute set and the quality metadata set, thereby forming a sample data package with unified semantics and complete context constraints.

[0029] After the sample data package is formed, the decision agent registers the sample data package to the processing status table shared by multiple agents, and binds the data integrity status and availability status to the sample data package. This serves as the sole input object for subsequent calls and processing by the quality assessment agent, knowledge reasoning agent, execution agent, and review agent, thereby completing the entire process of acquiring the temperature and pressure analysis dataset bound to the target sample, merging it, establishing sample identification, and forming the sample data package.

[0030] The decision-making agent consists of a task entry layer, a semantic parsing layer, a global state layer, an orchestration and control layer, and an output encapsulation layer. The task entry layer receives analysis requests and extracts the external identification information of the target sample. The semantic parsing layer parses the analysis request into internal semantic fields such as mineral pair identifiers, analysis scale type, and expected temperature and pressure inversion targets. The global state layer generates and maintains sample identifiers and sample processing contexts and processing state tables formed around the sample identifiers. The orchestration and control layer distributes instructions among multiple agents and maintains the dependencies and backtracking relationships between each processing stage. The output encapsulation layer registers, archives, and distributes sample data packets, task constraint sets, algorithm configuration sets, and structured evaluation results in a unified structure, thereby ensuring that all subsequent agents work collaboratively in the same process using the sample identifier as a unique index.

[0031] The quality assessment agent consists of a data access layer, a data normalization layer, a semantic consistency layer, an association and binding layer, a quality labeling layer, and a traceability record layer. The data access layer receives mineral chemical composition data, mineral spatial zoning data, sample structural background data, and analytical quality metadata from the decision-making agent. The data normalization layer unifies data fields, unit expressions, and missing data rules from different sources and outputs standardized records. The semantic consistency layer verifies the mapping relationships between mineral pair identifiers, measurement point identifiers, spatial zoning identifiers, and element identifiers and generates data status labels. The association and binding layer binds mineral chemical composition data and mineral spatial zoning data at the measurement point level under the sample identifier and adds structural background attributes and quality metadata. The quality labeling layer writes anomaly, missing data, and availability labels to composite data records to support subsequent balance label generation. The traceability record layer writes the reasons for each field transformation, record removal, and labeling occurrence into the traceability chain field, ensuring that the sample data package has a traceable data quality foundation before entering subsequent calculations.

[0032] The knowledge reasoning agent consists of a knowledge retrieval layer, a constraint generation layer, a conflict disambiguation layer, a candidate screening layer, a candidate solidification layer, and a version management layer. The knowledge retrieval layer accesses an expert knowledge base and retrieves the applicable premise set, theoretical boundary set, data requirement template, and empirical confidence parameters related to the thermobarometer by mineral pair identifier. The constraint generation layer maps equilibrium labels, non-equilibrium source labels, and sample construction background data into a task constraint set and forms a computable constraint expression. The conflict disambiguation layer identifies conflicts within the task constraint set and between the task constraint set and the thermobarometer theoretical boundary set, outputting consistent arbitration results and arbitration bases. The candidate screening layer filters and sorts the initial thermobarometer set under the constraints of the task constraint set to generate a candidate thermobarometer set. The candidate solidification layer generates a candidate description package set for each thermobarometer candidate and writes a construction sensitivity label and an applicability explanation field. The version management layer uniformly identifies the versions of knowledge entries, rule tables, and candidate description packages, ensuring that the generation process of the thermobarometer candidate set is reproducible and auditable.

[0033] The executing agent consists of an executability discrimination layer, a configuration generation layer, a parallel solution layer, a non-equilibrium correction layer, an uncertainty generation layer, and an anomaly management layer. The executability discrimination layer evaluates the candidate thermometer / barometer set based on the candidate description package set, assessing data completeness, premise consistency, theoretical boundary consistency, construction compatibility, and reproducibility, and outputs an executable thermometer / barometer set along with risk level labels. The configuration generation layer generates data mapping configurations, endmember solution configurations, activity model configurations, valence state processing configurations, non-equilibrium correction configurations, experimental anchoring configurations, and joint solution control configurations under configuration identifiers, forming an algorithm configuration set. The parallel solution layer... The system is used to encapsulate each configuration identifier into an independent solution task and execute the joint temperature and pressure solution in parallel under the resource isolation rule to output temperature and pressure results. The unbalance correction layer is used to call the unbalance correction configuration to correct the candidate point evaluation and result generation process when the balance label indicates an unbalance subset. The uncertainty generation layer is used to generate an uncertainty set based on data disturbance rules, parameter disturbance rules and boundary sensitivity rules and bind it to the result traceability chain. The anomaly management layer is used to record events such as data unsolvability, out-of-bounds handling triggers, abnormal termination and restart and form a solution anomaly record, thereby ensuring that the temperature and pressure calculation result set is both comparable and traceable.

[0034] The review agent consists of an reviewability screening layer, a standardization representation layer, a dispersion assessment layer, a compatibility assessment layer, a weight fusion layer, a score generation layer, and a report generation layer. The reviewability screening layer divides the temperature and pressure calculation results into a review result pool and a rejection result pool based on the solution status, uncertainty set completeness, and anomaly cause fields. The standardization representation layer converts the temperature, pressure, and uncertainty sets in the review result pool into standardized result representations under a unified semantic framework while retaining key traceability fields. The dispersion assessment layer calculates temperature and pressure dispersion indices on the standardized result representations and introduces risk level labels and uncertainty sets for robust weighting. The compatibility assessment layer... The first layer generates a compatibility matrix and compatibility index based on the applicable premise set, theoretical boundary set, task constraint set, and construction sensitivity label. The second layer, weight fusion layer, corrects the candidate weight set based on empirical confidence parameters and in combination with the compatibility matrix and traceability health. The third layer, score generation layer, integrates information such as dispersion index, compatibility index, candidate weight set concentration, and elimination ratio into a comprehensive confidence score while retaining decomposable fields. The fourth layer, report generation layer, outputs structured evaluation results bound to sample identifiers and simultaneously outputs a compatibility matrix summary, a candidate weight set summary, and aggregated information of key traceability fields, thereby making the final recommended value range and comprehensive confidence score interpretable and auditable.

[0035] S120: Perform data quality diagnosis based on sample data packets to generate a quality profile set, and divide the sample data packets into balanced subsets and unbalanced subsets according to the quality profile set and bind balance labels.

[0036] In one possible implementation, data quality diagnosis is performed based on sample data packets to generate a quality profile set. The sample data packets are then divided into balanced and non-balanced subsets based on the quality profile set and bound with balance labels. Specifically, this includes: determining the compatibility of mineral pair identifiers and mineral chemical composition data in the sample data packets in terms of elemental composition, stoichiometry, and endmember constraints, and marking incompatible records as anomalous records; after removing anomalous records, the remaining valid records are statistically analyzed using the key element requirement templates corresponding to the mineral pair identifiers to determine the missing proportion, substitution proportion, and reasonableness of measured values ​​of key elements, thereby generating a compositional integrity index; for valid records, by establishing charge conservation constraints for variable valence elements and combining this with the analysis of quality metadata, consistency judgment is performed on the valence state inference results of different measurement points in the same mineral spatial partition data to generate a valence state consistency index; and an endmember solvability assessment is performed on valid records to determine the compatibility between the compositional integrity index and the valence state... Within the feasible domain defined by the consistency index, endmember decomposition is performed on the mineral chemical composition data. An endmember solvability index is generated based on the proportion of feasible solutions, constraint conflicts, and the rationality of the endmember mole fraction. Spatial partition stability assessment is performed on valid records. By aligning the endmember mole fraction sets of the same mineral in different mineral spatial partitions, a cross-partition composition gradient characterization is constructed. The rebalancing characteristics and kinetic interpretability of the composition gradient are determined in conjunction with the sample tectonic background data, thereby generating a spatial partition stability index. A quality profile set is constructed based on the composition integrity index, valence consistency index, endmember solvability index, and spatial partition stability index. Based on the quality profile set, balance discrimination is performed on the sample data packets. Records that simultaneously meet the endmember solvability threshold, spatial partition stability threshold, and valence consistency threshold are assigned to the balanced subset and bound with a balance label. Records that do not meet any of the threshold conditions are assigned to the unbalanced subset and bound with an unbalanced source label.

[0037] Specifically, when the quality assessment agent performs compatibility determination on the mineral pair identifiers and mineral chemical composition data in the sample data package under sample identifier constraints, the decision agent first sends the mineral type dictionary and end-member rule table corresponding to the mineral pair identifiers to the quality assessment agent. Based on this, the quality assessment agent performs a set consistency check between the element identifier set in the mineral chemical composition data and the element identifier set allowed to appear in the mineral pair identifiers. Element composition compatibility is used to determine whether there are template prohibited elements, missing key elements, or incorrect element alias mapping. Subsequently, a stoichiometric relationship check is performed on the mineral chemical composition data of each measurement point. The stoichiometric relationship refers to the coordination and charge constraints that should be satisfied between the cation count and anion count derived from the element count under the selected normalized benchmark. During the check, the quality assessment agent marks measurement points with abnormal deviations as abnormal records and writes them into the abnormal reason field. Next, a consistency judgment is performed on the endmember constraints. Endmember constraints refer to the restrictions on the nonnegativity, summation, and certain coupling ratios of the set of endmember mole fractions when a mineral solid solution is composed of linear or nonlinear combinations of several endmember components. The quality assessment agent determines whether there is an unrealizable endmember combination or an endmember rule conflict at the measurement point through a rapid feasibility check, and uniformly marks incompatible measurement points as abnormal records to form a basis for subsequent removal. Abnormal records refer to measurement point records that are incompatible with the mineral pair identifier in any dimension of elemental composition, stoichiometry, or endmember constraints and do not have a repairable mapping relationship.

[0038] When generating the component integrity index after removing abnormal records, the quality assessment agent first provides a key element requirement template. This template is used to solidify the necessary element set, optional element set, and mapping rules for substitute elements for each mineral pair. The quality assessment agent then counts the missing elements in the necessary element set at each measurement point and generates the missing proportion. Simultaneously, it counts the substitution events triggered by the substitute element mapping rules to generate a substitution proportion. This substitution proportion characterizes the amplified uncertainty risk caused by approximating necessary elements with substitute elements. In the measurement value reasonableness judgment, the quality assessment agent performs interval verification on the measurement value of each element based on the allowable interval table provided by the key element requirement template. Measurement point fields exceeding the allowable interval are written into the abnormal field list for subsequent endmember solvability assessment to narrow the feasible region. The expression for the component integrity index is: in, This represents the completeness index of the components, with a value ranging from 0 to 1, and a larger value indicates more complete information. This represents the percentage of missing essential elements, and its value is obtained by dividing the number of missing essential elements by the total number of essential elements. This represents the substitution ratio, and its value is obtained by dividing the number of substitution events by the total number of required elements. This represents the out-of-bounds percentage, and its value is obtained by dividing the number of fields exceeding the allowed range by the total number of fields being validated. , , The weights are represented by a sum of 1. The indicator quantifies the availability of information required for end-member decomposition by applying weighted penalties to three types of incomplete information risks: missing, substitution, and out-of-bounds.

