Construction And Application Of 3D Training Image Of Multiple Point Geostatistics Oriented Mineral Resource Reserve Calculation

The mineral resources are abundant and various in China.However,the per capita possession is small.The exploitation and application of mineral resources are the key in the field of mineral resources.The resource reserves and spatial distribution of orebodies is helpful to rationally develop and utilize mineral resources and save manpower and material resources.Therefore,accurately calculating the resource reserves and spatial distribution of orebodies on the basis of exploration data before mining is necessary.At present,all existing calculation methods have defects such as inability to visualize the inner spatial structure of ore bodies,accurately calculate the reserves of ore bodies and the need for manual intervention.On the basis of the multi-point geostatistics,the resource reserves and spatial distribution of orebodies are calculated in paper.In multi-point geostatistics,the training image contains the geological knowledge of the research object,which is very important for the reserve calculation of orebody.There are few methods to construct and apply training images for orebodies.Therefore,the construction and application of3 D training image of multiple point geostatistics oriented mineral resource reserve calculation are proposed.The methods include the three-dimensional(3D)training images construction of orebody on the basis of borehole density and the property modeling of orebody on the basis of 3D training images.Compared with traditional methods,the calculation accuracy of the methods is improved.The main research results are as follows.(1)3D training images construction of orebody on the basis of borehole densityFirstly,three-dimensional geological modeling(3DGM)is used to process exploration data before mining to construct 3D geological model,which provides the basis for constructing 3D training images.Secondly,on the basis of borehole density,the 3D geological model is cut and meshed to construct the voxel model which accurately represents the internal spatial structure and boundary shape of orebody.Thirdly,the property characteristics of orebodies are obtained by interpolation methods on the basis of borehole data,that is,the property model is constructed.Finally,according to the geographical law,the property assignment method on the basis of Euclidean distance is used to assign the property characteristics of the property model to the voxel model,thereby constructing the 3D training images of orebody.3D training images of orebody accurately represent the spatial structure and property characteristics of ore bodies on the basis of borehole density,which provides accurate geological knowledge for calculating resource reserves and spatial distribution of orebodies.(2)The property modeling of orebody on the basis of 3D training imagesFirstly,the 3D geological model is divided by tetrahedral mesh generation method,and the property of boreholes is filled into the divided 3D geological model to construct the unevaluated voxel model which represents the spatial distribution of borehole in the orebody.Secondly,the voxel-oriented data structure and the data template with N-order neighbor voxels are defined for the unevaluated voxel model and the 3D training image of orebody.The data template and data structure provide the support for simulating the property of unevaluated voxels.Thirdly,the selfdefined data template is used to mine the 3D training image to construct the search tree,and the local conditional probability of the unevaluated voxel is obtained from the search tree,thereby the properties of unevaluated voxels are obtained.Finally,on the basis of the correlation degree among the regions where the 3D training images are located and the unevaluated voxel,the properties are fused to obtain the property of the unevaluated voxel,thereby constructing the property-voxel model representing properties of orebody.(3)Case analysis of calculating mineral resource reserves on the basis of multiple point geostatisticsAccording to the exploration data before mining in Heiniudong mining area,the grade-voxel model of I-3 orebody is constructed.According to the copper grade of voxels in the grade-voxel model,the copper reserves of voxels are calculated and added to obtain the copper reserves of orebodies.The 3D training images with a discrete step size of 5.3 meters,unevaluated voxel model with a discrete step size of 5 meters and data template with 8-order voxels are used.Comparing the average copper grade of I-3 orebody calculated by the paper method with the average copper grade of 2.90% after mining,the difference between them is small,which indicates that the method proposed in paper can be applied to the calculation of mineral resources reserves with high accuracy.In addition,the average grade of copper in I-3 orebody calculated by Kriging method is3.09%.Comparing the results of the two methods,the accuracy of the proposed method is higher than that of Kriging method.The discrete step size of the 3D training image and unevaluated voxel model and the order of the data template are discussed and analyzed.Multi-point geostatistics is introduced into the field of mineral resources reserves calculation,3D training images of orebody are constructed,the property modeling of orebody on the basis of3 D training images is performed,and the case analysis of calculating mineral resource reserves on the basis of multiple point geostatistics is made.The reserves and average grade of copper in I-3orebody are calculated accurately,and the accuracy of this method is higher than that of Kriging method.The proposed method expands the application field of multi-point geostatistics,provides a new idea for accurately calculating the reserves of orebody,enriches the calculation method system of reserves,and has theoretical significance and application value in the mining field.