Mine machine equipment reliability simulation method based on scene deep learning

Abstract

The invention discloses a mine machine equipment reliability simulation method based on scene deep learning, and belongs to the technical field of mine machine reliability analysis, and the method comprises the steps: S1, building a material discrete element model of a real mine field; s2, realizing oil cylinder driving displacement curve extraction of dynamics simulation; s3, a driving function database of different postures under typical working conditions is collected, a load spectrum database is established, the oil cylinder driving displacement curve obtained in the S2 is corrected, and dynamic simulation modeling of a real driving function is established; s4, establishing a coupling simulation model of the material particle discrete element and dynamics through a coupling interface; and S5, through an experiment, correcting data obtained through the material particle discrete element and the dynamic coupling simulation model under an actual working condition, and further generating a remote high-precision working condition load spectrum. The method has the advantage of high simulation precision.

Description

Technical Field

[0001] This invention relates to the field of reliability analysis technology for mining machinery, specifically to a reliability simulation method for mining equipment based on scenario deep learning. Background Technology

[0002] Mining excavators are important mining machines primarily used for open-pit mining and excavation / unloading. Their stable operation directly impacts mine safety and economic efficiency. Therefore, fatigue life assessment, reliability analysis, and performance optimization for mining excavators are currently key tasks.

[0003] Currently, these tasks are mainly accomplished through simulation. Existing technical solutions include: 1. Obtaining stress cloud maps of the equipment through static strength analysis of the structure; however, this typically only uses extreme working conditions for analysis and cannot consider specific real-world scenarios. 2. Exploring scenario analysis through coupled discrete element method (DEM) and dynamics simulation; however, the DEM model of materials is basically based on idealized spherical particles, which differs significantly from actual material information. 3. Analyzing the equipment's working scenarios through dynamics simulation; however, the power system only uses a simple and singular driving function, which is inconsistent with the actual driving operations of on-site workers.

[0004] All three of the above technical solutions suffer from the problem of simulation calculations failing to consider real-world scenarios, resulting in very poor accuracy. 1. Acquiring the motion posture of the mining excavator's working device relies on physical sensors, which are extremely costly to deploy and have limited accuracy, making it impossible to accurately obtain pose data. Furthermore, the lack of signals in the mining area prevents timely feedback of data information. 2. The lack of ore particle size distribution information makes accurate coupling with the simulation system impossible. 3. Mining companies, considering operational safety and production efficiency, find it difficult to shut down operations to install and read stress measurement devices. Even if the equipment is already in use in the mining area, there is still a lack of stress and strain test results from real-world scenarios, especially since operating load conditions are largely unavailable in overseas markets. Summary of the Invention

[0005] The present invention addresses the technical problems mentioned in the background section above by providing a method for simulating the reliability of mining equipment based on scene-based deep learning. This method combines remote visual recognition of mining scenes with particle-dynamics, and automatically acquires EDEM modeling parameters and kinematic characteristics of the working device through a combination of database and visual recognition. This enables high-fidelity coupled simulation to accurately predict the reliability of mechanical equipment, which helps improve the accuracy of structural reliability simulation, supports high-efficiency and lightweight product design, and creates scene-based customized and highly competitive mining products.

[0006] According to the present invention, the technical solution provided by the present invention is: a method for simulating the reliability of mining equipment based on scene deep learning, comprising the following steps: S1. Establish a discrete element model of materials in a real mine, and simultaneously perform step S2. S2. Use YOLO-pose to perform attitude analysis on the mining equipment on site, and extract the displacement curve of the drive device for dynamic simulation; S3. Collect a database of driving functions for different attitudes under typical working conditions, establish a load spectrum database, match, splice, and correct the displacement curves of the driving device obtained in S2, and establish a dynamic simulation model of the real driving function. S4. Through the coupling interface, a coupled simulation model of material particle discrete element and dynamics is established; S5. Install sensors on the mining equipment to measure the displacement, pressure, stress, etc. of the driving device of the mining machine, correct the data obtained by the material particle discrete element and dynamic coupling simulation model under actual working conditions, and then generate a remote high-precision working condition load spectrum.

