Unmanned mine car road intelligent identification device
By using transparent polyethylene protective cover and thread ring combination in the intelligent identification device of autonomous mine vehicle roads, the problem of difficult to quickly disassemble and dust damage in the mine vehicle detection device in the prior art is solved, and rapid disassembly and assembly are achieved and Dust protection improves the reliability and safety of the equipment.
Patent Information
- Application Number
- CN202420592363.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-26
- Publication Date
- 2024-11-01
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing mine car road drop detection device is difficult to disassemble and replace quickly when the environment inside the mine cave is complex, and it is easily damaged by dust and slag, which affects the reliability and safety of the equipment.
An intelligent identification device for roads of unmanned mine vehicles was designed, using a protective cover made of transparent polyethylene to protect the induction identification part, and the disassembly and assembly process of road intelligent identification components was simplified by the coordination of the external thread ring and the internal thread ring. At the same time, the design of protective panels and positioning blocks is used to reduce dust entry and improve the protection effect of components.
It realizes rapid disassembly and installation of intelligent road identification components, reduces the risk of damage caused by dust and debris, improves the reliability and safety of equipment, and ensures the stable operation of the mine car in the mine cave.
Smart Images

Figure CN221938223U_ABST
Abstract
Description
Technical Field
[0001] The utility model belongs to the field of mine car derailment detection, and specifically relates to an unmanned mine car road intelligent identification device. Background Art
[0002] The mine car in the prior art is a narrow-gauge railway transport vehicle for transporting bulk materials such as coal, ore and waste rock in mines. The environment inside the mine is complex, and the mine car is prone to derailment during movement. If it is not repaired in time, driving in a derailed state for a long time may easily cause the mine car to roll over and block the mine, posing a great safety hazard.
[0003] A Chinese patent with publication number CN218489656U discloses a mine car derailment detection device based on sound recognition, including a mine car bucket, a bottom wheel axle is fixedly connected to the bottom of the mine car bucket, a wheel axle rotating drum is fixedly connected to the left and right sides of the bottom wheel axle, a track wheel is rotatably connected to the outer ring of the wheel axle rotating drum, a mounting mechanism is fixedly connected to the outer side of the track wheel, a fixing mechanism is fixedly connected to the outer side of the track wheel at the center of the mounting mechanism, and a sound receiving device is fixedly connected to the left and right sides of the mine car bucket above the track wheel.
[0004] The above mine car derailment detection device is fixedly installed by bolts, and is arranged below the mine car, which makes it inconvenient for the staff to quickly disassemble and replace it when it is damaged.
[0005] In view of this, the present utility model is proposed. Utility Model Content
[0006] The technical problem to be solved by the utility model is to overcome the deficiencies of the prior art and provide an intelligent road recognition device for an unmanned mining vehicle.
[0007] In order to solve the above technical problems, the basic concept of the technical solution adopted by the utility model is:
[0008] An unmanned mine car road intelligent identification device comprises a mine car body, a track wheel is arranged at the bottom of the mine car body, a fixed frame is installed on opposite sides of the mine car body, a protective plate is arranged on the fixed frame, and a road intelligent identification component is arranged in the fixed frame;
[0009] A through groove is provided in the protective plate, a movable column is slidably fitted in the through groove, a clamping block is installed on one side of the movable column, a clamping groove is provided on one side of the fixed frame, the clamping block is clamped in the clamping groove, an internal thread ring is installed at the bottom of the fixed frame, an external thread ring is provided at the bottom of the road intelligent identification component, and the external thread ring is threadedly fitted in the internal thread ring.
[0010] Optionally, a protective cover is installed at the bottom of the fixed frame, the protective cover is located directly below the internal thread circle, and the protective cover is protruding from the bottom of the fixed frame bottom plate, and a pad is installed at the bottom of the top plate of the protective plate; the road intelligent identification component is threadedly fitted in the internal thread circle through the external thread circle, the protective cover is made of transparent polyethylene, and the sensing identification part at the bottom of the road intelligent identification component is located in the protective cover to protect it, thereby reducing the damage caused by dust and debris inside the unmanned mine car bouncing onto the sensing identification components of the road intelligent identification component during the driving of the unmanned mine car in the mine.
