Multi-source geological information fusion analysis method and system

Abstract

The present application relates to geological data processing and information fusion technical field, specifically to a kind of multi-source geological information fusion analysis method and system;The method comprises: systematized acquisition multi-source heterogeneous data, through adaptive correction, space-time coordinate unification, abnormal processing and feature extraction, form standardized multidimensional feature vector;Further construct space-time fusion system, feature mapping to uniform three-dimensional grid, combined with time series dynamic coding and adaptive weight optimization, generate multi-scale fusion feature matrix;Subsequently through state reconstruction, correlation mapping, dynamic deduction and result solidification mechanism, abnormal identification and evolution trend analysis are carried out to fusion feature;Finally, based on abnormal grade, evolution rate and space expansion index, intelligent decision matrix is constructed, abnormal area classification, strategy generation and adaptive resource allocation are realized, and decision is continuously optimized through closed-loop feedback.The present application constructs a complete closed-loop system from multi-source information perception, fusion analysis, abnormal deduction to intelligent decision.

Description

Technical Field

[0001] This invention relates to the field of geological data processing and information fusion technology, specifically to a method and system for multi-source geological information fusion analysis. Background Technology

[0002] With the continuous advancement of geological exploration technology and the widespread application of diversified sensing methods, the acquisition of geological data has entered a new stage characterized by multi-source, high-dimensionality, and real-time processing. Technologies such as drilling, geophysical exploration, remote sensing observation, and in-situ monitoring together constitute a comprehensive perception system for underground structures and geological phenomena. These multi-source heterogeneous data exhibit complementarity and complexity in terms of spatial coverage, time series, and physical attributes, providing a rich information foundation for a deeper understanding of geological structures, resource distribution, and environmental evolution.

[0003] Chinese invention patent application CN120123977A discloses a geological and mineral exploration intelligent system and method based on multi-source data fusion, belonging to the field of geological and mineral exploration technology. It includes a data acquisition module, a data preprocessing module, a data fusion module, and an intelligent analysis module. The data acquisition module collects geological and mineral data through multiple sensors; the data preprocessing module cleans, standardizes, and reduces the dimensionality of the collected data; the data fusion module uses a Bayesian network to fuse the preprocessed multi-source data; and the intelligent analysis module identifies potential mineral areas using the YOLO target detection algorithm, and then uses... Clustering algorithms classify geological units, and finally, genetic algorithms are used to optimize drilling paths and location selection.

[0004] Currently, the application of information fusion and intelligent analysis technologies in the geosciences is deepening, driving geological research to evolve from single-data interpretation to multi-source collaborative analysis. How to systematically integrate data from different sources, scales, and time phases, and extract comprehensive features with spatiotemporal consistency and physical interpretability, has become a key issue in improving the accuracy of geological cognition and the effectiveness of decision-making. Against this backdrop, constructing an integrated technical system capable of standardized access to multi-source data, spatiotemporal fusion modeling, dynamic anomaly identification, and intelligent decision support is of great significance for promoting the development of geological information processing towards intelligence, systematization, and operationalization, and also provides solid technical support for applications such as resource exploration, disaster early warning, and geological environment monitoring. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background technology by proposing a multi-source geological information fusion analysis method and system.

[0006] The technical solution of this invention: a multi-source geological information fusion analysis method, comprising the following specific implementation steps:

[0007] S1. Systematically collect geological exploration data, remote sensing image data, geophysical measurement data, and historical geological archive data; perform adaptive noise correction, spatiotemporal coordinate unification, outlier and missing value processing on each data source, and extract key geological features to generate standardized multidimensional feature vectors.

[0008] S2. The standardized multidimensional feature vectors are mapped to a unified three-dimensional geological grid. Combined with temporal dynamic coding and adaptive weight optimization based on data source reliability, spatial coverage and feature change magnitude, multi-source feature fusion is performed to generate a multi-scale fusion feature matrix.

[0009] S3. Based on the multi-scale fusion feature matrix, state reconstruction and correlation mapping are performed by calculating the baseline state, deviation degree, and cumulative deviation intensity of each grid unit; then, the anomaly evolution rate and spatial expansion index are analyzed to realize the dynamic identification and trend inference of geological anomalies.

[0010] S4. Based on the abnormal evolution rate and spatial expansion index, combined with the abnormal level classification results, a multi-dimensional intelligent decision matrix is ​​constructed to generate priority ranking, intervention strategies and resource allocation schemes for different abnormal regions; and by comparing the deviation between the strategy execution effect and the expected effect, the decision parameters are adaptively optimized through feedback learning to form a closed-loop decision system.

[0011] Preferably, in step S1, adaptive noise correction is achieved by introducing a dynamic correction function based on the noise level of each data source. This function corrects and performs preliminary filtering on the original acquired values.

[0012] Spatiotemporal coordinate unification includes using spatial coordinate transformation functions to transform the spatial coordinates of each source data to the same reference system, and mapping asynchronously acquired data to a standard time series defined by a unified time reference point.

