Mobile geological survey data real-time acquisition and analysis system for complex terrain

Abstract

The present invention discloses a mobile real-time geological survey data acquisition and analysis system for complex terrain, which relates to the field of geological survey technology. The acquisition and analysis system includes: a mobile acquisition terminal, an integrated multi-platform acquisition device and sensor that adapts to complex terrain, a historical data template library, which stores multi-dimensional historical geological data that has undergone standardization processing. The standardization processing includes data cleaning, feature extraction, and time-space coordinate calibration, and supports template retrieval and reuse based on feature similarity. A historical data reuse module realizes the reuse of historical templates, the semantic association of multi-source data, and the real-time adjustment of new data of the basic template through template matching, data fusion, and incremental updates. This system reuses the overlapping parts of historical geological survey data, thereby reducing the redundant analysis of newly collected geological data. It only needs to match the reused data model and update the model in combination with the new data to complete more efficient geological survey data acquisition and analysis.

Description

Technical Field

[0001] The present invention relates to the field of geological exploration technology, and in particular to a mobile geological exploration data real-time acquisition and analysis system for complex terrain. Background Art

[0002] Geological survey is a comprehensive technical activity that reveals the geological structure, material composition, hydrological conditions and other laws of the earth's surface and interior through systematic investigation, exploration and analysis, and provides a scientific basis for engineering construction, resource development, environmental protection, etc. Its core purpose is to identify the geological characteristics of the target area and evaluate the impact of geological conditions on human activities. The main contents include: investigating topography and surface material distribution, analyzing the physical and mechanical properties of strata and lithology, studying the distribution and activity of geological structures such as faults and folds, investigating the burial depth, water quality and movement of groundwater, identifying landslides and other geological phenomena. The potential risks of adverse geological phenomena such as slope failure and debris flow are determined by the use of technical means including remote sensing monitoring, geophysical exploration (such as seismic waves and electromagnetic methods), drilling sampling, indoor experiments (rock mechanics, water quality analysis, etc.) and on-site in-situ testing. By combining multiple methods to build a complete geological model, the application field is wide, from the site selection of projects such as house construction, bridges and tunnels, to the exploration of minerals, oil and gas resources, to the prevention and control of geological disasters, rational development of groundwater and ecological restoration, all of which need to be based on geological surveys, so as to achieve scientific decision-making, reduce engineering risks and ensure the sustainable use of resources.

[0003] At present, during geological surveys, various types of geological data of the target area are generally collected through equipment first, and then the data is transmitted to the analysis center. The data and collected samples are further analyzed and a visual model is generated. The collection and analysis process of geological surveys will generate a large amount of geological survey data, which is then transmitted back to the analysis center. The data is then analyzed and modeled to generate a visual geological model. However, the amount of data in this process is huge, so the analysis of geological data takes a long time and consumes a lot of resources. Summary of the Invention

[0004] In view of the shortcomings of the existing technology, the present invention provides a mobile real-time acquisition and analysis system for geological survey data in complex terrain, which solves the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a mobile real-time geological survey data acquisition and analysis system for complex terrain, the acquisition and analysis system comprising:

[0006] Mobile acquisition terminal: Integrates multi-platform acquisition equipment and sensors that adapt to complex terrain, and is used for real-time acquisition of multi-source heterogeneous geological data;

[0007] A historical data template library stores standardized multi-dimensional historical geological data, including data cleaning, feature extraction, and time-space coordinate calibration, and supports template retrieval and reuse based on feature similarity.

[0008] The historical data reuse module realizes the reuse of historical templates, semantic association of multi-source data, and real-time adjustment of new data of basic templates through template matching, data fusion, and incremental update;

[0009] The federated learning module is used for distributed joint modeling across institutions. It encrypts and aggregates the sub-models trained locally by each participant through a global aggregation node to generate a global analysis model.

[0010] An edge computing 5G collaboration module is deployed on the mobile acquisition terminal to implement data preprocessing and lightweight model reasoning, and to interact with the cloud server via a 5G communication link.

[0011] The environmental perception and reinforcement module integrates multiple types of environmental sensors and reinforcement learning algorithms to monitor environmental parameters and device status in real time and dynamically adjust collection strategies;

[0012] The spatial information grid collaboration module virtualizes distributed geological data resources into a grid pool and realizes cross-node parallel computing through dynamic resource scheduling.

