Intelligent prospecting multi-source data fusion analysis method and system for pegmatite lithium ore

Abstract

The invention relates to the technical field of intelligent detection and sensors, in particular to a pegmatite lithium ore intelligent prospecting multi-source data fusion analysis method and system. The method comprises the following steps: on the basis of a mineralization mechanism of pegmatite lithium ore, dividing a continuous mineralization process into a plurality of stages with definite geological significance, abstracting the stages into structured mineralization rhythm units, and constructing a mineralization rhythm constraint model for representing an evolution sequence and a spatial dependency relationship between the units; performing standardization processing on the multi-source prospecting data, and performing coupling mapping and sensitivity weighting on the multi-source prospecting data according to a rhythm unit to form a comprehensive rhythm response matrix; a comprehensive rhythm consistency index is generated through fusion by calculating spatial consistency and multi-source attribute consistency indexes, so that a high-potential metallogenic area is identified and delineated; and introducing actual exploration verification information to carry out dynamic correction and optimization on a prediction result to form a closed-loop feedback prospecting decision support. According to the invention, mechanism-driven intelligent prediction from multi-source data to the prospecting target area is realized.

Description

Technical Field

[0001] This invention relates to the field of intelligent detection and sensor technology, specifically to a method and system for intelligent mineral exploration of pegmatite lithium deposits using multi-source data fusion analysis. Background Technology

[0002] Currently, mineral resource exploration has entered a new stage characterized by data-driven and intelligent approaches. Multi-source data acquisition technologies, such as remote sensing, geophysics, geochemistry, and geological surveys, are becoming increasingly mature, and the volume and types of data are becoming increasingly rich, providing a solid information foundation for a comprehensive understanding of metallogenic systems.

[0003] Chinese invention patent application CN121069525A discloses a mineral exploration method based on multi-source geological relationship data. The method includes: performing spatiotemporal alignment and standardization processing on acquired geological history evolution data, geophysical field data, geochemical migration data, and remote sensing alteration information to generate a dynamic knowledge graph; dividing the geological history evolution data into mineralization stages to obtain mineralization stage division results; integrating the spatiotemporal path network of fluid migration and performing three-dimensional dynamic modeling processing to generate a phased three-dimensional mineralization evolution model; extracting stage-specific geological marker combinations from the phased three-dimensional mineralization evolution model, performing knowledge reasoning processing based on a preset historical deposit rule base, and outputting mineral exploration target areas and mineralization process constraints that meet preset confidence conditions, thereby improving the accuracy of the structural model, generating a phased fluid dynamic field, and enhancing the prediction confidence of the target area.

[0004] Currently, the widespread adoption of IoT technology has driven the extensive application of smart sensors in field environments, enabling dynamic, in-situ monitoring of mineralization-related physicochemical parameters and further expanding data dimensions. The rise of big data analysis and mining technologies has provided powerful technical means for extracting hidden patterns and effective information from massive, heterogeneous exploration data. Against this backdrop, how to deeply integrate multi-source, static, and dynamic exploration data with the specific mineralization theories and spatial evolution laws of pegmatite lithium deposits to construct an intelligent prospecting model that can both reflect geological mechanism constraints and fully leverage data value has become a key research direction for improving the efficiency and accuracy of lithium resource exploration, and also an important frontier for promoting the transformation and upgrading of geological exploration towards intelligence and refinement. 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 data fusion analysis method and system for intelligent prospecting of lithium deposits in pegmatites.

[0006] The technical solution of this invention: a multi-source data fusion and analysis method for intelligent prospecting of lithium deposits in pegmatites, comprising the following specific implementation steps: S1. Based on the mineralization process of pegmatite lithium deposits, identify key mineralization stages and abstract them as mineralization rhythm units, and construct a mineralization rhythm constraint model that characterizes the evolutionary order and spatial dependence between units; S2. Standardize the remote sensing data, geophysical measurement data, geochemical sampling data, geological survey data, and IoT sensor network monitoring data streams. Then, couple and map the processed multi-source mineral exploration data according to the metallogenic rhythm units and perform sensitivity weighting to form a comprehensive rhythm response matrix. S3. Based on the comprehensive rhythm response matrix, spatial consistency index and multi-source attribute consistency index are calculated and weighted and fused to generate a comprehensive rhythm consistency index. High potential mineralization areas are identified and delineated based on the comprehensive rhythm consistency index. S4. Introduce actual exploration verification information to dynamically correct and optimize the comprehensive rhythm consistency index and the high-potential mineralized areas delineated accordingly, and feed the new round of exploration results back to the aforementioned steps to form a closed-loop feedback mineral exploration decision support.