[0013] Preferably, in step S1, extracting key geological features specifically includes:

[0014] Extract stratigraphic thickness, lithological classification, gravity gradient, magnetic anomaly, and texture features from remote sensing images;

[0015] The extracted original values ​​of various features are normalized to form multidimensional feature vectors with uniform dimensions.

[0016] Preferably, in step S2, the standardized multidimensional feature vectors are mapped to a unified three-dimensional geological grid, specifically as follows:

[0017] For each grid cell, the inverse distance weight is calculated based on the Euclidean distance between all sampling points in its spatial neighborhood and the center of the cell.

[0018] The weights are then used to sum the feature vectors of each sampling point in the neighborhood, resulting in the spatially mapped feature vector of the grid cell.

[0019] Preferably, in step S2, the timing dynamic coding specifically includes:

[0020] For each grid cell, the characteristic change between adjacent time points is calculated as the instantaneous change characteristic;

[0021] Calculate the characteristic standard deviation within its recent time window as the volatility characteristic;

[0022] Instantaneous change features and fluctuation features are combined with preset weight coefficients to form a time-series dynamic feature vector. This time-series dynamic feature vector is then combined with the spatially mapped feature vector to generate a three-dimensional time dynamic coding matrix.

[0023] Preferably, in step S2, the adaptive weight optimization specifically includes:

[0024] For each data source, a reliability weight is calculated based on its average noise level, and a spatial coverage weight is calculated based on the ratio of the number of its covered grid cells to the total number of grid cells in the study area.

[0025] Then, by combining the characteristic change magnitude of the data source at the current moment, the final fusion weight of the data source at a specific grid cell and time node is obtained by comprehensively calculating the weighted adjustment parameters.

[0026] Preferably, in step S3, the state reconstruction specifically involves:

[0027] For each spatial grid cell, the mean value is calculated from the time interval with stable change amplitude selected from its historical fusion feature sequence, and the individualized baseline state vector of the cell is constructed.

[0028] Calculate the overall offset between the fused feature vector of the unit at the current moment and the reference state vector, as the instantaneous deviation.

[0029] Preferably, in step S3, the association mapping specifically involves:

[0030] Within a preset time accumulation window, the instantaneous deviation is summed by time decay weighting to obtain the cumulative deviation intensity;

[0031] The cumulative deviation intensity of each grid cell is compared with the mean and standard deviation of the cumulative deviation intensity of all cells in its region;

[0032] When a grid cell exceeds a threshold determined by the mean, standard deviation, and anomaly detection sensitivity coefficient, it is marked as an anomaly candidate cell.

[0033] Preferably, in step S3, the analysis of the abnormal evolution rate and spatial expansion index specifically includes:

[0034] The ratio of the cumulative deviation intensity difference of anomalous candidate units at adjacent time points to the time interval is calculated as the anomalous evolution rate.

[0035] The number of grid cells in the spatial neighborhood of the candidate anomalous cell that are also marked as anomalous candidates is counted, and the anomalous spatial expansion index is calculated in combination with the anomalous confirmation threshold.

[0036] The technical solution of the present invention: a multi-source geological information fusion analysis system, which is used to execute the above-mentioned multi-source geological information fusion analysis method, including: an edge terminal and a cloud terminal;

[0037] The edge end is equipped with a multi-source geological sensing and standardization access module, which is used to collect, temporally and spatially align, quality check and standardize the multi-source heterogeneous geological data from drilling records, geophysical measurements, remote sensing images and in-situ sensing monitoring.

[0038] The cloud platform is equipped with a geological feature fusion and anomaly characterization module, an anomaly evolution analysis and trend prediction module, and an intelligent decision generation and adaptive feedback module.

[0039] The geological feature fusion and anomaly characterization module is connected to the multi-source geological sensing and standardization access module. It is used to perform spatiotemporal fusion modeling on the standardized data, generate a multi-scale fusion feature matrix, and identify and quantify geological anomaly units that deviate from the background features.

[0040] The anomaly evolution analysis and trend prediction module is connected to the geological feature fusion and anomaly characterization module. It is used to predict the evolution direction, development rate and impact range of the geological anomaly unit based on the identified anomaly features and combined with time series and spatial neighborhood correlation information.

[0041] The intelligent decision generation and adaptive feedback module is connected to the anomaly evolution analysis and trend inference module. It is used to generate analysis conclusions and decision suggestions by comprehensively considering the anomaly level and evolution trend, and to adaptively adjust the analysis parameters and decision strategies based on subsequent data feedback.

