Intelligent car dumper residual coal cleaning method

Abstract

The invention provides an intelligent car dumper residual coal cleaning method which comprises the following steps: adopting a human body detection and tracking algorithm based on deep learning to monitor the positions and behaviors of personnel in an operation area in real time, and if it is detected that the personnel enter a dangerous area or have unsafe behaviors, triggering an alarm mechanism and controlling a cantilever to stop moving or transfer to a safe position; equipment operation parameters and vibration signal data are collected and analyzed, an equipment health state evaluation model based on a support vector machine is constructed, early warning and diagnosis of equipment faults are achieved, and if equipment abnormity is found, maintenance suggestions and schemes are generated; cantilever motion planning and control strategies are optimized, energy consumption and mechanical abrasion are reduced, the equipment state is monitored in real time, cleaning parameters are dynamically adjusted according to the health assessment result, and safe and efficient operation of equipment is ensured.

Description

Technical Field

[0001] The invention relates to the field of information technology, and in particular to an intelligent residual coal cleaning method for a dumper. Background Art

[0002] During the cleaning process of the empty car with residual coal, the modular cantilever design introduces new technical contradictions. The cantilever is composed of different functional modules, such as telescopic arms, folding arms and rotating joints, to adapt to the complex and changing cleaning environment. However, the change in the module combination leads to changes in the cantilever's center of gravity, length and movement mode, which puts forward the requirement of dynamic adjustment of the fulcrum structure. The fulcrum needs to provide the supporting torque and flexibility that matches the cantilever according to different cleaning scenarios to ensure the stability and safety of the cleaning operation.

[0003] This technical contradiction further extends to the space limitation of the fulcrum structure. The internal space of the dumper is small, and the rapid movement and precise adjustment of the fulcrum face space constraints. How to achieve dynamic adjustment of the fulcrum in a limited space while taking into account cleaning efficiency and stability is a technical problem that needs to be solved urgently.

[0004] In addition, the difference in the properties of residual coal also puts forward new requirements for the design of the cantilever structure. Dry coal powder is easy to raise dust, and wet coal blocks are easy to adhere. Different properties of residual coal require the use of module combinations and movement methods that are suitable for them. This requires the cantilever structure to be able to dynamically adjust the module combination and movement parameters according to the real-time detection results of the residual coal properties to optimize the cleaning effect.

[0005] In summary, the modular cantilever design improves the cleaning adaptability, but also introduces the technical contradiction of dynamic adjustment of the fulcrum. This contradiction involves the space limitation of the fulcrum structure, the difference in the properties of the residual coal, and the balance between cleaning efficiency and stability. It requires in-depth research and innovation in multiple aspects such as mechanical design, material selection, sensor detection and control algorithm to promote the breakthrough and development of empty car residual coal cleaning technology of dumpers. Summary of the invention

[0006] The present invention provides an intelligent method for cleaning residual coal of a dumper, which mainly comprises:

[0007] Obtain the internal space boundary information of the dumper and the cantilever motion trajectory planning results, and use the octree-based spatial collision detection algorithm to determine the feasibility of the fulcrum adjustment plan. If there is a risk of spatial interference, further optimize the adjustment plan until the spatial constraints are met;

[0008] Obtain the information on the properties of the residual coal. If the residual coal is dry coal powder, select a high-speed pneumatic cleaning tool with a dust collection device to control the cantilever to move at a low and uniform speed. If the residual coal is a wet coal block, select a cleaning tool with vibration and scraping functions to control the cantilever to move at a low speed.

[0009] Obtain the residual coal distribution and the current cantilever posture information, use the A-star algorithm to solve the optimal cleaning path under the obstacle avoidance constraint, set the cleaning completion threshold based on image processing, and judge whether the current area meets the predetermined cleaning standard. If not, return to the area for repeated cleaning until the requirements are met;

[0010] The deep learning-based human detection and tracking algorithm is used to monitor the position and behavior of people in the working area in real time. If it is detected that a person enters a dangerous area or engages in unsafe behavior, an alarm mechanism is triggered to control the cantilever to stop moving or move to a safe position.

