Ore rock mining and stripping amount acceptance method fusing point cloud-spectral data technology

Abstract

The invention provides an ore rock mining and stripping amount acceptance method fusing a point cloud-spectral data technology. The ore rock mining and stripping amount acceptance method comprises five steps of data acquisition, data preprocessing, stope mining and stripping volume calculation, ore rock boundary model establishment and ore rock mining and stripping amount calculation. The method specifically comprises the steps that firstly, an unmanned aerial vehicle is used for collecting stope point cloud data and stope spectral data, and meanwhile the collected point cloud-spectral data are preprocessed; secondly, the total mining and stripping volume of the ore rocks is calculated through the preprocessed stope point cloud data, finally, the mining and stripping volume of the stopes is calculated through an ore rock boundary model established by fusing spectral data, and support is provided for acceptance of the mining and stripping volume of the ore rocks. According to the method, spectral data information is fused, multiple algorithms such as progressive morphological filtering, the convolutional neural network and point cloud voxelization are nested, rapid, efficient and accurate calculation of different surface mine-rock volumes can be rapidly achieved, and the method has important engineering significance and theoretical value for acceptance of stope mining and stripping volumes and construction of intelligent mines.

Description

Technical Field

[0001] The present invention relates to the field of acceptance of mineral-rock mining volume in mines, and in particular to a method for acceptance of mineral-rock mining volume integrating point cloud-spectral data technology. Background Art

[0002] The acceptance calculation of the stripping volume of open-pit mines is an essential part of mining operations. Efficient and accurate acceptance helps to timely confirm whether the actual mining volume is consistent with the plan, so as to make rapid adjustments in the face of geological changes or production problems, and ensure the safety and stability of the mining process. At the same time, the efficient calculation of the ore-rock stripping volume of open-pit mines also facilitates the owner to reasonably settle the mining remuneration for the construction party in a timely manner, directly affecting the economic benefits and production efficiency of mining, and promoting the efficiency of open-pit mining.

[0003] At present, the acceptance of stripping volume in open-pit mines basically adopts the traditional ground measurement method, which mainly relies on surveyors to use measuring instruments to measure the mining area. However, this method has a large workload, is time-consuming, and has relatively low measurement accuracy. In addition, the judgment of the mine-rock geology of the mining site requires on-site investigation and survey by professionals. This survey method is not only inefficient, but also subject to greater subjective influence, which seriously affects the efficiency of acceptance.

[0004] With the development of computers and science and technology, the potential of drone aerial survey in mine volume acceptance has been gradually tapped. For example, drone-mounted lidar lenses can quickly and efficiently obtain large-area terrain data, generate high-resolution digital elevation models and three-dimensional terrain maps through aerial photography, and significantly improve the speed and accuracy of data acquisition. Compared with traditional ground surveying, drone aerial survey reduces the need for manual operation and reduces human errors. At the same time, it can work stably under various weather conditions, ensuring the continuity and consistency of data. In addition, drone aerial survey also has strong flexibility, can cover complex terrain areas, provide comprehensive perspectives and details, and provide reliable basic data for mine volume acceptance, optimizing the acceptance process. In the establishment of the mine-rock boundary model in the stope, the rich information content of hyperspectral lens data can meet the requirements for the establishment of the mine-rock model.

[0005] Based on this, adhering to the concept of smart mines and green mines, a fast, efficient and intelligent method for accepting the mining volume of ore and rock stripping that integrates point cloud-spectral data technology is urgently needed. This method has important theoretical and engineering significance for the efficient and intelligent acceptance of mining stripping. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present invention proposes a method for accepting and inspecting the mining volume of ore and rock by integrating point cloud-spectral data technology to solve the engineering problems described in the above background.

[0007] The present invention solves the problems of large workload and long time consumption in the acceptance calculation of the stripping volume of traditional open-pit mines, and is achieved through the following technical solutions.

