Mine mining plan making method and system

Abstract

The invention discloses a mining plan making method and system, and relates to the technical field of automatic management, and the method comprises the steps: obtaining a mine image and environment data; extracting multi-dimensional features of the mine image, and fusing the multi-dimensional features to obtain fused features; inputting the fusion features into a recognition model to obtain a mineral recognition result; combining the environment data with the mineral identification result to form a mine environment-resource distribution map; and inputting the mine environment-resource distribution map into the mining model to obtain a mine mining plan. According to the mining plan based on machine vision and a large model, real-time and high-definition monitoring of the mine environment is achieved by integrating the high-precision camera shooting function and the advanced video processing technology, subjectivity and errors of manual judgment are reduced, the mining process is accelerated, optimal configuration and efficient utilization of resources are achieved, and the mining efficiency is improved. And higher production efficiency and economic benefits are brought to mine enterprises.

Description

Technical Field

[0001] The present application relates to the field of automated management technology, and in particular to a method and system for formulating a mine mining plan. Background Art

[0002] In the traditional mining process, the distribution, grade and mining plan of minerals are mainly determined by manual experience and on-site surveys. This method is not only labor-intensive and inefficient, but also has high safety risks, such as geological disasters and harmful gas leaks. In recent years, although some mines have begun to introduce automated equipment and monitoring systems, most of them are still at the level of simple data collection and display, lacking in-depth data analysis and intelligent decision-making support, and are also accompanied by problems such as labor costs and safety accident risks.

[0003] In the field of machine vision, with the rapid development of image processing technology and deep learning algorithms, it has shown great potential in industrial automation, quality inspection, environmental monitoring, etc. However, in the complex application scenario of mining, the application of machine vision technology still faces many challenges, such as complex and changeable mining environment, different lighting conditions, and diverse mineral types with complex characteristics.

[0004] At the same time, the integrated application of big data and artificial intelligence technology has provided new ideas for mining. By building a large model knowledge base and deeply integrating and analyzing massive amounts of mining data (including videos, images, sensor data, etc.), a comprehensive understanding and accurate prediction of the mining environment can be achieved. However, how to effectively combine machine vision technology with large model technology to achieve accurate identification of minerals, rapid assessment of ore grade, and formulate scientific and reasonable mining plans based on this is still a key issue that needs to be urgently solved in current mining technology. Summary of the invention

[0005] The embodiments of the present application provide a method and system for formulating a mine mining plan to solve the problem in the prior art that big data and artificial intelligence technologies are insufficiently utilized, and accurate mineral identification and mining plan formulation cannot be performed.

[0006] On the one hand, an embodiment of the present application provides a method for formulating a mine mining plan, comprising: Acquire mine images; Obtaining environmental data on mines; Preprocess the mine image, extract the multi-dimensional features of the preprocessed mine image, and fuse the multi-dimensional features to obtain fused features; The fused features are input into the recognition model to obtain the mineral recognition results including the type information and taste information of the mineral; Combine environmental data and mineral identification results to form a mine environment-resource distribution map; Input the mine environment-resource distribution map into the mining model to obtain the optimal mine mining plan.

[0007] On the other hand, the embodiment of the present application also provides a mine mining plan formulation system, including: Mine inspection robots, used to obtain mine images and mine environmental data; The data center is used to preprocess the mine images, extract the multi-dimensional features of the preprocessed mine images, fuse the multi-dimensional features to obtain fused features, and input the fused features into the recognition model to obtain the mineral recognition results containing the type information and taste information of the minerals; The data background is used to combine environmental data and mineral identification results to form a mine environment-resource distribution map, and input the mine environment-resource distribution map into the mining model to obtain the optimal mine mining plan.

[0008] A mine mining plan formulation method and system in this application has the following advantages: Improve mining efficiency and accuracy: Mining plans based on machine vision and large models integrate high-precision camera functions and advanced video processing technology to achieve real-time, high-definition monitoring of the mining environment. Combined with multi-feature fusion recognition technology, it can accurately and quickly identify mineral types and grades, significantly improving the recognition accuracy and efficiency during the mining process. This not only reduces the subjectivity and errors of manual judgment, but also speeds up the mining process, bringing higher production efficiency and economic benefits to mining companies.