[0011] Optionally, a positioning groove is provided on one side of the fixed frame, and a positioning block is installed on one side of the side panel of the protective plate, the positioning block slides in the positioning groove, a placement groove is provided on one side of the fixed frame, and an L-shaped block is installed on the other side of the side panel of the protective plate, the L-shaped block cooperates with the placement groove, and a dust block is installed on one side of the top plate of the protective plate, and the dust blocks on both sides fit with the inner walls of the two side panels of the fixed frame; after the road intelligent identification component is installed in the fixed frame, the protective plate slides downward by locating the positioning block in the positioning groove, and the L-shaped block is located in the placement groove to limit and fix the side panels of the fixed frame, because the dust blocks on both sides fit with the inner walls of the two side panels of the fixed frame, the dust in the mine is reduced from entering through the gap between the protective plate and the fixed frame.
[0012] Optionally, a sliding groove is provided on one side of the protective plate, and the card block slides in the sliding groove, the bevel of the card block is set downward, and a limiting groove is provided on the other side of the protective plate, and a pull block is installed on the other side of the movable column, and the pull block slides in the limiting groove. A spring is sleeved on the movable column, and the spring is located in the sliding groove. When installing the protective plate, the positioning block of the protective plate is located in the positioning groove and slides downward, and the L-shaped block is located in the placement groove to limit and fix the side plate of the fixed frame. The bevel of the card block is facing downward, and the card block abuts against one side of the fixed frame. When passing through the card groove, the card block is located in the groove through the force of the spring to limit and fix the protective plate as a whole.
[0013] After adopting the above technical solution, the utility model has the following beneficial effects compared with the prior art. Of course, any product implementing the utility model does not necessarily need to achieve all the advantages described below at the same time:
[0014] 1. When the unmanned mining vehicle road intelligent identification device is installing the protective plate, the positioning block of the protective plate is located in the positioning slot and slides downward, the L-shaped block is located in the placement slot to limit and fix the side plate of the fixed frame, the bevel of the card block is facing downward, and the card block is abutted against one side of the fixed frame. When passing through the card slot, the card block is located in the slot through the force of the spring to limit and fix the protective plate as a whole. The card block is separated from the card slot by toggling the pull block, and then the protective plate is moved upward to disassemble it, so that the road intelligent identification component can be quickly disassembled and assembled.
[0015] 2. The unmanned mining vehicle road intelligent identification device fits the road intelligent identification component in the internal thread circle through the external thread circle. The protective cover is made of transparent polyethylene. The sensing identification part at the bottom of the road intelligent identification component is located in the protective cover to protect it, thereby reducing the damage caused by the dust and debris inside the unmanned mining vehicle bouncing onto the sensing identification components of the road intelligent identification component during the driving of the unmanned mining vehicle in the mine. The protective plate is installed on the fixed frame, and the pad block has a fixed abutment effect on the road intelligent identification component, which is convenient for quick disassembly and assembly of the road intelligent identification component.
