Hard rock type rare earth ore exploration method and system

Abstract

The invention relates to the technical field of mineral resource exploration, in particular to a hard rock type rare earth ore exploration method which comprises the following steps: step 1, performing regional geology and remote sensing preliminary screening, performing full-coverage aerial survey by adopting an unmanned aerial vehicle carrying a hyperspectral imager and an L DAR, interpreting a linear structure and an annular structure in combination with a satellite image, and performing remote sensing and remote sensing preliminary screening; an overlapping area of hyperspectral rare earth mineral abnormity, a fracture structure dense area and a granite / pegmatite exposure area is screened out to serve as a detailed investigation target area; compared with manual mapping, the unmanned aerial vehicle aerial survey coverage efficiency is improved by 10 times, the daily progress of a ground detection platform is improved by 3 times, the whole-process exploration period is shortened to 6-12 months, the hidden ore body prediction error is smaller than or equal to 15% through multi-source data fusion and a machine learning model, the drilling hit rate is improved to 70% or above, the whole process from data collection to target area decision making is automatic, and the method has the advantages of being high in practicability and high in practicability. Manual intervention is reduced, and complex terrain operation risks are reduced.

Description

Technical Field

[0001] This invention relates to the field of mineral resource exploration technology, specifically to a method and system for exploring hard rock rare earth deposits. Background Technology

[0002] Hard rock rare earth deposits are an important type of rare earth resource reserve in my country. They are characterized by complex occurrence (often associated with rock-forming minerals and uneven distribution), large burial depths (often reaching 50–500 meters), and are mostly distributed in areas with complex topography. Traditional exploration methods have significant limitations.

[0003] Low data acquisition efficiency: It relies on manual geological mapping, ground geophysical profile measurement and discrete borehole sampling. The average daily progress of a single survey line is less than 500 meters, and it is difficult to fully cover areas such as forests and steep slopes, resulting in poor data integrity.

[0004] Multi-source data integration is difficult: geological, geophysical (magnetic, gravity, induced polarization), geochemical (soil, stream sediments), and remote sensing data are independent of each other and lack a unified fusion and analysis method, resulting in problems such as "disconnection between geophysical anomalies and mineralization information" and "multiple interpretations of geochemical anomalies".

[0005] Low accuracy of deep prediction: Traditional methods rely on empirical models to infer the distribution of ore bodies, and the prediction error for concealed ore bodies (buried depth > 100 meters) can reach more than 30%, and the drilling verification hit rate is less than 40%.

[0006] The exploration cycle is long: it takes 2-3 years from regional general survey to detailed survey, and the subsequent data processing and modeling rely on manual interpretation, which is not timely and makes it difficult to respond quickly to the needs of resource exploration. Therefore, a method and system for exploring hard rock rare earth deposits is proposed. Summary of the Invention

[0007] In view of this, the present invention provides a method and system for exploring hard rock rare earth deposits to solve or alleviate the technical problems existing in the prior art, and at least provides a beneficial alternative.

[0008] The technical solution of this invention is implemented as follows: a method for exploring hard rock rare earth deposits, comprising the following steps:

[0009] Step 1: Regional geology and remote sensing preliminary screening. A drone equipped with a hyperspectral imager and LiDAR was used to conduct full-coverage aerial survey. By interpreting linear and ring structures in conjunction with satellite imagery, overlapping areas of "hyperspectral rare earth mineral anomalies + dense fault structures + granite / pegmatite outcrops" were selected as target areas for detailed investigation.

[0010] Step 2: Ground multi-parameter collaborative detection. Within the detailed investigation target area, magnetic, gravity, and induced polarization data are collected along the preset survey line using an autonomous mobile platform. At the same time, soil samples are collected using a grid-based sampling method, and the total rare earth content and characteristic element content are determined. The data are then subjected to terrain correction and regional background subtraction to extract relative outliers.

[0011] Step 3: Intelligent fusion and modeling of multi-source data. An improved CNN model is used to fuse hyperspectral images, geophysical anomaly maps, and geochemical anomaly maps to output a planar distribution map of mineralization potential index. Based on geophysical data, three-dimensional inversion is performed, and a three-dimensional geological structure model is constructed by combining borehole data. Then, the prediction model is trained by random forest algorithm to delineate key target areas.

[0012] Step 4: Intelligent drilling verification and data update. Optimize the borehole trajectory based on the three-dimensional geological structure model. During the drilling process, collect core images and elemental data in real time and transmit them to the system for analysis. Feed the drilling data back to the prediction model in Step 3 to update the algorithm parameters.