[0013] Preferably, the mobile acquisition terminal includes: a drone platform, a crawler vehicle-mounted platform and a portable sensor group;

[0014] The UAV platform is equipped with a multispectral camera, a laser radar and a barometric altimeter. The portable sensor group includes a cesium optical pump magnetometer, an X-ray fluorescence spectrometer and a fiber Bragg grating stress sensor. All devices are synchronously controlled through an industrial bus.

[0015] Preferably, the standardized processing dimensions of the historical data template library include: lithologic texture characteristics, geophysical field parameters, stratigraphic age data and mineralization anomaly indicators of geological exploration;

[0016] The template retrieval is performed through dual retrieval of spatiotemporal index and feature vector index, wherein the feature vector index is constructed based on lithologic spectral features and structural strike features.

[0017] Preferably, the historical data reuse module includes: a template matching submodule, a data fusion submodule and an incremental update submodule;

[0018] The template matching submodule uses a feature point matching algorithm to calculate the similarity between the newly collected geological data and the historical geological data template, and quickly identifies and marks the overlapping areas;

[0019] The data fusion submodule: aligns the spatiotemporal features of multi-source heterogeneous data, and then combines and splices historical geological data templates to construct a basic three-dimensional geological model;

[0020] The incremental update submodule: after the historical geological data templates are combined and spliced ​​to construct the basic three-dimensional geological model, for the new geological data collected but not matched, based on the constructed basic three-dimensional geological model, new data is added to the model to generate analysis results of the geological collection data.

[0021] Preferably, the incremental update submodule adopts a meta-learning framework to fine-tune the parameters of the basic template through a small amount of newly collected unmatched data, and the fine-tuning process is implemented based on model parameter migration and feature space mapping.

[0022] Preferably, the data fusion submodule adopts a multimodal feature fusion model when fusing data. The multimodal feature fusion model performs weighted association on the visual features of remote sensing images of geological exploration, the physical field features of geophysical data, and the core attribute features of drilling data through an attention mechanism, and outputs a semantic-level fusion result.

[0023] Preferably, the federated learning module includes: a local training node and a global aggregation node;

[0024] The local training node uses differential privacy to perturb the training data, and the global aggregation node uses a federated averaging algorithm based on homomorphic encryption, which supports joint modeling of cross-regional geological data without leaking the original data.

[0025] Preferably, the edge computing 5G collaborative module adopts an edge computing gateway with an ARM architecture. The gateway has a built-in data preprocessing module and a lightweight inference engine, which can perform noise filtering, feature dimensionality reduction and outlier detection in real time, and transmit the processed core data to the cloud server storage through 5G slicing.

[0026] Preferably, the environmental sensors of the environmental perception and enhancement module include: a temperature and humidity sensor, a wind speed sensor, a device vibration sensor, and a battery charge monitor;

[0027] The reinforcement learning of the environmental perception and reinforcement module utilizes a proximal strategy optimization algorithm to dynamically adjust equipment operating parameters and exploration paths based on real-time environmental parameters and collected data quality assessment results.

[0028] Preferably, the spatial information grid collaboration module divides geological data resources into 1km×1km grid units, each grid unit is associated with a corresponding computing node and storage node, and dynamically schedules computing resources across grid units through load balancing to support parallel rendering and multi-scale analysis of three-dimensional geological models.

[0029] The present invention provides a mobile real-time geological survey data acquisition and analysis system for complex terrains. It has the following beneficial effects:

[0030] (1) This application reuses historical geological survey data, uses deep learning to summarize and generalize historical data template libraries in previously collected geological data, collects historical data with high repetition as templates, and splits the templates into a large number of modular data models. When real-time geological survey data is collected, the data is overlapped and matched with the historical data templates, and the modular templates that match it in the template library are confirmed. Multiple matching templates are combined to form a basic geological survey model. Then, the new data that cannot be matched in the geological survey data collected this time is imported into the above-mentioned basic geological survey model. The geological model can be dynamically adjusted in real time in the basic geological survey model to generate a new geological model corresponding to this geological survey. There is no need to analyze and model the huge amount of collected data from scratch, which can effectively reduce resource usage and analysis time, realize real-time collection and analysis of geological survey data for complex terrain, and improve the efficiency of data analysis.