[0007] Preferably, in step S1, the construction process of the ore-forming rhythm constraint model specifically includes: Each identified mineralization stage is abstracted into a mineralization rhythm unit, which is jointly represented by spatial response state, set of control attributes, and stage identifier. Based on the stage identifier, the consistency index of the metallogenic stage evolution between different metallogenic rhythm units is calculated using the rhythm sequence consistency function; Based on the spatial response state, the spatial dependence index between different mineralization rhythm units is calculated using the rhythm spatial dependence function. Based on the consistency index of mineralization stage evolution and the spatial dependence index, a mineralization rhythm evolution constraint function is constructed. The mineralization rhythm evolution constraint function is used to quantify the comprehensive constraint strength between any two mineralization rhythm units in terms of mineralization evolution logic and spatial distribution relationship.

[0008] Preferably, in step S2, the standardization process specifically includes: Unified spatial coordinate registration, dimension normalization, and noise filtering are performed on remote sensing image data, aerial or ground geophysical measurement data, geochemical sampling data, and geological survey data. Time series analysis and feature extraction are performed on the time series data streams monitored by the Internet of Things sensor network acquired by smart sensors deployed in the exploration area, and the extracted feature values ​​are processed into spatial raster data that matches the analysis period of the mineralization rhythm unit.

[0009] Preferably, in step S2, the specific process of sensitivity weighting is as follows: For each metallogenic rhythm unit and each type of data source, calculate the sensitivity index of that type of data source to that metallogenic rhythm unit; Based on the calculated sensitivity index, determine the weighting coefficient of this type of data source on the corresponding metallogenic rhythm unit; The weighting coefficients are used to calculate the weighted response values ​​of various data sources mapped to the ore-forming rhythm unit, thereby obtaining the weighted rhythm response.

[0010] Preferably, in step S3, the specific method for generating the comprehensive rhythm consistency index is as follows: For each mineralization rhythm unit at spatial location (x,y), calculate the spatial consistency index of the response value within its local neighborhood; For the response of each mineralization rhythm unit at spatial location (x,y), calculate the multi-source attribute consistency index of the response values ​​from different data sources; The spatial consistency index and the multi-source attribute consistency index are multiplied by preset spatial consistency weight coefficients and attribute consistency weight coefficients, respectively, and then added together to obtain the comprehensive rhythm consistency index at that location.

[0011] Preferably, in step S3, identifying and delineating high-potential mineralized areas is achieved through the following methods: A high-potential threshold is set for the comprehensive rhythm consistency index, which is determined based on the percentile of the response values ​​of historical mining site data; For each mineralized rhythmic unit, all spatial locations with a comprehensive rhythmic consistency index higher than the high-potential threshold within its spatial range are extracted to form a high-potential region for that single rhythmic unit. All high-potential areas of single rhythmic units corresponding to the metallogenic rhythmic units are spatially superimposed and merged to generate the final comprehensive predicted metallogenic area.

[0012] Preferably, in step S4, the specific process of dynamic correction includes: The actual exploration verification points are mapped to the corresponding metallogenic rhythm units, and the prediction reliability of the metallogenic rhythm unit is calculated based on the mineralization result label at the verification point and the comprehensive rhythm consistency index. Based on the comparison between the prediction confidence and the average confidence of all units, the rhythm consistency correction factor of the metallogenic rhythm unit is calculated. The original comprehensive rhythm consistency index obtained in step S3 is corrected using a rhythm consistency correction factor to obtain the corrected rhythm consistency index.

[0013] Preferably, step S4 further includes the following closed-loop optimization process: Based on the change range of the corrected rhythm consistency index and the original index, high-potential areas with changes less than the preset stability threshold are selected as stable final preferred mineral exploration prediction areas. Based on the final optimized prospecting prediction area, an executable prospecting decision is generated, including drilling priority zoning and trench layout suggestions. The newly acquired exploration verification results after the mineral exploration decision are executed are added to the actual exploration verification information set, thereby triggering an iterative optimization process from the model parameter fine-tuning in step S1 to the re-correction in step S4.

[0014] Preferably, the IoT sensor network monitoring data in step S2 specifically includes time-series data collected by one or more of the following devices: high-precision gas sensors, soil temperature and humidity sensors, microseismic monitoring nodes, and groundwater chemical in-situ monitoring instruments.

[0015] The technical solution of this invention: A multi-source data fusion and analysis system for intelligent prospecting of lithium deposits in pegmatites, which is used to execute the above-mentioned multi-source data fusion and analysis method for intelligent prospecting of lithium deposits in pegmatites, comprising: The multi-source mineral exploration data collaborative acquisition and standardization module is used to collaboratively collect heterogeneous data from remote sensing, geophysics, geochemistry, geological surveys and IoT sensor networks, and to perform standardization and quality constraint processing under a unified spatiotemporal benchmark. The ore-forming indicator information structured expression and feature evolution module is used to construct ore-forming rhythm units based on the genetic characteristics of pegmatite lithium deposits, and to establish an evolution and spatial dependence constraint model between units; The multi-scale metallogenic response correlation and fusion module is used to map standardized multi-source data to metallogenic rhythm units and perform sensitivity-weighted fusion to generate a comprehensive rhythm response matrix, and then calculate the comprehensive rhythm consistency index and delineate high-potential metallogenic areas. The intelligent reasoning and verification feedback module for mineral exploration target areas is used to introduce actual exploration verification information to dynamically correct and optimize the prediction results, and drive the iterative update of the system model based on verification feedback to form closed-loop decision support.

