Machine vision-based control method and system for a chute mouth crusher and medium

Abstract

The application discloses a machine vision-based control method and system for a chute mouth crusher and a medium, relates to the technical field of automatic control, and constructs a visual monitoring area of a target chute mouth, generates a real-time crushing control strategy in real time according to visual data of the visual monitoring area to guide the work of the crusher; the crushing control strategy comprises a crusher position strategy, a crusher operation strategy, a crusher arm state strategy and a safety alarm strategy; the crusher can assist manual underground crushing operation, can send alarm information on the spot when dangerous conditions are encountered, and can avoid dangers and accidents; the crusher can realize intelligent perception and self-adaptive control capability for the ore crushing process, and can realize dynamic optimization and adjustment of crushing operation parameters based on the crusher operation strategy.

Description

Technical Field

[0001] This invention relates to the field of automatic control technology, specifically to a machine vision-based control method, system, and medium for a chute crusher. Background Technology

[0002] As a key industrial piece of equipment in the mining industry, crushers are mainly used for secondary crushing of large ore blocks to meet the particle size requirements of ore transportation systems and improve overall transportation efficiency. Traditional crushers typically use mechanical compression or impact crushing methods to crush large ore blocks at the underground chute, and their performance directly affects the continuity and stability of the mining production system.

[0003] With the advancement of mining automation technology, crushers at underground chutes are now generally equipped with remote operating systems. Operators can remotely control the crusher to perform crushing operations via a control panel in the dispatch room using manual control devices such as joysticks. While this remote control method avoids the risk of operators being directly exposed to the harsh underground working environment, it still has significant drawbacks in practical applications: First, manual control requires operators to maintain a high level of concentration to accurately identify ore blockage points and adjust crusher operating parameters. Especially when dealing with irregular ores, the response is delayed, resulting in low single-cycle crushing efficiency. Second, during peak mining periods, when ore output surges, companies typically increase operating shifts and assign multiple teams of operators to work in shifts. This not only significantly increases labor costs but also makes it difficult to effectively improve overall crushing efficiency due to the physical limitations of manual operation. Third, prolonged and intensive human-machine interaction can easily lead to operator visual fatigue and operational errors, potentially causing abnormal operating conditions such as equipment idling and overload, affecting equipment lifespan.

[0004] While remote control systems for crushers exist in existing technologies, they are mostly limited to simple command transmission and execution, lacking intelligent perception and adaptive control capabilities for the ore crushing process, making it difficult to dynamically optimize and adjust crushing operation parameters. Especially when facing complex and ever-changing ore conditions underground, existing systems still heavily rely on operator experience and manual adjustments, failing to form a complete intelligent operation loop. This has become a key technological bottleneck restricting the improvement of crushing operation efficiency in mines. Summary of the Invention

[0005] The technical problem this invention aims to solve is that while existing remote control systems for crushers exist, they are mostly limited to simple command transmission and execution, lacking intelligent perception and adaptive control capabilities for the ore crushing process, making it difficult to dynamically optimize and adjust crushing operation parameters. This invention specifically designs a machine vision-based crusher control method, system, and medium for crushing operations at ore chutes. It constructs a visual monitoring area for the target ore chute and generates real-time crushing control strategies based on visual data from this area to guide the crusher's operation. The crushing control strategies include: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy. This assists manual underground crushing operations, issuing alarm information on-site in case of danger to prevent accidents. It achieves intelligent perception and adaptive control capabilities for the ore crushing process, while simultaneously enabling dynamic optimization and adjustment of crushing operation parameters based on the crusher operation strategy.

[0006] This invention is achieved through the following technical solution:

[0007] This solution provides a machine vision-based control method for a chute crusher, including:

[0008] Construct a visual monitoring zone for the target wellhead and collect visual data from the monitoring zone in real time;

[0009] Preprocess the visual data;

[0010] The preprocessed visual data is input into the pre-built crusher control model to generate a real-time crushing control strategy; the crushing control strategy includes: crusher position strategy, crusher operation strategy, crusher arm status strategy and safety alarm strategy;

[0011] The crusher executes the crushing control strategy.