[0009] Enhance the safety of mine operations: The mine inspection robot not only has the function of mobile inspection, but also carries a variety of environmental sensors, which can monitor key safety indicators such as soil moisture, gas concentration, and carbon dioxide concentration in real time. This function effectively prevents safety accidents caused by environmental factors, such as gas explosions and collapses, and provides a safer and more reliable guarantee for mine operations. At the same time, through the analysis and processing of comprehensive data by large models, potential safety hazards can be warned in advance, providing a scientific basis for mine safety management.

[0010] Optimize resource allocation and mining plans: The large model knowledge base integrates a large amount of mining environment data and mining experience. Through retrieval and comparison, it can provide users with a more scientific and reasonable mining plan. This plan not only takes into account the type and grade of minerals, but also comprehensively considers environmental factors, mining costs, equipment performance and other aspects, achieving optimal allocation and efficient utilization of resources. In addition, the large model also has self-learning capabilities, which can continuously optimize the algorithm model as the mining process progresses, and improve the adaptability and accuracy of the mining plan.

[0011] Promoting the intelligent transformation of the mining industry: Through the integrated application of advanced technologies such as machine vision, big data processing, and artificial intelligence, intelligent monitoring and management of the entire mining process have been achieved. This not only reduces labor costs and improves work efficiency, but also promotes technological progress and industrial upgrading in the mining industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0013] Figure 1 A flowchart of a method for formulating a mine mining plan provided in an embodiment of the present application. DETAILED DESCRIPTION

[0014] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0015] Figure 1 A flowchart of a method for formulating a mining plan provided in an embodiment of the present application. An embodiment of the present application provides a method for formulating a mining plan, including: S100, acquiring a mine image.

[0016] S110, obtaining environmental data of the mine.

[0017] Exemplarily, mine images and environmental data are acquired by a mine inspection robot, which includes a mobile inspection mechanism, a camera and an environmental sensor. The camera and the environmental sensor are both arranged on the mobile inspection mechanism, and the camera and the environmental sensor are used to acquire mine images and environmental data, respectively.

[0018] Specifically, mine images include visible light images and infrared images of the mine. The camera is required to support WDR (Wide Dynamic Range) technology, have more than 4 million pixels, and support infrared night vision to capture on-site video images of the mine.

[0019] The mobile inspection organization can adopt a bipedal, quadrupedal, wheeled, tracked or crawling mobile platform. The environmental sensors include soil moisture sensors, gas concentration sensors and carbon dioxide concentration sensors. Therefore, the environmental data include soil moisture, gas concentration and carbon dioxide concentration data.

[0020] Furthermore, the mine inspection robot receives inspection instructions, which include inspection points and inspection time. The mine inspection robot builds a local map based on the inspection points, inspection time and known mine map based on SLAM (Simultaneous Localization and Mapping) technology, performs path planning in the local map, moves to each inspection point according to the planned path, and obtains mine images and environmental data at the inspection points.

[0021] Specifically, the mine map is an electronic map pre-built based on surveying and mapping data that contains the environmental geomorphological data of the mine. The environmental geomorphological data includes information such as the altitude and terrain of each location on the mine. The mine inspection robot constructs a local map based on the mine map and the scene data of the current location collected in real time. The scene data and the environmental geomorphological data contain the same specific data type. The local map is obtained after correcting the environmental geomorphological data according to the scene data within a certain range of the current location in the mine map, so that the local map can reflect the actual environmental conditions of the mine, and thus provide an accurate and reliable data basis for path planning.

[0022] Furthermore, the mine inspection robot also includes a laser radar, which is used to collect environmental topographic data. During the movement, the environmental topographic data is compared with the scene data of the current position in the mine map. If the environmental topographic data is consistent with the scene data, the mine inspection robot moves according to the planned path. If the environmental topographic data is inconsistent with the scene data, the mine inspection robot updates the mine map according to the environmental topographic data.

[0023] Specifically, since the mine map is established based on surveying and mapping data, and the frequency of surveying and mapping data collection is not very high, there is a situation where the surveying and mapping data does not match the actual situation of the mine after a period of time. At this time, if the environmental landform data collected by lidar, specifically the three-dimensional point cloud data, is different from the scene data of the current position in the mine map, it means that the geographical environment of the current position has changed. At this time, not only will the environmental landform data be corrected according to the scene data to establish a local map, but the corresponding data in the mine map will also be corrected and updated, so that the next time the mine map is used, it will be based on the latest data as much as possible, reducing the number of data updates.

[0024] Furthermore, after the mine inspection robot starts executing the inspection instruction, the ongoing production scheduling task will be stopped. If the safety index value determined according to the environmental data is lower than the set threshold, the already formulated production scheduling task will be adjusted.