[0013] Step 5: Resource estimation and target area classification. Geostatistical methods are used to calculate the rare earth resources in key target areas. Based on the resource quantity, mineralization continuity and mining conditions, the target areas are classified into Grade A, Grade B and Grade C, and an exploration report is output.

[0014] More preferably, in step 1, the hyperspectral imager has a wavelength range of 400-2500 nm, used to identify the characteristic absorption peaks of rare earth minerals; the satellite image is a Sentinel-2 satellite image with a resolution of 10 m; and the area of ​​the detailed survey target region is 10-50 km². 2 .

[0015] More preferably, in step 2, the autonomous mobile platform is a tracked type, adaptable to a 25° slope, equipped with a magnetic sensor with an accuracy of ±0.1nT, a gravimeter resolution of 1μGal, and an induced polarization meter error of <5%; the grid density of the grid-based sampling method is 100m×100m, the soil sample collection depth is 0-20cm, and the elemental content is determined on-site using an X-ray fluorescence spectrometer with a detection limit of <1ppm.

[0016] More preferably, in step 3, the improved CNN model takes hyperspectral images, magnetic anomaly maps, and geochemical anomaly maps as inputs, and fuses spatial and attribute features through a feature extraction layer to output a mineralization potential index of 0-100; the depth range of the three-dimensional geological structure model is 0-500 meters, with an accuracy of ±5 meters; the input parameters of the random forest algorithm include structural distance, rock mass type, and geochemical anomaly value, and output the probability of ore body existence.

[0017] More preferably, in step 4, the drilling trajectory is an inclined hole with an angle of 30-60° and a depth of 100-500 meters; the real-time analysis uses a laser ablation inductively coupled plasma mass spectrometer, with an analysis time of <1 minute / sample, and an automatic alarm is triggered when the total rare earth element ΣREE is detected to be >0.5%.

[0018] A hard rock rare earth mineral exploration system includes a geological information acquisition module, a multi-source data processing module, a three-dimensional modeling and prediction module, an intelligent drilling control module, and a decision support module that interact with each other in sequence.

[0019] The geological information acquisition module includes an UAV aerial survey submodule, a ground exploration submodule, and a drilling analysis submodule, which are used to acquire hyperspectral, LiDAR, geophysical, geochemical, and drilling data, respectively.

[0020] The multi-source data processing module includes a data preprocessing submodule and an intelligent fusion submodule, which are used to reduce noise and correct the raw data, and fuse multi-source data through an improved CNN model;

[0021] The 3D modeling and prediction module includes a 3D geological modeling submodule and an ore body prediction submodule, which are used to construct a 3D geological structure model and predict the ore body distribution through a random forest algorithm.

[0022] The intelligent drilling control module includes a drilling trajectory design submodule and a drilling feedback submodule, which are used to optimize the borehole trajectory and update the model in real time.

[0023] The decision support module includes a resource estimation submodule and a target classification submodule, which are used to calculate resource quantities and classify target areas into levels.

[0024] More preferably, the UAV aerial survey submodule is equipped with a hyperspectral imager with a resolution of 30cm and a point cloud density of 50 points / m². 2 The LiDAR has a battery life of ≥2 hours; the ground detection submodule integrates a magnetic sensor, gravimeter, induced polarization meter and X-ray fluorescence spectrometer, and supports automatic obstacle avoidance and path planning.

[0025] Further preferably, the data preprocessing submodule uses wavelet transform to denoise the hyperspectral data and converts all data into GeoTIFF format, achieving a processing efficiency of ≥10GB / hour; the intelligent fusion submodule supports GPU acceleration, with a fusion processing speed of ≥1km / h. 2 / 10 minutes.

[0026] More preferably, the three-dimensional geological modeling submodule supports dynamic cross-sectioning and attribute querying, and the ore body prediction submodule outputs an ore body existence probability distribution map with a resolution of 5 meters × 5 meters, with a prediction error of ≤15%.

[0027] Further preferably, the target area classification submodule classifies the target area into levels based on a multi-factor weighted score, with resource quantity accounting for 40%, mineralization continuity for 30%, and mining conditions for 30%, and outputs a visual exploration report.