[0042] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects:

[0043] This invention designs a multi-source geological information fusion analysis method and system. By constructing a three-dimensional fusion system of spatiotemporal network mapping + temporal dynamic coding + adaptive weight optimization, it elevates static data fusion to dynamic spatiotemporal state reconstruction. More importantly, it introduces an analysis and decision-making mechanism of state reconstruction-association mapping-dynamic inference-closed-loop feedback, realizing a leap from decentralized and isolated geological information processing to system integration. First, by introducing a cloud-edge collaborative system architecture, it significantly improves the overall efficiency and adaptability of data processing. At the edge, multi-source heterogeneous data is collected, corrected, and standardized in real time, effectively reducing noise and inconsistency of the original data and alleviating the transmission and storage load on the cloud. On the cloud, it focuses on spatiotemporal fusion, deep feature extraction, and trend inference, achieving optimized allocation of computing resources. This collaborative mechanism enables the system to respond to local real-time changes and perform global in-depth analysis and decision-making, ensuring the agility and reliability of the entire process. Second, by constructing a spatiotemporal fusion system and a dynamic weight optimization mechanism, it significantly improves the overall efficiency and adaptability of geological information processing. This approach enhances the accuracy and interpretability of information fusion by mapping multi-source features to a unified spatiotemporal grid and combining it with temporal dynamic coding to achieve a multi-dimensional and continuous expression of geological conditions. Through adaptive weight allocation, the system can dynamically adjust the contribution of each source information based on data quality, coverage, and sensitivity to change, making the fusion results more balanced and robust, and effectively uncovering the correlation patterns and anomalies hidden in heterogeneous data. Finally, an intelligent decision-making mechanism centered on state reconstruction, evolutionary deduction, and closed-loop feedback is established, improving the foresight and operability of geological anomaly identification and resource regulation. By transforming multi-source information into a computable and deducible system state and constructing a decision matrix based on anomaly level, evolution rate, and spatial expansion index, the system can achieve risk classification, strategy generation, and adaptive resource allocation. Continuous feedback learning further optimizes model parameters and decision logic, enabling the system to continuously evolve during operation, ultimately providing efficient and reliable intelligent decision support for geological monitoring, disaster early warning, and resource exploration. Attached Figure Description

[0044] Figure 1 This is a flowchart of a multi-source geological information fusion analysis method proposed in this invention;

[0045] Figure 2 This is a system architecture diagram of a multi-source geological information fusion analysis system proposed in this invention. Detailed Implementation

[0046] Example 1, as Figure 1 As shown, the multi-source geological information fusion analysis method proposed in this invention has the following specific implementation steps:

[0047] S1. Systematically collect, standardize, control data quality, and extract preliminary features from multi-source heterogeneous geological information to form a unified, multi-dimensional, and computable feature vector, providing high-quality input for subsequent multi-source fusion and anomaly prediction. The specific implementation process is as follows:

[0048] S11. Collect multi-source data including geological exploration, remote sensing imagery, geophysical surveys, and historical archives. Adaptive correction functions are used to eliminate noise and errors from different data sources. At the edge, point, area, and volume data are transformed into comparable standard data to provide a high-quality foundation for subsequent processing. Specifically:

[0049] Data is collected from multiple sources, including but not limited to:

[0050] Geological exploration data: point data such as borehole logging and core sample analysis;

[0051] Remote sensing image data: high-resolution satellite imagery, infrared thermal imagery, radar imagery, and other area data;

[0052] Geophysical measurement data: volumetric data including magnetic force, gravity, geoelectricity, and electromagnetic response;

[0053] Historical geological archives: records of mineral distribution, groundwater level, stratigraphic thickness, etc.

[0054] Introducing data source noise Dynamic correction function It adaptively adjusts data based on different collection conditions, for any data source. In time and spatial location The collected values ​​can be expressed as:

[0055] ;

[0056] in, This indicates that the data source i is located at spatial position p. k Time point t j Standardized data collection; This represents the raw values ​​originally collected from data source i, including point, area, or volume data; This indicates the observation error or noise level of data source i; This represents a data source-specific preprocessing function used for noise correction, unit conversion, and preliminary filtering.

[0057] It should be noted that data source-specific preprocessing functions It is an adaptive transformation function designed for different geological data sources (such as borehole logging, remote sensing imagery, geophysical measurements, etc.) and is used to transform raw acquired data. Noise correction, unit standardization, signal filtering, and preliminary correction of outliers are performed to achieve data standardization and comparability, while preserving key geological features. This provides a unified and high-quality input for multi-source fusion and is the core processing step to ensure that heterogeneous data can be fused in space, time, and dimensions.

[0058] S12. Unify the spatial coordinates of all source data to the same reference system, and map the asynchronously acquired data to a standard time series through interpolation or alignment to achieve consistency in spatial and temporal dimensions, providing reliable support for multi-source fusion and dynamic analysis. Specifically:

[0059] Spatial coordinates of data from different sources Transform to a unified reference system (such as WGS-84 or a local geological coordinate system): ;

[0060] Perform time interpolation or alignment on asynchronously acquired data: ;

[0061] in, Represents the standardized spatial coordinates; This represents a spatial coordinate transformation function that transforms the original coordinate system to a unified reference system (such as WGS-84 or a local geological coordinate system). Indicates the original coordinate reference parameters of data source i, such as original projection type, reference point, and coordinate offset; This represents a standardized point in time. This represents a unified time reference point or start time, set by the analysis system, and serves as the initial observation time for the dataset. Indicates the original collection time and reference time Time difference;