[0016] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: This invention designs a method and system for intelligent multi-source data fusion analysis in pegmatite lithium exploration. First, by constructing a mechanism-driven mineralization rhythm constraint model, the stage-based and spatially inherited geological mineralization process is transformed into a computable and constrainable quantitative framework. This ensures that the fusion analysis of multi-source exploration data is always controlled by the true mineralization logic, effectively overcoming the problem of mechanism disconnect caused by data piling up, and significantly improving the geological reliability and interpretability of mineralization predictions. Second, an IoT monitoring network containing various intelligent sensors is introduced to achieve dynamic, in-situ acquisition of mineralization-related physicochemical parameters. Combined with static data from remote sensing, geophysics, etc., an integrated air-space-ground multi-source information system is formed, providing multi-dimensional data support for identifying deep and concealed ore bodies. Building upon this foundation, and leveraging big data analytics and mining technologies, the system employs rhythmic response mapping, sensitivity weighting, and consistency fusion on massive heterogeneous data to accurately extract weak anomalies and composite mineral exploration indicators, thereby improving the reliability of information identification. Furthermore, the system utilizes a cloud-edge collaborative computing architecture, rationally distributing real-time data processing and model optimization tasks between the edge and the cloud, which not only enhances response speed but also achieves efficient resource utilization. By establishing a closed-loop optimization process from prediction and verification to feedback correction, the mineral exploration model can continuously improve itself as exploration progresses, dynamically enhancing prediction accuracy and the adaptability of mineral exploration decisions. This invention achieves a deep integration of mechanisms, data, and decision-making, driving lithium exploration towards intelligent and refined development. Attached Figure Description

[0017] Figure 1 This is a flowchart of a multi-source data fusion analysis method for intelligent prospecting of lithium deposits in pegmatite proposed in this invention; Figure 2 This is a system architecture diagram of a multi-source data fusion analysis system for intelligent prospecting of lithium deposits in pegmatite proposed in this invention. Detailed Implementation