[0031] (2) The efficiency of geological surveys is improved through the collaborative work of the edge computing 5G collaborative module and the historical data reuse module. By deploying the edge computing gateway on the mobile acquisition terminal, pre-processing such as noise filtering, feature dimensionality reduction and outlier detection can be performed in real time, and the core data can be quickly transmitted to the cloud through 5G slicing technology. At the same time, the historical data reuse module adopts template matching, data fusion and incremental update mechanisms, which can reuse historical data templates and adjust the model in real time, thereby reducing the amount of data transmission and cloud computing load, and speeding up the analysis speed. Compared with traditional methods, this type of data acquisition, transmission and analysis can greatly shorten the geological data analysis time, reduce resource consumption, and further realize the efficient acquisition and real-time analysis of geological data under complex terrain.

[0032] (3) The use of federated learning modules and environmental perception and enhancement modules effectively enhances data security and equipment adaptability. Federated learning uses a federated averaging algorithm based on differential privacy perturbation and homomorphic encryption, which does not leak original data during cross-institutional joint modeling. The environmental perception and enhancement module monitors environmental parameters in real time through multiple types of sensors and dynamically adjusts acquisition strategies and equipment parameters using a proximal strategy optimization algorithm, achieving the effect of ensuring data security sharing and improving the working stability of equipment in complex terrain environments. It not only meets the needs of cross-regional geological data joint analysis, but also can adapt to changing terrain and environmental conditions, ensuring the reliability of the acquisition process and data quality.

[0033] (4) The combination of the spatial information grid collaboration module and the historical data template library optimizes resource scheduling and model construction. The geological data resources are divided into grid units through the spatial information grid, and cross-grid computing resources are scheduled through dynamic load balancing to support parallel rendering and multi-scale analysis. At the same time, the historical data template library is standardized and supports feature similarity retrieval, providing fast matching and reuse for new data, thereby improving the utilization of computing resources and accelerating the construction of three-dimensional geological models. Therefore, it can efficiently integrate distributed geological data, reduce redundant calculations, realize the rapid generation and updating of geological models, and provide more timely and accurate decision support for geological surveys in complex terrains. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 This is a system block diagram of the mobile geological survey data real-time acquisition and analysis system for complex terrain according to the present invention;

[0035] Figure 2 This is a block diagram of a historical data reuse module of a mobile geological survey data real-time acquisition and analysis system for complex terrain according to the present invention;

[0036] Figure 3 This is a module block diagram of a mobile acquisition terminal of the mobile geological survey data real-time acquisition and analysis system for complex terrain according to the present invention. DETAILED DESCRIPTION

[0037] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention. Example 1

[0038] See also Figure 1 - Figure 3 The present invention provides a mobile geological survey data real-time acquisition and analysis system for complex terrain. To achieve the above purpose, the present invention is implemented through the following technical solutions: The mobile geological survey data real-time acquisition and analysis system for complex terrain includes:

[0039] Mobile acquisition terminal: Integrates multi-platform acquisition equipment and sensors that adapt to complex terrain, and is used for real-time acquisition of multi-source heterogeneous geological data;

[0040] The historical data template library stores standardized multi-dimensional historical geological data. The standardization process includes data cleaning, feature extraction, and time-space coordinate calibration. It supports template retrieval and reuse based on feature similarity.

[0041] The historical data reuse module realizes the reuse of historical templates, semantic association of multi-source data, and real-time adjustment of new data of basic templates through template matching, data fusion, and incremental update;

[0042] The federated learning module is used for distributed joint modeling across institutions. It encrypts and aggregates the sub-models trained locally by each participant through a global aggregation node to generate a global analysis model.

[0043] The edge computing 5G collaboration module is deployed on mobile acquisition terminals to implement data preprocessing and lightweight model reasoning, and interact with cloud servers through 5G communication links;

[0044] The environmental perception and reinforcement module integrates multiple types of environmental sensors and reinforcement learning algorithms to monitor environmental parameters and device status in real time and dynamically adjust collection strategies;

[0045] The spatial information grid collaboration module virtualizes distributed geological data resources into a grid pool and implements cross-node parallel computing through dynamic resource scheduling;

[0046] Mobile data collection terminals include: UAV platforms, tracked vehicle platforms and portable sensor groups;

[0047] The drone platform is equipped with a multispectral camera, lidar, and a barometric altimeter. The portable sensor suite includes a cesium optically pumped magnetometer, an X-ray fluorescence spectrometer, and a fiber Bragg grating stress sensor. All devices are synchronously controlled via an industrial bus.