[0062] S13. Perform outlier detection, missing value imputation, and noise filtering on the standardized data. Combine statistical methods with neighborhood interpolation to form a complete and reliable dataset, ensuring data accuracy and usability, and providing high-quality input for feature extraction. Specifically:

[0063] Standardized data Perform statistical analysis to identify outliers with significant deviations:

[0064] ;

[0065] when The time was marked as an outlier;

[0066] Missing data can be filled using neighborhood interpolation or prediction based on historical trends:

[0067] ;

[0068] in, Representing data points The standardized deviation is used for outlier identification; This represents the mean of data source i; This represents the standard deviation of data source i; This indicates the threshold for identifying outliers, which is adaptively set by the system designer based on the range of data fluctuations. This represents the data points after missing values ​​have been filled in; N represents the neighborhood set used during interpolation or completion, including valid data points in both the spatial and temporal neighborhoods.

[0069] S14. Extract key geological features from the processed multi-source data, such as stratigraphic thickness, lithology, gravity gradient, magnetic anomaly, and image texture, and normalize them to generate a unified multi-dimensional feature vector. This provides standardized and directly computable input for subsequent multi-source feature fusion, ensuring a unified representation of heterogeneous information. Specifically:

[0070] Extract key geological features based on different data types, including but not limited to: stratigraphic thickness. Lithological classification Gravity gradient Magnetic anomaly Remote sensing image texture features ;

[0071] The extracted features are normalized to form a unified vector:

[0072] ;

[0073] ;

[0074] in, This represents the preliminary feature vector, which contains key geological features extracted from multi-source data; This represents the normalized value of the l-th feature in vector F; This represents the l-th feature value extracted from the original sample (such as lithology, thickness, magnetic anomaly, etc.). and These represent the minimum and maximum values ​​of the l-th feature in the current dataset or historical observations, respectively.

[0075] S2. Based on the standardized features in step S1, a spatiotemporal fusion system of multi-source geological information is constructed in the cloud. Through spatial grid mapping, temporal dynamic coding, adaptive weight optimization, and multi-scale feature matrix generation, a unified representation and information complementarity of heterogeneous data are achieved, providing high-precision and operable input for subsequent anomaly identification and potential resource prediction. The specific implementation process is as follows:

[0076] S21. Map the multi-source standardized features output in step S1 onto a unified 3D geological grid. Calculate the fused features using inverse-weighted calculation based on the distance between neighboring points, and construct a local spatial topology matrix to reflect neighborhood correlations. This achieves spatial unification and preservation of local relationships for data from different sources, providing a physically interpretable foundation for subsequent dynamic feature encoding. Specifically:

[0077] The multi-source feature vector generated in step S1 Mapped to a unified three-dimensional geological grid middle: ; ;

[0078] Based on spatial neighborhood Construct the local topology matrix T: ;

[0079] in, Indicates time t j Below, the feature vector of a three-dimensional grid cell (x,y,z) after spatial mapping is a comprehensive representation of the features of multiple sampling points after spatial weighting; This represents the set of sampling points that are spatially adjacent to the grid cell (x,y,z), and this neighborhood is determined by a spatial distance threshold or the nearest neighbor rule; This indicates that the k-th sample point, which has undergone standardization, is output from step S1 at time t. j Multidimensional geological feature vectors; The spatial weight representing the contribution of sampling point k to the feature of the current grid cell is used to reflect the degree of influence of distance on the feature; Indicates sampling point p k The Euclidean distance between the grid cell center point and the grid cell center point is used to characterize the spatial proximity. This represents the sum of the reciprocals of the distances to all sampling points in the neighborhood, used to normalize the weights and ensure that the sum of all weights is 1; This represents the spatial topological association strength between sampling point i and sampling point j, used to characterize their correlation in spatial structure; This represents the spatial attenuation coefficient, used to control the rate at which distance affects the strength of topological associations. This parameter can be set according to the scale of the study area.

[0080] S22. Perform instantaneous change and fluctuation analysis on the time series of each grid cell, extract dynamic feature vectors, and combine historical and real-time features to generate a dynamic coding matrix. This ensures that the time series information is completely captured, reflects geological change trends, and forms a spatiotemporally continuous representation that can be used for subsequent fusion and prediction. This achieves a dynamic expression of unified spatial-temporal features, specifically:

[0081] Extract dynamic features from the time series of each grid cell:

[0082] ;

[0083] A three-dimensional-temporal dynamic coding matrix is ​​generated by combining spatial grids with temporal features. This forms a unified input format for subsequent fusion analysis;

[0084] in, Indicates the grid cell at time t j The temporal dynamic feature vector is used to reflect the state of geological features as time changes; This represents the characteristic changes of a grid cell between adjacent time points; Indicates from time t j-n to t j The characteristic standard deviation is used to characterize the degree of fluctuation over a short period of time; and Weighting coefficients representing instantaneous change characteristics and fluctuation characteristics are used to balance trend and stability information; This represents the joint feature representation after combining spatial fusion features and temporal dynamic features;