[0012] A further optimized solution involves constructing a visual monitoring zone for the target wellhead and collecting visual data from the monitoring zone in real time; including the following methods:

[0013] Based on the grid area at the target manhole opening, construct a vehicle identification zone, an emergency stop zone, and an alarm and deflection zone:

[0014] Based on the basic area, construct a vehicle recognition zone, an emergency stop zone, and an alarm-based deterrence zone:

[0015] Obtain the intersection area A of the base area, the total area occupied by the crusher in working state, and the total area occupied by the crusher in stopped state. Use the largest circumscribed quadrilateral of the intersection area A as the emergency stop area. The largest circumscribed quadrilateral is the quadrilateral that overlaps the most with the edge of the intersection area A. The crusher stopped state is the state in which the mechanical arm stops at a preset stopping position when the crusher is in the working position. The vertical projection of the preset stopping position onto the ground has no intersection with the base area or the intersection is minimal.

[0016] Obtain the working area B of the transport vehicle, and use the largest circumscribed quadrilateral of the working area B as the vehicle identification area.

[0017] Obtain the intersection area C of the vehicle recognition area and the emergency stop area, and use the intersection area C as the alarm and drive-away area;

[0018] Visual monitoring devices are deployed in the visual judgment area to collect visual data in real time.

[0019] A further optimization scheme is that the method for generating the crusher position strategy includes:

[0020] The position is determined based on the preprocessed visual data. If there are no blockages in the basic area and no vehicles in the vehicle recognition area within the preset time period T1, the crusher returns to the crusher stop area as the crusher position strategy; otherwise, the crusher keeps the working position as the crusher position strategy.

[0021] A further optimization is that the method for generating the crusher arm state strategy includes:

[0022] The arm status is determined based on the preprocessed visual data. When there are stationary or moving vehicles in the vehicle recognition area, the crusher arm stop status is used as the crusher arm status strategy.

[0023] A further optimization is that the method for generating the security alarm strategy includes:

[0024] Based on the preprocessed visual data, the safety status is judged. When there are non-transport vehicles or people in the alarm removal zone, the alarm removal is used as a safety alarm strategy.

[0025] A further optimization is that the method for generating the crusher operation strategy includes:

[0026] When a vehicle leaves the vehicle identification zone after the preset time period T2, the operation strategy judgment is initiated:

[0027] All blockage points in the basic area were detected;

[0028] Using an n×n grid as a window and a grid as the sliding distance, perform sliding window calculations on the basic area and calculate the blockage value of each window.

[0029] Construct a blockage heatmap based on the blockage values ​​of each window;

[0030] The working path of the crusher's robotic arm is generated based on the blockage heat map.

[0031] A further optimized solution is that the method for generating the working path of the crusher robotic arm based on the blockage heat map includes:

[0032] The blockage heatmap is divided into multiple sub-maps based on the blockage value; each sub-map includes one or more target color regions.

[0033] The working path of the crusher's robotic arm is generated using the blockage center of each sub-graph as the working point. The method for determining the blockage center includes: when the sub-graph includes a target color area, the center of the target color area is used as the blockage center; when the sub-graph includes multiple target color areas, the center of the target color area with the largest area is used as the blockage center.

[0034] A further optimized solution involves dividing the blockage heatmap into multiple sub-maps based on the blockage value, including the following method:

[0035] Obtain the target color region with the highest blockage value in the blockage heatmap. If the area of ​​the current target color region exceeds the first preset threshold, divide the current target color region into a separate sub-map.

[0036] If the area of ​​the current target color region is less than the second preset threshold, then determine whether there is a neighboring target color region that is less than L away from the current target color region, and the area of ​​the neighboring target color region is less than the second preset threshold; if there is, then divide the current target color region and the neighboring target color region into the same sub-graph; otherwise, divide the current target color region into a separate sub-graph.

[0037] This solution also provides a machine vision-based chute crusher control system to implement the aforementioned machine vision-based chute crusher control method; the system includes:

[0038] The region construction module is used to construct the visual monitoring area of ​​the target wellhead and collect the visual data of the visual monitoring area in real time.

[0039] The preprocessing module is used to preprocess visual data;

[0040] The strategy generation module is used to input preprocessed visual data into the pre-built crusher control model to generate real-time crushing control strategies; the crushing control strategies include: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy;

[0041] The execution module is used by the crusher to execute the crushing control strategy.

[0042] This solution also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, can implement the machine vision-based chute crusher control method described above.

