CN-121706617-B - Surface mine blasting effect optimization method and system based on video target recognition

CN121706617BCN 121706617 BCN121706617 BCN 121706617BCN-121706617-B

Abstract

The invention discloses a surface mine blasting effect optimization method and system based on video target recognition, which belong to the field of mining and image or video recognition, and comprise the steps of acquiring digital drilling design data and blasting design parameters of a target blasting area; the method comprises the steps of collecting video data from before blasting to finishing by an unmanned aerial vehicle, extracting pre-blasting video frames, identifying non-blasted holes by a first target detection model, determining blasthole numbers of the non-blasted holes, extracting post-blasting video sequences, identifying punching targets by a second target detection model, determining punching target tracks by combining a multi-target tracking algorithm, matching the positions of the punching targets with the non-blasted holes with matched blasthole numbers, determining blasthole numbers of the punching targets, obtaining actual coordinates, actual hole depths, blasting design parameters and characteristic parameters of punching events for the punching targets, and reasoning according to a preset diagnosis rule base to obtain blasthole diagnosis results. The invention can diagnose and optimize blasting effect with high efficiency and accuracy.

Inventors

LIU SIYUAN
WU SHANG
MAO JING
MA PINGPING
WANG SHUANG

Assignees

西安优迈智慧矿山科技有限公司

Dates

Publication Date: 20260505
Application Date: 20260213

Claims (9)

1. A surface mine blasting effect optimization method based on video target identification is characterized by comprising the following steps: The method comprises the steps of obtaining digital drilling design data and blasting design parameters of a target blasting area of an opencast mine, wherein the digital drilling design data comprise a blast hole number, a design coordinate and a design hole depth of each blast hole; Starting an unmanned aerial vehicle above the target blasting area to acquire video data from before blasting to after blasting; Extracting a video frame before blasting from the video data, and identifying non-blasted blasthole openings in the video frame by using a first target detection model which is trained in advance, and determining blasthole numbers corresponding to the identified non-blasted blasthole openings through position matching, wherein the first target detection model is used for identifying the non-blasted blasthole openings and pixel positions thereof in an input image; Extracting a post-blasting video sequence from the video data, identifying punching targets in the video sequence by using a second target detection model which is trained in advance, and determining the track of each punching target by combining a preset multi-target tracking algorithm; for each punching target track, using a nearest neighbor algorithm to match the initial appearance position of the punching target track with the pixel position of the non-blasted gun hole with the matched gun hole number, determining the non-blasted gun hole for generating the punching target, and determining the corresponding gun hole number, wherein the second target detection model is used for identifying the punching target and the pixel position thereof in the input image; For each punching target, corresponding actual coordinates, actual hole depths and blasting design parameters are obtained from a database according to the determined hole numbers, characteristic parameters of a punching event determined when a multi-target tracking algorithm determines the track of the punching target are obtained to form single-hole diagnosis basic data, the single-hole diagnosis basic data are inferred according to a preset diagnosis rule base to obtain diagnosis results of the corresponding holes of the punching target, the diagnosis results comprise the hole numbers, diagnosis conclusions and optimization suggestions, wherein the diagnosis results are used for storing the database as historical data to guide the blasting design of the holes under the future similar conditions, the similar conditions comprise similar geological conditions or adjacent spatial positions, the similar geological conditions mean that the difference of at least two geological parameters of a new blasting area and a historical blasting area meets the corresponding requirements, and the adjacent spatial positions mean that the new blasting area and the historical blasting area meet any one of the preset spatial position requirements.
2. The method of claim 1, wherein the first target detection model is a YOLOv-based blasthole identification model, trained using pre-blast sample images of a number of marked non-blasthole openings and their pixel locations.
3. The method of claim 2, wherein determining the number of blastholes corresponding to each identified non-blasted blast hole by location matching comprises: aiming at each identified non-blasted hole, converting the pixel position of the non-blasted hole from a pixel coordinate to a geodetic coordinate by using positioning and attitude determination data and camera internal reference data of the unmanned aerial vehicle and adopting a photogrammetry principle; and performing position matching on the obtained geodetic coordinates and the three-dimensional coordinates in the digital drilling design data by utilizing a nearest neighbor algorithm, and determining the blast hole number corresponding to one matched three-dimensional coordinate as the blast hole number corresponding to the non-blasted blast hole mouth.
4. The method of claim 2, wherein the second object detection model is a YOLOv-based punch recognition model, and is trained using post-blasting sample images of a plurality of marked punch objects and their pixel positions, wherein the punch objects are rock or smoke thrown vertically upward during blasting.
5. The method of claim 1, wherein the predetermined multi-target tracking algorithm comprises a multi-target tracking ByteTrack algorithm.
6. The method of claim 1 wherein the punch event characteristic parameters include punch duration, maximum punch height.
7. The method of claim 6, wherein the predetermined diagnostic rule base comprises a plurality of rules, each rule comprising two parts of conditions and conclusions, the conditions being based on a combination of related data of the blasting, the conclusions comprising diagnostic results of the blasting problem under the corresponding conditions.
8. A surface mine blasting effect optimizing system based on video target recognition is characterized by being used for realizing the surface mine blasting effect optimizing method based on video target recognition, which is disclosed in any one of claims 1-7, and comprises an unmanned plane module, a data management module, a visual analysis engine module, an intelligent diagnosis and optimizing module and a comprehensive management and control platform, The unmanned aerial vehicle module is used for acquiring video data of a target blasting area of the surface mine by using the unmanned aerial vehicle and carrying out data feedback; The data management module is used for storing and managing digital drilling design data, blasting design parameters, video data and historical data; The visual analysis engine module is integrated with a first target detection model, a second target detection model and a preset multi-target tracking algorithm, and is used for identifying non-blasted gun holes in a video frame before blasting in the video data by using the first target detection model, and determining gun hole numbers corresponding to the identified non-blasted gun holes through position matching; the method comprises the steps of detecting a video sequence of a shot hole, detecting a target in the video sequence after the shot hole is shot, identifying a punching target in the video sequence after the shot hole is shot in the video data by utilizing a second target detection model, determining each punching target track by combining the multi-target tracking algorithm, performing position matching on the initial appearance position of each punching target track and the pixel position of the non-shot hole with the matched shot hole number by utilizing a nearest neighbor algorithm, determining the non-shot hole for generating the punching target, and determining the corresponding shot hole number, wherein the first target detection model is used for identifying the non-shot hole and the pixel position thereof in an input image, and the second target detection model is used for identifying the punching target and the pixel position thereof in the input image; The intelligent diagnosis and optimization module is internally provided with a preset diagnosis rule base, and is used for acquiring corresponding actual coordinates, actual hole depths and blasting design parameters from a database according to the determined gun hole numbers for each punching target, acquiring determined punching event characteristic parameters when a multi-target tracking algorithm determines the punching target track to form single-hole diagnosis basic data; the comprehensive management and control platform is used for providing a man-machine interaction interface, and executing task scheduling, process monitoring, diagnostic result display and diagnostic result report output.
9. The system of claim 8, wherein the integrated management and control platform is further configured to provide an auxiliary blasting design interface, and output a history optimization suggestion similar to a current design area condition by performing history data matching in the database when a user performs a new area blasting design, wherein the condition similarity includes a geological condition similarity or a spatial position similarity, the geological condition similarity means that at least two geological parameters of the new blasting area and the history blasting area have differences meeting corresponding requirements, and the spatial position similarity means that the new blasting area and the history blasting area meet any one of preset spatial position requirements.