[0085] S23. Based on the reliability of the data source, spatial coverage, and the magnitude of local feature changes, dynamically calculate the fusion weight of each data source, and combine the weighted summation to generate the final fused feature vector. This achieves full-dimensional information complementarity, adaptively adjusts the contribution of different data sources to the results, improves fusion accuracy and stability, and ensures a balanced representation of multi-source features in spatial, temporal, and reliability dimensions. Specifically:

[0086] Calculate the reliability weight for each data source i and spatial coverage weight :

[0087] ; ;

[0088] The reliability, coverage, and temporal variations of the data source are combined to form the fusion weight. :

[0089] ;

[0090] The final fused features are generated based on the weights:

[0091] ;

[0092] in, This represents the set of all data sources that are effectively covered by the grid cell (x,y,z). Only the sources that are actually covered by data are summed to avoid interference from null values ​​or invalid data. This represents the reliability weight of the i-th data source; This represents the average noise or uncertainty level of data source i; Indicates the spatial coverage weight of data source i; This indicates the actual number of grid cells covered by the data source; This represents the total number of all grid cells within the study area; This represents the comprehensive fusion weight of the i-th data source under a specific grid and time point; This indicates the magnitude of the characteristic change of the data source at the current moment, used to reflect its sensitivity to abnormal changes; , and This represents the weighting adjustment parameter, used to control the relative impact of reliability, coverage, and dynamic sensitivity; This represents the final fused feature vector from multiple data sources at this grid cell and time point; This indicates the spatial mapping characteristics of data source i within the grid cell;

[0093] S24. Decompose the fused features according to different spatial scales, generate multi-scale features through neighborhood averaging, and combine them with dynamic temporal features to form an operable analysis matrix. This matrix reflects both local details and preserves the overall trend, providing high-dimensional computable input for subsequent anomaly identification and resource prediction, achieving multi-scale coverage and analytical flexibility. Specifically:

[0094] Fusion features Decomposed by spatial scale:

[0095] ;

[0096] Combine multi-scale features and dynamic temporal features to generate an analysis matrix:

[0097] ;

[0098] in, This represents the fusion feature result at scale s; This represents the set of spatial neighborhood grids corresponding to scale s; This indicates the number of grid cells contained in the neighborhood at this scale; This represents the final output multi-scale, multi-time, and multi-source fusion analysis matrix; Indicated in scale Below, regarding fusion features The scale feature vector obtained after spatial neighborhood aggregation; K represents the total number of spatial scale levels; This represents the k-th spatial scale level, corresponding to different sizes of spatial neighborhood ranges.

[0099] S3. Based on the structured intermediate results output in step S2, a data adaptive analysis and evolutionary decision-making mechanism for computer systems is constructed. Through a progressive processing chain of "state reconstruction - correlation mapping - dynamic deduction - result solidification", discrete and time-varying system information is transformed into a computable, feedback-capable, and continuously optimizable internal model, providing a stable and interpretable computational foundation for subsequent control or prediction. The specific implementation process is as follows:

[0100] S31. Perform semantic reconstruction on the output of step S2. By introducing contextual consistency constraints and time dependency constraints, the original results are mapped to a unified system state vector, solving the inconsistency issues in scale, granularity, and temporal sequence of results from different sources, and establishing a stable input for subsequent analysis. Specifically:

[0101] For each spatial grid cell (x, y, z), a relatively stable time interval is selected from its own historical fusion feature sequence to construct an individualized baseline state;

[0102] Construct the baseline state vector: ;

[0103] At the current time t j Calculate the overall offset of the fused feature of this unit relative to its own baseline:

[0104] ;

[0105] in, Represents the baseline state fusion feature vector of a spatial grid cell (x,y,z); This indicates the length of the historical time window used to calculate the baseline state; It represents the instantaneous deviation, measuring the difference between the current feature vector and the baseline state;

[0106] S32. Based on the state vector, an implicit association mapping mechanism is constructed. By jointly characterizing the sensitivity to state changes and covariance features, the inherent coupling relationship between each state component is explored, so that the system no longer depends on static rules, but forms an association structure that can evolve with the data. Specifically:

[0107] Within the sliding time window W, the deviation is weighted and accumulated:

[0108] ;

[0109] A grid cell is marked as an anomalous candidate when the following conditions are met:

[0110] ;

[0111] in, This represents the cumulative deviation intensity, where the deviation degree D is weighted and accumulated within the time window W; Represents the time decay weight, satisfying W represents the length of the time accumulation window, which is set according to the formation time of geological anomalies and the data sampling period. This represents the cumulative deviation intensity mean within the same region, i.e., the values ​​of each grid cell within the statistical region. The mean; This represents the cumulative deviation intensity standard deviation within the same area, and is used to statistically analyze the grid cells within the area. Standard deviation; This represents the sensitivity coefficient for anomaly detection, which is set empirically or optimized using historical data.