[0028] The embodiments of the present invention have the following advantages due to the adoption of the above technical solutions:

[0029] I. The UAV aerial survey coverage efficiency of this invention is 10 times higher than that of manual mapping, the daily progress of the ground detection platform is 3 times higher, the entire exploration cycle is shortened to 6-12 months, the multi-source data fusion and machine learning model make the prediction error of hidden ore bodies ≤15%, the drilling hit rate is increased to more than 70%, and the entire process from data acquisition to target area decision-making is automated, reducing manual intervention and reducing the risk of operation in complex terrain.

[0030] Second, this invention breaks through the traditional data silo problem, realizes the deep integration of remote sensing, geophysical exploration, geochemical exploration and drilling data, provides multi-dimensional evidence for ore body prediction, is applicable to various hard rock rare earth deposits (granite type, pegmatite type, etc.), and can be extended to the exploration of other metal deposits (such as tungsten and tin deposits), with broad application prospects.

[0031] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of the invention will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a flowchart of the method of the present invention;

[0034] Figure 2 This is a system module diagram of the present invention. Detailed Implementation

[0035] In the following description, only certain exemplary embodiments are briefly described. As those skilled in the art will recognize, the described embodiments can be modified in various ways without departing from the spirit or scope of the invention. Therefore, the drawings and description are considered to be exemplary in nature and not restrictive.

[0036] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0037] like Figure 1-2 As shown in the figure, this invention provides a method for exploring hard rock rare earth deposits, including the following steps:

[0038] Step 1: Regional geology and remote sensing preliminary screening. A drone equipped with a hyperspectral imager and LiDAR was used to conduct full-coverage aerial survey. By interpreting linear and ring structures in conjunction with satellite imagery, overlapping areas of "hyperspectral rare earth mineral anomalies + dense fault structures + granite / pegmatite outcrops" were selected as target areas for detailed investigation.

[0039] Step 2: Ground-based multi-parameter collaborative detection. Within the target area, magnetic, gravity, and induced polarization data are collected along the preset survey line using an autonomous mobile platform. At the same time, soil samples are collected using a grid-based sampling method, and the total rare earth content and characteristic element content are determined. Topographic correction and regional background subtraction are performed on the data to extract relative outliers.

[0040] Step 3: Intelligent fusion and modeling of multi-source data. An improved CNN model is used to fuse hyperspectral images, geophysical anomaly maps, and geochemical anomaly maps to output a planar distribution map of mineralization potential index. Based on geophysical data, three-dimensional inversion is performed, and a three-dimensional geological structure model is constructed by combining borehole data. Then, the prediction model is trained by random forest algorithm to delineate key target areas.

[0041] Step 4: Intelligent drilling verification and data update. Optimize the borehole trajectory based on the three-dimensional geological structure model. Collect core images and elemental data in real time during the drilling process and transmit them to the system for analysis. Feed the drilling data back to the prediction model in Step 3 to update the algorithm parameters.

[0042] Step 5: Resource estimation and target area classification. Geostatistical methods are used to calculate the rare earth resources in key target areas. Based on the resource quantity, mineralization continuity and mining conditions, the target areas are classified into Grade A, Grade B and Grade C, and an exploration report is output.

[0043] In one embodiment, in step 1, the hyperspectral imager has a wavelength range of 400-2500 nm and is used to identify the characteristic absorption peaks of rare earth minerals; the satellite image is a Sentinel-2 satellite image with a resolution of 10 m; and the area of ​​the target region to be investigated is 10-50 km². 2 .

[0044] In one embodiment, in step 2, the autonomous mobile platform is a tracked type, adaptable to a 25° slope, equipped with a magnetic sensor with an accuracy of ±0.1nT, a gravimeter with a resolution of 1μGal, and an induced polarization meter with an error of <5%; the grid density of the grid-based sampling method is 100m×100m, the soil sample collection depth is 0-20cm, and the element content is determined on-site using an X-ray fluorescence spectrometer with a detection limit of <1ppm.

[0045] In one embodiment, in step 3, the improved CNN model takes hyperspectral images, magnetic anomaly maps, and geochemical anomaly maps as inputs, fuses spatial and attribute features through a feature extraction layer, and outputs a mineralization potential index of 0-100; the depth range of the three-dimensional geological structure model is 0-500 meters, with an accuracy of ±5 meters; the input parameters of the random forest algorithm include structural distance, rock mass type, and geochemical anomaly value, and outputs the probability of ore body existence.