[0039] When generating the valence consistency index by the quality assessment agent, the decision agent first writes the calibration status, detection limit consistency, and outlier marking rules from the analyzed quality metadata into the valence inference context. The quality assessment agent then establishes charge conservation constraints for the set of variable valence elements in the mineral chemical composition data and infers the valence allocation for each measurement point. Variable valence elements refer to elements that may exhibit different valence values ​​under different redox environments and affect charge balance and endmember composition. The charge conservation constraint ensures that the total charge of cations and the total charge of anions are equal within the allowable error range, and based on this, inversely infers the valence ratio of variable valence elements. Under the same mineral spatial partitioning data, the quality assessment agent calculates the dispersion of the valence inference results for different measurement points and, combined with the analyzed quality metadata, reduces the weight of low-confidence measurement points, thereby generating the valence consistency index. The expression for the charge conservation residual is: in, This represents the charge conservation residual, and the smaller the residual, the more fully charge conservation is satisfied. Represents a set of cationic elements. This represents a set of anionic elements. This represents the normalized count of element i. Let i represent the price state value of element i. The expression for the price state consistency index is: in, This represents the price consistency index, with values ​​ranging from 0 to 1. A higher value indicates a more stable price inference. Indicates each measuring point within the same spatial partition The weighted average, This represents the tolerance parameter and is determined by the analysis quality metadata mapping. The coefficient of variation represents the result of valence state inference for variable valence elements. It is used to characterize the dispersion of valence state allocation across measurement points. The above index constrains the repeatability of the valence state processing strategy by simultaneously penalizing deviations from charge conservation and dispersion of valence state allocation.

[0040] When the quality assessment agent performs endmember solvability assessment, the execution agent first provides a unified endmember decomposition solver interface, and the knowledge reasoning agent provides the solid solution endmember set and endmember constraints corresponding to the mineral pair identifiers. The quality assessment agent performs endmember decomposition on the mineral chemical composition data within the feasible region defined by the composition integrity index and valence consistency index. The feasible region refers to the set of variables and the range of variable values ​​allowed to participate in the solution under constraints of missing elements, unstable valence states, and out-of-bounds fields. The goal of endmember decomposition is to find the endmember mole fraction set such that the element count obtained by linear combination of endmembers approximates the observed element count as closely as possible while satisfying the endmember constraints. After solving at each measurement point, the quality assessment agent records whether a feasible solution exists, the number of constraint conflicts, and the non-negativity and summation of the endmember mole fraction set, summing these to form the endmember solvability index. The expression for endmember decomposition is: in, Represents the set of endmember mole fractions. Let the vector of mole fractions of the endmembers be to be determined. Describes the feasible set of endmember constraint definitions and includes and Equal constraints, The mapping matrix representing endmembers to element counts is determined by the solid solution endmember set and stoichiometric rules. This represents the observed element count vector, derived from mineral chemical composition data under a selected normalization standard. The expression for the endmember solvability index is: in, This represents the endmember solvability index, with a value ranging from 0 to 1. A larger value indicates more stable and feasible endmember decomposition. This represents the proportion of feasible solutions, which is obtained by dividing the number of measurement points with feasible solutions by the total number of measurement points participating in the evaluation. This represents the average number of constraint conflicts. Indicates the conflict penalty coefficient. The term represents the degree of nonnegativity violation of endmembers. It is obtained by normalizing the aggregation of negative values ​​in the set of endmember mole fractions. The above index determines whether endmember decomposition has a basis for sustainable use by simultaneously rewarding feasible solution coverage and penalizing constraint conflicts and non-physical endmember results.

[0041] When the quality assessment agent performs spatial partition stability assessment, it first aligns the endmember mole fraction sets of the same mineral in different mineral spatial partitions under the sample identifier. Alignment refers to mapping the endmember mole fraction sets of the core measurement point set, edge measurement point set, fracture neighborhood measurement point set, and fine-grained zone measurement point set to the same endmember coordinate system using the spatial partition identifier as an index. Based on this, a cross-partition composition gradient characterization is constructed, which quantifies the magnitude and direction of change in endmember composition between different spatial partitions. Subsequently, the quality assessment agent combines the deformation intensity level and shear band location identifier in the sample structural background data to perform dynamic interpretation of the gradient morphology. Dynamic interpretation refers to whether the composition gradient better conforms to the smooth, monotonic change characteristics dominated by diffusion or the abrupt or multi-peak mixing characteristics caused by deformation-induced redistribution, thereby generating a spatial partition stability index. The expression for the cross-partition gradient magnitude is: in, This represents the gradient magnitude across partitions, and a larger value indicates more significant differences between spatial partitions. This represents a set of spatial partition pairs, such as combinations of the core and the edge, or the core and the fine-grained zone. The weights for spatial partitioning are determined by the number of measurement points, the quality labels of the measurement points, and the construction sensitivity mapping. Indicates spatial partitioning The robust center vector of the set of mole fractions of inner endpoints is obtained by robust aggregation of the set of mole fractions of inner endpoints of the partition. The expression for the spatial partition stability index is: in, This represents a spatial partitioning stability index, with values ​​ranging from 0 to 1, where a larger value indicates a higher degree of rebalancing. and Indicates the scaling factor. The gradient morphology instability is used to characterize whether there are abrupt changes or multi-peak mixing segments in the gradient. Its value is obtained by combining the robust curvature aggregation of the gradient sequence and the results of abrupt change detection. The above index characterizes the risk of redistribution and dynamic non-equilibrium under strong deformation conditions by simultaneously penalizing gradient amplitude and gradient morphology anomalies.

[0042] When constructing a quality profile set by the quality assessment agent, the component integrity index, valence consistency index, endmember solvability index, and spatial partitioning stability index are mapped according to sample identifiers and mineral pair identifiers to form a record-level quality profile. A quality profile refers to the joint representation of a single measurement point record or a set of measurement point records across four dimensions: data completeness, valence stability, endmember feasibility, and spatial consistency. Simultaneously, the quality assessment agent embeds the key element requirement template version identifier, endmember rule table version identifier, and construction background snapshot identifier into the quality profile to ensure version consistency. At the tracing level, the quality assessment agent writes a tracing field for each quality profile. This tracing field includes at least the abnormal record judgment result, a list of removed fields, spatial partitioning reference rules, and valence processing strategy candidates. This allows the quality profile set to be used not only for threshold discrimination but also for direct reference and interpretation when the subsequent knowledge reasoning agent constructs the task constraint set.

[0043] When the quality assessment agent performs balance judgment based on the quality profile set, the decision agent first issues the end-member solvability threshold, spatial partition stability threshold, and valence-state consistency threshold and writes them into the judgment rule table. The threshold is used to convert the multi-dimensional quality profile into balance labels that can be used for traffic splitting calculation paths. For each record, the quality assessment agent simultaneously determines whether the end-member solvability index is not lower than the end-member solvability threshold, the spatial partition stability index is not lower than the spatial partition stability threshold, and the valence-state consistency index is not lower than the valence-state consistency threshold. All records that meet the threshold conditions are included in the balanced subset and bound with balance labels. Records that do not meet the threshold conditions for any index are included in the unbalanced subset and bound with unbalanced source labels. Unbalanced source labels are used to identify the combination of index types that do not meet the threshold and serve as the triggering basis for the subsequent knowledge reasoning agent to generate the task constraint set and the execution agent to generate the unbalanced correction configuration. To ensure that the balanced and unbalanced subsets have interpretable boundaries, the quality assessment agent simultaneously writes a discriminative evidence field during the triage process. The discriminative evidence field includes at least the name of the indicator that triggered the threshold failure, the indicator value, the threshold value, and a summary of the corresponding source field. This ensures that the balance label forms a consistent constraint link in the subsequent screening of the thermometer and barometer candidate set and the generation of the algorithm configuration set, and avoids outputting misleading thermometer and barometer conclusions under strong deformation or dynamic non-equilibrium conditions.

[0044] S130, combining the balance label with the sample structure background data to construct a task constraint set, and generating a candidate set of thermobarometers that match the mineral pair identifiers under the constraints of the task constraint set.

[0045] In one possible implementation, a task constraint set is constructed by combining equilibrium labels and sample structural background data. Under the constraints of this task constraint set, a candidate set of thermobarometers matching mineral pair identifiers is generated. Specifically, this includes: performing joint alignment on the equilibrium labels and sample structural background data to form a structural equilibrium context. This structural equilibrium context simultaneously binds mineral pair identifiers, equilibrium labels, non-equilibrium source labels, structural unit identifiers, shear band location identifiers, deformation strength levels, and deformation indication features under the sample identifier; generating applicable mineral combination constraints based on the structural equilibrium context; mapping mineral pair identifiers to mineral combination rules in an expert knowledge base and combining them with equilibrium labels to exclude thermobarometer types lacking non-equilibrium fault tolerance; and based on deformation strength... Deformation strength constraints and strain rate sensitive constraints are generated based on the grade and deformation indication characteristics; rebalancing risk constraints are generated based on non-equilibrium source tags, and spatial zoning usage constraints are generated simultaneously; pressure range constraints and temperature range constraints are generated by combining structural unit identifiers, shear zone location identifiers, deformation strength grades, and equilibrium labels; applicable mineral combination constraints, deformation strength constraints, strain rate sensitive constraints, rebalancing risk constraints, spatial zoning usage constraints, pressure range constraints, and temperature range constraints are merged to form a task constraint set; based on the task constraint set, the initial set of thermobarometers is retrieved in the expert knowledge base using mineral pair identifiers as the primary search key; the initial set of thermobarometers is filtered and sorted according to the task constraint set to generate the aforementioned candidate set of thermobarometers.

[0046] Specifically, the decision-making agent, under sample identification constraints, reads equilibrium and non-equilibrium source labels from the quality assessment agent and then reads sample structural background data from the knowledge reasoning agent. The knowledge reasoning agent then performs joint alignment to form a structural equilibrium context. Joint alignment refers to mapping fields from different data channels using synonyms and ensuring primary key consistency. This ensures that mineral pair identifiers, equilibrium labels, and non-equilibrium source labels coexist in the same record with structural unit identifiers, shear zone location identifiers, deformation intensity levels, and deformation indication features, and can be referenced by the same rule. Structural unit identifiers uniquely identify the regional structural affiliation of the sample; shear zone location identifiers indicate the sample's position relative to the shear zone core, boundary, or outer region; deformation intensity levels quantify the degree of plastic deformation; and deformation indication features describe petrological evidence pointing to strain rate or deformation mechanism. The structural equilibrium context uses the sample identifier as a global index to ensure that subsequent constraint construction and thermobarometry retrieval can trace back to this context.

[0047] When the knowledge-reasoning agent generates applicable mineral combination constraints based on the constructed equilibrium context, it first uses mineral pair identifiers as rule indexes to retrieve the mineral combination rule table in the expert knowledge base and obtain a set of matching thermometer / barometer types. Mineral combination rules refer to knowledge entries that solidify the necessary, alternative, and incompatible mineral pair premises of the thermometer / barometer into computable rules. Subsequently, an equilibrium label is introduced as a secondary screening condition. The equilibrium label distinguishes between balanced and non-equilibrium subsets, thereby excluding thermometer / barometer types that explicitly rely on the complete equilibrium assumption and lack non-equilibrium fault tolerance. Non-equilibrium fault tolerance refers to the thermometer / barometer's ability to provide interpretable results through weighting, correction, or robust estimation even when there is component redistribution, diffusion restriction, or significant micro-region differences. The applicable mineral combination constraints are ultimately solidified into the constraint fields of the task constraint set in the form of "allowed sets and prohibited sets," while retaining the referenced mineral combination rule version identifier to support auditing.

[0048] When the knowledge-reasoning agent generates deformation strength constraints and strain rate sensitivity constraints based on deformation strength levels and deformation indicator features, the deformation strength constraints are used to limit the applicability boundary of the thermobarometer to strong deformation environments, while the strain rate sensitivity constraints are used to characterize the sensitivity of the thermobarometer to the dynamic non-equilibrium effects caused by high strain rates. Deformation indicator features are mapped to strain rate sensitivity levels through rules. For example, a sensitivity level label is formed by combining evidence such as dynamic recrystallization grain size, mylonitic fabric strength, and shear-oriented structures. This label is then matched with the structural sensitivity label of the thermobarometer, thus prioritizing the retention of thermobarometer types robust to high-strain shear band environments during the candidate stage, while excluding or downweighting thermobarometer types with high sensitivity and lacking correction channels. The expression for the sensitivity mapping, which maps the levels to computable constraints, is as follows: in, This represents a sensitivity score to the stress environment, ranging from 0 to 1, with higher values ​​indicating a stronger risk of high stress. This represents a mapping function that compresses linear combinations to a range of 0 to 1. The numerical value representing the deformation strength grade is obtained by mapping the discrete levels of the deformation strength grade. Indicates the first The triggering or intensity value of each deformation indicator feature is obtained from petrological interpretation results or statistical characteristics. arrive This represents the weighting coefficient, the value of which is given by the empirically calibrated entries of the expert knowledge base or obtained by fitting historical samples.