[0008] A method for accepting the mining volume of ore and rock by integrating point cloud-spectral data technology, characterized in that the method is divided into the following steps:

[0009] S1: Data collection. It is divided into stope point cloud data information and stope spectral data information. Stope point cloud data information is used to store and display the current three-dimensional form of the stope. Spectral data information is mainly the spectral curve of the entire stope, which is used to invert and establish the ore-rock boundary model of the stope area;

[0010] S2: Data preprocessing. Point cloud data preprocessing mainly includes the removal of abnormal points and noise points, and the removal of mine cars. The preprocessing methods of spectral data information include smoothing, denoising, baseline correction, and normalization of the spectral curve of the mining area;

[0011] S3: Calculation of mining volume of the stope. Unify the coordinate system of the two phases of point cloud, perform difference on the three-dimensional voxelized model presented by the pre-processed two phases of point cloud data, and then calculate the difference in volume of the mining area, that is, the volume of the acceptance mining area;

[0012] S4: Establishment of the mine-rock boundary model. Based on the pre-processed mine spectral data information, the characteristic bands of the mine-rock medium are extracted and the mine-rock boundary inversion model is established;

[0013] S5: Calculation of ore and rock stripping volume. The ore-rock boundary model established by integrating point cloud-spectral data information assigns ore-rock density attributes to each stripping area, and calculates the ore and rock stripping volume of the stope respectively.

[0014] Furthermore, the data collection in S1 is divided into mining site point cloud data and spectral data. Before data information collection, the onboard lens, shooting mode, flight altitude, take-off speed, route speed, lateral / heading overlap rate, return altitude, etc. need to be set. The mining site point cloud data needs to be collected by an airborne laser radar lens, and the spectral data needs to be collected by an airborne imaging spectral lens. The point cloud data information is used to calculate the acceptance volume of the mining area of ​​the mining site, and the spectral data information is used to demarcate the ore-rock boundary of the mining site.

[0015] Furthermore, the point cloud data preprocessing methods in S2 are mainly k-means clustering algorithm and progressive morphological filtering algorithm, and the spectral data preprocessing methods are mainly principal component analysis (PCA), convolutional neural network (CNN), spectral registration algorithm, etc.

[0016] Furthermore, the k-means clustering algorithm execution process is as follows:

[0017] S201: randomly select k samples from the original point cloud data as initial clustering centers;

[0018] S202: Calculate the Euclidean distance d between k center points of different initial cluster centers i , assign each point cloud to the nearest cluster center, and the Euclidean distance is expressed as follows:

[0019]

[0020] S203: Calculate the average coordinates of each cluster center point twice, and use the average value as the new center point of the cluster center. The center point distance expression is:

[0021]

[0022] S204: S202 and S203 are repeated in a loop, and the iteration ends when the center point no longer changes or the preset number of iterations is reached.

[0023] Furthermore, the principle of convolutional neural network (CNN) processing spectral images is:

[0024] First, input the three-dimensional tensor [H, W, N] as the input form of spectral image data, H and W represent the height and width of the stope image, N represents the number of spectral bands in the stope area, and each pixel has an N-dimensional spectral vector;

[0025] Secondly, a two-dimensional convolution kernel (such as 3×3 or 5×5) is used to slide on the spatial dimensions (H and W) to extract the spatial features of the image;

[0026] Then, the stope spectral image is pooled and the stope feature image is downsampled. After the convolution and pooling layers, the extracted multi-dimensional features are flattened into a one-dimensional vector and input into the fully connected layer.

[0027] Finally, an activation function such as Softmax is used to convert the output of the fully connected layer into a classification probability or regression value.

[0028] Furthermore, the steps of establishing the S3 three-dimensional voxelized model and calculating the mining volume are as follows:

[0029] S301: determining the overall boundary of the stope point cloud by using the MAX function, and establishing a minimum rectangular parallelepiped model that can contain the overall boundary of the stope;

[0030] S302: Set the size of the voxel grid of the stope point cloud model, and remove the voxel model outside the stope point cloud. x 、voxel y 、voxel z The dimensions are as follows:

[0031]

[0032] Among them, (x, y, z) represents the coordinates of the point, and voxel_size represents the side length of the voxel.

[0033] S303: Differentiate the point cloud data of the two periods to calculate the mining volume of the stope in that period.

[0034] Furthermore, the stope spectral data information preprocessed in S4 is mainly acquired through an imaging spectral lens and used to establish a stope ore-rock boundary model.

[0035] Furthermore, before calculating the mining volume of the mine rock in the S5, the coordinate system of the mine-rock boundary model and the mine point cloud data must be unified.

[0036] Furthermore, the method for accepting the mining volume of ore and rock by integrating point cloud-spectral data technology is characterized in that the mining volume calculation steps of the stope are as follows:

[0037] S501: Invert and establish a mine-rock boundary model through spectral data information;

[0038] S502: Align the stope ore-rock boundary model coordinate system with the point cloud data coordinate system;

[0039] S503: Fusing point cloud-spectral ore-rock boundary model data, and giving relevant density attributes to the stope geological model;

[0040] S504: Calculate the mining volume of ore and rock in the mining area respectively.

