A method for zoning ore fields based on regional uranium resources

Abstract

The present invention provides a method for zoning a mineral field based on regional uranium resources. The present invention is a method for utilizing regional uranium mines in a uranium mineral field to integrate resources based on a spatial clustering method. Specifically, the present invention adopts a two-step clustering method to achieve this. The first step is to cluster the uranium resources in each mineral deposit based on a three-dimensional block model of the mineral deposit to form clusters, and the number of clusters is much smaller than the number of block models. The second step is to cluster the mineral deposits based on the spatial distribution of the mineral deposits within the mineral field on the basis of the clustering of uranium resources in each mineral deposit to achieve mineral field zoning. After the mineral field is zoned using the method of the present invention, the mineral field resources can be fully utilized in the subsequent uranium mining process, the number of "idle mines" can be reduced, and after the regional uranium resources are integrated, the automation and intelligence level of the mining process and subsequent waste treatment can be improved, mining costs can be reduced, and environmental protection can be beneficial.

Description

Technical Field

[0001] The present invention relates to the technical field of uranium mining, in particular to a method for zoning an ore field based on regional uranium resources. Background Art

[0002] my country's hard rock regions are rich in uranium resources, with reliable reserves. The deposits are characterized by numerous, small, and scattered clusters, with a predominance of small and medium-sized deposits and a smaller proportion of large ones. Thousands of ore bodies exist within the ore fields, each with complex morphology and varying sizes, with varying numbers and spatial distribution characteristics.

[0003] At present, regional uranium resource mining adopts the traditional mining model of independent, scattered and small-scale mining, which has problems such as low utilization rate of mineral resources, high mining costs, and low level of automation and intelligence. Specifically, many "idle mines" are formed, making it difficult to form uranium resource aggregation and integration and increase production and expand capacity; each ore deposit has an independent mining model, the resource mining methods in the region are chaotic, there are multiple development and transportation systems, repeated investment in engineering construction, and high production costs, which restrict the improvement of the economic benefits of mines; it is not conducive to the centralization and minimization of waste, and brings many unfavorable factors to the environmental protection of ore deposit development. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for zoning a mine field based on regional uranium resources. The method integrates uranium resources within the mine field according to the resource domain endowment conditions, realizes the concentration of uranium resources in the region, optimizes the allocation of uranium resources, and lays the foundation for the overall planning of the mine field.

[0005] The present invention is achieved in that:

[0006] The method for zoning a mineral field based on regional uranium resources provided by the present invention comprises the following steps:

[0007] S1: Construct a three-dimensional block model of all ore bodies in the uranium resource-based ore field;

[0008] S2: extracting the three-dimensional block model attribute information of each ore body in the ore field as input data;

[0009] S3: Determine the adaptation parameters of each mineral deposit through the density-based clustering algorithm DBSCAN. The adaptation parameters of each mineral deposit include the neighborhood radius and the minimum number of points within the neighborhood radius to become a core object;

[0010] S4: Using the density-based clustering algorithm DBSCAN to cluster the ore bodies in each deposit, the clustering results of the ore bodies in the deposit are obtained;

[0011] S5: Using the clustering results of the ore bodies in each ore deposit in step S4 as input data, cluster the ore deposits in the ore field through the hierarchical clustering algorithm HAC, and finally realize the ore field zoning based on regional uranium resources.

[0012] Preferably, step S2 specifically comprises: using mining software to extract the three-dimensional spatial coordinates, grade, and size of the three-dimensional block model of each ore body of the ore deposit as input data.

[0013] Preferably, step S3 includes the following steps:

[0014] S301: Selecting the three-dimensional spatial coordinates and the grade of the three-dimensional block model to form four-dimensional attribute information data, and determining the k value in the k-nearest neighbor algorithm based on the four-dimensional attribute information data;

[0015] S302: Calculate the distance from each 3D block model to the k-th closest point, sort these distances from large to small, and draw a k-distance graph;

[0016] S303: Find the distance of the location where the change trend changes sharply from the k-distance graph, and use it as the value of the neighborhood radius;

[0017] S304: Set the minimum number of points within the neighborhood radius that become core objects to be k+1;

[0018] S305: Traverse all the ore deposits in the ore field according to the above steps, and determine the adaptation parameters of each ore deposit in the ore field based on the DBSCAN algorithm.

[0019] More preferably, in step S301 , the value of k is: k=2×dimension-1.