[0049] When the knowledge-reasoning agent generates rebalancing risk constraints based on non-equilibrium source labels and simultaneously generates spatial partition usage constraints, the rebalancing risk constraints are used to clarify the risk type of the non-equilibrium subset and limit the thermometer and barometer to have corresponding processing capabilities. The spatial partition usage constraints are used to specify how subsequent calculations reference mineral spatial partition data. Non-equilibrium source labels indicate combinations of indicator types that do not meet thresholds, such as valence consistency failure, endmember solvability failure, or spatial partition stability failure. The rebalancing risk constraints select different constraint strategies accordingly. For example, when spatial partition stability fails, the thermometer and barometer are required to allow partition modeling of the core and edge measurement point sets; when valence consistency fails, the thermometer and barometer are required to allow self-consistent iterative valence processing or allow the inclusion of valence uncertainty in the weights. The spatial partition usage constraints solidify the reference range of the core measurement point set, edge measurement point set, fracture neighborhood measurement point set, and fine-grained zone measurement point set into executable rules. These rules are then used as mandatory fields in the subsequent algorithm configuration set for data mapping configuration, thereby ensuring seamless integration of non-equilibrium correction and spatial partition referencing in the workflow.

[0050] When the knowledge-reasoning agent generates pressure and temperature range constraints by combining the building unit identifier, shear band location identifier, deformation strength level, and balance label, the building unit identifier provides a priori temperature and pressure intervals at the region scale. The shear band location identifier and deformation strength level are used to expand or shrink the priori intervals to reflect the increased periodicity and uncertainty caused by strong deformation. The balance label is used to determine the conservatism of the intervals and avoid false precision caused by excessively narrow intervals on unbalanced subsets. Pressure and temperature range constraints are written into the task constraint set in the form of closed intervals, and each interval records the source entry version and correction reason field. The expression for interval generation is: in, and This represents the lower and upper bounds of the temperature prior obtained by mapping from the construction unit identifiers. The values ​​are given by the construction prior entries in the expert knowledge base. This represents the temperature expansion, and its value is obtained by jointly mapping the shear band location identifier, deformation strength grade, and equilibrium label, and corresponds to the non-equilibrium subset. Take a larger value and This represents the lower and upper bounds of the pressure prior, obtained by mapping the structural unit identifiers. This represents the amount of pressure expansion, and its value is obtained by jointly mapping the shear band location identifier, deformation strength grade, and balance label.

[0051] When the knowledge-reasoning agent merges applicable mineral combination constraints, deformation strength constraints, strain rate sensitivity constraints, rebalancing risk constraints, spatial zoning constraints, pressure range constraints, and temperature range constraints to form a task constraint set, merging refers to standardizing each constraint into a unified set of constraint fields and establishing constraint priority rules to handle conflicts. Constraint priority rules are used to provide arbitration results in conflict situations such as "mineral combination allowed but structurally sensitive incompatible" or "theoretical boundaries intersectable but rebalancing risk not satisfied." Rebalancing risk constraints triggered by the balance label are used as high-priority constraints to avoid selecting thermobarometer types that are mathematically calculable but physically unreliable on non-equilibrium subsets. The task constraint set is also written to a conflict record field, which records the set of exclusion reasons for excluded candidates and the set of triggered constraint fields, allowing the review agent to explain candidate coverage differences in the final report.

[0052] When the knowledge-reasoning agent retrieves the initial set of thermometers and barometers from the expert knowledge base based on the task constraint set and using mineral pair identifiers as the primary search key, the mineral pair identifiers serve as the primary search key to quickly locate all thermometer and barometer entries that meet the necessary mineral combination prerequisites and form the initial set. The initial set of thermometers and barometers refers to the complete candidate set before the application of the task constraint set for filtering. During the retrieval process, the agent simultaneously reads the applicable prerequisite set, theoretical boundary set, data requirement template, empirical confidence parameters, and construction sensitivity labels for each thermometer and barometer entry, and encapsulates them into a candidate description package set for reuse in subsequent filtering, sorting, and executability analysis. To ensure reproducibility, the knowledge-reasoning agent embeds the knowledge base version identifier and entry version identifier in the retrieval results and writes them into the traceability field of the candidate description package set.

[0053] When the knowledge-reasoning agent performs filtering and sorting on the initial set of thermometers and barometers based on the task constraint set to generate a candidate set of thermometers and barometers, filtering is used to remove thermometer and barometer entries that violate any hard constraint field. Hard constraint fields include at least the applicable mineral assemblage constraint, rebalancing risk constraint, and spatial partitioning usage constraint. Sorting is used to calculate a matching score for thermometer and barometer entries that meet the hard constraints and output the candidate list in descending order of matching score. The matching score uses the empirical confidence parameter as the basic contribution, the degree of overlap between the theoretical boundary and the pressure range constraint and temperature range constraint as the compatibility contribution, and the degree of consistency between the construction sensitivity label and the strain rate sensitivity constraint as the construction compatibility contribution. Entries lacking correction channels under non-equilibrium subset conditions are penalized. The expression for the matching score is: Where R represents the thermo-barometer matching score, and C represents the normalized value of the empirical confidence parameter, which is provided by the expert knowledge base and mapped to 0 to 1. This represents the degree of overlap between the pressure range constraint and the theoretical boundary set, and its value is obtained by the ratio of the intersection length to the union length of the two intervals. This indicates the overlap between the temperature range constraint and the theoretical boundary set, and its value is determined by... Consistency is defined by K, which represents the construction compatibility, and its value is obtained by mapping the consistency between the construction sensitivity label and the strain rate sensitivity constraint, ranging from 0 to 1. H represents the risk penalty term, and its value is calculated by the penalty rule triggered by the rebalancing risk constraint, also ranging from 0 to 1. arrive The weight coefficient is given by the rule entries in the expert knowledge base or obtained by fitting historical samples. After sorting, the candidate set of thermometers and barometers formed is consistent with the mineral pair identifier, balance label, and task constraint set. It is registered by the decision-making agent to the processing status table for subsequent executability analysis by the execution agent.

[0054] S140, Perform executability analysis on the candidate set of thermobarometers to form an executable set of thermobarometers, and generate an algorithm configuration set for the executable set of thermobarometers.

[0055] In one possible implementation, an executability analysis is performed on the candidate thermobarometer set to form an executable thermobarometer set. Specifically, this includes: while maintaining the correspondence between the mineral pair identifiers and the candidate thermobarometer set, reading a candidate description package set for each candidate thermobarometer to form a candidate execution context. The candidate execution context includes an applicable prerequisite set, a theoretical boundary set, a data requirement template, and a construction sensitivity label. Based on the candidate execution context, the completeness of the candidate thermobarometer execution data is determined, and the data requirement template is compared with the mineral chemical composition data and mineral spatial partitioning data in the sample data package. The process involves matching data items and marking thermometer / barometer candidates that do not meet the data requirement template as data inoperability candidates. After removing data inoperability candidates, a premise consistency check is performed on the remaining thermometer / barometer candidates. The equilibrium assumption, solid solution assumption, and valence state treatment assumption in the applicable premise set are compared with the equilibrium label and non-equilibrium source label to determine whether there are any incompatibilities between theoretical assumptions and sample states. Thermometer / barometer candidates that lack non-equilibrium tolerance and rely on the complete equilibrium assumption are marked as premise conflict candidates. After removing premise conflict candidates, a theoretical boundary consistency check is performed on the remaining thermometer / barometer candidates. An intersection analysis is performed between the pressure and temperature ranges in the theoretical boundary set and the pressure and temperature range constraints in the task constraint set to determine the existence of a non-empty feasible region. Candidate thermometers that meet the conditions within this feasible region are marked as boundary risk candidates. After eliminating boundary risk candidates, a structural sensitivity consistency judgment is performed on the remaining thermometer candidates. The structural sensitivity labels are compared with the deformation strength level and shear band location markers in the sample structural background data to determine the suitability level of the thermometer candidates in the current structural environment. Candidates incompatible with the structural environment are marked as structurally incompatible. Candidates; After eliminating candidates with incompatible structures, a computational reproducibility assessment is performed on the remaining thermobarometer candidates. By applying perturbations to the input component parameters and model parameters under a unified computational semantics and analyzing the stability of temperature and pressure results, a reproducibility index is generated, and thermobarometer candidates sensitive to perturbations are marked as numerically unstable candidates. Based on the data completeness discrimination results, premise consistency discrimination results, theoretical boundary consistency discrimination results, construction sensitivity consistency discrimination results, and reproducibility index, thermobarometer candidates that are not marked as any unexecutable type are included in the above-mentioned executable thermobarometer set.

[0056] Specifically, under the scheduling constraint that the decision-making agent maintains the correspondence between mineral pairs and thermobarometer candidate sets, the executing agent reads the candidate description package set output by the knowledge reasoning agent one by one to form a candidate execution context. The candidate description package set refers to the encapsulation of rule-based meta-information for a single thermobarometer candidate, at least solidifying the applicable premise set, theoretical boundary set, data requirement template, and construction sensitivity label, thereby avoiding repeated access to the expert knowledge base in subsequent judgments and ensuring reproducibility. The candidate execution context refers to the context container used in the default configuration stage to carry all the prior fields required for executability judgment. The applicable premise set describes the equilibrium conditions, mineral solid solution conditions, and valence state processing conditions required for the thermobarometer candidate to be established; the theoretical boundary set describes the calibration applicable range and extrapolation forbidden zone of the thermobarometer candidate; the data requirement template describes the component fields and spatial partition fields necessary for the thermobarometer candidate calculation; and the construction sensitivity label describes the sensitive direction and tolerance boundary of the thermobarometer candidate to tectonic environments such as high-strain shear zones or mylonite systems.

[0057] When the execution agent determines the data completeness of candidate thermometer / barometers based on the candidate execution context, the quality assessment agent first provides a list of fields and a set of measurement point identifiers for the mineral chemical composition data and mineral spatial partitioning data in the sample data package. The execution agent then matches the required field set, optional field set, and spatial partitioning field set in the data requirement template with the field list of the sample data package item by item, and further checks the coverage and missing patterns of each required field on the referenced measurement point identifier set. The goal of the data completeness determination is to confirm that the thermometer / barometer candidate has the minimum information conditions for execution at the data level. Item-by-item matching includes field name consistency matching, field semantic consistency matching, and field precision consistency matching. Field precision consistency matching is used to determine whether the field meets the minimum resolution required by the thermometer / barometer candidate. For thermometer / barometer candidates that do not meet the data requirement template, the execution agent marks them as data inexecutable candidates and writes the reason for data inexecutability into the data inexecutability reason field to ensure that subsequent processes do not allocate any computing resources to this candidate. The expression for the data completeness score is: in, This represents the data completeness score, ranging from 0 to 1, with higher values ​​indicating a better fulfillment of the data requirements template. Indicates the set of required fields. This indicates the importance weight of field u. This is an indicator function that takes the value 1 if the condition is true and 0 otherwise. This represents the coverage of field u on the set of referenced measurement point identifiers, and its value is determined by the field... The number of non-missing test points divided by the total number of referenced test points is obtained. This represents the coverage threshold for field u, the value of which is given by the data requirement template or fixed by the decision agent in the task constraint set.