[0011] Collect and analyze equipment operating parameters and vibration signal data, build an equipment health status assessment model based on support vector machines, realize early warning and diagnosis of equipment failures, and generate maintenance suggestions and plans if equipment abnormalities are found;

[0012] Optimize cantilever motion planning and control strategies, reduce energy consumption and mechanical wear, monitor equipment status in real time, and dynamically adjust cleaning parameters based on health assessment results to ensure safe and efficient operation of the equipment.

[0013] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0014] The present invention discloses an intelligent method for cleaning residual coal of a dumper. The method first uses an octree spatial collision detection algorithm to determine the feasibility of a fulcrum adjustment scheme to ensure that there is no spatial interference in the cantilever motion. Then, according to the properties of the residual coal, a suitable cleaning tool and control strategy are selected, and the A-star algorithm is used to plan the optimal cleaning path. During the cleaning process, the present invention evaluates the cleaning completion degree through image processing technology, and repeats the cleaning when necessary. At the same time, a deep learning algorithm is used to monitor the safety of personnel in the working area in real time, and an abnormality is detected and an alarm is issued in time to take safety measures. In addition, the present invention also establishes an equipment health status assessment model based on a support vector machine to realize fault warning and diagnosis. By optimizing the motion control strategy, energy consumption and wear are reduced, and the safe and efficient operation of the equipment is ensured. The present invention realizes the intelligence and automation of residual coal cleaning of the dumper, and improves the working efficiency and safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The present invention is a flow chart of an intelligent method for cleaning residual coal of a car dumper.

[0016] Figure 2 It is a schematic diagram of an intelligent method for cleaning residual coal of a dumper according to the present invention.

[0017] Figure 3 It is another schematic diagram of an intelligent method for cleaning residual coal of a car dumper according to the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0019] like Figure 1-3 In this embodiment, an intelligent method for cleaning residual coal of a dumper may specifically include:

[0020] S101. Obtain the internal space boundary information of the tipping machine and the cantilever motion trajectory planning result, and use the octree-based space collision detection algorithm to determine the feasibility of the fulcrum adjustment plan. If there is a risk of space interference, further optimize the adjustment plan until the space constraint conditions are met.

[0021] After obtaining the internal space boundary information of the dumper and the cantilever motion trajectory planning results, the boundary information is preprocessed. The Gaussian filter algorithm is used to smooth and denoise the boundary point cloud data to improve the quality of the boundary information. Then, the internal space octree model of the dumper is constructed for subsequent spatial collision detection. The cantilever motion trajectory is discretized to obtain a series of cantilever posture points. For each cantilever posture point, the corresponding cantilever geometric model is obtained. The cantilever geometric model can be modeled by CAD software and converted into a triangular patch mesh model. Spatial collision detection is performed in the octree model to determine whether there is interference between the cantilever mesh model and the internal space of the dumper. Spatial collision detection is implemented using a method based on the axis-aligned bounding box tree (AABBTree). If there is spatial interference, the interference position information is obtained. According to the interference position and type, the A heuristic search algorithm is used to optimize the pivot adjustment scheme. The A algorithm takes the current pivot position as the starting point and aims to eliminate spatial interference, and searches for the optimal pivot position within the preset pivot feasible domain. The heuristic function is designed as the Euclidean distance from the current pivot position to the target position, which is used to estimate the cost. The optimized fulcrum adjustment scheme is applied to the cantilever motion trajectory planning to regenerate the motion trajectory. According to the new fulcrum position, the numerical iterative algorithm that cyclically utilizes the Jacobian matrix is ​​used to solve the cantilever inverse kinematics, calculate the motion parameters of each cantilever joint, and generate a new joint space trajectory. Then, through forward kinematics calculation, the joint space trajectory is converted to Cartesian space to form a new cantilever motion trajectory. Repeat the above collision detection and trajectory optimization process until there is no spatial interference on the entire motion trajectory, or the preset maximum number of iterations is reached. Output the fulcrum adjustment scheme and the optimized cantilever motion trajectory that meet the spatial constraints.