[0041] The present invention provides a method for accepting the mining volume of ore and rock by integrating point cloud-spectral data technology, which has the following technical effects:

[0042] 1. Based on point cloud-spectral data technology, a mine-rock classification acceptance method suitable for various open-pit mines is realized, which greatly reduces the time cost and labor cost of the traditional acceptance mode, thereby improving economic benefits;

[0043] 2. Compared with the traditional acceptance method of manual ground measurement, this method greatly improves the accuracy of ore-rock acceptance during mining, which helps to timely confirm whether the actual mining volume is consistent with the plan, so as to make rapid adjustments in the face of geological changes or production problems, and ensure the safety and stability of the mining process;

[0043] 3. The establishment of the geological model of the mining area does not require on-site inspection by professional geologists, thus reducing the workload of on-site staff;

[0044] 4. It provides important reference for the ore-rock stripping acceptance work of similar open-pit mines, and has important theoretical and engineering significance for the construction of green mines, intelligent mines and ore-rock stripping acceptance. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0046] Figure 1 It is a flow chart of a method for accepting and checking the mining volume of ore and rock by integrating point cloud-spectral data technology according to the present invention;

[0047] Figure 2 The original point cloud information image of the mine collected in Example 1;

[0048] Figure 3 This is the image of the point cloud information after preprocessing in Example 1;

[0049] Figure 4 It is a model diagram of the ore-rock boundary inverted by the imaging spectrum data in Example 1; DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.

[0052] The purpose of the present invention is to provide a method for accepting and checking the mining volume of ore and rock by integrating point cloud-spectral data technology.

[0053] Embodiment 1:

[0054] Take the mining and rock stripping acceptance of an open-pit mine in Anshan City as an example.

[0055] S1: Data collection. It is divided into stope point cloud data information and stope spectral data information. Stope point cloud data information is used to store and display the current three-dimensional form of the stope. Spectral data information is mainly the spectral curve of the entire stope, which is used to invert and establish the ore-rock boundary model of the stope area;

[0056] S2: Data preprocessing. Point cloud data preprocessing mainly includes the removal of abnormal points and noise points, and the removal of mine cars. The preprocessing methods of spectral data information include smoothing, denoising, baseline correction, and normalization of the spectral curve of the mining site. The generation of abnormal points and noise points will affect the accuracy of the establishment of the mining site point cloud voxel model. The presence of mine cars will affect the accuracy of the calculation of the mining volume and ore-rock acceptance volume. The smoothing, denoising, baseline correction, and normalization of the spectral curve of the mining site are to improve the accuracy of the establishment of the ore-rock boundary model.

[0057] S3: Calculation of mining volume of the mining area. Unify the coordinate system of the two phases of point cloud, and perform differential analysis on the three-dimensional voxelized models presented by the two phases of point cloud data after preprocessing, that is, perform differential analysis on the voxelized mining area point cloud elevation data of the nth phase and the n+1th phase, and then calculate the difference in the volume of the mining area, that is, the volume of the mining area to be accepted;

[0058] S4: Establishment of the mine-rock boundary model. Based on the pre-processed mine spectral data information, the characteristic bands of the mine-rock medium are extracted and the mine-rock boundary inversion model is established;

[0059] S5: Calculation of ore and rock stripping volume. The ore-rock boundary model established by integrating point cloud-spectral data information assigns ore-rock density attributes to each stripping area, and calculates the ore and rock stripping volume of the stope respectively.

[0060] Data collection is divided into mining site point cloud data and spectral data. Before data information collection, it is necessary to set the onboard lens (L1 lens and imaging spectrum lens), photo mode (equal interval continuous shooting mode), flight altitude (120m), take-off speed (10m / s), route speed (12m / s), lateral / heading overlap rate (80%, 80%), return altitude (120m), etc. The mining site point cloud data needs to be collected by the airborne laser radar lens (L1), and the spectral data needs to be collected by the airborne imaging spectrum lens. The point cloud data information is used to calculate the acceptance volume of the mining area of ​​the mining site, and the spectral data information is used for the mine-rock boundary of the mining site.

[0061] The collected original point cloud information imaging of the mine is as follows Figure 2 shown.

