CN-122023945-A - Ore formation prediction method, device, equipment and storage medium based on neural network

CN122023945ACN 122023945 ACN122023945 ACN 122023945ACN-122023945-A

Abstract

The disclosure provides an ore-forming prediction method, device and equipment based on a neural network and a storage medium. The method comprises the steps of obtaining a multi-modal geological image of a target area, determining one-dimensional multi-modal data corresponding to center points and two-dimensional multi-modal data of a neighborhood of the center points from the multi-modal geological image by means of a pre-trained ore-forming prediction model, extracting local attribute features from the one-dimensional multi-modal data, extracting spatial context features from the two-dimensional multi-modal data, carrying out self-adaptive fusion processing on the local attribute features and the spatial context features to obtain target fusion features, determining ore-forming probability values of the center points based on the target fusion features, and determining ore-forming prediction results of the target area according to the ore-forming probability values of all the center points in the target area and an integration strategy. Thus, the ore forming possibility of each center point can be estimated more accurately, and the accuracy of the ore forming prediction result of the target area is improved.

Inventors

WANG MENG
Ju Hongkuang
LIU LEI
ZHANG ZHIWU
HU JIE
LIU YANQUN
LIN LUJUN
WU HAORAN
ZHANG BORUI
ZHANG JINLING

Assignees

中国冶金地质总局矿产资源研究院

Dates

Publication Date: 20260512
Application Date: 20260409

Claims (10)

1. An ore-forming prediction method based on a neural network, the method comprising: Acquiring a multi-modal geological image of a target area; determining one-dimensional multi-modal data corresponding to the center point and two-dimensional multi-modal data of the neighborhood of the center point from the multi-modal geological image by utilizing a pre-trained ore-forming prediction model; Extracting local attribute features from the one-dimensional multi-mode data, and extracting spatial context features from the two-dimensional multi-mode data; Performing self-adaptive fusion processing on the local attribute characteristics and the spatial context characteristics to obtain target fusion characteristics; and determining the mining probability values of the central points based on the target fusion characteristics, and determining the mining prediction result of the target area according to the mining probability values of all the central points in the target area and an integration strategy.
2. The method of claim 1, wherein determining the one-dimensional multi-modal data corresponding to the center point and the two-dimensional multi-modal data of the center point neighborhood from the multi-modal geologic image comprises: determining a central point for ore-forming prediction from the target area; Determining a plurality of first geological parameters corresponding to the central point from the multi-modal geological image by using a first extraction module in a pre-trained ore-forming prediction model, and converting the plurality of first geological parameters into one-dimensional multi-modal data corresponding to the central point; And determining a plurality of second geological parameters of the central point neighborhood from the multi-modal geological image by using a second extraction module in the pre-trained ore-forming prediction model, and converting the plurality of second geological parameters into two-dimensional multi-modal data of the central point neighborhood.
3. The method of claim 1, wherein the extracting local attribute features from the one-dimensional multimodal data and extracting spatial context features from the two-dimensional multimodal data comprises: extracting a first geological feature from the one-dimensional multi-modal data by utilizing a one-dimensional convolution layer in a pre-trained ore-forming prediction model, and carrying out enhancement processing on the first geological feature through a channel attention module to obtain a local attribute feature; And extracting a second geological feature from the two-dimensional multi-modal data by utilizing a two-dimensional convolution layer in the pre-trained ore-forming prediction model, and carrying out enhancement processing on the second geological feature through a spatial attention module to obtain a spatial context feature.
4. The method according to claim 1, wherein the adaptively fusing the local attribute feature and the spatial context feature to obtain a target fused feature comprises: carrying out self-adaptive fusion on the local attribute characteristics and the spatial context characteristics based on a channel attention mechanism to obtain first fusion characteristics; Splicing the local attribute features and the spatial context features to obtain spliced features, and processing the spliced features by using a gating unit to obtain second fusion features; and determining target fusion characteristics according to the first fusion characteristics and the second fusion characteristics.
5. The method of claim 4, wherein the adaptively fusing the local attribute feature and the spatial context feature based on the channel attention mechanism to obtain a first fused feature comprises: determining a first channel descriptor of the local attribute feature and a second channel descriptor of the spatial context feature; Determining a first dynamic weight corresponding to the local attribute feature according to the first channel descriptor, and determining a second dynamic weight corresponding to the spatial context feature according to the second channel descriptor; and fusing the local attribute feature and the spatial context feature based on the first dynamic weight and the second dynamic weight to obtain a first fused feature.
6. The method of claim 1, wherein the determining the mine probability values for the center points based on the target fusion feature, determining mine prediction results for the target region based on mine probability values for all center points in the target region in combination with an integration strategy, comprises: Predicting whether the central point is a mine point based on the target fusion characteristics to obtain a classification prediction result, and converting the classification prediction result into an ore probability value of the central point; Determining the mining probability values of all the center points in the target area by utilizing a sliding window, and determining a mining probability map of the target area based on the mining probability values of all the center points; And weighting a plurality of ore-forming probability maps output by the ore-forming prediction models to obtain an ore-forming prediction result of the target area.
7. The method of claim 1, wherein the pre-training process of the mineralisation model comprises: acquiring sample geological data of a target area, and determining a positive sample and a negative sample from the sample geological data; dividing the positive sample into first training data and first test data according to a preset proportion, and dividing the negative sample into second training data and second test data according to a preset proportion; taking the first training data and the second training data as training sample data, and taking the first test data and the second test data as test sample data; inputting the training sample data into an initial prediction model, and determining an ore formation prediction graph according to the training sample data by utilizing the initial prediction model; Determining a cross entropy loss function based on the actual ore forming label corresponding to the ore forming prediction graph and the training sample data, and updating model parameters of the initial prediction model according to the cross entropy loss function to obtain an updated prediction model; and verifying the updated prediction model by using the test sample data, and taking the updated prediction model as an ore-forming prediction model in response to determining that the updated prediction model passes the verification.
8. An ore-forming prediction apparatus based on a neural network, comprising: An acquisition module configured to acquire a multi-modal geologic image of a target area; The geological data determining module is configured to determine one-dimensional multi-modal data corresponding to the center point and two-dimensional multi-modal data of the neighborhood of the center point from the multi-modal geological image by utilizing a pre-trained ore-forming prediction model; The feature extraction module is configured to extract local attribute features from the one-dimensional multi-mode data and extract spatial context features from the two-dimensional multi-mode data; the feature fusion module is configured to perform self-adaptive fusion processing on the local attribute features and the spatial context features to obtain target fusion features; The ore formation prediction module is configured to determine an ore formation probability value of the central point based on the target fusion characteristic, and determine an ore formation prediction result of the target area according to the ore formation probability values of all central points in the target area and an integration strategy.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and running on the processor, the processor implementing the method of any one of claims 1 to 7 when executing the program.
10. A non-transitory computer readable storage medium storing computer instructions for causing a computer to perform the method of any one of claims 1 to 7.