[0043] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0044] 1. The present invention provides a machine vision-based method, system, and medium for controlling a crusher at a chute; it constructs a visual monitoring area for the target chute, and generates real-time crushing control strategies based on the visual data of the monitoring area to guide the crusher's operation; the crushing control strategies include: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy; it assists manual operation in underground crushing operations, and can issue alarm information on-site in case of dangerous situations to avoid danger and accidents; it realizes intelligent perception and adaptive control capabilities for the ore crushing process, and simultaneously achieves dynamic optimization and adjustment of crushing operation parameters based on the crusher operation strategy.

[0045] 2. The present invention provides a machine vision-based control method, system and medium for a chute crusher; the working path of the crusher's robotic arm is generated based on the blockage heat map, and the working point is the blockage center of each sub-map; the crusher's robotic arm is generated by prioritizing the crushing of areas with dense blockage points, thereby improving the working efficiency of the crusher and the chute. Attached Figure Description

[0046] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0047] Figure 1 This is a schematic diagram of a machine vision-based control method for a chute crusher.

[0048] Figure 2 This is a schematic diagram showing the construction result of the visual monitoring area;

[0049] Figure 3 This is a schematic diagram of the assembly of a visual monitoring device;

[0050] Figure 4 This is a schematic diagram of coordinate system mapping;

[0051] Figure 5 This is a schematic diagram of a sliding window;

[0052] Figure 6 This is a schematic diagram of the control process of the chute crusher in Example 3. Detailed Implementation

[0053] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0054] The main components of a crusher are: the main structure, the electrical control cabinet, and the operating device. The crusher can quickly process large stones accumulated on the crushing screen, rapidly clearing and loosening the material piled up on the screen to ensure smooth feed.

[0055] While remote control systems for crushers exist in the existing technology, they are mostly limited to simple command transmission and execution, lacking intelligent perception and adaptive control capabilities for the ore crushing process, making it difficult to dynamically optimize and adjust crushing operation parameters. In view of this, this solution provides the following embodiments to solve the above-mentioned technical problems.

[0056] Example 1

[0057] This embodiment provides a machine vision-based control method for a chute crusher, such as... Figure 1 As shown, it includes:

[0058] Step 1: Construct a visual monitoring zone for the target wellhead and collect visual data from the monitoring zone in real time; this step specifically includes the following methods:

[0059] S11, such as Figure 2 Based on the grid area at the target manhole opening, construct a vehicle identification zone, an emergency stop zone, and an alarm and deflection zone:

[0060] Based on the basic area, a vehicle identification area, an emergency stop area, and an alarm drive-away area are constructed: The intersection area A of the basic area, the total area occupied by the crusher in working state, and the total area occupied by the crusher in stopped state is obtained. The largest circumscribed quadrilateral of the intersection area A is used as the emergency stop area; the largest circumscribed quadrilateral represents the quadrilateral that overlaps most with the edge of the intersection area A; the crusher stopped state represents the state where the robotic arm is stopped at a preset stopping position when the crusher is in the working position, and the vertical projection of the preset stopping position onto the ground has no intersection with or minimal intersection with the basic area.

[0061] The total occupied area of ​​the crusher in working state refers to the total occupied area corresponding to all postures of the robotic arm when the crusher is in working position.

[0062] Obtain the working area B of the transport vehicle, and use the largest circumscribed quadrilateral of the working area B as the vehicle identification area.

[0063] Obtain the intersection area C of the vehicle recognition area and the emergency stop area, and use the intersection area C as the alarm and drive-away area;

[0064] S12, Deploy visual monitoring devices in the visual judgment area to collect visual data in real time. Specifically, this is done by deploying network cameras in the visual judgment area to collect visual data in real time.

[0065] The crusher is generally installed centered on one side of the chute opening; the specific side it is installed on depends on the actual situation and does not affect the implementation of this method. Figure 3 As shown, in this embodiment, the camera is installed on the top of the chute, and the position is selected so that it can completely capture the screen and the transport vehicle to be unloaded, while minimizing image distortion. The specific installation method and angle can be as simple as possible to meet the image requirements of the crusher control model.

[0066] Step two involves preprocessing the visual data; the specific preprocessing process mainly includes data cleaning and coordinate system unification.