[0022] Exemplarily, after obtaining the boundary information of the internal space of the dumper and the planning result of the cantilever motion trajectory, the boundary information is preprocessed. The Gaussian filtering algorithm is used to smooth and denoise the boundary point cloud data to improve the quality of the boundary information. Then, an octree model of the internal space of the dumper is constructed for subsequent spatial collision detection. The cantilever motion trajectory is discretized to obtain a series of cantilever posture points. For each cantilever posture point, the corresponding cantilever geometric model is obtained. The cantilever geometric model can be modeled by CAD software and converted into a triangular patch mesh model. Spatial collision detection is performed in the octree model to determine whether there is interference between the cantilever mesh model and the internal space of the dumper. Spatial collision detection is implemented using a method based on the axis-aligned bounding box tree (AABB Tree). If there is spatial interference, the interference position information is obtained. According to the interference position and type, the A heuristic search algorithm is used to optimize the pivot adjustment scheme. The A algorithm takes the current pivot position as the starting point and aims to eliminate spatial interference, and searches for the optimal pivot position within the preset pivot feasible domain. The heuristic function is designed as the Euclidean distance from the current pivot position to the target position, which is used to estimate the cost. The optimized fulcrum adjustment scheme is applied to the cantilever motion trajectory planning to regenerate the motion trajectory. According to the new fulcrum position, the numerical iterative algorithm that cyclically utilizes the Jacobian matrix is ​​used to solve the cantilever inverse kinematics, calculate the motion parameters of each cantilever joint, and generate a new joint space trajectory. Then, through forward kinematics calculation, the joint space trajectory is converted to Cartesian space to form a new cantilever motion trajectory. Repeat the above collision detection and trajectory optimization process until there is no spatial interference on the entire motion trajectory, or the preset maximum number of iterations is reached. Output the fulcrum adjustment scheme and the optimized cantilever motion trajectory that meet the spatial constraints.

[0023] S102, obtaining the property information of the residual coal. If the residual coal is dry coal powder, select a high-speed pneumatic cleaning tool with a dust collection device to control the cantilever to move at a low and uniform speed; if the residual coal is a wet coal block, select a cleaning tool with vibration and scraping functions to control the cantilever to move at a low speed.

[0024] The humidity and particle size property information of the residual coal is obtained; the type of the residual coal is determined according to the property information; if the residual coal is dry coal powder, a high-speed pneumatic cleaning tool with a dust collection device is selected from a preset cleaning tool database, and the maximum speed threshold of the cantilever's low-speed uniform motion is determined by finite element analysis according to the physical parameters of the cantilever, and the high-speed pneumatic cleaning tool is installed at the end of the cantilever, and the cantilever is controlled to move at a low-speed uniform speed at the maximum speed threshold, and the high-speed pneumatic cleaning tool is started to perform cleaning operations; if the residual coal is a wet coal block, a cleaning tool with vibration and scraping functions is selected from the cleaning tool database, and the cleaning tool is installed at the end of the cantilever, and the cantilever is controlled to move at the lowest speed, and the vibration and scraping device of the cleaning tool are started to loosen and remove the wet coal block; during the cleaning process, the residual coal surface image is collected in real time by an industrial camera, and the cleaning degree of the residual coal is identified by an image processing algorithm. According to the image recognition result, the cantilever movement speed and the cleaning tool vibration frequency parameters are dynamically adjusted by the control algorithm until the residual coal is cleaned.