[0112] S33. Based on the correlation mapping results, a dynamic inference mechanism is introduced. Through the linkage of continuous state transitions and constraints, the evolution path of the system under different disturbances or input conditions is simulated, thereby identifying potential trends and abnormal directions in advance and enhancing the system's adaptability to complex scenarios. Specifically:

[0113] Time gradient analysis of cumulative deviation intensity:

[0114] ;

[0115] Calculate the anomaly expansion index by combining the neighborhood grid:

[0116] ;

[0117] in, Indicates the rate of anomalous evolution; Indicates the interval between adjacent time points; Indicates the exponent of anomaly space expansion; Indicates the threshold for confirming anomalies; This indicates an indicator function; it outputs 1 if an exception is detected, and 0 otherwise.

[0118] S34. The inference results are subjected to stability screening and result solidification processing. High-confidence evolution conclusions are transformed into standardized calculation outputs and fed back into the internal model of the system to form a closed-loop update relationship. Specifically:

[0119] An anomaly level function is constructed by combining deviation intensity, evolution rate, and expansion index:

[0120] ;

[0121] in, Indicates the comprehensive level of abnormality; , and The weighting of the overall abnormality level can be optimized using historical data or set based on expert experience.

[0122] S4. Based on the anomaly level, evolution rate, and spatial expansion index output in step S3, a multi-dimensional intelligent decision matrix is ​​constructed to realize the priority ranking, strategy generation, adaptive optimization, and comprehensive operable output of anomaly areas. Through a closed-loop feedback mechanism, the strategy is continuously adjusted, so that geological anomaly management and prediction form a dynamic, implementable, and feasible intelligent decision-making closed loop. The specific implementation process is as follows:

[0123] S41. The abnormal units identified in step S3 are divided into different risk areas according to their comprehensive level. Regional priorities are calculated based on evolution rate and spatial expansion index to ensure priority intervention in high-risk areas, aligning resource allocation with risk severity. Simultaneously, the level thresholds are dynamically adjusted to adapt to different geological backgrounds. Specifically:

[0124] According to the abnormality level index Classify spatial grid cells:

[0125] ;

[0126] Each Corresponding to different risk levels: potential anomaly, developmental anomaly, and high-risk anomaly;

[0127] To ensure spatial continuity, spatial neighborhood connectivity analysis is used to merge adjacent anomalous units into a complete region, avoiding misjudgment of isolated units;

[0128] Calculate dynamic threshold ;

[0129] And sort by region priority: ;

[0130] in, Indicates at time t j The set of spatial regions divided into the i-th anomaly level; This represents the threshold value corresponding to the i-th anomaly level; This represents the average value of the anomaly level index L within a historical time window. This represents the historical standard deviation of the anomaly level index L; This represents the anomaly level adjustment coefficient, used to adjust the sensitivity and boundary width between different anomaly levels; Indicates abnormal area Overall priority score; , and These represent abnormal regions. The average of internal anomaly level, evolution rate, and expansion index; , and Represents the weighting coefficients of anomaly intensity, evolution rate, and spatial expansion in priority calculation;

[0131] S42. Generate a unit-level strategy matrix based on each anomaly region and its priority, including measure type, intervention intensity, and resource allocation. Calculate weight ratios to allocate resources to high-priority regions, ensuring that anomaly intervention plans are both spatially and temporally reasonable and targeted, while also considering the complexity of the geological environment. Specifically:

[0132] Generate the policy matrix:

[0133] ;

[0134] Calculate resource allocation weights: ;

[0135] For high-level anomaly areas: implement emergency intervention, increase monitoring frequency, and deploy mobile sensors;

[0136] For areas with abnormal development: increase predictive sampling and extend the data collection window;

[0137] For potentially abnormal areas: maintain basic monitoring and collect historical data;

[0138] in, This indicates the total number of currently identified anomalous regions; This represents the total amount of resources available for allocation; Indicates the region Total number of grid cells; This represents the initial strategy matrix generated at a specified spatial location and time. Indicates the strategy generation function; This indicates the proportion of resources allocated to this spatial unit;

[0139] It should be noted that the policy generation function This function is used to transform the comprehensive anomaly degree, evolution trend, and spatial impact characteristics of anomalous areas into executable control and response strategies. Its core function is to establish a mapping relationship between analysis results and decision-making behavior. By comprehensively evaluating the relative importance and urgency of different anomaly indicators, the function dynamically adjusts the strategy strength, execution priority, and resource allocation ratio, so that high-risk, rapidly evolving, or expanding areas receive priority intervention support, while avoiding excessive intervention in low-risk areas. The strategy generation function fully considers historical execution effects and real-time feedback information during operation, and can adaptively correct the generated results, thereby ensuring that the strategy output is consistent with the current geological conditions and has continuity and stability, providing a reliable foundation for subsequent dynamic optimization and closed-loop updates.