[0067] Computer vision (crusher control model) coordinate systems are generally image-based. However, the actual on-site conditions, captured by cameras and input into the computer, suffer from scaling and distortion issues, meaning the computer vision coordinate system cannot reflect real-world coordinates. Furthermore, the crusher itself is rotatable, and its working area is a fan-shaped region; its coordinate position cannot be represented by simple x and y coordinates, so the computer vision coordinate system cannot directly generate commands to control the crusher. Therefore, this solution establishes a real-world coordinate system based on the screen grid. From the crusher's perspective, the lower left position is the origin, the horizontal axis is the y-axis, and the vertical axis is the x-axis. The computer converts the pixel coordinates of the blockage points detected in the image into screen grid hole coordinates. After sending these coordinates to the crusher, the crusher converts the screen grid information into control parameter information for each joint to execute the crushing and unblocking operation.

[0068] In summary, as Figure 4 As shown, a real-world coordinate system is established based on the grid screen. From the perspective of the crusher, the origin of the coordinate system is the lower left position, the horizontal axis is the y-axis, and the vertical axis is the x-axis. The pixel coordinates of the blockage points detected in the image are converted into grid screen coordinates. After the coordinates are sent to the crusher, the crusher converts the grid screen information into control parameter information for each joint to perform crushing and unblocking operations.

[0069] Step 3: Input the preprocessed visual data into the constructed crusher control model to generate a real-time crushing control strategy; the crushing control strategy includes: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy.

[0070] In step three, the method for generating the crusher position strategy includes:

[0071] The position is determined based on the preprocessed visual data. If there are no blockages in the basic area and no vehicles in the vehicle recognition area within the preset time period T1, the crusher returns to the crusher stop area as the crusher position strategy; otherwise, the crusher keeps the working position as the crusher position strategy.

[0072] The method for generating the crusher arm state strategy includes:

[0073] The arm status is determined based on the preprocessed visual data. When there are stationary or moving vehicles in the vehicle recognition area, the crusher arm stop status is used as the crusher arm status strategy.

[0074] The method for generating the security alarm strategy includes:

[0075] Based on the preprocessed visual data, the safety status is judged. When there are non-transport vehicles or people in the alarm removal zone, the alarm removal is used as a safety alarm strategy.

[0076] In this embodiment, the determination of position, arm status, operation strategy, and safety status mainly rely on image processing technology in machine vision technology.

[0077] The method for generating the crusher operation strategy includes:

[0078] When a vehicle leaves the vehicle identification zone after the preset time period T2, the operation strategy judgment is initiated:

[0079] S31, detects all blockage points in the basic area;

[0080] S32, using an n×n grid as a window and a grid as the sliding distance, perform sliding window calculations on the basic area and calculate the blockage value of each window;

[0081] S33, Construct a blockage heatmap based on the blockage values ​​of each window;

[0082] S34, Generate the working path of the crusher robotic arm based on the blockage heat map. This step specifically includes the following methods:

[0083] S341, Divide the blockage heatmap into multiple sub-maps according to the blockage value; each sub-map includes one or more target color regions; this step specifically includes the following method:

[0084] Obtain the target color region with the highest blockage value in the blockage heatmap. If the area of ​​the current target color region exceeds the first preset threshold, divide the current target color region into a separate sub-map.

[0085] If the area of ​​the current target color region is less than the second preset threshold, then determine whether there is a neighboring target color region that is less than L away from the current target color region, and the area of ​​the neighboring target color region is less than the second preset threshold; if there is, then divide the current target color region and the neighboring target color region into the same sub-graph; otherwise, divide the current target color region into a separate sub-graph.

[0086] S342, using the blockage center of each sub-graph as the work point, generate the work path of the crusher robotic arm; the method for determining the blockage center includes: when the sub-graph includes a target color area, the center of the target color area is used as the blockage center; when the sub-graph includes multiple target color areas, the center of the target color area with the largest area is used as the blockage center.

[0087] In this embodiment, the sliding window calculation process specifically includes the following steps:

[0088] like Figure 5 As shown, starting with the top left corner as the origin, select the first 3x3 grid area (red) as the first window. Calculate how many of the 9 points in the current area are blocked. Then, move the entire area one grid to the right (yellow area). After calculating, move it one more grid, and so on, until it reaches the far right. After calculating the blockage in the top three rows, return the x-axis to the origin and move the y-axis down one row (blue area), and so on, calculating to the right again. Repeat the above process until the window slides to cover the last point in the bottom right corner. After the calculation is complete, a heatmap of the blockage is obtained.