[0025] Specifically, the mines targeted by this application are mines that are already in mining. Since the types and grades of minerals in the mines may change at any time during the mining process, it is necessary to update the mining plan in a timely manner to maximize and most efficiently utilize the mining equipment and personnel in the mine. Before the mine inspection robot executes the inspection instructions, the production scheduling center of the mine has generated detailed production scheduling tasks based on the mining plan formulated last time. The equipment and personnel in the mine execute the production scheduling tasks, so that the previous mining plan can be implemented. As time goes by, the situation of the mine, specifically the type and grade of minerals, may change. In order to cope with such changes, it is necessary to collect relevant data of the mine again and formulate a mining plan.

[0026] Under this demand, the mine inspection robot will work regularly or irregularly to collect mine images and environmental data required for the development of mining plans. For environmental data, it can reflect the safety indicators of the mine at the current moment. After obtaining the environmental data, the data center will first clean the data, filter out the abnormal data, and then use SQL (Structured Query Language) to query the corresponding calculation formula, and use the calculation formula to calculate the safety index value of the current mine. If the safety index value is greater than the first threshold but less than or equal to the second threshold, a corresponding alarm notification is generated to remind the production scheduling center to adjust the production scheduling task to avoid unnecessary losses due to safety issues. If the safety index value is less than or equal to the first threshold, it means that the safety of the mine is too low, and the production scheduling center will be reminded to stop executing the production scheduling task and execute the production scheduling task after the safety problem is resolved.

[0027] For alarm notifications, the embodiment of the present application also uploads various abnormal situations and corresponding processing solutions in the large model knowledge base in advance based on the LangChain framework. The LLM (Large Language Model) retrieves the processing solution that best matches the alarm notification in the large model knowledge base, and finally feeds back the retrieved processing solution to the staff of the production scheduling center, so that the staff can guide the disposal personnel to quickly and accurately solve the safety problems so that the production scheduling tasks can continue to be executed.

[0028] In the embodiment of the present application, after the mine inspection robot collects the mine image and environmental data, it will transmit it to the data center through wireless transmission technology, such as 5G, Wi-Fi 6, etc. The data center processes the data and then transmits it to the data background, and finally the data background makes a decision. Both the data center and the data background can run on a cloud server.

[0029] Furthermore, after the mine inspection robot is turned on, it will first perform a self-check of each module, including the camera, mobile inspection mechanism, environmental sensor and data transmission module, to ensure that all components are working properly. In the data transmission verification, the signal stability is detected by the IoT edge layer, and then the data center confirms the correctness of the data through image recognition algorithms, high-frequency data detection, and abnormal data analysis steps. When everything is ready, the robot starts to execute the inspection instructions.

[0030] S120, preprocessing the mine image, extracting multi-dimensional features of the preprocessed mine image, and fusing the multi-dimensional features to obtain fused features.

[0031] Exemplarily, the preprocessing of the mine image includes denoising, color correction and color enhancement, and the multi-dimensional features include color features, texture features and shape features.

[0032] Specifically, image denoising techniques such as median filtering and Gaussian filtering can be used to remove noise in the mine image and maintain the true color of the mine image. Taking Gaussian filtering as an example, Gaussian filtering is performed according to the following formula:

[0033] In the formula, f ( x ) is the result after Gaussian filtering, is the standard deviation, e is a natural constant, x is the input value, i.e. the pixel value of a pixel in the mine image, is the mean, usually 0.

[0034] Before using Gaussian filtering, you need to first construct a discretized Gaussian kernel, which is also called a Gaussian filter. This Gaussian kernel will be used for the convolution operation of the mine image to achieve the denoising effect.

[0035] Create one n × n Gaussian kernel of size , where n Is an odd number, such as 3, 5, 7, etc., to ensure that there is a center point. Each element of the Gaussian kernel is calculated according to the Gaussian function. Here, a two-dimensional Gaussian function is used to calculate the value of each element:

[0036] In the formula, G ( x , y ) is the position coordinate relative to the center point ( x , y ) elements, ranging from - ( n -1) / 2 to ( n -1) / 2.