[0140] S43. Real-time comparison of the strategy execution effect with the initial strategy, calculation of deviation, and iterative update of the strategy matrix according to the feedback learning rate to achieve adaptive strategy optimization. This allows for continuous adjustment of resource allocation, intervention intensity, and measure type to match abnormal evolution trends, forming a closed-loop feedback and improving decision-making accuracy. Specifically:

[0141] Calculate the feedback error: ;

[0142] Configure policy update rules:

[0143] ;

[0144] in, This represents the quantitative result of the strategy's actual effect after execution, obtained by the monitoring system through data collection and evaluation of the execution results; It represents the deviation between the expected results of a strategy and the actual results, and is used to measure the rationality of the current strategy; It represents the policy feedback learning rate, controls the policy update magnitude, and prevents over-adjustment or response lag; This represents the updated optimization strategy matrix at the next time step.

[0145] S44. Integrate anomaly level, evolution rate, spatial expansion, and optimization strategy information to generate a comprehensive decision matrix, and visualize the spatial distribution map, time trend curve, and intervention suggestion table. This provides actionable guidance for engineering operations, supports continuous monitoring and strategy iteration, and ensures that decisions are executable and capable of continuous optimization. Specifically:

[0146] Generate a comprehensive decision matrix:

[0147] ;

[0148] Visualize the matrix information and generate:

[0149] Spatial distribution map: Highlighting high-risk areas and the direction of abnormal expansion;

[0150] Time trend curve: shows the rate of abnormal development and the effectiveness of the strategy;

[0151] Decision Recommendation Form: Clearly define the intervention type, priority, and resource allocation ratio for each abnormal area;

[0152] in, This represents the final integrated decision matrix, which encapsulates abnormal states, evolutionary characteristics, and optimization strategies in a unified manner for system output and visualization.

[0153] Example 2, as Figure 2As shown, the present invention proposes a multi-source geological information fusion analysis system, which is used to execute a multi-source geological information fusion analysis method proposed in Embodiment 1, including: edge terminal and cloud terminal.

[0154] The edge terminal includes: a multi-source geological sensing and standardized access module;

[0155] The multi-source geological sensing and standardized access module is used to uniformly collect, temporally and spatially align, and verify the quality of multi-source heterogeneous geological data from drilling records, geophysical measurements, remote sensing images, and in-situ sensing monitoring at the edge. It also transforms data from different sources and scales into a standardized and fusionable representation, providing a consistent data foundation for subsequent analysis.

[0156] The cloud platform includes: a geological feature fusion and anomaly characterization module, an anomaly evolution analysis and trend prediction module, and an intelligent decision generation and adaptive feedback module;

[0157] The geological feature fusion and anomaly characterization module is used to jointly model and fuse geological structural features, physical property change features and environmental response features based on standardized data, construct a comprehensive feature expression reflecting the state of geological units, and identify and quantify anomalous units that deviate from background features, thereby realizing a systematic characterization of geological anomalies.

[0158] The anomaly evolution analysis and trend prediction module is used to continuously predict and dynamically update the evolution direction, development rate and impact range of anomalies based on the identified anomaly characteristics, combined with time series change relationships and spatial neighborhood correlation information, thereby forming interpretable analysis results of the geological risk evolution process.

[0159] The intelligent decision generation and adaptive feedback module is used to generate targeted analysis conclusions and decision suggestions by comprehensively considering anomaly level, evolution trend and spatial correlation information. It also adaptively adjusts analysis parameters and strategies by feeding back the results of subsequent data updates, enabling the system to continuously optimize analysis accuracy and decision reliability during continuous operation, thereby realizing a complete closed loop of multi-source geological information from perception, fusion, analysis to decision output.

[0160] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for multi-source geological information fusion analysis, characterized in that, The specific implementation steps include the following: S1. Systematically collect geological exploration data, remote sensing image data, geophysical measurement data, and historical geological archive data; perform adaptive noise correction, spatiotemporal coordinate unification, outlier and missing value processing on each data source, and extract key geological features to generate standardized multidimensional feature vectors. S2. The standardized multidimensional feature vectors are mapped to a unified three-dimensional geological grid. Combined with temporal dynamic coding and adaptive weight optimization based on data source reliability, spatial coverage and feature change magnitude, multi-source feature fusion is performed to generate a multi-scale fusion feature matrix. S3. Based on the multi-scale fusion feature matrix, state reconstruction and association mapping are performed by calculating the baseline state, deviation degree, and cumulative deviation intensity of each grid cell. Furthermore, the evolution rate and spatial expansion index of the anomalies are analyzed to achieve dynamic identification and trend prediction of geological anomalies; S4. Based on the abnormal evolution rate and spatial expansion index, combined with the abnormal level classification results, a multi-dimensional intelligent decision matrix is ​​constructed to generate priority ranking, intervention strategies and resource allocation schemes for different abnormal regions; and by comparing the deviation between the strategy execution effect and the expected effect, the decision parameters are adaptively optimized through feedback learning.