[0046] In one embodiment, in step 4, the drilling trajectory is an inclined hole with an angle of 30-60° and a depth of 100-500 meters; real-time analysis is performed using a laser ablation inductively coupled plasma mass spectrometer, with an analysis time of <1 minute / sample, and an automatic alarm is triggered when the total rare earth element ΣREE is detected to be >0.5%.

[0047] A hard rock rare earth mineral exploration system includes a geological information acquisition module, a multi-source data processing module, a three-dimensional modeling and prediction module, an intelligent drilling control module, and a decision support module that interact with each other in sequence.

[0048] The geological information acquisition module includes an UAV aerial survey submodule, a ground exploration submodule, and a drilling analysis submodule, which are used to acquire hyperspectral, LiDAR, geophysical, geochemical, and drilling data, respectively.

[0049] The multi-source data processing module includes a data preprocessing submodule and an intelligent fusion submodule, which are used to reduce noise and correct the raw data, and fuse multi-source data through an improved CNN model;

[0050] The 3D modeling and prediction module includes a 3D geological modeling submodule and an ore body prediction submodule, which are used to construct 3D geological structure models and predict ore body distribution through the random forest algorithm;

[0051] The intelligent drilling control module includes a drilling trajectory design submodule and a drilling feedback submodule, which are used to optimize the borehole trajectory and update the model in real time;

[0052] The decision support module includes a resource estimation submodule and a target classification submodule, which are used to calculate resource quantities and classify target areas.

[0053] In one embodiment, the UAV aerial survey submodule is equipped with a hyperspectral imager with a resolution of 30 cm and a point cloud density of 50 points / m². 2 The LiDAR has a battery life of ≥2 hours; the ground detection submodule integrates a magnetic sensor, gravimeter, induced polarization meter and X-ray fluorescence spectrometer, and supports automatic obstacle avoidance and path planning.

[0054] In one embodiment, the data preprocessing submodule uses wavelet transform to denoise the hyperspectral data and converts all data into GeoTIFF format, achieving a processing efficiency of ≥10GB / hour; the intelligent fusion submodule supports GPU acceleration, with a fusion processing speed of ≥1km / h. 2 / 10 minutes.

[0055] In one embodiment, the 3D geological modeling submodule supports dynamic cross-sectioning and attribute querying, and the ore body prediction submodule outputs a ore body existence probability distribution map with a resolution of 5 meters × 5 meters and a prediction error of ≤15%.

[0056] In one embodiment, the target area classification submodule classifies target areas into different levels based on a multi-factor weighted score, with resource quantity accounting for 40%, mineralization continuity for 30%, and mining conditions for 30%, and outputs a visual exploration report.

[0057] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in the present invention, and these should all be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method of exploration for hard rock type rare earth ore, characterized by: The method comprises the following steps: Step 1: regional geology and remote sensing preliminary screening, using a UAV equipped with a hyperspectral imager and Li DAR for full coverage aerial survey, combining satellite image interpretation of linear structures and ring structures, screening out the overlapping area of "hyperspectral rare earth mineral anomaly + fracture structure dense area + granite / pegmatite outcrop area" as the detailed target area; Step 2: ground multi-parameter cooperative detection, in the detailed target area, collecting magnetic, gravity, induced polarization data along the preset survey line by the autonomous mobile platform, simultaneously collecting soil samples by grid point method and measuring total rare earth content and characteristic element content, and performing terrain correction and regional background subtraction on the data to extract relative anomaly values; Step 3: intelligent fusion and modeling of multi-source data, using an improved CNN model to fuse hyperspectral images, geophysical anomaly maps and geochemical anomaly maps, and outputting a mineralization potential index plane distribution map; based on geophysical data, three-dimensional inversion is performed, a three-dimensional geological structure model is constructed in combination with drilling data, and a prediction model is trained through a random forest algorithm to delineate key target areas; Step 4: intelligent drilling verification and data updating, optimizing the drilling trajectory based on the three-dimensional geological structure model, collecting core images and element data in real time during drilling and transmitting them to the system for analysis, and feeding back the drilling data to the prediction model of step 3 to update the algorithm parameters; Step 5: resource quantity estimation and target area classification, using a geological statistical method to calculate the rare earth resource quantity of the key target area, and classifying the target area into A, B and C levels according to the resource quantity, mineralization continuity and mining conditions, and outputting an exploration report.