[0025] Exemplarily, the humidity and particle size of the residual coal are obtained by a temperature and humidity sensor and a laser particle size analyzer to determine whether the residual coal type is dry coal powder or wet coal block. If the residual coal is dry coal powder, a high-speed pneumatic cleaning tool with a dust collection device is selected from the preset cleaning tool database. According to the physical parameters such as the length and weight of the cantilever, the maximum speed threshold of the cantilever's low-speed uniform motion is calculated by finite element analysis to ensure that the cantilever is evenly stressed and avoids deformation. The high-speed pneumatic cleaning tool with a dust collection device is installed at the end of the cantilever, and the cantilever is controlled by a servo motor to move at a low speed uniform speed at the calculated maximum speed threshold. At the same time, the high-speed pneumatic device of the cleaning tool is started to perform cleaning operations, and the dust collection device collects and cleans the fallen coal powder. If the residual coal is a wet coal block, a cleaning tool with vibration and scraping functions is selected from the preset cleaning tool database. The cleaning tool with vibration and scraping functions is installed at the end of the cantilever, and the cantilever is controlled by a servo motor to move at the lowest speed, and the vibration and scraping devices of the cleaning tool are started to loosen and remove the wet coal blocks. During the cleaning process, the industrial camera collects the surface image of the residual coal in real time, and the edge detection and threshold segmentation algorithms of the OpenCV image processing library are used to identify the degree of residual coal cleaning. According to the image recognition results, the PID control algorithm dynamically adjusts the cantilever movement speed and the vibration frequency of the cleaning tool until the residual coal is cleaned.

[0026] S103, obtain the distribution of residual coal and the current position information of the cantilever, use the A-star algorithm to solve the optimal cleaning path under the obstacle avoidance constraint, set the cleaning completion threshold based on image processing, and judge whether the current area meets the predetermined cleaning standard. If not, return to the area for repeated cleaning until the requirements are met.

[0027] After obtaining the residual coal distribution information and the current cantilever posture information, the obtained information is preprocessed to extract effective information that can be used for the A algorithm. The preprocessed information is input into the A algorithm, and the obstacle avoidance constraints are considered to plan and generate the optimal cleaning path. According to the preset cleaning completion threshold, the image of the cleaned area is processed using image processing techniques such as image segmentation and binarization in OpenCV to extract the residual coal area after cleaning. The area ratio of the residual coal area after cleaning is calculated and used as the current cleaning completion. It is judged whether the current cleaning completion reaches the preset threshold. If it reaches, the area is considered to be cleaned and enters the next area; if it does not reach, the A algorithm is used to generate a path to return to the uncompleted area, and the cantilever is controlled to return to the area. According to the return path, the cantilever is controlled to return to the unqualified area and the cleaning operation is repeated. The cleaned image is obtained again, the cleaning completion is recalculated, and it is judged whether it meets the standard. Repeat the above cleaning, return, and judgment process until the cleaning completion of the current area meets the preset threshold requirements or the number of repetitions reaches the set upper limit. The area whose cleaning completion degree reaches the preset threshold is marked as a cleaned area, and it is removed from the residual coal distribution information, and the residual coal distribution information is updated. The image of the cleaned area is replaced with a coal-free image, and the residual coal distribution map is updated. According to the updated residual coal distribution information, the cleaning operation of the next area is started until all areas are cleaned.

[0028] Exemplarily, after obtaining the residual coal distribution information and the current position information of the cantilever, the obtained information is preprocessed to extract valid information that can be used for the A algorithm. The preprocessed information is input into the A algorithm, and the obstacle avoidance constraint is considered to plan and generate the optimal cleaning path. According to the preset cleaning completion threshold, the image of the cleaned area is processed by image processing techniques such as image segmentation and binarization in OpenCV to extract the residual coal area after cleaning. The area ratio of the residual coal area after cleaning is calculated and used as the current cleaning completion. It is judged whether the current cleaning completion reaches the preset threshold. If it reaches, it is considered that the cleaning of the area is completed and enters the next area; if it does not reach, the A algorithm is used to generate a path to return to the uncompleted area, and the cantilever is controlled to return to the area. According to the return path, the cantilever is controlled to return to the unqualified area and the cleaning operation is repeated. The image after cleaning is obtained again, the cleaning completion is recalculated, and it is judged whether it meets the standard. Repeat the above cleaning, return, and judgment process until the cleaning completion of the current area meets the preset threshold requirement or the number of repetitions reaches the set upper limit. The area whose cleaning completion degree reaches the preset threshold is marked as a cleaned area, and it is removed from the residual coal distribution information, and the residual coal distribution information is updated. The image of the cleaned area is replaced with a coal-free image, and the residual coal distribution map is updated. According to the updated residual coal distribution information, the cleaning operation of the next area is started until all areas are cleaned.