[0062] The point cloud data preprocessing methods mainly include k-means clustering algorithm and progressive morphological filtering algorithm. The spectral data preprocessing methods mainly include principal component analysis (PCA), convolutional neural network (CNN), spectral registration algorithm, etc.

[0063] This embodiment uses the k-means clustering algorithm to remove noise points and the progressive morphological filtering algorithm to separate ground points from non-ground points, thereby eliminating non-ground points such as mining cars, personnel, and electric poles.

[0064] The k-means clustering algorithm removes noise execution process as follows:

[0065] S201: randomly select k samples from the original point cloud data as initial clustering centers;

[0066] S202: Calculate the Euclidean distance d between k center points of different initial cluster centers i , assign each point cloud to the nearest cluster center, and the Euclidean distance is expressed as follows:

[0067]

[0068] S203: Calculate the average coordinates of each cluster center point twice, and use the average value as the new center point of the cluster center. The center point distance expression is:

[0069]

[0070] S204: S202 and S203 are repeated in a loop, and the iteration ends when the center point no longer changes or the preset number of iterations is reached.

[0071] The point cloud information image after preprocessing is shown in the figure Figure 3 shown.

[0072] The principle of convolutional neural network (CNN) processing spectral images is:

[0073] First, input the three-dimensional tensor [H, W, N] as the input form of spectral image data, H and W represent the height and width of the stope image, N represents the number of spectral bands in the stope area, and each pixel has an N-dimensional spectral vector;

[0074] Secondly, a two-dimensional convolution kernel (such as 3×3 or 5×5) is used to slide on the spatial dimensions (H and W) to extract the spatial features of the image;

[0075] Then, the stope spectral image is pooled and the stope feature image is downsampled. After the convolution and pooling layers, the extracted multi-dimensional features are flattened into a one-dimensional vector and input into the fully connected layer.

[0076] Finally, an activation function such as Softmax is used to convert the output of the fully connected layer into a classification probability or regression value.

[0077] The steps for establishing a 3D voxelized model and calculating the excavation volume in S3 are as follows:

[0078] S301: determining the overall boundary of the stope point cloud by using the MAX function, and establishing a minimum rectangular parallelepiped model that can contain the overall boundary of the stope;

[0079] S302: Set the size of the voxel grid of the stope point cloud model, and remove the voxel model outside the stope point cloud. x 、voxel y 、voxel z The dimensions are as follows:

[0080]

[0081] Among them, (x, y, z) represents the coordinates of the point, and voxel_size represents the side length of the voxel.

[0082] In this embodiment, the side length of the voxel is set to 0.1 m.

[0083] S303: Differentiate the point cloud data of the two periods to calculate the mining volume of the stope in that period.

[0084] The preprocessed stope spectral data information in S4 is mainly obtained through imaging spectral lens and used to establish the stope ore-rock boundary model.

[0085] Before calculating the mining volume of the mine in S5, the coordinate system of the mine-rock boundary model and the mine point cloud data must be unified.

[0086] A method for accepting the mining volume of ore and rock that integrates point cloud-spectral data technology is characterized in that the mining volume calculation steps of the stope are as follows:

[0087] S501: Invert and establish a mine-rock boundary model through spectral data information;

[0088] S502: Align the stope ore-rock boundary model coordinate system with the point cloud data coordinate system;

[0089] S503: Establishing a mine-rock boundary model by integrating point cloud and spectrum, and giving relevant density attributes to the mining field geological model;

[0090] The ore-rock boundary models in different areas are shown below:

[0091] Table 1

[0092]

[0093] S504: Calculate the mining volume of ore and rock in the mining area respectively.

[0094] The ore-rock boundary model based on point cloud-spectrum fusion is shown in Figure 2. Figure 4 shown.

[0095] A method for accepting ore and rock mining volume that integrates point cloud and spectral data technology. The method calculates the total mass of mineral products in that month as 1558023.9t, while the original method actually measured the total mass of mineral products as 1526524.2t, with a relative error of 2.06%; the method calculates the total mass of rock as 4923522.1t, while the original method actually measured the total mass of rock as 5012115.6t, with a relative error of -1.77%, and the total error of ore and rock mining volume measured by the method is -0.87%.

[0096] Meet the requirements of mining and stripping acceptance.