[0007] Furthermore, S1. Establish a discrete element model of materials in a real mine. S11. Conduct geological exploration, experiments and data collection in the mining area, and establish a database of rock material and mechanical properties for the mining area; S12. Collect images of materials in the mining area, place size references when taking pictures, and display the area where the photos were taken; obtain rock particle size and shape information in the mining area through machine vision methods, quickly determine relevant rock material, mechanical properties and other parameters based on ore database and geographic coordinates, and establish a material shape and particle size database by recognizing geometric shape and particle size distribution. S13. Combine rock material and mechanical property databases to establish a discrete element model of real mine materials in EDEM.

[0008] Furthermore, S11. Geological exploration, experimentation and data collection in mining areas refers to obtaining static friction coefficient, rolling friction coefficient, collision recovery coefficient, etc. for typical rock materials.

[0009] Furthermore, S2. Use YOLO-pose to perform attitude analysis on the mining equipment on site, and extract the displacement curve of the drive device for dynamic simulation; S21. Collect video footage of the mining equipment operating at the site; S22. Train YOLO-pose to detect key points of the working device of mining equipment; S23. Analyze the attitude of the mining equipment in the working condition video to obtain the working process of the mining equipment's working device; at the same time, by capturing angle changes, use the DH method to complete the conversion between the moving angle of the mining equipment's working device and the displacement of the driving component, and realize the extraction of the driving device displacement curve for dynamic simulation.

[0010] Furthermore, S4. Through the coupling interface, a coupled simulation model of material particle discrete element and dynamics is established; S41. Based on the material shape and particle size database obtained in step S12, irregular particles and dynamic modeling are obtained. That is, by dividing the material into four size segments: ≤60mm, 60mm-150mm, 150mm-300mm, and ≥300mm, the material shape and particle size are initially determined according to the area, perimeter, length and width of each segment. Among them, large materials are mainly blocky, flake-shaped, and strip-shaped. The actual shape is simplified and approximated according to the template. Among them, materials with a particle size ≤40mm are mainly spherical particles. S42. Dynamic simulation modeling of the real driving function, that is, by matching the load spectrum driving function database, the driving function is obtained, and a dynamic model of the working device of the mining equipment is established. The busing connection is adopted, and a single side uses a rotary joint. The model of material particle-dynamic coupling is established through the coupling interface.

[0011] Furthermore, S11. Conduct geological exploration, experiments, and data collection in the mining area, establish a rock material and mechanical properties database for the mining area, and combine the geographical coordinates of the specific mining area as search entries to complete the establishment of the rock regional database.

[0012] Furthermore, it also includes: S6. Simulation is performed using the generated high-precision working condition load spectrum to predict and analyze the stress and wear of the working device.

[0013] Furthermore, it also includes: S7. Simulation is performed using the generated high-precision working condition load spectrum to achieve predictive analysis of the reliability and lifespan of mechanical equipment.

[0014] Furthermore, it also includes: S8. Simulation analysis and optimization of the bucket full rate and digging energy consumption of mining excavators are carried out by simulating the generated high-precision working condition load spectrum.

[0015] Furthermore, it also includes: S9. Simulate using the generated high-precision working condition load spectrum to perform scenario-based customized product optimization.