[0029] S104. Use a human detection and tracking algorithm based on deep learning to monitor the position and behavior of people in the working area in real time. If it is detected that a person enters a dangerous area or engages in unsafe behavior, an alarm mechanism is triggered to control the cantilever to stop moving or move to a safe position.

[0030] The video image in the working area is obtained, and the pre-trained human body detection model is used to detect and locate the personnel in the image to obtain the position coordinate information of the personnel; the position coordinates of the personnel are compared with the preset dangerous area coordinate range to determine whether the personnel enters the dangerous area; if the position coordinates of the personnel enter the dangerous area, the alarm mechanism is triggered; the key body parts of the personnel are detected by using a human posture estimation model based on key point detection to obtain the coordinate information of the key points of the human skeleton; the key point coordinates are input into the pre-trained behavior recognition model to classify and recognize the behavior of the personnel; if unsafe behavior is identified, the alarm mechanism is triggered; the alarm mechanism includes sending an audible and visual alarm signal and sending the alarm information to the cantilever control system; after receiving the alarm information, the cantilever control system controls the cantilever to stop the current movement and records the position coordinates when it stops; the nearest safe coordinate is selected from the preset safe position coordinate list, and the cantilever is controlled to move to the safe coordinate position; when it is detected that the person has left the dangerous area and has not entered again for a certain period of time, the cantilever is controlled to return to the position coordinate recorded when it stopped, and the previous operation process is continued.

[0031] Exemplarily, a video image in the working area is collected by a camera, and the image is analyzed in real time using computer vision technology. A pre-trained human detection model, such as YOLO or Faster R-CNN, is used to detect and locate the person in the image, and obtain the position coordinate information of the person. After obtaining the position coordinates of the person, it is compared with the preset coordinate range of the dangerous area. The coordinates of the dangerous area are calibrated in advance by the management personnel according to the actual situation of the working site and entered into the system. If it is detected that the position coordinates of the person enter the dangerous area range, the alarm mechanism is triggered. At the same time, a human posture estimation model based on key point detection, such as OpenPose, is used to detect the key body parts of the person and obtain the coordinate information of the key points of the human skeleton. The key point coordinates are input into a pre-trained behavior recognition model, such as LSTM, to classify and recognize the behavior of the person. The behavior recognition model is trained with a large amount of sample data of normal behavior and unsafe behavior, and can determine whether the current person's behavior belongs to the predefined unsafe behavior category. If unsafe behavior is identified, the alarm mechanism is also triggered. The alarm mechanism includes sending an audible and visual alarm signal to remind the operator to pay attention to safety, and sending the alarm information to the cantilever control system. After receiving the alarm information, the control system immediately controls the cantilever to stop the current movement and records the position coordinates when it stops. After the cantilever stops, the control system selects the nearest safe coordinate from the preset safe position coordinate list and controls the cantilever to move to the safe coordinate position. The safe position coordinates are calibrated and entered into the system in advance by the management personnel according to the layout of the work site to ensure that the cantilever is away from the dangerous area and the operators when it is in this position. When it is detected that a person has left the dangerous area and has not entered again for a certain period of time, the site is confirmed to be safe. The control system issues a recovery operation command to control the cantilever to return to the position coordinates recorded when it stopped before, and continue the previous operation process. During the whole process, the camera continuously collects video images of the working area, cyclically executes steps such as personnel detection, position judgment, and behavior recognition, and monitors the safety status of personnel in real time. Once it is detected that a person enters the dangerous area or an unsafe behavior occurs again, an alarm is immediately triggered and the operation is stopped to ensure the safety of the operator.