[0048] In this embodiment, the mobile data collection terminal uses a three-level collaborative geological inspection data collection device model consisting of drones, crawlers, and portable personnel to achieve multi-source synchronous cross-platform data collection. The drone platform performs large-scale scanning with an 800m cruising radius. The multispectral camera acquires a frame of 400-1000nm band image every 3 seconds. The lidar generates a terrain point cloud at a density of 500 points / m2. The barometric altimeter corrects the altitude error to ±0.3m in real time. The crawler-mounted platform travels along the bottom of the gully and uses a robotic arm to deliver the portable sensor group to the cliff outcrop. The cesium optical pump magnetometer measures magnetic anomalies with a resolution of 0.1nT. The X-ray fluorescence spectrometer completes 30 element analysis of soil samples within 2 minutes. The fiber Bragg grating stress sensor records tiny deformations of the cliff in real time. All devices achieve microsecond time synchronization through the bus to ensure that the data's temporal and spatial coordinate deviation is less than 0.5m.

[0049] The historical data template library has a special sub-library built for geological characteristics, containing more than 3,000 sets of geological data from historical data. Data cleaning removes anomalies caused by environmental interference. The feature extraction process uses wavelet transform to extract the texture characteristics of geological vertical joints. Time and space calibration combines Beidou positioning with measured geological profile data. In a subsequent geological survey, feature similarity retrieval can be used to match similar geological templates in neighboring areas within 5 seconds, directly reuse and eliminate overlapping data, reducing the amount of repeated data collection and analysis.

[0050] Through multi-platform collaboration and the construction of special template libraries, all-round data coverage and efficient reuse of complex terrains have been achieved. The three-level coordination of drones, tracked vehicles and portable equipment has filled the blind spots of data collection in terrains such as gullies and steep cliffs. The template library processing mechanism tailored to regional characteristics allows historical data to be quickly adapted to new survey scenarios, achieving the effect of no blind spots in data collection and efficient reuse of historical information. In complex terrain areas such as the Loess Plateau, more comprehensive geological data can be obtained at a lower cost, providing accurate data support for soil and water conservation projects and geological disaster prevention and control. Example 2

[0051] Specific: Refer to Figure 1 and Figure 2 ,The historical data reuse module includes : template matching submodule, data fusion submodule, and incremental update submodule;

[0052] Template matching submodule: uses feature point matching algorithm to calculate the similarity between newly collected geological data and historical geological data templates, and quickly identifies and marks overlapping areas;

[0053] Data fusion submodule: Based on the alignment of the spatiotemporal characteristics of multi-source heterogeneous data, the historical geological data templates are combined and spliced ​​to construct a basic 3D geological model;

[0054] Incremental update submodule: After the basic 3D geological model is constructed by combining and splicing historical geological data templates, the new geological data collected but not matched is added to the constructed basic 3D geological model as the basis to generate the analysis results of the geological data;

[0055] The incremental update submodule adopts a meta-learning framework to fine-tune the parameters of the basic template through a small amount of newly collected unmatched data. The fine-tuning process is achieved through model parameter migration and feature space mapping.

[0056] The data fusion submodule uses a multimodal feature fusion model to fuse data. The multimodal feature fusion model uses an attention mechanism to perform weighted association on the visual features of remote sensing images for geological exploration, the physical field features of geophysical data, and the core attribute features of drilling data, and outputs a semantic-level fusion result.

[0057] In this embodiment, the geological environment of a certain area is used as an example. The historical data template library processes shale gas exploration data of a certain basin from 2010 to 2022 using a deep learning algorithm, screens out characteristic data such as lithology distribution and gas content distribution with a repetition rate of more than 90%, and constructs more than 800 sets of modular templates. These templates are divided into 15 basic modules according to geological elements (such as the floor interface module and the siliceous shale distribution module in a certain area of ​​the basin). Each module contains standardized data such as spectral feature vectors and structural trend parameters.