[0018] Example 1, as Figure 1 As shown, the present invention proposes a multi-source data fusion analysis method for intelligent prospecting of lithium deposits in pegmatites, which includes the following specific implementation steps: S1. Based on the staged and rhythmic characteristics of the lithium mineralization process in pegmatite deposits, this study first identifies mineralization stages with stable directional significance from the actual mineralization mechanism. Then, each mineralization stage is formalized into computable and constrained mineralization rhythmic units. Furthermore, the evolutionary sequence and spatial dependency constraints between rhythmic units are established, ultimately forming a unified mineralization rhythmic constraint model. This model provides a basic framework constrained by the mineralization mechanism for subsequent rhythmic response mapping and fusion analysis of multi-source mineral exploration data. The specific implementation process is as follows: S11. Through comprehensive analysis of known pegmatite lithium deposit mineralization examples and regional geological data, this study systematically reviews mineralization manifestations such as late-stage magmatic differentiation, pegmatite vein distribution, tectonic control of ore deposits, and lithium enrichment. It extracts key mineralization stages that repeatedly occur in different mining areas and can be stably responded to by multi-source exploration data. Therefore, the continuous mineralization process is divided into several mineralization rhythmic stages with clear geological significance, specifically: Given that pegmatite lithium deposits are characterized by clear magmatic evolution stages, well-controlled spatial distribution, and gradual element enrichment, this study systematically reviews and compares the geological phenomena related to lithium mineralization in the study area, based on regional-scale geological survey results, mineralization examples of discovered pegmatite lithium deposits, and exploration experience accumulated during the exploration process. The focus is on identifying mineralization patterns that repeatedly appear in different mining areas and can be stably responded to by multi-source exploration data; For example, late-stage magmatic differentiation indicators, spatial occurrence patterns of lithium-rich pegmatite veins, tectonic assemblages controlling vein emplacement, and evolutionary characteristics of lithium and associated rare metal elements from background to anomalous. Based on this, the mineral exploration process is divided into several phased units with relatively independent geological significance and which may have superimposed or inherited relationships in space, thereby forming a rhythmic understanding of the mineralization process of pegmatite lithium deposits and providing a clear basis for phase division for subsequent model construction. S12. Based on the identification of mineralization stages, each mineralization stage is abstractly defined as a mineralization rhythm unit. This rhythm unit is then structurally expressed through spatial response states, mineralization control attributes, and stage sequence identifiers. This transforms the mineralization stages, which were originally described based on experience, into model objects with clear state parameters that can participate in subsequent calculations and constraint analyses. Specifically: After identifying the mineralization stages, each mineralization stage is further abstracted into a mineralization rhythm unit and structurally defined so that it can serve as the basic analytical object for subsequent multi-source data fusion and inference. Define any mineralization rhythm unit Its state is characterized by spatial response, control attributes, and stage sequence, and its expression is defined as: ; in, This represents the i-th mineralization rhythm unit, used to indicate a stage node in the mineralization process of pegmatite lithium deposits. It is extracted from actual geological exploration data, known mineralization mechanisms of pegmatite lithium deposits, and regional geological patterns. The spatial response state of the i-th metallogenic rhythm unit is represented by the analysis of actual spatial data such as the distribution of magma veins, rare metal anomaly zones, and structural ore-controlling units. This represents the set of control attributes for the i-th metallogenic rhythm unit, extracted based on magmatic evolution chemistry, structural geology, and geochemical element enrichment mechanisms. This represents the stage identifier of the i-th metallogenic rhythm unit in the metallogenic evolution sequence, which is determined based on the metallogenic evolution law and the experience of the spatiotemporal evolution of ore deposits; S13. To address the sequential evolution and spatial inheritance relationships among different mineralization rhythm units, an evolutionary and spatial dependency constraint model is constructed between rhythm units. Rhythmic combinations that do not conform to the mineralization logic of pegmatite lithium deposits are restricted and weakened. This ensures, at the model level, that the combination results of multi-stage mineralization information conform to the actual mineralization process. Specifically: Based on the general evolutionary sequence of late-stage magmatic differentiation, pegmatite vein emplacement, and lithium enrichment during the mineralization of pegmatite lithium deposits, stage markers defined in rhythmic units are used. and A rhythmic sequence consistency function is constructed to obtain the consistency index of the mineralization stage evolution. Used to describe mineralization rhythm units and The rationality of the sequence of mineralization evolution: ; Introducing the spatial dependency function, we obtain the spatial dependency. Units for characterizing mineralization rhythm and The spatial relationships reflect whether there is a continuity, superposition, or control relationship between different mineralization stages in space: ; For any two mineralization rhythm units and The correlation between mineralization stages and spatial dependence is characterized, and a mineralization rhythm evolution constraint function is constructed, which is defined as: ; in, Representing mineralization rhythm units and The comprehensive constraint strength between them in terms of metallogenic evolution logic and spatial distribution relationship is used to measure whether the two can coexist reasonably in the same metallogenic process; The weighting coefficient of the consistency of the evolution of mineralization rhythm in the comprehensive constraint function is used to adjust the degree of influence of the rationality of the mineralization stage sequence on the credibility of the rhythm combination. The weighting coefficient of the spatial dependence of mineralization rhythm in the comprehensive constraint function is used to adjust the degree of influence of spatial inheritance or adjacency relationship on the reliability of rhythm combination; This represents a unit describing the ore-forming rhythm. and The degree of consistency in the mineralization evolution sequence reflects whether the two conform to the stage evolution logic of the pegmatite lithium mineralization process; Representing mineralization rhythm units and The degree of dependence in spatial distribution is used to measure whether there is a relationship of inheritance, adjacency or control between the two in space; Represents a function for rhythmic order consistency; This represents the rhythmic spatial dependency function, used to calculate the strength of the dependency relationship between the spatial response states of different mineralization rhythmic units; It should be noted that the rhythm sequence consistency function This function is used to characterize the degree of reasonable matching of different metallogenic rhythm units in the metallogenic evolution sequence. Its core idea is that as the sequence of metallogenic rhythm stages transitions from unreasonable to reasonable, the corresponding constraint strength continuously increases without reverse fluctuation. The function takes the relative stage relationship of metallogenic rhythm units as input, and strengthens the rhythm connection that conforms to the metallogenic evolution law of pegmatite lithium deposits by suppressing unreasonable combinations caused by stage inversion or jump. This allows the model to prioritize the retention of rhythmic responses with continuous evolution significance during the fusion analysis process, thereby reducing abnormal interferences that are inconsistent with the actual metallogenic process at the source. It should be noted that the rhythm spatial dependency function This function is used to characterize the spatial distribution dependency of different metallogenic rhythm units. Its core idea is to start from the spatial inheritance and control phenomena commonly seen in mineral exploration practice, and comprehensively evaluate the degree of correlation of different rhythm units in terms of spatial location, distribution direction and continuity characteristics. By analyzing the adjacency relationship, overlap characteristics and the continuation trend along the structure or vein direction between rhythm response areas, this function can determine whether subsequent metallogenic rhythms are controlled or inherited by previous rhythms, thereby reflecting the spatial coherence and rationality of the metallogenic process, and providing a basis for excluding spatially isolated anomalies that lack metallogenic directional significance. S14. All metallogenic rhythmic units and their inter-constraint relationships within the study area are integrated to form a complete metallogenic rhythmic constraint model for pegmatite lithium deposits. This model serves as the foundational input framework for subsequent multi-source mineral exploration data rhythmic response mapping and fusion analysis, ensuring that subsequent analysis results are always constrained by the overall metallogenic mechanism and evolutionary laws. Specifically: After defining the metallogenic rhythm units and modeling their mutual constraints, all metallogenic rhythm units identified in the study area and their corresponding rhythm constraint functions are integrated to construct a metallogenic rhythm constraint model for pegmatite lithium deposits. Its overall form is expressed as follows: ; in, This represents a complete constrained model of the mineralization rhythm of pegmatite lithium deposits.