[0020] Preferably, step S4 includes the following steps:

[0021] S401: Bringing the adaptation parameters of each ore deposit into the DBSCAN algorithm for solution, and obtaining clusters based on the three-dimensional block model;

[0022] S402: Determine the central spatial coordinates, average grade, number of blocks included, and resource weight attribute indicators of the ore body clusters of each ore deposit, where the resource weight is the sum of the product of the volume of each block model included in the cluster and the corresponding block grade.

[0023] Preferably, step S5 includes the following steps: using the results of ore body clustering within each ore deposit as input data, using the Ward method in the linkage function of the hierarchical clustering algorithm HAC to perform resource spatial clustering within the ore field to obtain ore deposit aggregation results.

[0024] More preferably, the Ward method is to find the best cluster merging method by minimizing the sum of variances within clusters, specifically:

[0025] ① Calculate the ESS of each cluster;

[0026] ② Enumerate all possible cases of clusters being grouped in pairs, and calculate the increment of ESS after merging all possible two clusters;

[0027] ③ Select the two clusters with the smallest incremental △ESS and merge them;

[0028] ④ Repeat the above process until the number of clusters is reduced to 1;

[0029] The calculation formula of ESS is as follows.

[0030]

[0031] Where, X ij represents the jth eigenvalue of the i-th cluster, is the mean of the j-th dimension feature, ρ is the feature dimension, and n is the number of clusters.

[0032] The present invention realizes the resource integration and utilization of regional uranium mines in a uranium ore field based on a spatial clustering method. Specifically, a two-step clustering method is adopted to achieve this. The first step is to cluster the uranium resources in each ore deposit based on a three-dimensional block model of the ore deposit to form clusters, and the number of clusters is much smaller than the number of block models. The second step is to cluster the ore deposits based on the spatial distribution of the ore deposits within the ore field on the basis of the clustering of uranium resources in each ore deposit to realize ore field zoning.

[0033] After the mine field is divided into zones using the method of the present invention, the mine field resources can be fully utilized in the subsequent uranium mining process, the number of "idle mines" can be reduced, and after the regional uranium resources are integrated, the automation and intelligence level of the mining process and subsequent waste treatment can be improved, thereby reducing mining costs and benefiting environmental protection.

[0034] The present invention can realize the reasonable zoning and integration of regional uranium resources, and provide technical support for the unified planning and development of regional uranium resources and the improvement of resource utilization level. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flow chart of the method for zoning a mineral field based on regional uranium resources of the present invention.

[0036] Figure 2 It is the k-distance graph of the No. 10 ore deposit in the embodiment of the present invention.

[0037] Figure 3 This is a tree structure diagram formed after spatial clustering of mineral resources in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The method for zoning a uranium field based on regional resources, provided by this invention, consists of two phases: the first phase involves the initial aggregation of ore bodies within a deposit, and the second phase involves the clustering of ore deposits within the field. In the first phase, ore bodies within each deposit are clustered using the DBSCAN clustering method to identify densely populated areas. The output of the first phase serves as input for the second phase, which uses the HAC clustering algorithm to cluster and partition the field.

[0039] Combine Figure 1 The steps of the method for zoning a mineral field based on regional uranium resources provided by the embodiment of the present invention are as follows:

[0040] S1: Use mining software (3DMine) to construct a three-dimensional block model of all ore bodies in the uranium resource-based ore field.

[0041] S2: Use mining software to extract the spatial coordinates (x, y, z), grade (u), and size (size_x, size_y, size_z) data of the ore block model of each ore deposit in turn.

[0042] S3: The DBSCAN algorithm (density-based clustering algorithm) is used to determine the adaptation parameters of each mineral deposit. The adaptation parameters of each mineral deposit include the neighborhood radius and the minimum number of points within the neighborhood radius to become a core object.

[0043] Each ore body within a deposit is considered a point. The distance from each point to its k-th closest point is calculated (k-nearest neighbor algorithm). These distances are then sorted from largest to smallest and plotted as a k-distance graph. The distance to the location where the trend changes sharply is found in the k-distance graph, which serves as the neighborhood radius (Eps). If a point's neighborhood radius has a sufficient number of points (at least MinPts, denoted as the threshold), the area is considered high-density and can be expanded into a cluster. Cluster expansion is then performed using density-connected points. MinPts is used here to determine whether a point is a core object. For example, if MinPts is set to 4, then a point is considered a core object if there are at least four points close to it. If there are fewer than four points near a point, it is considered a non-core object.