[0058] During a real-time survey of a shale gas field in a certain area, multispectral images and seismic data acquired by a mobile acquisition terminal were first fed into a template matching submodule. This submodule used an improved algorithm to compare the lithologic spectral characteristics of the real-time data with the feature vectors of historical templates. Within 10 seconds, it identified eight modular templates with a matching degree exceeding 85%. The overlap between the shale distribution template in a certain area and the actual drilling data reached 92%.

[0059] The data fusion submodule immediately combines the eight matching templates. Through spatiotemporal coordinate calibration and semantic association, it stitches the modules together into a basic geological model covering 80% of the survey area. This model includes 23 key parameters such as formation thickness and porosity, and is built in just one-fifth the time of traditional modeling methods.

[0060] For newly acquired data from three fault fracture zones (30% different from the historical template), the incremental update submodule, based on a meta-learning framework, converted the new data into feature gradient vectors. Through five rounds of parameter migration, each with 20 iterations, the basic model was dynamically adjusted. The adjusted model automatically updated the lithologic strength parameters within the fault's influence range. The resulting new geological model achieved a 90% agreement with actual drilling verification, reducing computing resource consumption by 60% compared to traditional modeling and shortening the model update cycle from 1.5 days to 6 hours.

[0061] Through the reuse and modular combination of historical geological data templates, the modeling cost and time have been greatly reduced. Highly repetitive historical data are split into reusable modules. Real-time data only needs to be matched and combined to form a basic model. New data is integrated into the model through incremental adjustments, avoiding modeling from scratch and reducing resource usage. This modular reuse can realize real-time data analysis and dynamic model updating in complex terrain surveys, significantly improving the efficiency and economy of geological exploration. Example 3

[0062] Specific: Refer to Figure 1 ,The standardized processing dimensions of the historical data template library include: lithologic texture characteristics, geophysical field parameters, stratigraphic age data and mineralization anomaly indicators of geological exploration;

[0063] Template retrieval is performed through dual retrieval of spatiotemporal index and feature vector index. The feature vector index is constructed based on the lithologic spectral characteristics and structural strike characteristics.

[0064] The spatial information grid collaboration module divides geological data resources into 1km×1km grid cells. Each grid cell is associated with a corresponding computing node and storage node. Through load balancing, computing resources across grid cells are dynamically scheduled to support parallel rendering and multi-scale analysis of 3D geological models.

[0065] In this embodiment, the following uses the geological environment of a certain place as an example to illustrate the standardized processing of the historical data template library in the karst area of ​​a certain region, forming a complete process: lithologic texture features are extracted through high-resolution remote sensing imagery, and the texture entropy value of the cave development zone is obtained using a grayscale co-occurrence matrix. Geophysical field parameters cover 15 indicators such as gravity and seismic wave velocity, which are converted into spectral features through Fourier transform. Stratigraphic age data are combined with paleontological fossils and carbon-14 dating to establish a standardized timeline from the Devonian to the Quaternary. Mineralization anomaly indicators are used to delineate mineralized zones using the element ratio method. Therefore, the subsequent dual search mechanism is effective in the exploration of a certain lead-zinc mine. The spatiotemporal index quickly locates mineralization data from 2015 to 2020 at 103°-104° east longitude. The feature vector index encodes the 380nm spectral reflectance of the skarn and the northeast structural strike into feature vectors through the network. Finally, the top five matching templates are retrieved from more than 100,000 data sets in just 1 second with an accuracy rate of 91%.

[0066] In a landslide monitoring project, the spatial information grid collaboration module divided the area into 1,000 1km×1km grids. Each grid was equipped with a computing device as a computing node and a 2TB storage node. The module monitored node status in real time and automatically migrated some rendering tasks to less-loaded nodes when the CPU utilization of a node exceeded 80%. When generating a 1:10,000 scale landslide model, parallel computing reduced rendering time from 45 minutes for a single node to 5 minutes. For multi-scale analysis, switching from a 1:50,000 scale to a 1:10,000 scale took only 1.8 seconds.