[0019] S2. For the mineralization rhythm constraint model generated in step S1, a rhythm response parsing and mapping scheme for multi-source mineral exploration data is designed. A cloud-edge collaborative computing architecture is adopted, aiming to rationally allocate tasks such as data preprocessing, real-time response, and model optimization between the edge and the cloud, thereby optimizing the utilization of computing resources and achieving near real-time mineral exploration response. Through spatial registration, normalization preprocessing, rhythm unit mapping, sensitivity weighting, and comprehensive response matrix construction, the data is made to conform to the logic of mineralization rhythm, providing a reliable and interpretable multi-source data foundation for subsequent rhythm consistency fusion and mineral exploration prediction. The specific implementation process is as follows: S21. Perform unified spatial coordinate registration, dimension normalization, and noise filtering on multi-source data from remote sensing, geophysics, geochemistry, geological surveys, and IoT sensor network monitoring. The IoT data primarily includes time-series data streams acquired by various intelligent sensors deployed in the exploration area (such as high-precision gas sensors (measuring Rn, CO2, etc.), soil temperature and humidity sensors, microseismic monitoring nodes, and in-situ groundwater chemical monitoring instruments). For this type of dynamic data, perform time-series analysis and feature extraction (such as calculating daily averages, trend terms, and identifying abnormal pulses), and interpolate or aggregate its feature values ​​into spatial raster data matching the analysis period of the rhythm unit. This ensures the spatial and numerical comparability of all data, providing reliable and standardized input for rhythm mapping. Specifically: For the multi-source mineral exploration data acquired in the study area, including remote sensing imagery, airborne / ground geophysical survey data, geochemical sampling data, existing geological survey data, and time-series data monitored by the Internet of Things sensor network, each type of data was standardized, including but not limited to unified spatial coordinate registration, dimensional normalization, and noise filtering, to obtain standardized data. ; in, This represents the result of preprocessing the k-th class of data (spatial registration, normalization, noise filtering); S22. Using the mineralization rhythm units defined in step S1, the preprocessed multi-source data is mapped to the spatial region corresponding to each rhythm unit, and the data response is adjusted in conjunction with the rhythm control attributes to form a preliminary spatial response matrix for each rhythm unit, thereby realizing the coupling mapping between data and mineralization logic. Specifically: For each rhythm unit Determine its spatial coverage area ; Calculate the data response function for each raster cell (x, y): ; in, This indicates that the k-th type of data is in the rhythm unit. The initial response value corresponding to the spatial unit (x,y); Represents the rhythm matching function; Representing rhythmic units The spatial coverage state of the corresponding spatial unit (x,y); It should be noted that the rhythm matching function This is used to couple multi-source data with metallogenic rhythm units. It determines whether the grid containing the data is within the range of the rhythm unit by the spatial coverage status. At the same time, it adjusts the validity of the data in the rhythm unit by combining control attributes, so that the response value is only effectively output in the area that meets the spatial distribution and geological conditions, weakening the signals that are not related to the rhythm. This ensures that the data mapping conforms to the spatial law of metallogenesis and reflects the control characteristics of the metallogenic process, so as to achieve the physical and mechanistic consistency mapping of multi-source data to the rhythm unit. S23. To address the differences in sensitivity of different data sources to various rhythm units, calculate rhythm sensitivity indices and generate weights. Adjust the response intensity of each data type to the rhythm unit using weighted averages to ensure that high-contribution data dominates, reduce interference from low-sensitivity or noisy data, and improve the reliability and physical rationality of the rhythm response. Specifically: Calculate the sensitivity of each rhythm unit to each data source. : ; Calculate weights based on sensitivity: ; Received weighted response: ; in, Represents the k-th type of data pair for rhythm units Sensitivity index; Indicates the rhythm control attribute correction factor; This represents the spatial coverage of the k-th type of data in the rhythm unit. The variance of the internal response values; This indicates that the k-th type of data is in the rhythm unit. The weighting coefficient; K represents the total number of data sources (such as remote sensing, magnetism, geochemistry, geology, etc.); This represents the weighted rhythmic response; S24. The weighted multi-source data is superimposed on each rhythm unit to generate a comprehensive rhythm response matrix, and then standardized to form a unified spatial response map that can be used for rhythm consistency analysis. This achieves effective fusion of multi-source data at the rhythm unit dimension and visualization of preliminary mineral exploration information. Specifically: Overlay all data sources onto the rhythm unit: ; Generate rhythm response matrix The spatial distribution map is used to initially determine the spatial significance of the rhythm; And the rhythm response matrix The standardized comprehensive response matrix is ​​obtained by performing standardization. To ensure comparability of responses from different rhythms; in, Representing rhythmic units The comprehensive response matrix.