[0089] The working path of the crusher robotic arm obtained by the above steps is still based on the pixel information of the image. The pixel information needs to be converted into grid screen coordinate information through mapping relationship calculation before it is sent to the crusher. The crusher performs crushing operation according to the order of the points.

[0090] The method for generating the crusher operation strategy also includes:

[0091] During crusher operation, operation decisions are made based on pre-processed visual data: when there are vehicles or people in the emergency stop zone, emergency stop operation is adopted as the crusher operation strategy; when there are no vehicles or people in the emergency stop zone, the crusher resumes operation.

[0092] Step four: The crusher executes the crushing control strategy.

[0093] Example 2

[0094] This embodiment provides a machine vision-based control system for a chute crusher, used to implement the machine vision-based control method for a chute crusher described in Embodiment 1; the system includes:

[0095] The region construction module is used to construct the visual monitoring area of ​​the target wellhead and collect the visual data of the visual monitoring area in real time.

[0096] The preprocessing module is used to preprocess visual data;

[0097] The strategy generation module is used to input preprocessed visual data into the pre-built crusher control model to generate real-time crushing control strategies; the crushing control strategies include: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy;

[0098] The execution module is used by the crusher to execute the crushing control strategy.

[0099] Example 3

[0100] This embodiment provides a computer-readable medium storing a computer program, which, when executed by a processor, can implement the machine vision-based chute crusher control method as described in Embodiment 1; specifically, the following steps are performed:

[0101] Step 1: Construct a visual monitoring area for the target wellhead and collect visual data of the monitoring area in real time;

[0102] Step two: Preprocess the visual data;

[0103] Step 3: Input the preprocessed visual data into the constructed crusher control model to generate a real-time crushing control strategy; the crushing control strategy includes: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy.

[0104] Step four: The crusher executes the crushing control strategy.

[0105] like Figure 6 As shown, the crusher control model stores data models of the screen and transport vehicles. At the same time, it connects to the camera to obtain real-time video stream information, identify whether people or vehicles are in the alarm drive-away area, and if so, execute the alarm drive-away according to the safety alarm strategy; identify whether people or vehicles are in the emergency stop area, and if so, control the crusher to stop operation in an emergency to avoid accidents.

[0106] When a transport vehicle is detected about to begin dumping operations (when there is a stationary or moving vehicle in the vehicle identification area), the crusher arm is stopped according to the crusher arm status strategy to avoid collisions between the robotic arm and the transport vehicle, and also to prevent the robotic arm from blocking the screen and causing misidentification. After the vehicle leaves the vehicle identification area for a preset time period T2 (T2 is 5 seconds in this embodiment, as the ore is generally stationary 5 seconds after the vehicle leaves), the operation strategy judgment is initiated: the blockage status of the screen in 25 frames of images is continuously detected, the detection results of the 25 frames of images are summarized and deduplicated to obtain the pixel values ​​of the blockage points in the images, a blockage heat map is constructed based on the blockage values ​​of each window, and the operation path of the crusher robotic arm is generated according to the blockage heat map. After receiving the crushing point information, the crusher converts the point information into parameter information for the control of each joint, and controls each joint to perform unblocking operations according to the parameter information.

[0107] Repeat the above process. When the crusher has cleared all the blockage points or can skip any points that have not yet been worked on, the current blockage clearing operation is considered complete, and the crusher moves to the stop point.

[0108] After the crushing process is stopped, the screen is checked for debris. If any is found, the robotic arm is controlled to move the debris to the edge of the screen, and then it is removed manually.

[0109] The crusher control model evaluates the identification results of each screen blockage point. When the model's identification accuracy decreases, the system automatically saves the vehicle unloading process as an image and issues a prompt in the system. After seeing the prompt, the system administrator enters the image warehouse to confirm whether the saved image truly failed to identify the screen blockage point. If the blockage was indeed not identified, the blockage situation needs to be labeled and saved. When the number of labeled images reaches 500, the system automatically uses the newly labeled images to perform incremental training on the model and automatically replaces the original model for real-time analysis.