[0044] Based on the characteristics of uranium resources and the block model attribute information, four attributes, namely three-dimensional spatial coordinates (x, y, z) and grade (u), are selected in the DBSCAN algorithm clustering process in this embodiment, that is, the dimension is 4; and the value of k is usually k=2×dimension-1, so k=7.

[0045] Generally speaking, MinPts ≥ the dimension of the data set + 1. In this embodiment, MinPts = k + 1; since the value of k is 7, MinPts = 8.

[0046] In this example, block models of 21 ore deposits in the mining field were selected, and the DBSCAN algorithm adaptation parameters (domain radius and threshold) of the 21 ore deposit block models were obtained according to the above method, as shown in Table 1 below.

[0047] Table 1 Algorithm adaptation parameters for the 21 ore body models of the present embodiment

[0048]

[0049] like Figure 2 As shown, Figure 2 This is the k-distance diagram of deposit No. 10 when k=7. Figure 2 It can be seen that the distance of the inflection point in the k-distance graph is the value of Eps. Figure 2 The middle inflection point position is 2, the inflection point value is 4.0, so Eps=4.0.

[0050] S4: After obtaining the DBSCAN algorithm adaptation parameters for 21 mineral deposits, the DBSCAN algorithm is used to cluster the ore bodies within the mineral deposits, generate clusters, and greatly reduce the number of blocks. After clustering the mineral deposits in turn, the central spatial coordinates of the ore body clusters of each mineral deposit are obtained. , average grade, number of included blocks (count), resource weight (weight), and other attribute information, resulting in a total of 4076 clusters of varying sizes. The resource weight (weight) is the sum of the product of the volume of each block model contained in the cluster and the corresponding block grade.

[0051] S5: The clustering results for each ore body in the first stage are used as input for the second stage clustering, using the Hierarchical Agglomerative Clustering (HAC) algorithm. The HAC algorithm uses the linkage function to solve the model. linkage() is a function model used to generate a clustering hierarchy. Its input data is preprocessed data. Hierarchical clustering algorithms often rely on data scale, so data is usually normalized before clustering. The output of the hierarchical clustering algorithm is a hierarchical structure of clustering results.

[0052] The core parameter of the HAC algorithm is the similarity metric, which is used to specify the criteria for merging clusters in hierarchical clustering. The Ward method seeks the optimal inter-cluster merging strategy by minimizing the sum of variances within all clusters. This makes it tend to generate spherical clusters of similar size, which can maximize clustering quality. Therefore, the HAC algorithm of the present invention chooses to use Ward to determine cluster merging. The Ward method minimizes the total variance of each cluster as much as possible, so it will choose to merge the two clusters that minimize the intra-cluster variance, ultimately generating more uniform clusters.

[0053] At the same time, the present invention chooses to use the Euclidean distance measurement method in combination with the Ward similarity measurement method, and the two are combined to calculate the square error distance between clusters.

[0054] The Ward method minimizes intra-cluster variance. At each step, it considers every possible combination of two clusters and merges the two clusters that minimize the increase in the error sum-of-squares criterion (ESS). Here, the increase in the error sum-of-squares is represented by ΔESS. For example, if clusters A and B are merged, the increase in the error sum-of-squares criterion ΔESS is:

[0055] △ESS=ESS A∪B -(ESS A +ESS B )

[0056] ESS A∪B It represents the sum of squared errors after cluster A and cluster B are merged, ESS A and ESS B Represent the sum of squared errors of cluster A and cluster B respectively.

[0057] The optimal control of intra-cluster variance is achieved by minimizing △ESS.

[0058] The calculation formula of ESS is as follows:

[0059]

[0060] Where, X ij represents the jth eigenvalue of the i-th cluster, is the mean of the j-th dimension feature, ρ The dimension used in this embodiment is the central space coordinate and resource weight, so ρ is 4; n is the number of clusters. ESS is essentially the sum of the squares of the Euclidean distances from a cluster to its mean point, reflecting the degree of dispersion of the data in the feature space.

[0061] The Ward method requires that the increment of ESS after each merger is the smallest. The specific steps are as follows:

[0062] ① Calculate the ESS of each cluster. Initially, each cluster is a separate class, and the ESS of all clusters is 0.

[0063] ② Group the clusters into pairs and enumerate all possible grouping scenarios. When the number of clusters is n, there are n×(n-1) / 2 possible pairwise grouping scenarios. Calculate the increment △ESS of the ESS after merging all possible two clusters.

[0064] ③ Select the two clusters with the smallest incremental △ESS to merge, and use the merged cluster as the new cluster.

[0065] ④ Repeat the above process until the number of clusters n is reduced to 1.