[0058] When the executing agent performs a consistency check on the remaining thermobarometer candidates after eliminating unexecutable candidates, it compares the set of applicable premises in the candidate execution context with the balance label and non-equilibrium source label output by the quality assessment agent. It also performs consistency checks on the balance assumption, solid solution assumption, and valence state treatment assumption. The balance assumption refers to whether the thermobarometer candidate requires the sample to be in a single thermodynamic equilibrium state or allows for incomplete reequilibrium and handles it using a weighted approach. The solid solution assumption refers to the requirements of the thermobarometer candidate on endmember solvability, endmember mixing mechanism, and endmember coupling constraints. The valence state treatment assumption refers to whether the thermobarometer candidate allows a fixed valence state strategy or requires self-consistent iterative valence state treatment and the tolerance range for valence state uncertainty. When premise consistency judgment identifies a thermometer / barometer candidate that strictly relies on the perfect equilibrium assumption and lacks non-equilibrium tolerance, and the balance label indicates a non-equilibrium subset or the non-equilibrium source label indicates non-equilibrium risks such as spatial partition stability failure or valence state consistency failure, the executing agent marks this candidate as a premise conflict candidate and records the conflict entry and conflict source label. This prevents subsequent stages from explicitly relying on this candidate to output misleading thermometer / barometer conclusions. The expression for premise consistency scoring is: in, This represents the premise consistency score, with a value ranging from 0 to 1, where a larger value indicates greater consistency. This indicates a conflict indicator of the balance assumption. This indicates a conflict indicator of the solid solution hypothesis. This indicates the conflict indicator quantity for handling the valence state assumption. The value of is determined by the set of applicable premises and the rules for comparing balance labels and unbalanced source labels. It takes 1 when there is an irreconcilable conflict and 0 otherwise, thus realizing that any conflict of key assumptions is judged as inconsistent through the product structure.

[0059] When the executing agent performs theoretical boundary consistency judgment on the remaining thermobarometer candidates after eliminating candidates with conflicting premises, it extracts the theoretical boundary set in the candidate execution context as theoretical boundary intervals and performs intersection analysis with the pressure range constraints and temperature range constraints in the task constraint set to determine whether a non-empty feasible region exists. The theoretical boundary set refers to the set of temperature and pressure intervals that can guarantee interpretive consistency under experimental calibration, theoretical derivation, or empirical fitting conditions. The pressure range constraints and temperature range constraints refer to the sample-level prior intervals generated by the knowledge reasoning agent based on the constructed equilibrium context. A non-empty feasible region refers to the intersection of the theoretical boundary interval and the sample prior interval in both the temperature and pressure dimensions, with the intersection width not less than the minimum stable width threshold, thus avoiding numerical instability or strong extrapolation caused by solving within extremely narrow intervals. For thermobarometer candidates determined to have a non-empty feasible region, the executing agent marks them as boundary risk candidates and writes them into the boundary risk level field for subsequent configuration generation phase adjustment of convergence strength and out-of-bounds handling strategies. The expression for feasible region overlap is: in, Indicates the degree of overlap in temperature ranges. This indicates the degree of overlap between pressure zones, with values ​​ranging from 0 to 1, and a larger value indicates greater overlap. and This represents the temperature and pressure ranges extracted from the theoretical boundary set. and This represents the temperature range constraint and pressure range constraint in the task constraint set. The length of the interval is represented by the intersection length of 0, which indicates that there is no feasible region. Thus, the overlap measure is used to quantify the consistency between the theoretical boundary and the sample prior.

[0060] After eliminating boundary risk candidates, the executing agent performs structural sensitivity consistency judgment on the remaining thermobarometer candidates. The agent compares the structural sensitivity labels in the candidate execution context with the deformation intensity level and shear band location identifiers in the sample structural background data. The structural sensitivity labels describe the thermobarometer candidate's sensitivity to high-strain environments, the allowable dynamic deviation range, and whether additional correction channels are needed. The core of structural sensitivity consistency judgment is determining whether the thermobarometer candidate meets the applicability level requirements in the current structural environment. For example, when the shear band location identifier indicates the shear band core and the deformation intensity level is high, only candidates whose structural sensitivity labels declare applicability to high strain or can be used in conjunction with non-equilibrium correction methods are retained. For thermobarometer candidates incompatible with the structural environment, the executing agent marks them as structurally incompatible candidates and records the incompatibility reason field to ensure that subsequent thermobaric co-solution does not output seemingly reasonable but geologically uninterpretable results when the structural background is mismatched. The expression for structural compatibility scoring is: in, This represents the compatibility score, with values ​​ranging from 0 to 1, where a larger value indicates greater compatibility. This represents the strain environment sensitivity score obtained by mapping deformation strength level to deformation indicator characteristics. The candidate tolerance score is obtained by constructing a sensitivity label mapping. The value is obtained by mapping the candidate's tolerance level to high-strain environments. The smaller the difference between the two, the better the candidate's tolerance matches the environmental risk.

[0061] After eliminating incompatible candidates, the executing agent performs computational reproducibility evaluation on the remaining thermobarometer candidates. Under unified computational semantics, the agent constructs controlled perturbation experiments and repeatedly executes simplified solutions to evaluate the stability of temperature and pressure results. Unified computational semantics refers to uniformly constraining the endmember solution method, activity model selection method, valence state processing strategy invocation method, and solution control rules, ensuring that the reproducibility evaluation of different thermobarometer candidates is under comparable conditions. Input composition parameter perturbation refers to applying perturbations to the measured values ​​of key elements in the mineral chemical composition data according to the error rules given by the analytical quality metadata. Model parameter perturbation refers to applying perturbations to activity model parameters, valence state processing parameters, or non-equilibrium correction related parameters within their allowable ranges. Stability analysis refers to comparing the changes in temperature and pressure results before and after perturbation and forming a reproducibility index. For perturbation-sensitive thermobarometer candidates, the executing agent marks them as numerically unstable candidates and records the sensitive source field to avoid significant deviations in subsequent parallel solutions due to numerical ill-conditioning amplifying small input errors. The expression for the reproducibility index is: in, This represents a reproducibility index, with values ​​ranging from 0 to 1, and a larger value indicates greater stability. This indicates the degree of dispersion of temperature results under controlled perturbation testing. Its value is obtained from the standard deviation of temperature results obtained from multiple perturbation solutions or a robust dispersion estimate. This indicates the degree of dispersion of pressure results under controlled disturbance testing, and its value is determined by... Consistent and The stability scale parameter is obtained by mapping the temperature range constraint, pressure range constraint, or empirical rule in the task constraint set. It is used to normalize the degree of dispersion to a comparable scale, thereby penalizing the sensitivity to temperature and pressure through exponential decay.

[0062] When the executing agent performs a comprehensive executability merging based on data completeness, premise consistency, theoretical boundary consistency, construction sensitivity consistency, and reproducibility indicators, it aggregates the data inexecutability candidate tags, premise conflict candidate tags, boundary risk candidate tags, construction incompatibility candidate tags, and numerical instability candidate tags for each thermobarometer candidate. The thermobarometer candidates not marked as any inexecutable type are then included in the executable thermobarometer set. The executable thermobarometer set refers to the candidate set that simultaneously satisfies the execution conditions in four dimensions: data, theory, construction, and numerical stability. Each candidate retains an executability label and a risk level label for subsequent algorithm configuration set generation phases to adjust the solution control strength and uncertainty generation strategy. To ensure continuity with the previous task constraint set and the subsequent result traceability chain, the executing agent registers the set of executable thermobarometers under the sample identifier to the processing status table and maintains consistent binding with the mineral pair identifier, equilibrium label, non-equilibrium source label, and candidate description package set version identifier, thereby forming an interpretable screening closed loop from the thermobarometer candidate set to the set of executable thermobarometers.

[0063] In one possible implementation, the algorithm configuration set for generating the executable thermobarometer set specifically includes: generating a data mapping configuration based on sample identifiers for each executable thermobarometer candidate read data requirement template in the executable thermobarometer set; determining an endmember decomposition strategy based on the solid solution endmember set, endmember constraints, and task constraints, and selecting endmember decomposition constraints according to the aforementioned balance labels to output the endmember solution configuration; generating an activity model configuration based on a matching activity model selected from the applicable premise set and the theoretical boundary set; and generating a fixed valence state configuration based on a fixed valence state strategy selected from the valence state consistency index and the non-equilibrium source label. The configuration includes: valence state processing configuration; determining the activation status of the non-equilibrium correction channel based on the balance label to generate the non-equilibrium correction configuration; extracting calibration data from the theoretical boundary set and combining it with the task constraint set to form the calibration feasible region and generate the experimental anchoring configuration; unifying and solidifying the selection of solution variables, initial value generation, search strategy, convergence criterion, abnormal termination condition, and restart strategy to generate the joint solution control configuration; and converging the data mapping configuration, endmember solution configuration, activity model configuration, valence state processing configuration, non-equilibrium correction configuration, experimental anchoring configuration, and joint solution control configuration and performing consistency verification to form the algorithm configuration set.

[0064] Specifically, when the executing agent generates data mapping configuration for each executable thermobarometer candidate in the executable thermobarometer set under the sample identifier constraint, it first reads the data requirement template corresponding to the executable thermobarometer candidate from the candidate description package set, and solidifies the set of required component fields, the set of required spatial partition fields, and the set of optional fields in the data requirement template into field matching rules. Then, the executing agent calls the sample data package field directory and the set of measuring point identifiers provided by the quality assessment agent, aligns the mineral chemical composition data and mineral spatial partition data at the row level according to the set of measuring point identifiers, and performs partition aggregation according to the spatial partition identifiers, thereby forming a set of standard variables under the configuration identifier. The data mapping configuration addresses the differences in naming, normalization benchmarks, and reference ranges for the same data field among different thermometer / barometer candidates. The standard variable set refers to a variable container that consistently names elemental contents, endmember-related variables, and spatial partition statistics under a unified normalization benchmark. Field matching rules refer to the combined rules of field name mapping and field semantic mapping. Spatial partition fields refer to the reference identifiers and aggregation methods for the core measurement point set, edge measurement point set, fracture neighborhood measurement point set, and fine-grained zone measurement point set. To ensure the integrity of the subsequent result traceability chain, the executing agent writes the referenced measurement point identifier set, spatial partition reference rules, field missing handling rules, and normalization benchmark identifier into the data mapping configuration, thereby enabling the input variables of any thermometer / barometer candidate to be traced back to the specific measurement point records in the sample data package.

[0065] When the execution agent generates the endmember solution configuration, the knowledge reasoning agent first provides the solid solution endmember set and endmember constraints from the candidate description package set. The execution agent then reads the task constraint set to obtain indirect constraints on the endmember decomposition strategy based on spatial partitioning constraints, pressure range constraints, and temperature range constraints. The solid solution endmember set refers to the set of endmember types used to characterize the composition of mineral solid solutions. Endmember constraints refer to the set of constraints that the endmember mole fraction set must satisfy, including nonnegativity, summation, and several endmember coupling relationships. The endmember decomposition strategy refers to the solution method and regularization approach used to map mineral chemical composition data to the endmember mole fraction set. The execution agent determines the solver type and constraint expression form of the endmember decomposition strategy based on the solid solution endmember set and endmember constraints. It selects endmember decomposition constraints based on the balance label. When the balance label indicates a balanced subset, strict endmember constraints are solidified to obtain stronger endmember uniqueness. When the balance label indicates a non-balanced subset, regularized endmember constraints are solidified to suppress ill-conditioned solutions caused by spatial partitioning differences. The expression for solving endmember decomposition is: in, Let represent the set of endmember mole fractions, and let x represent the vector of endmember mole fractions to be determined. Represents the feasible set of endmember constraint definitions and includes and Under the constraints, A represents the mapping matrix from endmembers to element counts, which is determined by the solid solution endmember set and stoichiometry rules; y represents the observed element count vector, which is derived from mineral chemical composition data under a normalized benchmark configured by the data mapping; and L represents the regularization structure matrix, used to characterize the smoothness or coupling relationship of the endmember mole fraction set. This represents the regularization coefficient, whose value is obtained by jointly mapping the balanced label and the unbalanced source label, with the unbalanced subset corresponding to a larger value. To enhance stability.