2. The multi-source geological information fusion analysis method according to claim 1, characterized in that, In step S1, adaptive noise correction is achieved by introducing a dynamic correction function based on the noise level of each data source. This function corrects and performs preliminary filtering on the original acquired values. Spatiotemporal coordinate unification includes using spatial coordinate transformation functions to transform the spatial coordinates of each source data to the same reference system, and mapping asynchronously acquired data to a standard time series defined by a unified time reference point.

3. The multi-source geological information fusion analysis method according to claim 2, characterized in that, In step S1, extracting key geological features specifically includes: Extract stratigraphic thickness, lithological classification, gravity gradient, magnetic anomaly, and texture features from remote sensing images; The extracted original values ​​of various features are normalized to form multidimensional feature vectors with uniform dimensions.

4. The multi-source geological information fusion analysis method according to claim 3, characterized in that, In step S2, the standardized multidimensional feature vectors are mapped to a unified three-dimensional geological grid, specifically as follows: For each grid cell, the inverse distance weight is calculated based on the Euclidean distance between all sampling points in its spatial neighborhood and the center of the cell. The weights are then used to sum the feature vectors of each sampling point in the neighborhood, resulting in the spatially mapped feature vector of the grid cell.

5. The multi-source geological information fusion analysis method according to claim 4, characterized in that, In step S2, the timing dynamic coding specifically involves: For each grid cell, the characteristic change between adjacent time points is calculated as the instantaneous change characteristic; Calculate the characteristic standard deviation within its recent time window as the volatility characteristic; Instantaneous change features and fluctuation features are combined with preset weight coefficients to form a time-series dynamic feature vector. This time-series dynamic feature vector is then combined with the spatially mapped feature vector to generate a three-dimensional time dynamic coding matrix.

6. The multi-source geological information fusion analysis method according to claim 5, characterized in that, In step S2, the adaptive weight optimization specifically includes: For each data source, a reliability weight is calculated based on its average noise level, and a spatial coverage weight is calculated based on the ratio of the number of its covered grid cells to the total number of grid cells in the study area. Then, by combining the characteristic change magnitude of the data source at the current moment, the final fusion weight of the data source at a specific grid cell and time node is obtained by comprehensively calculating the weighted adjustment parameters.

7. The multi-source geological information fusion analysis method according to claim 6, characterized in that, In step S3, the state reconstruction specifically involves: For each spatial grid cell, the mean value is calculated from the time interval with stable change amplitude selected from its historical fusion feature sequence, and the individualized baseline state vector of the cell is constructed. Calculate the overall offset between the fused feature vector of the unit at the current moment and the reference state vector, as the instantaneous deviation.

8. The multi-source geological information fusion analysis method according to claim 7, characterized in that, In step S3, the association mapping is specifically as follows: Within a preset time accumulation window, the instantaneous deviation is summed by time decay weighting to obtain the cumulative deviation intensity; The cumulative deviation intensity of each grid cell is compared with the mean and standard deviation of the cumulative deviation intensity of all cells in its region; When a grid cell exceeds a threshold determined by the mean, standard deviation, and anomaly detection sensitivity coefficient, it is marked as an anomaly candidate cell.

9. The multi-source geological information fusion analysis method according to claim 8, characterized in that, Step S3, analyzing the anomalous evolution rate and spatial expansion index specifically includes: The ratio of the cumulative deviation intensity difference of anomalous candidate units at adjacent time points to the time interval is calculated as the anomalous evolution rate. The number of grid cells in the spatial neighborhood of the candidate anomalous cell that are also marked as anomalous candidates is counted, and the anomalous spatial expansion index is calculated in combination with the anomalous confirmation threshold.

10. A multi-source geological information fusion analysis system, used to execute the multi-source geological information fusion analysis method according to any one of claims 1 to 9, characterized in that, include: Edge and cloud; The edge end is equipped with a multi-source geological sensing and standardization access module, which is used to collect, temporally and spatially align, quality check and standardize the multi-source heterogeneous geological data from drilling records, geophysical measurements, remote sensing images and in-situ sensing monitoring. The cloud platform is equipped with a geological feature fusion and anomaly characterization module, an anomaly evolution analysis and trend prediction module, and an intelligent decision generation and adaptive feedback module. The geological feature fusion and anomaly characterization module is connected to the multi-source geological sensing and standardization access module. It is used to perform spatiotemporal fusion modeling on the standardized data, generate a multi-scale fusion feature matrix, and identify and quantify geological anomaly units that deviate from the background features. The anomaly evolution analysis and trend prediction module is connected to the geological feature fusion and anomaly characterization module. It is used to predict the evolution direction, development rate and impact range of the geological anomaly unit based on the identified anomaly features and combined with time series and spatial neighborhood correlation information. The intelligent decision generation and adaptive feedback module is connected to the anomaly evolution analysis and trend inference module. It is used to generate analysis conclusions and decision suggestions by comprehensively considering the anomaly level and evolution trend, and to adaptively adjust the analysis parameters and decision strategies based on subsequent data feedback.

Images