[0016] The advantages of this invention compared to existing technologies are as follows: 1. Based on remotely transmitted material photos and operation videos, the material distribution and equipment posture information are obtained through computer vision technology. Matching actual working conditions, this enables discrete element method (DEM)-dynamic coupling simulation of materials, thereby obtaining a high-precision load spectrum and completing accurate simulation of the reliability of mining equipment. This helps reduce costs and improve simulation accuracy. 2. Based on ore particle size, shape, and material, geological data is collected from major mining areas to establish a corresponding ore database. Combined with geographic coordinates, the distribution of relevant rock database geological parameters is quickly determined, supplementing, calibrating, and correcting missing ore material data. Based on the database parameters, discrete element method information of materials from a real mine is established, solving the problems of idealized particle shape and uniform material in EDEM modeling. 3. By collecting a database of driving functions for different postures under typical working conditions in advance, and using YOLO-pose posture analysis based on on-site large mining excavator operation videos, the operation process of the excavator's working device is obtained. Matching with the existing load spectrum database, the cylinder driving function for dynamic simulation is extracted. Based on specific experiments, a load spectrum database is established, and the displacement curves analyzed by the DH method are spliced ​​and corrected, improving the accuracy of dynamic simulation of the working device.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the reliability simulation method for mining excavators according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating material photo acquisition and feature extraction in an embodiment of the present invention; Figure 3 This is a schematic diagram of a mining area retrieval based on character recognition, according to an embodiment of the present invention. Figure 4 This is a schematic diagram illustrating the acquisition of the cylinder displacement curve of the excavator working device according to an embodiment of the present invention; Figure 5 A schematic diagram illustrating the modeling of material particles in an embodiment of the present invention; Figure 6 This is a schematic diagram of the excavator working device according to an embodiment of the present invention; Figure 7 This is a schematic diagram of a sensor installed on an excavator during load spectrum correction according to an embodiment of the present invention; Figure 8 This is a schematic diagram illustrating the comparison and correction between the displacement load spectrum obtained by the DH method and the load spectrum measured by the wire sensor in an embodiment of the present invention. Figure 9This is a schematic diagram comparing the force load spectrum based on particle and dynamics simulation with sensor test data in an embodiment of the present invention; Figure 10 This is a schematic diagram illustrating the simulation of structural strength and fatigue life based on a high-precision load spectrum according to an embodiment of the present invention. Figure 11 This is a schematic diagram illustrating the simulation analysis of full bucket rate and excavation energy consumption based on high-precision working condition load spectrum in an embodiment of the present invention. Figure 12 This is a schematic diagram illustrating the technical problems and solutions mentioned in the background section of this invention. Detailed Implementation

[0019] The present invention will now be described in further detail.

[0020] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments will be clearly and completely described below with reference to the accompanying drawings. The following embodiments are used to illustrate the present invention, but are not intended to limit the scope of the present invention.

[0021] The reliability simulation method for mining equipment based on scene deep learning of the present invention, taking a mining excavator as an example, is based on the following problem-solving approach: Figure 12 As shown, combined with Figure 1 As shown in the figure, the reliability simulation method for mining excavators in this embodiment is as follows: I. Data Collection Database establishment: Geological exploration, experiments, and data collection were conducted in major mining areas, such as Wuhai Dongyuan Technology Open-pit Mine and Xinjiang Zhundong Hongshaquan Open-pit Mine. For typical rock materials, such as white sandstone, mudstone, coarse sandstone, and marble, static friction coefficient, rolling friction coefficient, and collision recovery coefficient were obtained to establish a rock material and mechanical property database. Combined with the geographical coordinates of specific mining areas as search entries, the regional rock database was established.

[0022] Multimodal feature extraction of mine materials: Photographs are taken of rock materials in the mining area, with size references placed during photography and the area where the photos were taken displayed. Machine vision methods are used to obtain information such as rock particle size and shape. Figure 2 As shown; simultaneously, based on the ore database and geographic coordinates, relevant rock material, mechanical properties, and other parameters are quickly determined, such as... Figure 3 As shown.

[0023] By identifying geometric shapes and particle size distributions, a database is established, and then corresponding mechanical parameter mappings are set according to the database, thereby establishing a discrete element model of materials from a real mine in EDEM.

[0024] YOLO-pose attitude analysis of the working device driving function: Based on the on-site operation video of a large mining excavator, YOLO-pose is used for training to detect key points of the working device, thereby analyzing the excavator's attitude in the video and obtaining the operating process of the excavator's working device; simultaneously, by capturing angle changes, the DH method is used to complete the conversion between angle and cylinder displacement, realizing the extraction of cylinder drive curves from dynamic simulation, such as... Figure 4 As shown.

[0025] However, YOLO-pose attitude analysis has certain errors. For example, during the excavation process, there will be continuous impacts, and the bucket will push against the rock resistance to dig deeper, resulting in a period of oscillation on the cylinder displacement curve. In addition, the data identified by YOLO-pose will have jitter and distortion. Therefore, based on experience and specific experiments, it is necessary to collect a database of driving functions for different attitudes under typical working conditions in advance, establish a load spectrum database, and match, stitch together, and correct the displacement curves analyzed by the DH method to improve the accuracy of dynamic simulation of the working device.