[0066] By traversing and calculating the △ESS after merging every two clusters, we find the two clusters with the smallest △ESS after merging, connect the clusters to form a new cluster, and describe the cluster by the maximum distance required to connect each part. Different clusters are formed at different distances, which can be presented using a dendrogram. In the dendrogram, the y-axis marks the distance of cluster mergers, and the deposits are placed along the x-axis to prevent clusters from mixing.

[0067] Based on the results of the first phase of internal resource clustering in the mining area, the HAC algorithm is used to perform spatial clustering of resources within the mining area to obtain the clustering results of the mineral deposits. The tree structure formed by the final clustering is as follows: Figure 3 As shown. Figure 3 As can be seen from the tree diagram, the mineral resources are roughly divided into two parts. Considering the requirements of subsequent production scale and intensive development, the optimal partitioning scheme is to select four clustering areas (as shown by the dotted lines in the figure).

[0068] In the embodiment of the present invention, the HAC clustering algorithm is used to form a tree structure, and the 21 mineral deposits in the mining field are finally integrated into 4 areas, as shown in Table 2 below.

[0069] Table 2 HAC clustering results

[0070]

[0071] Depend on Figure 3 As shown in Table 2, the two-step clustering method of the present invention can eventually realize the partition clustering of uranium resource fields, which lays the foundation for subsequent mining and intelligent processing.

Claims

1. A method for zoning a mine field based on regional uranium resources, characterized in that: The steps include: S1: Construct a three-dimensional block model of all ore bodies in the uranium resource-based ore field; S2: extracting the three-dimensional block model attribute information of each ore body in the ore field as input data; S3: Determine the adaptation parameters of each mineral deposit through the density-based clustering algorithm DBSCAN. The adaptation parameters of each mineral deposit include the neighborhood radius and the minimum number of points within the neighborhood radius to become a core object; S4: Using the density-based clustering algorithm DBSCAN to cluster the ore bodies in each deposit, the clustering results of the ore bodies in the deposit are obtained; S5: using the clustering results of the ore bodies in each ore deposit in step S4 as input data, clustering the ore deposits in the ore field through the hierarchical clustering algorithm HAC, and finally realizing the ore field zoning based on regional uranium resources; Step S2 specifically comprises: using mining software to extract the three-dimensional spatial coordinates, grade, and size of the three-dimensional block model of each ore body of the ore deposit as input data; Step S3 includes the following steps: S301: Selecting the three-dimensional spatial coordinates and grade of the three-dimensional block model to form four-dimensional attribute information data, and determining the k value in the k-nearest neighbor algorithm based on the four-dimensional attribute information data; k=2×dimension-1; S302: Calculate the distance from each 3D block model to the k-th closest point, sort these distances from large to small, and draw a k-distance graph; S303: Find the distance of the location where the change trend changes sharply from the k-distance graph, and use it as the value of the neighborhood radius; S304: Set the minimum number of points within the neighborhood radius that become core objects to be k+1; S305: Traverse all the ore deposits in the ore field according to the above steps, and determine the adaptation parameters of each ore deposit in the ore field based on the DBSCAN algorithm; Step S4 includes the following steps: S401: Bringing the adaptation parameters of each ore deposit into the DBSCAN algorithm for solution, and obtaining clusters based on the three-dimensional block model; S402: Determine the central spatial coordinates, average grade, number of blocks included, and resource weight attribute indicators of each ore body cluster of each deposit, where the resource weight is the sum of the product of the volume of each block model included in the cluster and the corresponding block grade; Step S5 includes the following steps: using the clustering results of the ore bodies in each ore deposit as input data, using the Ward method in the linkage function of the hierarchical clustering algorithm HAC to perform spatial clustering of uranium resources within the ore field to obtain the ore deposit clustering results; The Ward method is to find the best cluster merging method by minimizing the sum of variance within the cluster. Specifically: ① Calculate the ESS of each cluster; ② Enumerate all possible clusters and group them into pairs, and calculate the increment of ESS after merging all possible two clusters. ③ Select the two clusters with the smallest incremental △ESS and merge them; ④ Repeat the above process until the number of clusters is reduced to 1; The calculation formula of ESS is as follows: Where, X ij represents the jth eigenvalue of the i-th cluster, is the mean of the j-th dimension feature, ρ is the feature dimension, n is the number of clusters; the feature dimension is four-dimensional, that is ρ =4; the four feature dimensions are: three dimensions corresponding to the central space coordinates obtained in step S402 and one dimension corresponding to the resource weight.

Images