Description

Ore formation prediction method, device, equipment and storage medium based on neural network Technical Field The disclosure relates to the technical field of data processing, in particular to an ore forming prediction method, device and equipment based on a neural network and a storage medium. Background With the continuous development of economy, the demands on mineral resources are increasingly abundant, and regional ore formation prediction is an important link of geological prospecting work. In general, whether a certain position point is a mine point is directly predicted, and the problem of inaccurate ore forming prediction results in an area exists. In view of this, how to improve the accuracy of the in-area ore formation prediction results is a technical problem to be solved. Disclosure of Invention Accordingly, an objective of the present disclosure is to provide an ore-forming prediction method, apparatus, device and storage medium based on a neural network, which are used for solving or partially solving the above technical problems. Based on the above object, a first aspect of the present disclosure proposes an ore-forming prediction method based on a neural network, the method comprising: Acquiring a multi-modal geological image of a target area; determining one-dimensional multi-modal data corresponding to the center point and two-dimensional multi-modal data of the neighborhood of the center point from the multi-modal geological image by utilizing a pre-trained ore-forming prediction model; Extracting local attribute features from the one-dimensional multi-mode data, and extracting spatial context features from the two-dimensional multi-mode data; Performing fusion processing on the local attribute characteristics and the spatial context characteristics to obtain target fusion characteristics; And determining the mining probability values of the central points based on the target fusion characteristics, and determining the mining prediction result of the target area according to the mining probability values of all the central points in the target area. Based on the same inventive concept, a second aspect of the present disclosure proposes an ore-forming prediction apparatus based on a neural network, comprising: An acquisition module configured to acquire a multi-modal geologic image of a target area; The geological data determining module is configured to determine one-dimensional multi-modal data corresponding to the center point and two-dimensional multi-modal data of the neighborhood of the center point from the multi-modal geological image by utilizing a pre-trained ore-forming prediction model; The feature extraction module is configured to extract local attribute features from the one-dimensional multi-mode data and extract spatial context features from the two-dimensional multi-mode data; The feature fusion module is configured to fuse the local attribute features and the spatial context features to obtain target fusion features; And the ore formation prediction module is configured to determine an ore formation probability value of the central point based on the target fusion characteristic, and determine an ore formation prediction result of the target area according to the ore formation probability values of all the central points in the target area. Based on the same inventive concept, a third aspect of the present disclosure proposes an electronic device comprising a memory, a processor and a computer program stored on the memory and executable by the processor, the processor implementing the method as described above when executing the computer program. Based on the same inventive concept, a fourth aspect of the present disclosure proposes a non-transitory computer-readable storage medium storing computer instructions for causing a computer to perform the method as described above. From the foregoing, it can be seen that the present disclosure provides a neural network-based ore-forming prediction method, apparatus, device, and storage medium. A multi-modal geologic image of the target region is acquired. And determining one-dimensional multi-mode data corresponding to the center point and two-dimensional multi-mode data of the neighborhood of the center point from the multi-mode geological image by utilizing a pre-trained ore-forming prediction model, wherein the one-dimensional multi-mode data can cover local detailed geological attributes corresponding to the center point, the two-dimensional multi-mode data can cover a larger range of spatial geological information of the neighborhood of the center point, and the one-dimensional multi-mode data and the two-dimensional multi-mode data have rich data dimensions and can accurately perform ore-forming prediction. The local attribute features are extracted from the one-dimensional multi-modal data, the spatial context features are extracted from the two-dimensional multi-modal data, the local attribute features can accurately c