[0097] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the system disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0098] This article uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only used to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for accepting the mining volume of ore and rock by integrating point cloud-spectral data technology, characterized in that: The method is divided into the following steps: S1: Data collection. It is divided into stope point cloud data information and stope spectral data information. Stope point cloud data information is used to store and display the current three-dimensional form of the stope. Spectral data information is mainly the spectral curve of the entire stope, which is used to invert and establish the ore-rock boundary model of the stope area; S2: Data preprocessing. Point cloud data preprocessing mainly includes the removal of abnormal points and noise points, and the removal of mine cars. The preprocessing methods of spectral data information include smoothing, denoising, baseline correction, and normalization of the spectral curve of the mining area; S3: Calculation of mining volume of the stope. Unify the coordinate system of the two phases of point cloud, perform difference on the three-dimensional voxelized model presented by the pre-processed two phases of point cloud data, and then calculate the difference in volume of the mining area, that is, the volume of the acceptance mining area; S4: Establishment of the mine-rock boundary model. Based on the pre-processed mine spectral data information, the characteristic bands of the mine-rock medium are extracted and the mine-rock boundary inversion model is established; S5: Calculation of ore and rock stripping volume. The ore-rock boundary model established by integrating point cloud-spectral data information assigns ore-rock density attributes to each stripping area, and calculates the ore and rock stripping volume of the stope respectively.

2. According to claim 1, a method for accepting the amount of ore and rock stripping that integrates point cloud and spectral data technology is characterized in that: The data collection in S1 is divided into mining site point cloud data and spectral data. Before data information collection, the onboard lens, shooting mode, flight altitude, take-off speed, route speed, lateral / heading overlap rate, return altitude, etc. need to be set. The mining site point cloud data needs to be collected by an airborne laser radar lens, and the spectral data needs to be collected by an airborne imaging spectral lens. The point cloud data information is used to calculate the acceptance volume of the mining area of ​​the mining site, and the spectral data information is used to demarcate the ore-rock boundary of the mining site.

3. The method for accepting the mining volume of ore and rock by integrating point cloud and spectral data technology according to claim 1 is characterized in that: The point cloud data preprocessing methods in S2 mainly include k-means clustering algorithm and progressive morphological filtering algorithm, and the spectral data preprocessing methods mainly include principal component analysis (PCA), convolutional neural network (CNN), spectral registration algorithm, etc. The execution process of the k-means clustering algorithm is as follows: S201: randomly select k samples from the original point cloud data as initial clustering centers; S202: Calculate the Euclidean distance d between k center points of different initial cluster centers i , assign each point cloud to the nearest cluster center, and the Euclidean distance is expressed as follows: S203: Calculate the average coordinates of each cluster center point twice, and use the average value as the new center point of the cluster center. The center point distance expression is: S204: S202 and S203 are repeated in a loop, and the iteration ends when the center point no longer changes or the preset number of iterations is reached.

4. The method for accepting the mining volume of ore and rock by integrating point cloud and spectral data technology according to claim 1 is characterized in that: The steps of establishing the S3 three-dimensional voxelized model and calculating the excavation volume are as follows: S301: determining the overall boundary of the stope point cloud by using the MAX function, and establishing a minimum rectangular parallelepiped model that can contain the overall boundary of the stope; S302: Set the size of the voxel grid of the stope point cloud model, and remove the voxel model outside the stope point cloud. x 、voxel y 、voxel z The dimensions are as follows: Among them, (x, y, z) represents the coordinates of the point, and voxel_size represents the side length of the voxel. S303: Differentiate the point cloud data of the two periods to calculate the mining volume of the stope in that period.

5. The method for accepting the mining volume of ore and rock by integrating point cloud and spectral data technology according to claim 1 is characterized in that: The stope spectral data information preprocessed in S4 is mainly acquired through an imaging spectral lens and is used to establish a stope ore-rock boundary model.

6. The method for accepting the mining volume of ore and rock by integrating point cloud and spectral data technology according to claim 1 is characterized in that: Before calculating the mining volume of the mine rock in the S5, the coordinate system of the mine-rock boundary model and the mine point cloud data must be unified.

7. The method for accepting the mining volume of ore and rock by integrating point cloud and spectral data technology according to claim 6 is characterized in that: The steps for calculating the mining volume of the mine are as follows: S501: Inverting and establishing a mine-rock boundary model through spectral data information; S502: Align the stope ore-rock boundary model coordinate system with the point cloud data coordinate system; S503: Fusing point cloud-spectral ore-rock boundary model data, and giving relevant density attributes to the stope geological model; S504: Calculate the mining volume of ore and rock in the mining area respectively.

Images