Description

Surface mine blasting effect optimization method and system based on video target recognition Technical Field The invention belongs to the field of mining, digital blasting and image or video recognition, and particularly relates to a surface mine blasting effect optimization method and system based on video target recognition. Background In the production of surface mines, the blasting effect is directly related to the subsequent loading, transportation efficiency and comprehensive cost. At present, the evaluation and optimization of the blasting effect mainly depend on manual field investigation and measurement after blasting, such as the evaluation of the form, the block distribution, the backlash distance and the like of a blasting stack. The method has obvious disadvantages that firstly, the method is seriously lagged, instant feedback cannot be provided for procedures such as charging, networking and the like of current blasting, secondly, the subjectivity is strong, the coverage rate is low, the high-risk area is difficult to be carefully checked, thirdly, macroscopic blasting quality problems (such as large blocks, root bottoms and excessive flying stones) cannot be precisely attributed to a specific blast hole, the follow-up blasting parameter optimization is lack of pertinence, and large-area adjustment is often only carried out by experience, the efficiency is low, and the effect is unstable. With the development of unmanned aerial vehicles and image processing technologies, methods for evaluating the variance and the blockiness distribution of post-blasting terrain scanning by using unmanned aerial vehicles are available. But these methods are still post-evaluation and fail to capture critical dynamic information during blasting. The blasting "punching" is the most direct dynamic phenomenon reflecting the blocking effect of the blast hole and the rationality of the charge. If the punching holes can be automatically identified and positioned to specific blast holes in real time, revolutionary technical means are provided for realizing accurate and instant diagnosis and optimization of blasting effect. Disclosure of Invention In order to solve the problems in the prior art, the invention provides a surface mine blasting effect optimization method and system based on video target recognition. The technical problems to be solved by the invention are realized by the following technical scheme: In a first aspect, an embodiment of the present invention provides a method for optimizing an effect of blasting in a surface mine based on video target recognition, where the method includes: The method comprises the steps of obtaining digital drilling design data and blasting design parameters of a target blasting area of an opencast mine, wherein the digital drilling design data comprise a blast hole number, a design coordinate and a design hole depth of each blast hole; Starting an unmanned aerial vehicle above the target blasting area to acquire video data from before blasting to after blasting; Extracting a video frame before blasting from the video data, identifying non-blasted gun holes in the video frame by using a first target detection model which is trained in advance, and determining gun hole numbers corresponding to the identified non-blasted gun holes through position matching; extracting a post-blasting video sequence from the video data, identifying punching targets in the video sequence by utilizing a second target detection model which is trained in advance, and determining all punching target tracks by combining a preset multi-target tracking algorithm; For each punching target, corresponding actual coordinates, actual hole depths and blasting design parameters are obtained from a database according to the determined hole numbers, characteristic parameters of a punching event determined when a multi-target tracking algorithm determines the track of the punching target are obtained to form single-hole diagnosis basic data, the single-hole diagnosis basic data are inferred according to a preset diagnosis rule base to obtain diagnosis results of the corresponding holes of the punching target, the diagnosis results comprise the hole numbers, diagnosis conclusions and optimization suggestions, wherein the diagnosis results are used for storing the database as historical data to guide the blasting design of the holes under the future similar conditions, the similar conditions comprise similar geological conditions or adjacent spatial positions, the similar geological conditions mean that the difference of at least two geological parameters of a new blasting area and a historical blasting area meets the corresponding requirements, and the adjacent spatial positions mean that the new blasting area and the historical blasting area meet any one of the preset spatial position requirements. In a second aspect, the embodiment of the invention provides a surface mine blasting effect optimization system ba