[0016] The specific implementation of the utility model is further described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] The drawings described below are only some embodiments. For those skilled in the art, other drawings can be obtained based on these drawings without creative work. In the drawings:
[0018] Figure 1 A schematic diagram of the GhostNet design core of the utility model embodiment;
[0019] Figure 2 This is a schematic diagram of the structure of GhostConv in an embodiment of the utility model;
[0020] Figure 3 It is a schematic diagram of the structure of GhostBottelneck according to an embodiment of the utility model;
[0021] Figure 4 This is a schematic diagram of the C2fGhost structure of an embodiment of the utility model;
[0022] Figure 5 Schematic diagram of the GAM attention module of the embodiment of the utility model;
[0023] Figure 6 Schematic diagram of a channel attention submodule according to an embodiment of the present utility model;
[0024] Figure 7 This is a schematic diagram of the structure of the spatial attention submodule of an embodiment of the utility model;
[0025] Figure 8 It is a schematic diagram of the overall structure of the embodiment of the utility model;
[0026] Fig. 9 This is a structural schematic diagram of a cross-sectional view of the installation of a fixed frame and a protective plate according to an embodiment of the utility model;
[0027] Fig.10 This is a schematic diagram of the structure of the protective plate of an embodiment of the utility model;
[0028] Fig.11 This is a schematic diagram of the structure of the intelligent road identification components connected in the embodiment of the utility model;
[0029] In the accompanying drawings, the components represented by the reference numerals are listed as follows:
[0030] 1- mining car body, 2- track wheel, 3- fixed frame, 4- protective plate, 5- road intelligent identification component, 6- moving column, 301- card slot, 302- protective cover, 303- positioning slot, 304- placement slot, 305- internal thread ring, 401- dust block, 402- positioning block, 403- L-shaped block, 404- sliding slot, 405- limiting slot, 406- pad, 407- through slot, 501- external thread ring, 601- pull block, 602- spring, 603- card block.
[0031] It should be noted that these drawings and textual descriptions are not intended to limit the conceptual scope of the present invention in any way, but rather to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION
[0032] The utility model will now be described in further detail with reference to the accompanying drawings.
[0033] Example 1: Please refer to Figure 8-11 As shown, in this embodiment, an unmanned mine car road intelligent identification device is provided, comprising a mine car body 1, a track wheel 2 is arranged at the bottom of the mine car body 1, a fixing frame 3 is installed on opposite sides of the mine car body 1, a protective plate 4 is arranged on the fixing frame 3, and a road intelligent identification component 5 is arranged in the fixing frame 3;
[0034] A through groove 407 is provided in the protective plate 4, and a movable column 6 is slidably fitted in the through groove 407. A clamping block 603 is installed on one side of the movable column 6. A clamping groove 301 is provided on one side of the fixed frame 3, and the clamping block 603 is clamped in the clamping groove 301. An internal thread ring 305 is installed at the bottom of the fixed frame 3, and an external thread ring 501 is provided at the bottom of the road intelligent identification component 5, and the external thread ring 501 is threadedly fitted in the internal thread ring 305.
[0035] A protective cover 302 is installed at the bottom of the fixed frame 3 of this embodiment. The protective cover 302 is located directly below the internal thread ring 305, and the protective cover 302 is protruding from the bottom of the bottom plate of the fixed frame 3. A cushion block 406 is installed at the bottom of the top plate of the protective plate 4. The road intelligent identification component 5 is threadedly fitted in the internal thread ring 305 through the external thread ring 501. The protective cover 302 is made of transparent polyethylene. The inductive identification part at the bottom of the road intelligent identification component 5 is located in the protective cover 302 to protect it and reduce the damage caused by the dust and debris inside the unmanned mine car bouncing onto the inductive identification components of the road intelligent identification component 5 during the driving process in the mine. The protective plate 4 is installed on the fixed frame 3, and the cushion block 406 fixes and abuts the road intelligent identification component 5, which is convenient for quick disassembly and assembly of the road intelligent identification component 5.
[0036] In this embodiment, a positioning groove 303 is provided on one side of the fixed frame 3, and a positioning block 402 is installed on one side of the side plate of the protective plate 4. The positioning block 402 slides in the positioning groove 303. A placement groove 304 is provided on one side of the fixed frame 3, and an L-shaped block 403 is installed on the other side of the side plate of the protective plate 4. The L-shaped block 403 cooperates with the placement groove 304. A dust block 401 is installed on one side of the top plate of the protective plate 4, and the dust blocks 401 on both sides are in contact with the inner walls of the two side plates of the fixed frame 3; after the road intelligent identification component 5 is installed in the fixed frame 3, the protective plate 4 slides downward by locating the positioning block 402 in the positioning groove 303, and the L-shaped block 403 is located in the placement groove 304 to limit and fix the side plate of the fixed frame 3. Because the dust blocks 401 on both sides are in contact with the inner walls of the two side plates of the fixed frame 3, the dust in the mine is reduced from entering through the gap between the protective plate 4 and the fixed frame 3, thereby improving the protection of the road intelligent identification component 5.