[0026] II. Simulation Modeling Irregular particle and dynamic modeling based on visual recognition and database acquisition: By dividing the material into four size segments—≤60mm, 60mm-150mm, 150mm-300mm, and ≥300mm—the material shape and particle size are initially determined based on the identified area, perimeter, length, and width of each segment. Larger materials are mainly blocky, flake-shaped, and elongated, with simplified approximations of the actual shape based on templates. Materials ≤40mm are mainly spherical particles. Figure 5 As shown.

[0027] Dynamic simulation modeling of real driving functions: By matching the load spectrum driving function database, the driving functions are obtained, and a dynamic model of the excavator's working device is established, such as... Figure 6 As shown. In order to obtain the load spectrum of each hinge point and consider the off-center loading scenario, a bushing connection is preferred to avoid under-constraint. A revolute joint is used on one side, and the model of material particle-dynamic coupling is established through the coupling interface layer.

[0028] III. Reliability Assessment Remote high-precision load spectrum generation: Based on the simulation methods provided in Parts 1 and 2, on-site personnel remotely transmit material photos and operational videos. By matching these with actual working conditions, high-precision hydraulic cylinder pressure and stress load spectra are obtained, thereby completing accurate discrete element and dynamic simulations of the material. Simultaneously, a drawwire sensor is installed on the local prototype to measure the hydraulic cylinder displacement, a hydraulic cylinder pressure interface is installed to test the hydraulic cylinder pressure, and strain gauges are applied to test the weld stress, thus verifying and correcting the test results of the relevant methods. Figure 7 As shown.

[0029] This invention also involves experimental calibration, whereby the analyzed curves are spliced ​​and corrected based on specific experimental results. The results demonstrate that the corrected load spectrum trend is essentially consistent with the test data, showing a significant improvement in accuracy. This accuracy currently meets the preliminary requirements for reliability simulation. (Example: Comparison of displacement load spectrum identification and test data) Figure 8 As shown, the force load spectrum based on particle and dynamics simulation is compared with the test data, for example... Figure 9 As shown, the efficiency of load spectrum acquisition is significantly improved.

[0030] The results are as follows Figure 10 , Figure 11 As shown, by simulating the generated high-precision working condition load spectrum, it is possible to realize the stress and wear analysis of the working device, the reliability and life prediction analysis of the mechanical equipment, the simulation analysis of the bucket full rate and digging energy consumption, etc. The accuracy of the simulation results is significantly improved, and it is also possible to formulate customized optimization schemes for products in different scenarios.

[0031] Summarize The advantages of the technical solutions reflected in the embodiments of the present invention are as follows: 1. This invention realizes a remote high-precision working condition load spectrum generation technology: by acquiring material distribution and equipment attitude information, matching actual working conditions, realizing material discrete element-dynamic coupling simulation, thereby obtaining a high-precision load spectrum, completing accurate simulation of the reliability of mining equipment, which helps to reduce costs and improve simulation accuracy.

[0032] 2. This invention realizes a multimodal feature recognition technology for mining materials based on visual recognition and database: material information of actual working conditions is established through vision and ore database, improving the accuracy and authenticity of material modeling.

[0033] 3. This invention realizes the inversion and correction of the driving function of the non-contact working device, which greatly improves the simulation accuracy of the dynamics of the working device.

[0034] 4. This invention realizes a customized design and optimization technology for mine structures based on remote scene recognition, which can support product customization, differentiated design and optimization scheme design, and has universal applicability.

[0035] This invention enables simulation analysis and optimization of structural strength, fatigue life, bucket capacity, and excavation energy consumption of mining equipment based on real-world scenarios. It also supports customized and differentiated product design and optimization, particularly for overseas markets, improving product development progress and competitiveness. Currently, it has achieved a 10% increase in bucket capacity and a 200% increase in structural fatigue life for 120T mining excavators.