[0037] In order to ensure that the brightness of the image remains unchanged after convolution, the Gaussian kernel needs to be normalized. Normalization is to add up the element values ​​of all Gaussian kernels and then divide by this sum so that the sum of all elements is equal to 1. This can be achieved by the following formula:

[0038] In the formula, G ( i , j ) is the position coordinate relative to the center point ( i , j ). Once we have the normalized Gaussian kernel, we can perform a convolution operation on the mine image. For each pixel in the mine image, we use the normalized Gaussian kernel to perform a convolution operation with the pixels around it. The convolution process is as follows: (1) Overlay the normalized Gaussian kernel on each pixel of the mine image.

[0039] (2) For each pixel in the coverage area, multiply its value by the value in the corresponding Gaussian kernel.

[0040] (3) Add all the products together and use the result to replace the value of the center pixel in the original image.

[0041] (4) Repeat the above steps until all pixels have been processed.

[0042] Colors may be distorted due to factors such as light and camera settings, so color correction is required to ensure color consistency and accuracy. During color correction, the RGB (Red Green Blue) color space is converted to the HSV (Hue, Saturation, Value) color space, and the hue and saturation can be adjusted in the HSV space.

[0043] Because the mining environment, especially the underground environment, is dark and dusty, the color of the mining image needs to be enhanced. The color contrast of the image is enhanced through histogram equalization, contrast stretching and other technologies, making it easier to distinguish different parts of the mineral.

[0044] For example, contrast stretching can expand the dynamic range of mining images by remapping the grayscale of mining images, thereby improving the visual effect of the image. Contrast stretching is particularly suitable for images with grayscale levels concentrated in a narrow range. By stretching these grayscale levels, the features in mining images can be made more prominent. This application uses linear contrast stretching, and the darkest and brightest pixel values ​​in the image can be selected as the starting and end points of the mapping, and then all pixel values ​​between these two points are mapped to between 0 and 255 (for 8-bit grayscale images) according to a certain ratio. The specific formula is as follows:

[0045] In the formula, g ( x , y ) is the pixel value after contrast stretching, L =256, A is the minimum gray value of the mine image, B is the maximum gray value, f ( x , y ) is the position in the mine image ( x , y ), To round down.

[0046] After converting the RGB color space to the HSV color space, the color features of the mine image, including hue, saturation, and color brightness values, can be extracted. For texture features, methods such as gray-level co-occurrence matrix (GLCM) and local binary patterns (LBP) can be used for extraction. For shape features, the geometric features of the shape, size, edge, etc. of the minerals in the mine image can be analyzed as a supplement to color features and texture features.

[0047] S130, inputting the fused features into the recognition model to obtain a mineral recognition result including the type information and taste information of the mineral.

[0048] Exemplarily, the recognition model is trained using data in the large model knowledge base. The data in the large model knowledge base includes recognition sample sets, which are obtained by experts manually annotating mine images collected by mine inspection robots. During the annotation process, experts use experience to determine the type and taste of minerals based on the color, texture, shape and other characteristics of minerals in the mine images. After obtaining the recognition sample set, a part of the samples will be used to train the recognition model. The recognition model can specifically use commonly used models such as convolutional neural networks. After the training is completed, another part of the samples in the recognition sample set will be used for testing and verification.

[0049] Specifically, the type information of a mineral refers to which type of mineral the mineral belongs to, including: silicate minerals, oxide minerals, sulfide minerals, halide minerals, carbonate minerals, phosphate minerals, sulfate minerals, and the like.

[0050] The grade information of a mineral refers to the content of useful components in the mineral (such as metal elements or useful oxides in the mineral), which is generally expressed as the mass percentage of the useful components.

[0051] Furthermore, since the recognition model has certain requirements on the format of input data, it is necessary to ensure that the input fusion features meet the format requirements of the recognition model. Moreover, if the fusion features extracted from the mine images in the recognition sample set are standardized or normalized during the training process, then in actual use, the fusion features extracted from the mine images to be recognized also need to be standardized or normalized to maintain data consistency.

[0052] S140, combining the environmental data and the mineral identification results to form a mine environment-resource distribution map.

[0053] For example, a mine environment-resource distribution map can be displayed on a mine map, and the mineral identification results obtained for each inspection point will be displayed at the corresponding position of the mine map, and the corresponding environmental data will also be displayed at that position.

[0054] S150, inputting the mine environment-resource distribution map into the mining model to obtain the optimal mine mining plan.