[0020] S3. Based on the rhythmic response matrix generated in step S2, a comprehensive rhythmic consistency index is constructed through spatial consistency and multi-source attribute consistency analysis to identify high-potential mineralized areas and achieve multi-source data fusion prediction. The specific implementation process is as follows: S31. Based on the rhythmic response matrix generated in step S2, a comprehensive rhythmic consistency index is constructed through spatial consistency and multi-source attribute consistency analysis to identify high-potential mineralized areas and achieve multi-source data fusion prediction, specifically: For each grid cell (x,y), a neighborhood window N(x,y) is defined, which can be a 3×3 or 5×5 grid to ensure local spatial continuity capture; Calculate local spatial consistency: ; Where N(x,y) represents the set of neighborhood windows centered at (x,y); Indicates the number of raster cells contained in the neighborhood window; Representing rhythmic units Spatial consistency index at location (x,y); S32. Calculate the weighted multi-source response variance at each grid point and convert it into an attribute consistency index through normalization. This quantifies the response consistency of different data sources at that point, thereby filtering out anomalous single-source signals and realizing mechanism-driven fusion of multi-source data within the rhythm unit, ensuring prediction reliability. Specifically: Calculate attribute variance: ; ; Standardize the attribute consistency index: ; in, Representing rhythmic units The average value of the multi-source response at location (x,y); Representing rhythmic units Variance of multi-source response at location (x,y); Representing rhythmic units Maximum multi-source response variance within the spatial range; Representing rhythmic units Indicators of attribute consistency at location (x,y); S33. Spatial consistency indicators and attribute consistency indicators are integrated according to weights to form a comprehensive rhythmic consistency index. This index is used to quantitatively assess the mineralization potential of each grid point, realizing a prediction logic that combines spatial and attribute dimensions. This provides a direct basis for the subsequent extraction of high-potential areas. Specifically: Calculate the weighted fusion: ; And standardize it: ; in, The spatial consistency weighting coefficient is set based on experience regarding the spatial continuity of ore-forming bodies. This represents the attribute consistency weighting coefficient, which is set based on experience with the reliability of multi-source data. Representing rhythmic units The overall rhythm consistency index at position (x,y); This represents the standardized comprehensive rhythm consistency index; and Representing rhythm units The minimum and maximum values ​​of the overall consistency index within the range; S34. By setting a threshold, high-consistency regions of single rhythmic units are extracted, and all rhythmic units are superimposed and fused to generate a final high-potential mineralization prediction area map of multi-source data, which can be used for field exploration and drilling site selection. Specifically: Set a high potential threshold (Historical mining point percentiles (such as the top 20% response values) can be used); Extracting the high-response region of a single-rhythm unit: ; And perform multi-rhythm unit superposition and fusion: ; in, Representing rhythmic units A set of high-potential mineralization spatial points; This represents the comprehensive predicted mineralization region after the superposition of multiple rhythmic units; n represents the total number of mineralization rhythmic units.

[0021] S4. Utilizing the multi-rhythm fusion prediction results generated in step S3, and with the gradual introduction of limited verification information (trenching, drilling, engineering exposure, etc.), the prediction results are dynamically corrected, and a decision optimization output that can be used for the next round of mineral exploration deployment is generated. The specific implementation process is as follows: S41. By mapping actual exploration and verification information to the rhythm consistency space, a rhythm reliability description of the prediction results is constructed, providing a quantitative basis for subsequent correction, specifically: Comprehensive prediction of mineralized regions after superposition of multiple rhythmic units Internally, a phased set of exploration and verification points is introduced. : ; At each verification point, the overall consistency index of the corresponding rhythm unit is extracted. : ; Constructing rhythmic units Credibility function: ; in, Indicates the spatial location of the j-th verification point; The label represents the verification result; a value of 1 indicates the presence of minerals or significant mineralization, while a value of 0 indicates the absence of minerals or weak mineralization; m represents the number of spatial locations of the verification points. Indicates falling into a rhythm unit The number of verification points within; This indicates the reliability of the rhythm unit, used to quantify the degree of consistency between prediction and reality; S42. Based on the rhythm unit reliability, the rhythm consistency index obtained in step S3 is adaptively corrected so that the prediction results gradually converge with the verification information, specifically as follows: Constructing rhythm correction factors: ; Correcting the rhythm consistency index: ; The corrected exponent is normalized to ensure consistent scaling. in, This represents the mean confidence level of all rhythm units. This represents the adjustment coefficient, used to control the correction amplitude; This represents the rhythm consistency correction factor, reflecting the relative enhancement or weakening of the predictive reliability of the rhythm unit; This represents the corrected rhythm consistency index; S43. By comparing the changes in the rhythm consistency index before and after correction, stable and reliable prediction areas are screened to improve the reliability of mineral exploration decisions. Specifically: Construct stability metrics: ; Extracting the high-potential threshold set in step S3 Define the stable prediction region: ; All rhythm units are superimposed to form the preferred prediction region: ; in, This indicates the magnitude of change in the consistency index, used to measure the stability of predictions. This represents the stability threshold, which limits the allowable range of correction variations. This indicates a stable, high-potential region within the rhythmic unit; This indicates the final preferred mineral exploration prediction area; S44. Transform the preferred prediction area into an executable mineral exploration decision, and feed the new round of exploration results back to steps S1-S3 to form a closed-loop optimization, specifically as follows: according to Output drilling priority zones, trench layout suggestions, and verification sequence; The newly acquired exploration results will be added to the interim exploration verification point set. ; Trigger the fine-tuning of the rhythm unit constraint parameters in step S1, and update steps S2-S4 in sequence; It should be noted that during this process, edge nodes can receive on-site verification data in real time (such as trench exploration results recorded by handheld devices) and immediately make initial local consistency judgments and warnings; at the same time, these verification data are synchronized to the cloud, triggering global fine-tuning of the rhythm constraint model and fusion parameters; the updated model parameters can be sent to edge nodes as appropriate, thereby completing the closed-loop optimization process of cloud training-edge inference-on-site feedback.