[0110] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A machine vision-based control method for a chute crusher, characterized in that, include: Construct a visual monitoring zone for the target wellhead and collect visual data from the monitoring zone in real time; Preprocess the visual data; The preprocessed visual data is input into the constructed crusher control model to generate a real-time crushing control strategy; The crushing control strategy includes: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy; The crusher executes the aforementioned crushing control strategy; The method for constructing a visual monitoring zone for the target wellhead and collecting visual data from the monitoring zone in real time includes: Based on the grid area of ​​the target manhole opening, a vehicle identification zone, an emergency stop zone, and an alarm and deflection zone are constructed. Obtain the union region A of the basic region, the total area occupied by the crusher in working state, and the total area occupied by the crusher in stopped state. Use the largest circumscribed quadrilateral of the union region A as the emergency stop zone. The largest circumscribed quadrilateral is the quadrilateral that overlaps the most with the edge of the union region A. The crusher stopped state refers to the state in which the mechanical arm stops at a preset stopping position when the crusher is in the working position. The vertical projection of the preset stopping position onto the ground has no intersection with the basic region or the intersection is minimal. Obtain the working area B of the transport vehicle, and use the largest circumscribed quadrilateral of the working area B as the vehicle identification area. Obtain the union area C of the vehicle recognition area and the emergency stop area, and use the union area C as the alarm drive-away area; Visual monitoring devices are deployed in the visual monitoring area to collect visual data in real time. The method for generating the crusher operation strategy includes: When a vehicle leaves the vehicle identification zone after a preset time period T2, the operation strategy is initiated to determine: all congestion points in the basic area are detected; using an n×n grid as a window and a grid as the sliding distance, a sliding window calculation is performed on the basic area to calculate the congestion value of each window; a congestion heat map is constructed based on the congestion value of each window. The working path of the crusher's robotic arm is generated based on the blockage heat map; The method for generating the working path of the crusher robotic arm based on the blockage heat map includes: The blockage heatmap is divided into multiple sub-maps based on the blockage value; each sub-map includes one or more target color regions. The working path of the crusher's robotic arm is generated using the blockage center of each sub-graph as the working point; the method for determining the blockage center includes: when the sub-graph includes a target color area, the center of the target color area is taken as the blockage center; when the sub-graph includes multiple target color areas, the center of the target color area with the largest area is taken as the blockage center. The method for dividing the blockage heatmap into multiple sub-maps according to the blockage value includes: Obtain the target color region with the highest blockage value in the blockage heatmap. If the area of ​​the current target color region exceeds the first preset threshold, divide the current target color region into a separate sub-map. If the area of ​​the current target color region is less than the second preset threshold, then determine whether there is a neighboring target color region that is less than L away from the current target color region, and the area of ​​the neighboring target color region is less than the second preset threshold; if there is, then divide the current target color region and the neighboring target color region into the same sub-graph; otherwise, divide the current target color region into a separate sub-graph.

2. The machine vision-based control method for a chute crusher according to claim 1, characterized in that, The method for generating the crusher position strategy includes: The position is determined based on the preprocessed visual data. When there are no blockages in the basic area and no vehicles in the vehicle recognition area within the preset time period T1, the crusher returns to the crusher stop area as the crusher position strategy; otherwise, the crusher keeps the working position as the crusher position strategy.

3. The machine vision-based control method for a chute crusher according to claim 1, characterized in that, The method for generating the crusher arm state strategy includes: The arm status is determined based on the preprocessed visual data. When there are stationary or moving vehicles in the vehicle recognition area, the crusher arm stop status is used as the crusher arm status strategy.

4. The machine vision-based control method for a chute crusher according to claim 1, characterized in that, The method for generating the security alarm strategy includes: Based on the preprocessed visual data, the safety status is judged. When there are non-transport vehicles or people in the alarm removal zone, the alarm removal is used as a safety alarm strategy.

5. A machine vision-based control system for a chute crusher, characterized in that, The system is used to implement the machine vision-based control method for a chute crusher as described in any one of claims 1-4; the system includes: The region construction module is used to construct the visual monitoring area of ​​the target wellhead and collect the visual data of the visual monitoring area in real time. The preprocessing module is used to preprocess visual data; The strategy generation module is used to input preprocessed visual data into the pre-built crusher control model to generate real-time crushing control strategies; the crushing control strategies include: crusher position strategy, crusher operation strategy, crusher arm status strategy, and safety alarm strategy; The execution module is used by the crusher to execute the crushing control strategy.

6. A computer-readable medium having a computer program stored thereon, characterized in that, The computer program, executed by a processor, can implement the machine vision-based chute crusher control method as described in any one of claims 1-4.

Images