[0037] A sliding groove 404 is provided on one side of the protective plate 4 of the present embodiment, and a block 603 is slidably fitted in the sliding groove 404, and the oblique side of the block 603 is set downward. A limiting groove 405 is provided on the other side of the protective plate 4, and a pull block 601 is installed on the other side of the moving column, and the pull block 601 is slidably fitted in the limiting groove 405. A spring 602 is sleeved on the moving column 6, and the spring 602 is located in the sliding groove 404; when the protective plate 4 is installed, the positioning block 402 of the protective plate 4 is located in the positioning groove 303 and slides downward. The L-shaped block 403 is located in the placement slot 304 to limit and fix the side plate of the fixed frame 3. The hypotenuse of the block 603 is facing downward, and the block 603 is in contact with one side of the fixed frame 3. When passing through the slot 301, through the force of the spring 602, the block 603 is located in the slot 301 to limit and fix the protective plate 4 as a whole. By toggling the pull block 601 to separate the block 603 from the slot 301 and then moving the protective plate 4 upward to disassemble it, the road intelligent recognition component 5 can be quickly disassembled and assembled.
[0038] Example 2: Figure 1-7 As shown, in this embodiment, a road intelligent recognition component 5 is provided. This road intelligent recognition component 5 has the function of taking the transporting mining trucks, loading mining trucks, passenger vehicles and pedestrians in the open-pit mine as the objects of target detection, while ensuring the accuracy of the detection and considering the real-time detection. Ensure that the unmanned mining trucks can operate safely when encountering obstacles, oncoming vehicles and other events during driving. In view of the large differences in target sizes in open-pit mines, as well as the problems of algorithm network structure and training methods, an open-pit mine target detection optimization algorithm based on yolov8 is proposed. The algorithm proposes different solutions according to the following problems:
[0039] 1. Improve the problem of insufficient feature extraction capabilities for various targets in mining environments. If target feature information cannot be extracted in a timely and accurate manner, the safety of unmanned mining trucks will be lost.
[0040] The solution to this problem is to design the C2f_Ghost module to replace the original C2f, thereby improving the feature extraction capability and making the model lightweight.
[0041] GhostNet is a lightweight network designed by Huawei Noah's Ark Laboratory in 2020. Ghostconv is a convolutional module in the GhostNet network that can replace ordinary convolution. Figure 1 As shown in the figure, the GhostNet lightweight network model can reduce the amount of network calculation and parameters while maintaining the size of the original convolution output feature map and the channel size. First, a small number of ordinary convolution kernels are used to extract feature information from the input feature map, and then a linear change operation is performed on the feature map, which is cheaper than ordinary convolution. Finally, the final feature map is generated through Concat. Figure 2 As shown in the figure, Cheapoperation is a cheap linear operation. GhostConv first uses a convolution with half the size of the original convolution to generate half of the feature map, and then continues to use a 5×5 convolution kernel and a cheap Cheapoperation with a step size of 1 to obtain the other half of the feature map. Finally, the two parts of the feature map are spliced into a complete feature map through the Concat operation.
[0042] like Figure 3 As shown in the figure, GhostBottelneck first passes through the first GhostConv as an expansion layer to increase the number of channels, then passes through regularization and SiLU activation function, and then passes through the second GhostConv to reduce the number of channels of the output feature map to match the number of input channels, and finally adds the feature map obtained in the previous step to the residual edge for feature fusion.
[0043] like Figure 4 As shown in the figure, the designed new module C2fGhost is obtained by replacing all the Bottelneck in the C2f module in the original network with GhostBottelneck. This structure uses the cross-stage feature fusion strategy and truncated gradient flow technology to enhance the variability of learning features between different network layers, thereby reducing the impact of redundant gradient information and enhancing the learning ability of the network. The introduction of GhostConv and C2fGhost modules reduces a large number of 3×3 ordinary convolutions in the original structure, greatly compresses the model size of the network, reduces the number of parameters and the amount of calculation, and enables the model to be deployed on the mobile terminal.