2. The method for exploring hard rock type rare earth ore according to claim 1, characterized in that: In the step 1, the wave band range of the hyperspectral imager is 400-2500nm, which is used to identify the characteristic absorption peak of rare earth minerals; the satellite image is a Sentinel-2 satellite image, and the resolution is 10m; the area of the detailed investigation target area is 10-50km 2 .

3. The method according to claim 1, characterized in that: In step 2, the autonomous mobile platform is a tracked type, which can adapt to a 25° slope, and is equipped with a magnetic sensor with an accuracy of ±0.1nT, a gravimeter with a resolution of 1μGal, and an induced polarization instrument with an error of less than 5%; In step 2, the grid point method has a grid density of 100m x 100m, and the soil sample collection depth is 0-20cm, and the element content is determined by an X-ray fluorescence spectrometer on site, with a detection limit of less than 1ppm.

4. The method for exploring hard rock type rare earth ore according to claim 1, characterized in that: In step 3, the improved CNN model takes hyperspectral images, magnetic anomaly maps and geochemical anomaly maps as input, fuses spatial and attribute features through a feature extraction layer, and outputs a mineralization potential index of 0-100; the three-dimensional geological structure model has a depth range of 0-500 meters and an accuracy of ±5 meters; the input parameters of the random forest algorithm include structural distance, rock type and geochemical anomaly value, and the output is the probability of the existence of ore bodies.

5. The method for exploring hard rock type rare earth ore according to claim 1, characterized in that: In step 4, the drilling trajectory is an inclined hole with an angle of 30-60° and a depth of 100-500 meters; the real-time analysis uses a laser ablation inductively coupled plasma mass spectrometer, and the analysis time is less than 1 minute / sample, and an automatic alarm is triggered when the total rare earth content ΣREE is greater than 0.5%.

6. A hard rock type rare earth ore exploration system, which is matched with the hard rock type rare earth ore exploration method according to any one of claims 1-5, characterized in that: The method comprises the following steps: The geological information acquisition module comprises a UAV aerial survey submodule, a ground detection submodule and a drilling analysis submodule, which are respectively used for acquiring hyperspectral, Li DAR, geophysical, geochemical and while-drilling data; The geological information acquisition module comprises a UAV aerial survey submodule, a ground detection submodule and a drilling analysis submodule, which are respectively used for acquiring hyperspectral, Li DAR, geophysical, geochemical and while-drilling data; The multi-source data processing module comprises a data preprocessing submodule and an intelligent fusion submodule, and is used for denoising and correcting original data and fusing multi-source data through an improved CNN model; The three-dimensional modeling and prediction module comprises a three-dimensional geological modeling submodule and a ore body prediction submodule, and is used for constructing a three-dimensional geological structure model and predicting ore body distribution through a random forest algorithm; The intelligent drilling control module comprises a drilling trajectory design submodule and a while-drilling feedback submodule, and is used for optimizing a drilling trajectory and updating a model in real time; The decision support module comprises a resource quantity estimation submodule and a target area grading submodule, and is used for calculating resource quantity and grading target areas.

7. The system for exploration of hard rock type rare earth ore according to claim 6, characterized in that: The unmanned aerial vehicle photogrammetry submodule is equipped with a hyperspectral imager with a resolution of 30cm and a LiDAR with a point cloud density of 50 points / m 2 The endurance time is greater than or equal to 2 hours; the ground detection submodule is integrated with a magnetic sensor, a gravimeter, a induced polarization instrument and an X-ray fluorescence spectrometer, and supports automatic obstacle avoidance and path planning.

8. The system for exploration of hard rock type rare earth ore according to claim 6, characterized in that: The data preprocessing submodule adopts wavelet transform to denoise hyperspectral data, converts all data into GeoTIFF format, and the processing efficiency is greater than or equal to 10 GB / hour; the intelligent fusion submodule supports GPU acceleration, and the fusion processing speed is greater than or equal to 1 km 2 / 10 minutes.

9. The system for exploration of hard rock type rare earth ore according to claim 6, characterized in that: The three-dimensional geological modeling submodule supports dynamic sectioning and attribute query, and the ore body prediction submodule outputs an ore body existence probability distribution map with a resolution of 5m x 5m, and a prediction error is less than or equal to 15%.

10. The system for exploration of hard rock type rare earth ore according to claim 6, characterized in that: The target area grading submodule grades target areas based on multi-factor weighted scoring, wherein resource quantity accounts for 40%, mineralization continuity accounts for 30%, and mining conditions account for 30%, and outputs a visualized exploration report.

Images