[0067] Through multi-dimensional standardization and grid collaboration, efficient reuse of geological data and computational acceleration are achieved. Standardized processing allows data from different sources to be compared, dual retrieval quickly locates similar templates, and grid scheduling maximizes the use of computing resources, achieving the effects of increased data reuse rate and shortened model rendering time. It can provide efficient data support for geological surveys in complex areas such as karst areas and reservoir areas, reduce exploration costs, and speed up decision-making. Example 4

[0068] Specific: Refer to Figure 1The federated learning module includes: local training nodes and global aggregation nodes. The local training nodes use differential privacy to perturb the training data, and the global aggregation nodes use a federated averaging algorithm based on homomorphic encryption, which supports joint modeling of cross-regional geological data without leaking the original data.

[0069] The edge computing 5G collaboration module uses an ARM-based edge computing gateway with a built-in data pre-processing module and a lightweight inference engine. It can perform noise filtering, feature dimensionality reduction, and outlier detection in real time, and transmit the processed core data to cloud server storage via 5G slicing.

[0070] The environmental sensors of the environmental perception and enhancement module include: temperature and humidity sensors, wind speed sensors, device vibration sensors and battery power monitors;

[0071] The reinforcement learning of the environmental perception and reinforcement module uses a proximal policy optimization algorithm to dynamically adjust equipment operating parameters and exploration paths based on real-time environmental parameters and collected data quality assessment results;

[0072] In this embodiment, the federated learning module achieves cross-institutional collaborative modeling through a distributed architecture. Local training nodes perform differential privacy processing on the original data. By adding perturbation factors to the data set, individual data features are blurred while the overall statistical characteristics are retained, ensuring that sensitive information is not leaked when the data is involved in modeling. Each node independently trains a sub-model and only encrypts the model parameters using homomorphic encryption technology before uploading them to the global aggregation node. The global node uses federated averaging to aggregate the encrypted parameters. The generated global model can integrate the characteristics of multi-source data while avoiding the cross-institutional flow of original data, meeting the requirements of data security and privacy protection.

[0073] The edge computing 5G collaboration module localizes data processing through the edge computing gateway on the terminal side. The gateway's built-in pre-processing module processes the collected raw data in real time, eliminating signal distortion caused by environmental interference through noise filtering, using feature dimensionality reduction technology to extract key information to reduce data redundancy, and using an outlier detection mechanism to identify and eliminate invalid data. The processed core data is transmitted through 5G slicing technology, which divides independent communication channels to ensure low latency and high reliability of data transmission, while reducing the data flow interacting with the cloud and improving the response speed of the entire system.

[0074] The environmental perception and enhancement module builds an environmental perception network using multiple types of sensors to capture environmental parameters such as temperature, humidity, and airflow velocity in the geological survey area, as well as equipment operating status information in real time. Reinforcement learning optimizes data collection quality and equipment operating stability. It continuously learns the mapping relationship between environmental parameters and equipment adjustment strategies through a proximal strategy optimization algorithm. When environmental parameters or equipment status change, the algorithm automatically adjusts the equipment's operating parameters and motion path, enabling the equipment to maintain stable operation in complex environments, ensuring the continuity and validity of collected data while reducing the risk of equipment loss.

[0075] The synergy of distributed security modeling, real-time processing at the edge, and adaptive environmental adjustments enhances the system's security, real-time performance, and adaptability. Federated learning ensures the security of data sharing. Edge computing and 5G collaborate to accelerate data processing and transmission efficiency, reducing the transmission of some collected geological data that duplicates historical data back to the analysis center, thereby ensuring data accuracy and simplification. Reinforcement learning enables dynamic adaptation of equipment to complex environments, achieving cross-institutional data security joint modeling, real-time, efficient data processing, and stable equipment operation. Therefore, the system can achieve efficient, secure, and stable data analysis and equipment control in complex terrain surveys, providing technical support for intelligent geological surveys.

[0076] While the embodiments of the present invention have been shown and described, it will be apparent to those skilled in the art that various changes, modifications, substitutions, and alterations can be made to the embodiments without departing from the principles and spirit of the invention.