[0044] 2. Improve the problem of mutual occlusion of various targets in the mining environment. Due to the limited camera angle, there are problems of mutual occlusion between mining trucks, between mining trucks and loading vehicles, between mining trucks and passenger vehicles, and even between pedestrians. Due to the poor air quality in the mine, the environment is full of dust, which makes the camera full of stains, and this factor will also cause the problem of detection targets being blocked.
[0045] The solution to this problem is to introduce the GAM attention mechanism to improve the feature expression ability and the feature fusion ability of the model.
[0046] Since the mine scene is relatively empty and the background occupies most of the image, in order to improve the model's ability to express small targets, the GAM (global attention mechanism) attention mechanism is introduced into the algorithm model of this study. GAM is a new attention module after CBAM. Its structure is as follows Figure 5 As shown in Figure 1, it is a module that combines the channel attention mechanism and the spatial attention mechanism. It is an attention mechanism that can reduce diffuse information while amplifying the global dimension feature interaction. Given a feature map F1, after two independent attention sub-modules, the original feature map is first corrected by the channel attention mechanism module to obtain the intermediate state F2, and then corrected again by the spatial attention mechanism module to obtain the final feature map F3. The channel attention sub-module is shown in Figure 1. Figure 6 As shown in Figure 1, a three-dimensional arrangement is used to preserve the three-dimensional information, and then a two-layer multi-layer perceptron MLP is used to amplify the cross-dimensional channel spatial dependency. The intermediate state is defined as (where MC is the channel attention map, represents element-wise multiplication):
[0047]
[0048] The spatial attention submodule is as follows Figure 7As shown, the size of the input feature map F2 is C×H×W, and two 7×7 convolutions are used to model the nonlinear relationship between pixels in the 7×7 square, so the parameters can better capture the relationship between pixel spaces. The intermediate state is defined as follows (where MS is the spatial attention map):
[0049]
[0050] 3. Improve the problem of false detection and missed detection of small targets in the distance and targets in dense areas. Due to the large mining environment, distant targets are smaller than nearby targets, and there may be problems of missed detection and false detection.
[0051] The solution to this problem is to introduce the SIOU loss function to solve the problem of mismatch between the true box and the predicted box direction, so that the model can better learn the location information.
[0052] The original YOLOv8 model uses DFLLoss+CIoULoss as regression loss, but CIoU has some ambiguity in the relative value of the intersection and parallel ratio description, and does not consider the balance of difficult and easy samples. Therefore, this paper introduces GIoULoss to replace CIoULoss. GIoU, as an improved form of IoU, also has the advantage of scale invariance. In addition, compared with IoU, GIoU also considers the non-overlapping area of the target, which makes up for the shortcoming that the IoU boundary loss function cannot quantify the non-intersection between the real box and the predicted box. GIoU is defined as:
[0053]
[0054] In the formula, C is the minimum rectangular box surrounding A and B, the value range of IoU is [0, 1], and the value range of GIoU is [-1, 1]. The definition of IoU is:
[0055]
[0056] GIoU is used as a loss function and is expressed as:
[0057] L GIoU =1-GIoU
[0058] Research on road segmentation of unmanned mining trucks in open-pit mines
[0059] With an open-pit mine as the background, the mine road is segmented. Improve the accuracy of edge segmentation to ensure the safety of mine trucks and avoid rollovers due to false detection and other problems. The DeepLabV3+ model has good performance in segmentation tasks of various data sets, so this research task on the segmentation of unmanned mine trucks in open-pit mines uses DeepLabV3+ as the basic model and optimizes it on this basis. The algorithm proposes different solutions based on the following two problems:
[0060] Improve the completeness of drivable area segmentation and strengthen the ability to extract road features.
[0061] Since the color of the open-pit mine road is similar to the background, there are many potholes, and the road captured by the camera when the mining truck is reversing and turning is limited, these factors will lead to the problem of incompleteness in the segmented drivable area.