[0066] When the execution agent generates the activity model configuration, it reads the applicable premise set and theoretical boundary set in the candidate execution context, and determines the family of activity models allowed for the executable thermobarometer in the candidate description package set. Then, combining the pressure range constraint and temperature range constraint in the task constraint set and the calibration feasible region to be formed by the experimental anchoring configuration, it selects the activity model that matches the solid solution endmember set and is interpretable within the calibration feasible region. The activity model refers to the functional form that maps the set of endmember mole fractions to activity variables. The activity variables refer to the effective concentration characterization of the equilibrium constant and reaction quotient. The activity model configuration is used to solidify the activity model version identifier, parameter source identifier, and solution domain restriction rules, so that subsequent joint solutions are only performed within the range allowed by the theoretical boundary set and extrapolation is avoided. To ensure consistency with the endmember solution configuration, the executing agent unifies the variable naming of the endmember mole fraction set with the variable naming of the activity model parameter set, and establishes a reference relationship between the activity model configuration and the data mapping configuration through configuration identifiers, thereby ensuring that the endmember input required for activity calculation is consistent with the spatial partitioning reference rules.

[0067] When the executing agent generates the valence state processing configuration, it reads the valence state consistency index and non-equilibrium source label output by the quality assessment agent, and reads the set of allowed strategies for valence state processing for the executable thermobarometer candidate from the candidate description package set. Then, based on the valence state consistency index, it selects a fixed valence state strategy and fixes the valence state correction parameters. The fixed valence state strategy refers to using a locking rule for the valence state values ​​of variable valence elements, ensuring that the valence state correction parameters remain consistent across all referenced measurement points and spatial partitions, thereby reducing the convergence uncertainty caused by self-consistent iteration. The valence state correction parameters refer to the set of parameters used to decompose the total amount of elements into different valence state shares or to correct charge balance residuals. Under the fixed valence state strategy, the executing agent selects the value range of the valence state correction parameters based on the numerical level of the valence state consistency index and writes this value range into the valence state processing configuration for parameter perturbation during subsequent uncertainty set generation. To maintain consistent coupling with the activity model configuration, the executing agent uses the valence correction parameter as a correction term in the activity model parameter set and fixes the reference link within the configuration, so that the impact of valence treatment on the activity variable can be consistently transmitted throughout the joint solution process.

[0068] When the executing agent generates the unbalanced correction configuration, it determines the enabled status of the unbalanced correction channel based on the balance label and uses the unbalanced source label to select the composition type and value rules of the unbalanced correction parameters. The unbalanced correction channel refers to the correction mechanism introduced during the joint solution candidate point evaluation and result generation process to address dynamic deviations, component redistribution, and spatial partitioning differences. The enabled status is used to indicate whether to correct the balance consistency index on the unbalanced subset. When the balance label indicates a balanced subset, the executing agent fixes the unbalanced correction configuration to disabled and records the reason for disabling it. When the balance label indicates an unbalanced subset, the agent fixes the unbalanced correction configuration to enabled and reads the spatial partitioning reference rules and partitioning endmember association mapping from the data mapping configuration, further fixing the diffusion length parameter, component gradient parameter, and rebalancing weight parameter. The diffusion length parameter is used to characterize the effective scale of element exchange under the current tectonic environment. Its value is obtained by mapping the deformation intensity level and the statistical characteristics of spatial partitions. The composition gradient parameter is used to characterize the difference magnitude of the endmember mole fraction set between the core and edge partitions. Its value is obtained by the robust norm of the partition endmember difference vector. The rebalancing weight parameter is used to weight and fuse the contributions of each spatial partition when evaluating candidate points. Its value ranges from 0 to 1 and satisfies the weight normalization rule.

[0069] When the execution agent generates the experimental anchoring configuration, it extracts calibration data from the theoretical boundary set and combines it with the task constraint set to form the calibration feasible region. This unifies the theoretical applicable range and sample-level prior range of the executable thermobarometer candidates into computable constraints at the configuration level. Calibration data refers to the boundary information used to define extrapolation risks within the experimental calibration range, empirical fitting coverage range, or theoretical derivation validity range of the thermobarometer candidates. The calibration feasible region refers to the intersection range that simultaneously satisfies the theoretical boundary set and the task constraint set in both the temperature and pressure dimensions. The execution agent embeds outbound judgment rules and outbound handling rules in the experimental anchoring configuration. The outbound judgment rules are used to detect whether candidate temperature and pressure values ​​exceed the calibration feasible region during the joint solution process. The outbound handling rules are used to trigger search boundary contraction or risk level label updates when an outbound is detected, thereby ensuring that subsequent joint temperature and pressure solutions do not produce misleading convergence results in regions lacking calibration support. The expression for the calibration feasible region is: in, This indicates the feasible region to be defined. and This represents the temperature and pressure ranges extracted from the theoretical boundary set. and This represents the temperature range constraint and pressure range constraint in the task constraint set. When the intersection interval is empty, it indicates that there is no feasible region and triggers the executability flag writeback.

[0070] When the execution agent generates the joint solution control configuration, it uniformly solidifies the selection of solution variables, initial value generation, search strategy, convergence criterion, abnormal termination condition, and restart strategy under the sample identification, so that different executable thermometer and barometer candidates have comparable solution control semantics during parallel solution. Solution variable selection refers to the candidate temperature and pressure values ​​and their parameterization forms as free variables in the joint solution. Initial value generation refers to the rules for generating initial temperature and pressure candidate values ​​within the calibrated feasible domain. Search strategy refers to the rules for step size control, boundary handling, and exploration intensity control of candidate points within the solution domain. Convergence criterion refers to the criteria used to determine whether the equilibrium consistency index has reached a stable threshold and allows the output of temperature and pressure results. Abnormal termination condition refers to the termination rules when numerical divergence, failure to handle out-of-bounds errors, or iteration stagnation occur. Restart strategy refers to the rules for adjusting the initial values ​​and search boundaries and restarting the solution after abnormal termination. The executing agent incorporates risk level labels into the parameter table of the search strategy and convergence criteria, enabling candidates with higher risk level labels to adopt more conservative search boundaries and stricter convergence criteria, thereby reducing the risk of false convergence caused by non-equilibrium effects under high-strain shear zones or mylonite systems.

[0071] When the executing agent aggregates the data mapping configuration, endmember solution configuration, activity model configuration, valence state processing configuration, non-equilibrium correction configuration, experimental anchoring configuration, and joint solution control configuration to form an algorithm configuration set, aggregation refers to merging each configuration object into a unified configuration package under the configuration identifier and establishing a cross-configuration reference index table. Consistency verification refers to jointly verifying field completeness, cross-configuration reference consistency, and configuration and task constraint set consistency. Field completeness is used to confirm that the required fields of each configuration object have been written and can be executed and called. Cross-configuration reference consistency is used to confirm that the valence state correction parameters have been referenced by the activity model configuration and the non-equilibrium correction parameters have been called by the joint solution control configuration. Configuration and task constraint set consistency is used to confirm that the calibration feasible domain does not exceed the pressure range constraint and temperature range constraint and that the spatial partitioning reference rules do not violate the spatial partitioning usage constraint. When the consistency check fails, the executing agent writes the reason for the failure into the configuration exception field and writes back the executability tag to prevent the creation of subsequent solution tasks. When the consistency check passes, the algorithm configuration set is registered to the processing status table and is kept traceable with the sample identifier, mineral pair identifier, equilibrium tag and result traceability chain, so that the algorithm configuration set becomes the only standardized configuration input for subsequent temperature and pressure joint solution.

[0072] S150 performs joint temperature and pressure solving in parallel based on the set of executable thermobarometers and the set of algorithm configurations, generates temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate, introduces non-equilibrium correction on the non-equilibrium subset, and outputs a set of temperature and pressure calculation results.

[0073] In one possible implementation, a joint temperature and pressure solution is performed in parallel based on an executable set of thermometers and a set of algorithm configurations. While maintaining the result traceability chain, this generates temperature results, pressure results, and uncertainty sets corresponding to each thermometer candidate. Non-equilibrium corrections are introduced on the non-equilibrium subset. Specifically, this includes: assembling variables in the sample data package based on the data mapping configuration to generate a candidate variable package; performing endmember solving based on the endmember solving configuration to generate an endmember mole fraction set; generating an activity variable package based on the activity model configuration; performing valence state processing on the activity variable package based on the valence state processing configuration to generate a corrected activity variable package; and initiating the joint solution main process based on the joint solution control configuration, within the solution domain constraints, processing the temperature and pressure candidate values. A joint search is performed, and temperature and pressure results are determined at each candidate point based on the modified activity variable package. When the equilibrium label indicates a non-equilibrium subset, a non-equilibrium correction is introduced in the main joint solution process. The non-equilibrium correction is weighted and fused based on the diffusion length parameter, composition gradient parameter, and rebalancing weight parameter in the non-equilibrium correction configuration, considering the endmember contributions of the core and edge regions. After obtaining the temperature and pressure results, an uncertainty set bound to the configuration identifier is generated. The uncertainty set includes data perturbation uncertainty, parameter perturbation uncertainty, and model boundary uncertainty. After the parallel solution is completed, the temperature and pressure results, uncertainty set, and solution status are combined to form a temperature and pressure calculation result set under the sample identifier.

[0074] Specifically, the executing agent invokes the data mapping configuration under the constraints of sample identifiers and configuration identifiers to perform variable assembly on the sample data package to generate a candidate variable package. Variable assembly refers to aligning, filtering, and transforming the mineral chemical composition data and mineral spatial partition data in the sample data package according to the field matching rules, normalization benchmark identifiers, and spatial partition reference rules fixed in the data mapping configuration, thereby obtaining a set of standard variables that can be directly used for subsequent endmember solving and activity calculation. The candidate variable package refers to a variable container encapsulated under the configuration identifier, which at least contains the set of referenced measuring point identifiers, the set of core measuring points, the set of edge measuring points, spatial partition statistics, element content vectors, and field missing handling results. A traceability field is written into the candidate variable package to record the measuring point identifier and field mapping path from which each variable originates, so that the subsequent solution results can be traced back to the original record of the sample data package along the traceability field.

[0075] The executing agent performs endmember solving on the candidate variable package based on the endmember solving configuration to generate an endmember mole fraction set. Endmember solving refers to finding endmember mole fraction vectors within the feasible set defined by endmember constraints, ensuring that these vectors, through the mapping matrix corresponding to the solid solution endmember set, can reconstruct the observation element count vectors in the candidate variable package while minimizing reconstruction residuals. Simultaneously, strict endmember constraints or regularized endmember constraints are selected based on the balance label to improve stability under non-equilibrium subset conditions. The endmember mole fraction set refers to the proportion vector of each endmember component in the mineral solid solution, satisfying nonnegativity and summation constraints, and serves as the core input variable for subsequent activity model configuration. The expression for endmember solving is: in, Let represent the set of endmember mole fractions, and let x represent the vector of endmember mole fractions to be determined. Represents the feasible set of endmember constraint definitions and includes and The constraints are as follows: A represents the mapping matrix from endmembers to element counts, which is determined by the solid solution endmember set and stoichiometry rules; y represents the observed element count vector, which is derived from the candidate variable package under the constraint of normalized benchmark identification; and L represents the regularization structure matrix, used to express endmember coupling relationships or endmember smoothness constraints. represents the regularization coefficient, whose value is obtained by jointly mapping the balanced label and the non-balanced source label, and takes a larger value under the condition of non-balanced subset to suppress ill-conditioned solutions.