[0055] Exemplarily, the mining model is also trained using data from the large model knowledge base. The large model knowledge base also contains a mining sample set, which contains a variety of different mineral identification results and environmental data. Experts have developed corresponding mining plans for each mineral identification result and environmental data, including: mining area, mining depth, mining method, mining sequence, and corresponding equipment and personnel configuration. These mining plans will serve as labels for mineral identification results and environmental data. After the mineral identification results and environmental data are input into the mining model, the mining model is trained with the corresponding labels as the target, so that the mining model learns the relationship between the mineral identification results and environmental data and the mining plan. After the training is completed, it can be applied to actual decision-making after testing and verification.

[0056] The mineral identification results and environmental data of each inspection point in the mine environment-resource distribution map are input into the mining model in turn, and the mining plan of each inspection point will be obtained. The mining plan of the mine can be obtained by integrating the mining plans of all inspection points.

[0057] The present application also provides a mining plan formulation system, including: Mine inspection robots, used to obtain mine images and mine environmental data; The data center is used to preprocess the mine images, extract the multi-dimensional features of the preprocessed mine images, fuse the multi-dimensional features to obtain fused features, and input the fused features into the recognition model to obtain the mineral recognition results containing the type information and taste information of the minerals; The data background is used to combine environmental data and mineral identification results to form a mine environment-resource distribution map, and input the mine environment-resource distribution map into the mining model to obtain the optimal mine mining plan.

[0058] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0059] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalents, the present application is also intended to include these modifications and variations.

Claims

1. A method for formulating a mining plan, characterized in that: include: Acquire mine images; Obtaining environmental data on mines; Preprocessing the mine image, extracting multi-dimensional features of the preprocessed mine image, and fusing the multi-dimensional features to obtain fused features; Inputting the fusion features into a recognition model to obtain a mineral recognition result including type information and taste information of the mineral; Combining the environmental data with the mineral identification results to form a mine environment-resource distribution map; The mine environment-resource distribution map is input into the mining model to obtain the optimal mine mining plan.

2. A method for formulating a mining plan according to claim 1, characterized in that: The mine image and the environmental data are acquired by a mine inspection robot, which includes a mobile inspection mechanism, a camera and an environmental sensor. The camera and the environmental sensor are both arranged on the mobile inspection mechanism, and the camera and the environmental sensor are used to acquire the mine image and the environmental data respectively.

3. A method for formulating a mining plan according to claim 2, characterized in that: The mobile inspection mechanism adopts a bipedal, quadrupedal, wheeled, tracked or crawling mobile platform, and the environmental sensors include a soil moisture sensor, a gas concentration sensor and a carbon dioxide concentration sensor.

4. A method for formulating a mining plan according to claim 2, characterized in that: The mine inspection robot receives an inspection instruction, which includes an inspection point and an inspection time. The mine inspection robot builds a local map based on the SLAM technology according to the inspection points, the inspection time and a known mine map, performs path planning in the local map, moves to each of the inspection points according to the planned path, and obtains the mine image and the environmental data at the inspection points.

5. A method for formulating a mining plan according to claim 4, characterized in that: The mine inspection robot also includes a laser radar, which is used to collect environmental topography data. During the movement, the environmental topography data is compared with the scene data of the current position in the mine map. If the environmental topography data is consistent with the scene data, the mine inspection robot moves according to the planned path. If the environmental topography data is inconsistent with the scene data, the mine inspection robot updates the mine map according to the environmental topography data.

6. A method for formulating a mining plan according to claim 4, characterized in that: After the mine inspection robot starts to execute the inspection instruction, the ongoing production scheduling task will be stopped. If the safety index value determined according to the environmental data is lower than the set threshold, the already formulated production scheduling task will be adjusted.

7. A method for formulating a mining plan according to claim 1, characterized in that: The preprocessing of the mine image includes denoising, color correction and color enhancement.

8. A method for formulating a mining plan according to claim 1, characterized in that: The multi-dimensional features include color features, texture features and shape features.

9. A method for formulating a mining plan according to claim 1, characterized in that: The identification model and the mining model are both trained using data in a large model knowledge base.

10. A system using a method for formulating a mining plan according to any one of claims 1 to 9, characterized in that: include: Mine inspection robots, used to obtain mine images and mine environmental data; A data center is used to preprocess the mine image, extract multi-dimensional features of the preprocessed mine image, fuse the multi-dimensional features to obtain fused features, and input the fused features into a recognition model to obtain a mineral recognition result including type information and taste information of the mineral; The data background is used to combine the environmental data with the mineral identification results to form a mine environment-resource distribution map, and input the mine environment-resource distribution map into the mining model to obtain the optimal mine mining plan.

Images