[0062] The solution to this problem is to design the MobileNetV2 module to enhance feature extraction capabilities while achieving lightweight.
[0063] MobileNetV2 is a lightweight convolutional neural network with a simple model structure and small number of parameters. MobileNetV2 adopts two innovative designs: the reverse residual structure and the linear bottleneck layer. The reverse residual structure can more effectively solve the model degradation problem caused by the increase in the number of network layers and increase the feature extraction ability of the model; the linear bottleneck can improve the accuracy of the model while reducing the model parameters and the amount of calculation.
[0064] Improved the accuracy of drivable area boundaries.
[0065] There is no obvious color difference between the drivable area and the non-drivable area, and the boundary is not clear enough, which easily leads to inaccurate boundaries in the segmentation of the drivable area.
[0066] The solution to this problem is to add a CBAM attention module to improve feature expression and feature fusion capabilities.
[0067] The attention module CBAM combines the channel attention module and the spatial attention module and processes them in series. It combines the information of the spatial dimension with the channel characteristics, which can not only strengthen the target features but also consider the target position information.
[0068] The present invention is not limited to the above-mentioned implementation modes. Anyone should be aware of the structural changes made under the inspiration of the present invention. Any technical solution that is the same or similar to the present invention falls within the protection scope of the present invention. The technology, shape, and structure that are not described in detail in the present invention are all known technologies.
Claims
1. An intelligent road recognition device for unmanned mining vehicles, characterized in that: include: A mine car body (1), wherein a track wheel (2) is arranged at the bottom of the mine car body (1), a fixing frame (3) is installed on opposite sides of the mine car body (1), a protective plate (4) is arranged on the fixing frame (3), and a road intelligent identification component (5) is arranged in the fixing frame (3); A through slot (407) is provided in the protective plate (4), a movable column (6) is slidably fitted in the through slot (407), a clamping block (603) is installed on one side of the movable column (6), a clamping slot (301) is provided on one side of the fixed frame (3), the clamping block (603) is clamped in the clamping slot (301), an internal thread ring (305) is installed at the bottom of the fixed frame (3), an external thread ring (501) is provided at the bottom of the intelligent road recognition component (5), and the external thread ring (501) is threadably fitted in the internal thread ring (305).
2. The unmanned mine car road intelligent identification device according to claim 1 is characterized in that: A protective cover (302) is installed at the bottom of the fixed frame (3), the protective cover (302) is located directly below the internal thread ring (305), and the protective cover (302) is protruding from the bottom of the bottom plate of the fixed frame (3), and a cushion block (406) is installed at the bottom of the top plate of the protective plate (4).
3. The intelligent road recognition device for unmanned mining vehicles according to claim 1 is characterized in that: A positioning groove (303) is provided on one side of the fixing frame (3), and a positioning block (402) is installed on one side of the side plate of the protection plate (4), and the positioning block (402) is slidably fitted in the positioning groove (303).
4. The unmanned mine car road intelligent identification device according to claim 1 is characterized in that: A placement groove (304) is provided on one side of the fixed frame (3), an L-shaped block (403) is installed on the other side of the side plate of the protective plate (4), the L-shaped block (403) is matched with the placement groove (304), a dustproof block (401) is installed on one side of the top plate of the protective plate (4), and the dustproof blocks (401) on both sides are in contact with the inner walls of the two side plates of the fixed frame (3).
5. The unmanned mine car road intelligent identification device according to claim 1 is characterized in that: A sliding groove (404) is provided on one side of the protection plate (4), and the clamping block (603) is slidably fitted in the sliding groove (404), with the oblique edge of the clamping block (603) being arranged downward.
6. The intelligent road recognition device for unmanned mining vehicles according to claim 5 is characterized in that: A limiting groove (405) is provided on the other side of the protective plate (4), a pulling block (601) is installed on the other side of the movable column, and the pulling block (601) is slidably fitted in the limiting groove (405). A spring (602) is sleeved on the movable column (6), and the spring (602) is located in the sliding groove (404).