[0076] The executing agent generates an activity variable package based on the activity model configuration. The activity variable package refers to an intermediate container for activity calculation encapsulated under a configuration identifier. It consists of a set of endmember mole fractions, an activity model version identifier, a set of activity model parameters, calibration feasible region constraints, and solution domain restrictions. Each activity variable is assigned a variable identifier and a source identifier to support subsequent traceability. The activity model configuration determines the family of functions and parameterization that map the endmember mole fraction set to activity variables. It also solidifies the pressure range constraints, temperature range constraints, and calibration feasible region constraints as applicable domain restrictions for the activity model, thereby avoiding the generation of activity variables lacking interpretability in regions exceeding the theoretical boundary set.

[0077] The executing agent performs valence state processing on the activity variable package according to the valence state processing configuration to generate a modified activity variable package. Valence state processing refers to the uncertainty of valence state assignment for variable valence elements in the candidate variable package. By fixing the valence state strategy, the valence state correction parameters are injected into the activity model parameter set and the valence state values ​​are locked. This ensures that the charge balance constraint and the endmember solution results are consistent under the spatial partitioning reference rule, thereby reducing the transmission bias of valence state fluctuations to activity variables. The modified activity variable package refers to the activity variable container obtained after valence state processing, which includes valence state correction parameters, valence state locking rules, modified activity variables, and correction effect records. The correction effect records are used to identify which activity variables are affected by the valence state correction parameters and the direction of the effect, so that the subsequent uncertainty set can decompose and attribute valence state-related uncertainties as part of the parameter perturbation uncertainty.

[0078] The executing agent initiates the joint solution main process based on the joint solution control configuration, and performs a joint search for candidate temperature and pressure values ​​within the solution domain constraints to determine the temperature and pressure results. The joint solution main process refers to generating initial candidate temperature and pressure values ​​according to the initial value generation rules within the search space jointly defined by the calibrated feasible domain constraints and the solution domain constraints. It then iteratively updates the candidate points according to the search strategy, and simultaneously constructs an equilibrium consistency index at each candidate point using a modified activity variable package to evaluate the feasibility of that candidate point, until the convergence criterion is met and the temperature and pressure results are output. The equilibrium consistency index characterizes the residual level derived from the relationship between the activity variable and the reaction equilibrium at the candidate point. The convergence criterion determines whether the stability and improvement magnitude of the residuals reach the allowable threshold. The abnormal termination condition terminates the current solution and restarts it according to the restart strategy when numerical divergence, out-of-bounds errors cannot be handled, or iteration stalls, ensuring that the solution process for each configuration identifier is reproducible and auditable under unified control semantics.

[0079] When the balance label indicates a non-equilibrium subset, the executing agent introduces a non-equilibrium correction during the joint solution process and weights and fuses the endmember contributions of the core and edge regions based on the non-equilibrium correction configuration. The non-equilibrium correction addresses common component redistribution and dynamic non-equilibrium phenomena in high-strain shear zones or mylonite systems. It uses diffusion length parameters, component gradient parameters, and reequilibrium weight parameters to jointly model the endmember mole fraction sets of the core and edge measurement point sets during the candidate point evaluation stage. This ensures that the balance consistency index reflects the correction effect of incomplete reequilibrium rather than forcibly assuming all measurement points are in the same equilibrium state. The expression for the weighted fusion of endmember contributions is: in, This represents the fusion endmember vector used for candidate point evaluation. This represents the robust center vector of the set of endmember mole fractions in the core region, whose values ​​are obtained by robust aggregation of the set of endmember mole fractions of the core measurement points. Let represent the robust center vector of the set of mole fractions of the edge partitions, whose values ​​are obtained by robust aggregation of the set of mole fractions of the edge measurement points. This represents the rebalancing weight parameters for the core region. Describes the edge partition rebalancing weight parameters and satisfies The rebalancing weight parameter is obtained by jointly mapping the diffusion length parameter and the composition gradient parameter and is used to reflect the relative contribution of the core and edge to the characterization of the "near equilibrium" composition.

[0080] After obtaining the temperature and pressure results, the executing agent generates an uncertainty set bound to a configuration identifier. This uncertainty set includes data perturbation uncertainty, parameter perturbation uncertainty, and model boundary uncertainty. Data perturbation uncertainty refers to the dispersion of the temperature and pressure results after applying perturbations to the measured values ​​of key elements in the candidate variable package under the error rules constraints given by the analysis of the mass metadata, and repeatedly executing the controlled solution. Parameter perturbation uncertainty refers to the dispersion of the temperature and pressure results after applying perturbations within the allowable value ranges of the valence correction parameter, the activity model parameter set, and the non-equilibrium correction parameter set, and repeatedly executing the controlled solution. Model boundary uncertainty refers to the sensitivity of the temperature and pressure results to boundary changes after repeatedly executing the controlled solution after probing the boundary contraction or expansion of the solution domain near the calibration feasible region constraint boundary. The expression for the combined uncertainty is: Where U represents the overall uncertainty, This represents the uncertainty of data perturbation, and its value is obtained by mapping the dispersion of temperature and pressure results obtained from multiple data perturbation solutions. This represents the uncertainty of parameter perturbation, and its value is obtained by mapping the dispersion of temperature and pressure results obtained from multiple parameter perturbation solutions. The model boundary uncertainty is represented by a value obtained by mapping the sensitivity of the temperature and pressure results obtained from the boundary trial solution. The sum of squares and square root structure is used to synthesize the uncertainty contributions from three independent sources and maintain a decomposable record for the review agent to perform confidence fusion.

[0081] After parallel solving, the executing agent aggregates the temperature results, pressure results, uncertainty sets, and solution states corresponding to each configuration identifier under the sample identifier to form a set of temperature and pressure calculation results. Parallel execution refers to establishing an independent solution task for each executable thermometer / barometer candidate in the executable thermometer / barometer set and distributing it to the parallel solution queue. This ensures that candidate variable package generation, endmember solving, activity variable package generation, valence state processing, joint solution of the main process, non-equilibrium correction, and uncertainty set generation are completed in mutually isolated task environments, thereby avoiding cross-contamination caused by sharing intermediate states among different thermometer / barometer candidates. The result traceability chain refers to the set of traceability fields bound to the configuration identifier, which records at least the spatial partitioning reference rules, field mapping paths, endmember solving constraint selection, activity model version identifier, valence state processing strategy, non-equilibrium correction activation status, calibration feasible region constraint triggering record, and convergence criterion hit record. This allows the temperature and pressure calculation result set to be directly used by the review agent for result dispersion evaluation, theoretical compatibility evaluation, and confidence weighted fusion, and supports the location and interpretation of potentially misleading conclusions under strong deformation conditions.

[0082] S160 performs a consistency review on the set of temperature and pressure calculation results. It generates a recommended value range and a comprehensive confidence score by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

[0083] In one possible implementation, a consistency review is performed on the temperature and pressure calculation result set. A recommended value range and a comprehensive confidence score are generated through a weighted fusion of result dispersion, theoretical compatibility, and confidence level, outputting a structured evaluation result bound to the sample identifier. Specifically, this includes: performing reviewability screening on the temperature and pressure calculation result set to form a review result pool; constructing a standardized result representation for each temperature and pressure gauge candidate in the review result pool, whereby the standardized result representation solidifies the temperature result, pressure result, uncertainty set, and source decomposition under a unified temperature and pressure dimension; calculating the result dispersion based on the standardized result representation, generating temperature dispersion indices and pressure dispersion indices respectively. During the dispersion calculation process, an uncertainty set is introduced to apply weight adjustment to the result items, and a dispersion penalty weight is applied to result items with high risk level labels or a high number of out-of-bounds triggers; and finally, the results are reviewed and approved. After obtaining the temperature and pressure dispersion indices, theoretical compatibility is determined. Based on the applicable premise set, theoretical boundary set, task constraint set, balance label, and non-balance source label, the theoretical assumption compatibility, applicable interval overlap, and construction sensitivity consistency among the candidate thermometers and barometers in the review result pool are jointly determined, and a compatibility matrix and compatibility index are generated. A candidate weight set is generated based on the compatibility matrix and empirical confidence parameters, and the candidate weight set is corrected by combining risk level label, price state processing strategy stability, non-balance correction activation status, and calibration feasible region constraint status to form a fusion weight. After the temperature dispersion index, pressure dispersion index, compatibility index, and candidate weight set are processed, a weighted fusion is performed to generate weighted center values ​​in the temperature and pressure dimensions, respectively, and expands them into recommended temperature and pressure value ranges by combining the uncertainty set.

[0084] Specifically, the review agent receives the set of temperature and pressure calculation results output by the execution agent under the sample identification constraint, and first performs reviewability screening on the set of temperature and pressure calculation results to form a review result pool. Reviewability screening refers to determining whether the result item meets the comparability and interpretability requirements based on the solution status, anomaly cause field, and uncertainty set completeness field of each result item. The solution status is used to indicate whether the candidate thermometer and pressure gauge has completed the joint solution main process and output temperature and pressure results. The anomaly cause field is used to indicate unusable reasons such as data insolvability, premise conflict, failure to handle out-of-bounds errors, abnormal termination, or restart failure. The uncertainty set completeness field is used to indicate whether the data disturbance uncertainty, parameter disturbance uncertainty, and model boundary uncertainty have all been generated and bound to the configuration identifier. The review agent will include the results that meet the reviewability criteria into the review result pool and the results that do not meet the reviewability criteria into the elimination result pool. At the same time, it will retain the reference relationship between mineral pair identifiers, balance labels, non-equilibrium source labels, risk level labels and result traceability chains in the review context to ensure that subsequent dispersion calculations, compatibility judgments and weight corrections can be traced back to specific candidates and their solution process evidence.

[0085] The review agent constructs a standardized result representation for each thermometer / barometer candidate in the review result pool, and solidifies the temperature result, pressure result, uncertainty set, and source decomposition under a unified temperature and pressure dimension. The standardized result representation refers to a structured record with thermometer / barometer candidates as the basic unit. Its fields include at least the temperature center value, pressure center value, temperature uncertainty, pressure uncertainty, and uncertainty source decomposition vector. It also includes traceability fields such as configuration identifier, activity model version identifier, valence state handling strategy identifier, non-equilibrium correction activation status, calibration feasible region constraint trigger record, and out-of-bounds trigger count. The unified temperature and pressure dimensions mean using the same numerical expression semantics for all thermometer / barometer candidates. For example, both temperature and pressure results are represented as combinations of center values ​​and interval endpoints, ensuring consistent interval endpoint generation rules to avoid incomparability caused by different interval definitions used by different candidates. Source decomposition means decomposing the uncertainty set into data disturbance uncertainty contribution, parameter disturbance uncertainty contribution, and model boundary uncertainty contribution, and writing corresponding disturbance rule identifiers and disturbance magnitude identifiers for each contribution, enabling subsequent weight adjustments to differentiate the handling of uncertainties from different sources.