[0036] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for simulating the reliability of mining equipment based on scene deep learning, characterized in that, Includes the following steps: S1. Establish a discrete element model of materials in a real mine, and simultaneously perform step S2. S2. Use YOLO-pose to perform attitude analysis on the mining equipment on site, and extract the displacement curve of the drive device for dynamic simulation; S3. Collect a database of driving functions for different attitudes under typical working conditions, establish a load spectrum database, match, splice, and correct the displacement curves of the driving device obtained in S2, and establish a dynamic simulation model of the real driving function. S4. Through the coupling interface, a coupled simulation model of material particle discrete element and dynamics is established; S5. Install sensors on the mining equipment to measure the displacement, pressure, stress, etc. of the driving device of the mining machine, correct the data obtained by the material particle discrete element and dynamic coupling simulation model under actual working conditions, and then generate a remote high-precision working condition load spectrum.

2. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that: S1. Establish a discrete element model of materials in a real mine. S11. Conduct geological exploration, experiments and data collection in the mining area, and establish a database of rock material and mechanical properties for the mining area; S12. Collect images of materials in the mining area, place size references when taking pictures, and display the area where the photos were taken; obtain rock particle size and shape information in the mining area through machine vision methods, quickly determine relevant rock material, mechanical properties and other parameters based on ore database and geographic coordinates, and establish a material shape and particle size database by recognizing geometric shape and particle size distribution. S13. Combine rock material and mechanical property databases to establish a discrete element model of real mine materials in EDEM.

3. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that: S11. Geological exploration, experimentation and data collection in mining areas refers to obtaining static friction coefficient, rolling friction coefficient, collision recovery coefficient, etc. for typical rock materials.

4. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that: S2. Use YOLO-pose to perform attitude analysis on the mining equipment on site, and extract the displacement curve of the drive device for dynamic simulation; S21. Collect video footage of the mining equipment operating at the site; S22. Train YOLO-pose to detect key points of the working device of mining equipment; S23. Analyze the attitude of the mining equipment in the working condition video to obtain the working process of the mining equipment's working device; at the same time, by capturing angle changes, use the DH method to complete the conversion between the moving angle of the mining equipment's working device and the displacement of the driving component, and realize the extraction of the driving device displacement curve for dynamic simulation.

5. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that: S4. Through the coupling interface, a coupled simulation model of material particle discrete element and dynamics is established; S41. Based on the material shape and particle size database obtained in step S12, irregular particles and dynamic modeling are obtained. That is, by dividing the material into four size segments: ≤60mm, 60mm-150mm, 150mm-300mm, and ≥300mm, the material shape and particle size are initially determined according to the area, perimeter, length and width of each segment. Among them, large materials are mainly blocky, flake-shaped, and strip-shaped. The actual shape is simplified and approximated according to the template. Among them, materials with a particle size ≤40mm are mainly spherical particles. S42. Dynamic simulation modeling of the real driving function, that is, by matching the load spectrum driving function database, the driving function is obtained, and a dynamic model of the working device of the mining equipment is established. The busing connection is adopted, and a single side uses a rotary joint. The model of material particle-dynamic coupling is established through the coupling interface.

6. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 2, characterized in that: S11. Conduct geological exploration, experiments, and data collection in the mining area, establish a rock material and mechanical properties database for the mining area, and combine the geographical coordinates of the specific mining area as search entries to complete the establishment of the rock regional database.

7. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that, Also includes: S6. Simulation is performed using the generated high-precision working condition load spectrum to predict and analyze the stress and wear of the working device.

8. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that, Also includes: S7. Simulation is performed using the generated high-precision working condition load spectrum to achieve predictive analysis of the reliability and lifespan of mechanical equipment.

9. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that, Also includes: S8. Simulation analysis and optimization of the bucket full rate and digging energy consumption of mining excavators are carried out by simulating the generated high-precision working condition load spectrum.

10. The method for simulating the reliability of mining equipment based on scene deep learning according to claim 1, characterized in that, Also includes: S9. Simulate using the generated high-precision working condition load spectrum to perform scenario-based customized product optimization.

Images