[0022] Example 2, as Figure 2 As shown, the present invention proposes a multi-source data fusion analysis system for intelligent prospecting of lithium deposits in pegmatites. This system is used to execute a multi-source data fusion analysis method for intelligent prospecting of lithium deposits in pegmatites proposed in Example 1. The system includes: a multi-source prospecting data collaborative acquisition and standardization module, a mineralization indicator information structured expression and feature evolution module, a multi-scale mineralization response correlation fusion module, and a prospecting target area intelligent reasoning and verification feedback module.

[0023] The multi-source mineral exploration data collaborative acquisition and standardization module is used to collaboratively acquire and process quality constraints on remote sensing images, geochemical sampling data, geophysical measurement data, existing geological survey results, and real-time monitoring data streams from IoT sensor networks under a unified spatiotemporal benchmark. It automatically eliminates the differences in resolution, sampling density, and measurement accuracy among different data sources, providing an aligned and comparable data foundation for subsequent analysis. The ore-forming indicator information structured expression and feature evolution module is used to transform the extracted lithological assemblage, tectonic distribution, elemental anomalies and physical property responses into ore-forming indicator units with associated semantics based on the genetic characteristics of pegmatite lithium deposits. It also adaptively characterizes the spatial evolution features of these units by combining regional scale changes, thus realizing the transition from raw observation data to ore-forming expression features. The multi-scale mineralization response correlation and fusion module is used to establish the intrinsic relationship between mineralization responses across different spatial scales and data types. Through the collaborative analysis of local anomalies and regional background, it enhances the comprehensive identification capability of mineralization control factors in pegmatite lithium deposits and avoids mineral exploration bias caused by single data dominance. The intelligent reasoning and verification feedback module for mineral exploration target areas is used to form potential mineral exploration target area identification results based on the aforementioned fusion results, and to dynamically correct them by combining existing mineral deposit examples and newly acquired verification data. The verification results are then fed back in reverse to achieve the overall self-correction and continuous optimization of the system.

[0024] 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 intelligent mineral exploration of lithium deposits in pegmatite using multi-source data fusion and analysis, characterized in that, The specific implementation steps include the following: S1. Based on the mineralization process of pegmatite lithium deposits, identify key mineralization stages and abstract them as mineralization rhythm units, and construct a mineralization rhythm constraint model that characterizes the evolutionary order and spatial dependence between units; S2. Standardize the remote sensing data, geophysical measurement data, geochemical sampling data, geological survey data, and IoT sensor network monitoring data streams. Then, couple and map the processed multi-source mineral exploration data according to the metallogenic rhythm units and perform sensitivity weighting to form a comprehensive rhythm response matrix. S3. Based on the comprehensive rhythm response matrix, spatial consistency index and multi-source attribute consistency index are calculated and weighted and fused to generate a comprehensive rhythm consistency index. High potential mineralization areas are identified and delineated based on the comprehensive rhythm consistency index. S4. Introduce actual exploration verification information to dynamically correct and optimize the comprehensive rhythm consistency index and the high-potential mineralized areas delineated accordingly, and feed the new round of exploration results back to the aforementioned steps to form a closed-loop feedback mineral exploration decision support.

2. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 1, characterized in that, In step S1, the construction process of the ore-forming rhythm constraint model specifically includes: Each identified mineralization stage is abstracted into a mineralization rhythm unit, which is jointly represented by spatial response state, set of control attributes, and stage identifier. Based on the stage identifier, the consistency index of the metallogenic stage evolution between different metallogenic rhythm units is calculated using the rhythm sequence consistency function; Based on the spatial response state, the spatial dependence index between different mineralization rhythm units is calculated using the rhythm spatial dependence function. Based on the consistency index of mineralization stage evolution and the spatial dependence index, a mineralization rhythm evolution constraint function is constructed. The mineralization rhythm evolution constraint function is used to quantify the comprehensive constraint strength between any two mineralization rhythm units in terms of mineralization evolution logic and spatial distribution relationship.

3. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 2, characterized in that, In step S2, the standardization process specifically includes: Unified spatial coordinate registration, dimension normalization, and noise filtering are performed on remote sensing image data, aerial or ground geophysical measurement data, geochemical sampling data, and geological survey data. Time series analysis and feature extraction are performed on the time series data streams monitored by the Internet of Things sensor network acquired by smart sensors deployed in the exploration area, and the extracted feature values ​​are processed into spatial raster data that matches the analysis period of the mineralization rhythm unit.

4. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 3, characterized in that, In step S2, the specific process of sensitivity weighting is as follows: For each metallogenic rhythm unit and each type of data source, calculate the sensitivity index of that type of data source to that metallogenic rhythm unit; Based on the calculated sensitivity index, determine the weighting coefficient of this type of data source on the corresponding metallogenic rhythm unit; The weighting coefficients are used to calculate the weighted response values ​​of various data sources mapped to the ore-forming rhythm unit, thereby obtaining the weighted rhythm response.

5. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 4, characterized in that, In step S3, the specific method for generating the comprehensive rhythm consistency index is as follows: For each mineralization rhythm unit at spatial location (x,y), calculate the spatial consistency index of the response value within its local neighborhood; For the response of each mineralization rhythm unit at spatial location (x,y), calculate the multi-source attribute consistency index of the response values ​​from different data sources; The spatial consistency index and the multi-source attribute consistency index are multiplied by preset spatial consistency weight coefficients and attribute consistency weight coefficients, respectively, and then added together to obtain the comprehensive rhythm consistency index at that location.

6. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 5, characterized in that, In step S3, high-potential mineralization areas are identified and delineated, specifically through the following methods: A high-potential threshold is set for the comprehensive rhythm consistency index, which is determined based on the percentile of the response values ​​of historical mining site data; For each mineralized rhythmic unit, all spatial locations with a comprehensive rhythmic consistency index higher than the high-potential threshold within its spatial range are extracted to form a high-potential region for that single rhythmic unit. All high-potential areas of single rhythmic units corresponding to the metallogenic rhythmic units are spatially superimposed and merged to generate the final comprehensive predicted metallogenic area.

7. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 6, characterized in that, In step S4, the specific process of dynamic correction includes: The actual exploration verification points are mapped to the corresponding metallogenic rhythm units, and the prediction reliability of the metallogenic rhythm unit is calculated based on the mineralization result label at the verification point and the comprehensive rhythm consistency index. Based on the comparison between the prediction confidence and the average confidence of all units, the rhythm consistency correction factor of the metallogenic rhythm unit is calculated. The original comprehensive rhythm consistency index obtained in step S3 is corrected using a rhythm consistency correction factor to obtain the corrected rhythm consistency index.

8. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 7, characterized in that, Step S4 also includes the following closed-loop optimization process: Based on the change range of the corrected rhythm consistency index and the original index, high-potential areas with changes less than the preset stability threshold are selected as stable final preferred mineral exploration prediction areas. Based on the final optimized prospecting prediction area, an executable prospecting decision is generated, including drilling priority zoning and trench layout suggestions. The newly acquired exploration verification results after the mineral exploration decision are executed are added to the actual exploration verification information set, thereby triggering an iterative optimization process from the model parameter fine-tuning in step S1 to the re-correction in step S4.

9. The method for intelligent multi-source data fusion analysis of pegmatite lithium deposits according to claim 3, characterized in that, The IoT sensor network monitoring data in step S2 specifically includes time-series data collected by one or more of the following devices: high-precision gas sensors, soil temperature and humidity sensors, microseismic monitoring nodes, and groundwater chemical in-situ monitoring instruments.

10. A multi-source data fusion and analysis system for intelligent prospecting of lithium deposits in pegmatite, used to execute the multi-source data fusion and analysis method for intelligent prospecting of lithium deposits in pegmatite as described in any one of claims 1 to 9, characterized in that, include: The multi-source mineral exploration data collaborative acquisition and standardization module is used to collaboratively collect heterogeneous data from remote sensing, geophysics, geochemistry, geological surveys and IoT sensor networks, and to perform standardization and quality constraint processing under a unified spatiotemporal benchmark. The ore-forming indicator information structured expression and feature evolution module is used to construct ore-forming rhythm units based on the genetic characteristics of pegmatite lithium deposits, and to establish an evolution and spatial dependence constraint model between units; The multi-scale metallogenic response correlation and fusion module is used to map standardized multi-source data to metallogenic rhythm units and perform sensitivity-weighted fusion to generate a comprehensive rhythm response matrix, and then calculate the comprehensive rhythm consistency index and delineate high-potential metallogenic areas. The intelligent reasoning and verification feedback module for mineral exploration target areas is used to introduce actual exploration verification information to dynamically correct and optimize the prediction results, and drive the iterative update of the system model based on verification feedback to form closed-loop decision support.

Images