[0086] The review agent calculates the dispersion of results based on standardized results, generating temperature dispersion and pressure dispersion indices. During the dispersion calculation, an uncertainty set is introduced to apply weight adjustment to the result items, and a dispersion penalty weight is applied to result items with high risk level labels or a high number of out-of-bounds triggers. Result dispersion refers to the degree of consistency of temperature or pressure results from different thermometer / barometer candidates within the review result pool at the overall level. The temperature dispersion index quantifies the dispersion of temperature results, and the pressure dispersion index quantifies the dispersion of pressure results. Weight adjustment assigns an effective weight to each result item, weakening the contribution of result items with large uncertainty sets to the overall dispersion, thus preventing low-quality items from dominating the dispersion evaluation. The dispersion penalty weight further reduces the effective weight of corresponding items with high risk level labels or a high number of out-of-bounds triggers, reflecting a stronger extrapolation risk or numerical instability risk for that item. The expression for the weighted robustness center value is: in, This represents the temperature-weighted center value. This represents the stress-weighted center value, and N represents the number of items in the review result pool. This represents the center value of the temperature result for the i-th candidate thermobarometer. This represents the center value of the pressure result for the i-th candidate thermobarometer. The effective weight of the i-th entry is represented by a value obtained through a joint mapping of the uncertainty set, risk level label, and number of out-of-bounds triggers. The expression for the dispersion index is: in, This represents the temperature dispersion index. The values ​​represent pressure dispersion indices; the larger these values ​​are, the worse the consistency among the multiple thermobarometer candidates. The expression for the effective weight is: in, Let represent the combined uncertainty of the i-th item, whose value is obtained by combining the data disturbance uncertainty, parameter disturbance uncertainty, and model boundary uncertainty according to the same combination rule. This represents the numerical value of the risk level label, which is obtained from the risk level mapping table that is fixed by the executing agent. This represents the normalized value of the number of out-of-bounds triggers, obtained by dividing the number of out-of-bounds triggers by the number of iterations or by a preset upper limit. This represents the basic confidence weight, and its value is obtained from the normalized result of the empirical confidence parameter. , , This represents the penalty coefficient, the value of which is fixed by the review rules table of the review agent. The above weight structure, by penalizing uncertainty, risk, and out-of-bounds behavior, makes the dispersion calculation more focused on the concentration of high-confidence entries.

[0087] After obtaining the temperature dispersion index and pressure dispersion index, the review agent performs theoretical compatibility judgment. Based on the set of applicable premises, the set of theoretical boundaries, the set of task constraints, the balance label and the non-balance source label, it jointly judges the theoretical assumption compatibility, the overlap of applicable intervals and the consistency of construction sensitivity among the candidate thermometers and barometers in the review result pool, thereby generating a compatibility matrix and compatibility index. Theoretical compatibility criterion determines whether different thermometer / barometer candidates output comparable results within the same physical assumption framework and applicable range. Theoretical assumption compatibility is used to determine whether there are irreconcilable conflicts between equilibrium assumptions, solid solution assumptions, and valence state treatment assumptions. For example, a decrease in compatibility occurs when one candidate requires complete equilibrium while another allows non-equilibrium corrections and the equilibrium label indicates a non-equilibrium subset. Applicable interval overlap is used to determine whether the calibration feasible region formed by the theoretical boundary sets and task constraint sets of each candidate has sufficient common coverage in the temperature and pressure dimensions, thus avoiding "superficially computable but incomparable" results caused by different candidates converging in almost non-overlapping intervals. Construction sensitivity consistency is used to determine whether the applicability level of the construction sensitivity label is consistent under the constraints of deformation strength level and shear band location, thus avoiding merging candidates sensitive to high-strain environments with candidates robust to high-strain environments on an equal footing. The expression for the compatibility matrix is: in, Represents the compatibility matrix. This indicates the compatibility between the i-th thermometer / barometer candidate and the j-th thermometer / barometer candidate. This represents the theoretical assumption compatibility score, whose value is determined by comparing the set of applicable premises with the rules for balanced and unbalanced source labels and mapped to 0 to 1. The score represents the applicability interval overlap, and its value is obtained by combining the overlap of the two candidate calibration feasible regions in the temperature and pressure dimensions and mapping it to 0 to 1. The construction sensitivity consistency score is determined by comparing the construction sensitivity label with the sample construction background data and mapping it to a range of 0 to 1. The expression for the compatibility index is: in, The compatibility index represents the i-th thermometer / barometer candidate, ranging from 0 to 1, with a larger value indicating better compatibility with other candidates. N represents the number of entries in the review result pool. This structure allows the compatibility matrix to both preserve pairwise relationships for subsequent connectivity assessment and provide a candidate-level summary index for weight correction.

[0088] The review agent generates a candidate weight set based on the compatibility matrix and empirical confidence parameters. This candidate weight set is then corrected using risk level labels, price-state processing strategy stability, unbalanced correction activation status, and calibrated feasible region constraint status to form the fusion weights. The candidate weight set refers to the weight assigned to each thermometer / barometer candidate for subsequent fusion. The empirical confidence parameters refer to the prior quantitative results of the reliability of thermometer / barometer candidates under typical application conditions from the expert knowledge base. The compatibility matrix provides posterior consistency evidence to suppress isolated candidates incompatible with the majority of candidates. Price-state processing strategy stability refers to whether the price-state processing configuration exhibits stable locking or frequent boundary handling stability during candidate solving. Unbalanced correction activation status distinguishes whether candidates use unbalanced correction channels on unbalanced subsets. Calibrated feasible region constraint status distinguishes whether candidates frequently trigger out-of-bounds judgment rules or are on the edge of the feasible region during solving. The expression for the fusion weights is: in, This represents the fusion weight of the i-th thermobarometer candidate. This represents the normalization operator, used to make the sum of all fused weights equal to 1. This represents the normalized value of the empirical confidence parameter. Represents the compatibility index, η and The exponential coefficient representing the moderating strength of prior and posterior contributions. This represents the numerical value of the risk level label. The normalized value representing the number of out-of-bounds triggers. This represents the instability of the price state handling strategy, and its value is obtained by mapping signals such as "frequent boundary crossings, boundary handling triggering, and increased sensitivity" through the price state handling strategy stability rules. , , Indicates the penalty coefficient. This represents the unbalanced correction factor, whose value is jointly determined by the balance label and the unbalanced correction activation status. The unbalanced correction factor is activated when the balance label indicates an unbalanced subset, the candidate unbalanced correction channel is activated, and the rebalancing risk constraint is met. A larger value is taken when the candidate does not have an unbalanced correction channel enabled or its calibrated feasible region constraint state indicates a strong risk of extrapolation. The values ​​are smaller. The above weight structure achieves systematic suppression of "calculable but unreliable" candidates by jointly encoding prior confidence, posterior compatibility and process health into fusion weights.

[0089] After obtaining the temperature dispersion index, pressure dispersion index, compatibility index, and candidate weight set, the review agent performs weighted fusion to generate weighted center values ​​in both the temperature and pressure dimensions. These center values ​​are then expanded into recommended temperature and pressure value ranges by incorporating the uncertainty set. Weighted fusion refers to weighting the center values ​​of each candidate with fusion weights to obtain the recommended center value, and then weighting and aggregating the uncertainties of each candidate with fusion weights to obtain the half-width of the recommended range. Simultaneously, the half-width of the range is adaptively amplified based on the temperature and pressure dispersion indices, ensuring that the greater the dispersion between candidates, the wider the recommended range, reflecting the increased uncertainty of the conclusion. The expressions for the recommended center values ​​of temperature and pressure are: in, This indicates the recommended center value for temperature. This represents the recommended central value for stress. Indicates the fusion weight. and Let represent the center values ​​of the temperature and pressure results for the i-th candidate. The expression for the recommended interval half-width is: in, This indicates the half-width of the recommended temperature range. This indicates a recommended range of half width for pressure. Let represent the temperature uncertainty of the i-th candidate, whose value is obtained by synthesizing the temperature-related components in the uncertainty set. Let represent the pressure uncertainty of the i-th candidate, whose value is obtained by synthesizing the pressure-related components in the uncertainty set. This represents the temperature dispersion index. This represents the pressure dispersion index. and This represents the dispersion amplification factor, the value of which is fixed by the review rules table of the review agent. The expressions for the recommended temperature and pressure ranges are: in, and This indicates the lower and upper bounds of the recommended temperature range. and This represents the lower and upper bounds of the recommended pressure value interval. The aforementioned interval generation mechanism integrates weighted uncertainty and adaptively adjusts the interval width using a dispersion index. This ensures that the recommended value interval simultaneously reflects the central tendency and consistency level of multiple candidate results, while maintaining a traceable connection with sample identification, thermometer / barometer candidates, and the result traceability chain. This embodiment also discloses a geological temperature and pressure analysis device based on multi-agent collaborative decision-making, referring to... Figure 2 The device includes an acquisition module 201, a processing module 202, and an output module 203. It is used to execute any of the above-described multi-agent collaborative decision-making-based geothermal and pressure analysis methods, wherein: The acquisition module 201 is used to acquire the temperature and pressure analysis dataset bound to the target sample, merge it to establish a sample identifier, and form a sample data package; The processing module 202 is used to perform data quality diagnosis based on the sample data packets to generate a quality profile set, and to divide the sample data packets into balanced subsets and unbalanced subsets according to the quality profile set and bind balance labels. Processing module 202 is used to construct a task constraint set by combining the balance label and the sample structure background data, and generate a set of thermobarometer candidates that match the mineral pair identifiers under the constraints of the task constraint set. The processing module 202 is used to perform executability analysis on the candidate set of thermobarometers to form an executable set of thermobarometers, and generate an algorithm configuration set for the executable set of thermobarometers; Processing module 202 is used to perform joint temperature and pressure solving in parallel based on the set of executable thermobarometers and the set of algorithm configurations, generate temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate, introduce non-equilibrium correction on the non-equilibrium subset, and output the set of temperature and pressure calculation results. Output module 203 is used to perform a consistency review on the set of temperature and pressure calculation results. It generates a recommended value range and a comprehensive confidence score by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

[0090] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0091] This embodiment also discloses an electronic device, as shown in the reference. Figure 3The electronic device may include: at least one processor 301, at least one communication bus 302, user interface 303, network interface 304, and at least one memory 305.

[0092] The communication bus 302 is used to enable communication between these components.

[0093] The user interface 303 may include a display screen and a camera. Optionally, the user interface 303 may also include a standard wired interface and a wireless interface.

[0094] The network interface 304 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).

[0095] The processor 301 may include one or more processing cores. The processor 301 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 305, and by calling data stored in memory 305. Optionally, the processor 301 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array (PLA). The processor 301 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also not be integrated into the processor 301 and may be implemented as a separate chip.

[0096] The memory 305 may include random access memory (RAM) or read-only memory. Optionally, the memory may include a non-transitory computer-readable storage medium. The memory 305 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 305 may include a program storage area and a data storage area. The program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch functionality, sound playback functionality, image playback functionality, etc.), instructions for implementing the various method embodiments described above, etc.; the data storage area may store data involved in the various method embodiments described above, etc. Optionally, the memory 305 may also be at least one storage device located remotely from the aforementioned processor 301. As a computer storage medium, the memory 305 may include an operating system, a network communication module, a user interface 303 module, and an application program for a geological temperature and pressure analysis method based on multi-agent collaborative decision-making.

[0097] exist Figure 3 In the electronic device shown, the user interface 303 is mainly used to provide an input interface for the user and to obtain the user input data; while the processor 301 can be used to call an application program stored in the memory 305 that is a geological temperature and pressure analysis method based on multi-agent collaborative decision-making. When executed by one or more processors 301, the electronic device executes one or more methods as described in the above embodiments.

[0098] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that the present invention is not limited to the described order of actions, as some steps can be performed in other orders or simultaneously according to the present invention. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions and modules involved are not necessarily essential to the present invention.

[0099] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0100] In the several embodiments provided by this invention, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.

[0101] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0102] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0103] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 305 and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned memory 305 includes various media capable of storing program code, such as a USB flash drive, external hard drive, magnetic disk, or optical disk.

[0104] The present invention also discloses a non-transitory computer-readable storage medium storing instructions. When executed by one or more processors 301, these instructions cause an electronic device to perform one or more methods as described in the above embodiments.

[0105] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Those skilled in the art will readily conceive of other embodiments of this disclosure upon considering the specification and the disclosure of practical truths. This invention is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.