Claims

1. A mobile real-time geological survey data acquisition and analysis system for complex terrain, featuring: The acquisition and analysis system includes: Mobile acquisition terminal: Integrates multi-platform acquisition equipment and sensors that adapt to complex terrain, and is used for real-time acquisition of multi-source heterogeneous geological data; A historical data template library stores standardized multi-dimensional historical geological data, including data cleaning, feature extraction, and time-space coordinate calibration, and supports template retrieval and reuse based on feature similarity. The historical data reuse module realizes the reuse of historical templates, semantic association of multi-source data, and real-time adjustment of new data of basic templates through template matching, data fusion, and incremental update; The federated learning module is used for distributed joint modeling across institutions. It encrypts and aggregates the sub-models trained locally by each participant through a global aggregation node to generate a global analysis model. An edge computing 5G collaboration module is deployed on the mobile acquisition terminal to implement data preprocessing and lightweight model reasoning, and to interact with the cloud server via a 5G communication link. The environmental perception and reinforcement module integrates multiple types of environmental sensors and reinforcement learning algorithms to monitor environmental parameters and device status in real time and dynamically adjust collection strategies; The spatial information grid collaboration module virtualizes distributed geological data resources into a grid pool and realizes cross-node parallel computing through dynamic resource scheduling.

2. The mobile geological survey data real-time acquisition and analysis system for complex terrain according to claim 1 is characterized by: The mobile acquisition terminal includes: a UAV platform, a tracked vehicle-mounted platform and a portable sensor group; The UAV platform is equipped with a multispectral camera, a laser radar and a barometric altimeter. The portable sensor group includes a cesium optical pump magnetometer, an X-ray fluorescence spectrometer and a fiber Bragg grating stress sensor. All devices are synchronously controlled through an industrial bus.

3. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 1 is characterized by: The standardized processing dimensions of the historical data template library include: lithologic texture characteristics, geophysical field parameters, stratigraphic age data and mineralization anomaly indicators of geological exploration; The template retrieval is performed through dual retrieval of spatiotemporal index and feature vector index, wherein the feature vector index is constructed based on lithologic spectral features and structural strike features.

4. The mobile geological survey data real-time acquisition and analysis system for complex terrain according to claim 1 is characterized by: The historical data reuse module includes: a template matching submodule, a data fusion submodule and an incremental update submodule; The template matching submodule uses a feature point matching algorithm to calculate the similarity between the newly collected geological data and the historical geological data template, and quickly identifies and marks the overlapping areas; The data fusion submodule: aligns the spatiotemporal features of multi-source heterogeneous data, and then combines and splices historical geological data templates to construct a basic three-dimensional geological model; The incremental update submodule: after the historical geological data templates are combined and spliced ​​to construct the basic three-dimensional geological model, for the new geological data collected but not matched, based on the constructed basic three-dimensional geological model, new data is added to the model to generate analysis results of the geological collection data.

5. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 4 is characterized by: The incremental update submodule adopts a meta-learning framework to fine-tune the parameters of the basic template through a small amount of newly collected unmatched data. The fine-tuning process is implemented based on model parameter migration and feature space mapping.

6. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 4 is characterized by: The data fusion submodule adopts a multimodal feature fusion model when fusing data. The multimodal feature fusion model uses an attention mechanism to perform weighted association on the visual features of remote sensing images of geological exploration, the physical field features of geophysical data, and the core attribute features of drilling data, and outputs a semantic-level fusion result.

7. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 1 is characterized by: The federated learning module includes: a local training node and a global aggregation node; The local training node uses differential privacy to perturb the training data, and the global aggregation node uses a federated averaging algorithm based on homomorphic encryption, which supports joint modeling of cross-regional geological data without leaking the original data.

8. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 1 is characterized by: The edge computing 5G collaboration module adopts an ARM architecture edge computing gateway. The gateway has a built-in data preprocessing module and a lightweight inference engine, which can perform noise filtering, feature dimensionality reduction and outlier detection in real time, and transmit the processed core data to the cloud server storage through 5G slicing.

9. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 1 is characterized by: The environmental sensors of the environmental perception and enhancement module include: temperature and humidity sensors, wind speed sensors, equipment vibration sensors and battery power monitors; The reinforcement learning of the environmental perception and reinforcement module utilizes a proximal strategy optimization algorithm to dynamically adjust equipment operating parameters and exploration paths based on real-time environmental parameters and collected data quality assessment results.

10. The mobile real-time geological survey data acquisition and analysis system for complex terrain according to claim 1, characterized in that: The spatial information grid collaboration module divides geological data resources into 1km×1km grid units. Each grid unit is associated with a corresponding computing node and storage node. Through load balancing, computing resources across grid units are dynamically scheduled to support parallel rendering and multi-scale analysis of 3D geological models.

Images