Claims

1. A geological temperature and pressure analysis method based on multi-agent collaborative decision-making, characterized in that, The method includes: Acquire the temperature and pressure analysis datasets bound to the target sample, merge them to establish sample identifiers, and form a sample data package; Data quality diagnosis is performed based on the sample data packets to generate a quality profile set, and the sample data packets are divided into balanced subsets and unbalanced subsets according to the quality profile set and bound with balance labels. A task constraint set is constructed by combining the balance label with the sample structure background data, and a candidate set of thermobarometers matching the mineral pair identifiers is generated under the constraints of the task constraint set. An executability analysis is performed on the candidate thermometer / barometer set to form an executable thermometer / barometer set, and an algorithm configuration set is generated for the executable thermometer / barometer set; Based on the set of executable thermobarometers and the set of algorithm configurations, the temperature and pressure are solved in parallel to generate temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate. Non-equilibrium correction is introduced on the non-equilibrium subset to output the set of temperature and pressure calculation results. A consistency review is performed on the set of temperature and pressure calculation results. A recommended value range and a comprehensive confidence score are generated by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

2. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 1, characterized in that, The step of performing data quality diagnosis based on the sample data packets to generate a quality profile set, and dividing the sample data packets into balanced subsets and unbalanced subsets according to the quality profile set and binding balance labels thereon, specifically includes: The mineral pair identifiers and mineral chemical composition data in the sample data package are judged to be compatible in terms of elemental composition, stoichiometry, and endmember constraints, and incompatible records are marked as abnormal records. After removing the abnormal records, the remaining valid records are compared with the key element requirement template corresponding to the mineral pair identifier to calculate the missing ratio, substitution ratio and reasonableness of the key elements, thereby generating a component integrity index. By establishing charge conservation constraints for variable valence elements and combining them with the analysis of mass metadata, the consistency of valence state inference results at different measurement points in the same mineral spatial partition data is judged to generate a valence state consistency index. An endmember solvability assessment is performed on the valid records. The mineral chemical composition data is solved by endmember decomposition within the feasible region defined by the composition integrity index and the valence state consistency index. An endmember solvability index is generated based on the proportion of feasible solutions, the constraint conflict situation and the rationality of the endmember mole fraction. Spatial partition stability assessment is performed on the valid records. By aligning the endmember mole fraction sets of the same mineral in different mineral spatial partitions, a cross-partition composition gradient characterization is constructed. The rebalancing characteristics and kinetic interpretability of the composition gradient are determined by combining the sample construction background data, thereby generating a spatial partition stability index. The quality profile set is constructed based on the component integrity index, the valence consistency index, the endmember solvability index, and the spatial partition stability index. Based on the quality profile set, the sample data packets are subjected to balance discrimination. Records that simultaneously meet the end-member solvability threshold, spatial partition stability threshold, and valence consistency threshold are included in the balanced subset and bound with balance labels. Meanwhile, records that do not meet any of the threshold conditions are included in the unbalanced subset and bound with unbalanced source labels.

3. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 1, characterized in that, The process of constructing a task constraint set by combining the balance label with the sample tectonic background data, and generating a candidate set of thermobarometers matching the mineral pair identifiers under the constraints of the task constraint set, specifically includes: The balance label and the sample tectonic background data are jointly aligned to form a tectonic balance context, which is simultaneously bound to mineral pair identifiers, balance labels, non-equilibrium source labels, tectonic unit identifiers, shear band location identifiers, deformation strength grades, and deformation indication features under the sample identifier. Based on the constructed equilibrium context, applicable mineral combination constraints are generated. The mineral combination rules in the expert knowledge base are mapped by mineral pair identifiers and combined with the equilibrium label to exclude thermometer and barometer types that do not have non-equilibrium fault tolerance. Deformation strength constraints and strain rate sensitive constraints are generated based on the deformation strength level and the deformation indication features; Based on the aforementioned unbalanced source labels, rebalancing risk constraints are generated, and spatial partition usage constraints are generated simultaneously. The pressure range constraint and temperature range constraint are generated by combining the structural unit identifier, the shear band location identifier, the deformation strength grade, and the balance label; The applicable mineral combination constraint, the deformation strength constraint, the strain rate sensitivity constraint, the rebalancing risk constraint, the spatial partitioning usage constraint, the pressure range constraint, and the temperature range constraint are combined to form the task constraint set; Based on the task constraint set, the initial set of thermometers is retrieved in the expert knowledge base using mineral pair identifiers as the main search key. The initial set of thermometers is filtered and sorted according to the task constraint set to generate the candidate set of thermometers.

4. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 1, characterized in that, The step of performing executability analysis on the candidate thermometer / barometer set to form an executable thermometer / barometer set specifically includes: While maintaining the correspondence between the mineral pair identifiers and the thermobarometer candidate set, a candidate description package set is read for each thermobarometer candidate to form a candidate execution context; Based on the candidate execution context, the completeness of the candidate execution data of the thermometer and barometer is judged, and the thermometer and barometer candidates that do not meet the data requirement template are marked as data unexecutable candidates; After eliminating the data non-executable candidates, the thermobarometer candidates that do not have non-balance fault tolerance and rely on the assumption of complete balance are marked as premise conflict candidates. After eliminating the aforementioned conflict candidates, the theoretical boundary consistency judgment is performed on the remaining thermobarometer candidates, and the thermobarometer candidates that meet the conditions within the feasible domain are marked as boundary risk candidates. After eliminating the boundary risk candidates, a construction sensitivity consistency judgment is performed on the remaining thermobarometer candidates, and thermobarometer candidates that are incompatible with the construction environment are marked as construction incompatible candidates. After eliminating the incompatible candidates, a computational reproducibility evaluation is performed on the remaining thermobarometer candidates, and the thermobarometer candidates that are sensitive to disturbances are marked as numerically unstable candidates. Based on the data completeness judgment results, premise consistency judgment results, theoretical boundary consistency judgment results, construction sensitivity consistency judgment results, and reproducibility index, the thermobarometer candidates in the thermobarometer candidate set that are not marked as any unexecutable type are included in the executable thermobarometer set.

5. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 1, characterized in that, The configuration set for the executable thermobarometer set generation algorithm specifically includes: Based on the sample identifier, a data mapping configuration is generated for each candidate read data requirement template of the executable thermobarometer set; The endmember decomposition strategy is determined based on the solid solution endmember set, endmember constraint conditions, and task constraint set, and the endmember decomposition constraints are selected according to the balance label to output the endmember solution configuration. The activity model configuration is generated by selecting a matching activity model based on the set of applicable premises and the set of theoretical boundaries. Based on the price state consistency index and the non-equilibrium source label, a fixed price state strategy is selected to generate a price state processing configuration; The unbalanced correction channel is enabled based on the balance label, and an unbalanced correction configuration is generated. The calibration data is extracted from the theoretical boundary set and combined with the task constraint set to form the calibration feasible domain, generating the experimental anchoring configuration. The selection of solution variables, initial value generation, search strategy, convergence criterion, abnormal termination conditions and restart strategy are uniformly and solidified to generate joint solution control configuration; The data mapping configuration, the endmember solving configuration, the activity model configuration, the valence state processing configuration, the non-equilibrium correction configuration, the experimental anchoring configuration, and the joint solving control configuration are aggregated and a consistency check is performed to form the algorithm configuration set.

6. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 5, characterized in that, The method involves parallel execution of joint temperature and pressure solving based on the executable thermobarometer set and the algorithm configuration set. While maintaining the result traceability chain, it generates temperature results, pressure results, and uncertainty sets corresponding to each thermobarometer candidate. Furthermore, it introduces unbalanced corrections on the unbalanced subset, specifically including: Based on the data mapping configuration, variable assembly is performed on the sample data package to generate a candidate variable package; Perform endmember solving according to the endmember solving configuration to generate a set of endmember mole fractions; Generate an activity variable package based on the activity model configuration; Based on the valence state processing configuration, valence state processing is performed on the activity variable package to generate a modified activity variable package; Based on the joint solution control configuration, the joint solution main process is started, and a joint search is performed on the candidate values ​​of temperature and pressure within the solution domain constraint. At each candidate point, the temperature result and pressure result are determined according to the modified activity variable package. When the balance label indicates an unbalanced subset, an unbalanced correction is introduced in the joint solution master process. The unbalanced correction is based on the diffusion length parameter, composition gradient parameter and rebalancing weight parameter in the unbalanced correction configuration to weight and fuse the endmember contributions of the kernel partition and the edge partition. After obtaining the temperature and pressure results, an uncertainty set bound to the configuration identifier is generated; After the temperature results, pressure results, uncertainty set, and solution status are solved in parallel, the temperature and pressure calculation result set is formed under the sample identifier.

7. The geological temperature and pressure analysis method based on multi-agent collaborative decision-making according to claim 1, characterized in that, The consistency review of the temperature and pressure calculation results set is performed. A recommended value range and a comprehensive confidence score are generated through a weighted fusion of result dispersion, theoretical compatibility, and confidence level. This results in a structured evaluation linked to the sample identifier, specifically including: The set of temperature and pressure calculation results is subjected to reviewability screening to form a review result pool; For each thermometer / barometer candidate in the review result pool, a standardized result representation is constructed; Based on the standardized results, the dispersion of the calculation results is represented, and temperature dispersion index and pressure dispersion index are generated respectively. After obtaining the temperature dispersion index and the pressure dispersion index, a theoretical compatibility judgment is performed. Based on the set of applicable premises, the set of theoretical boundaries, the set of task constraints, the balance label and the non-balance source label, the theoretical assumption compatibility, the overlap of applicable intervals and the consistency of construction sensitivity among the candidate thermometers and barometers in the review result pool are jointly judged, and a compatibility matrix and compatibility index are generated. A candidate weight set is generated based on the compatibility matrix and empirical confidence parameters. The candidate weight set is then corrected by combining risk level labels, price state processing strategy stability, unbalanced correction activation status, and calibration feasible region constraint status to form fused weights. The temperature dispersion index, the pressure dispersion index, the compatibility index, and the candidate weight set are then weighted and fused to generate weighted center values ​​in the temperature and pressure dimensions, respectively. These values ​​are then combined with the uncertainty set to expand the range of recommended temperature and pressure values.

8. A geological temperature and pressure analysis device based on multi-agent collaborative decision-making, characterized in that, The device is used to execute a geothermal and pressure analysis method based on multi-agent collaborative decision-making as described in any one of claims 1-7. The device includes an acquisition module, a processing module, and an output module, wherein: The acquisition module is used to acquire and merge the temperature and pressure analysis dataset bound to the target sample to establish a sample identifier, thereby forming a sample data package; The processing module is used to perform data quality diagnosis based on the sample data package to generate a quality profile set, and to divide the sample data package into a balanced subset and an unbalanced subset according to the quality profile set and bind a balance label. The processing module is used to construct a task constraint set by combining the balance label and the sample structure background data, and generate a set of thermobarometer candidates that match the mineral pair identifiers under the constraints of the task constraint set. The processing module is used to perform executability analysis on the candidate set of thermobarometers to form an executable set of thermobarometers, and generate an algorithm configuration set for the executable set of thermobarometers; The processing module is used to perform joint temperature and pressure solving in parallel based on the set of executable thermobarometers and the set of algorithm configurations, generate temperature results, pressure results and uncertainty sets corresponding to each thermobarometer candidate, introduce non-equilibrium correction on the non-equilibrium subset, and output a set of temperature and pressure calculation results. The output module is used to perform a consistency review on the set of temperature and pressure calculation results, and generate a recommended value range and a comprehensive confidence score by weighted fusion of result dispersion, theoretical compatibility and confidence level, so as to output a structured evaluation result bound to the sample identifier.

9. An electronic device, characterized in that, The device includes a processor, a communication bus, a user interface, a network interface, and a memory. The memory is used to store instructions. The user interface and the network interface are both used to communicate with other devices. The communication bus is used to enable communication between the components within the electronic device. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.

10. A non-transitory computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.

Images