[0032] S105. Collect and analyze equipment operating parameters and vibration signal data, build an equipment health status assessment model based on support vector machine, realize early warning and diagnosis of equipment failure, and generate maintenance suggestions and plans if equipment abnormalities are found.

[0033] The equipment operation parameters and vibration signal data are obtained, and the data are preprocessed to obtain a preprocessed data set; the preprocessed data set is divided into a training set and a test set; a support vector machine algorithm is used to build an equipment health status assessment model on the training set, and the hyperparameters of the assessment model are optimized through grid search; the operation parameters and vibration signal data collected by the equipment in real time are input into the assessment model to obtain the health status score of the equipment; if the health status score is lower than a preset threshold, it is judged that the equipment is abnormal; according to the abnormal parameters and signal characteristics of the equipment, combined with the pre-established expert knowledge base and fault case library, the expert system is used for rule reasoning, and the optimal maintenance plan is determined in combination with the maintenance cost and equipment downtime; the optimal maintenance plan is sent to relevant personnel to guide them to carry out equipment maintenance and maintenance, and the maintenance process and results are fed back to the expert knowledge base and fault case library for updating; the equipment operation status is continuously monitored, the equipment data is regularly input into the assessment model, and the equipment health status is dynamically assessed; the equipment maintenance strategy is adjusted according to the assessment results, and when the equipment health status decreases, the fault diagnosis and maintenance decision process is triggered.

[0034] Exemplarily, the operating parameters and vibration signal data of the equipment are collected by sensors, and the collected data are preprocessed, including data cleaning, feature extraction, and data normalization, to obtain a data set suitable for modeling. The preprocessed data set is divided into a training set and a test set, and a support vector machine algorithm is used to build an equipment health status assessment model on the training set. The model hyperparameters are optimized by grid search and other methods, and the model performance is evaluated on the test set, so that the model can accurately assess the health status of the equipment. The operating parameters and vibration signal data collected by the equipment in real time are input into the trained assessment model, and the model outputs the health status score of the equipment. If the score is lower than the preset threshold, it is judged that the equipment is abnormal. According to the abnormal parameters and signal characteristics, combined with the expert knowledge base and the fault case library, the expert system is used for rule reasoning to generate targeted equipment maintenance suggestions and plans. The generated maintenance suggestions and plans are feasibility analyzed and optimized using a multi-objective optimization algorithm, and the optimal maintenance plan is determined by comprehensively considering factors such as maintenance costs and equipment downtime. The optimized maintenance plan is sent to relevant personnel to guide them to repair and maintain the equipment, and the maintenance process and results are fed back to the system for updating the expert knowledge base and the fault case library. Continuously monitor the equipment operation status, regularly input equipment data into the evaluation model, and dynamically evaluate the equipment health status. Adjust the equipment maintenance strategy based on the evaluation results, and trigger the fault diagnosis and maintenance decision-making process when the equipment health status declines. Through this closed-loop control strategy, predictive maintenance and life management of equipment can be achieved, improving equipment operation reliability and production efficiency.

[0035] S106. Optimize cantilever motion planning and control strategies, reduce energy consumption and mechanical wear, monitor equipment status in real time, and dynamically adjust cleaning parameters based on health assessment results to ensure safe and efficient operation of the equipment.

[0036] Acquire the cantilever vibration and temperature status data, which are collected in real time by sensors; construct an equipment health assessment model based on the status data using a support vector regression algorithm; determine the equipment health status level through the equipment health assessment model; if the equipment health status level is lower than a preset threshold, dynamically adjust the cleaning motion parameters, which include operating speed, acceleration and amplitude; acquire energy consumption data under different equipment operating parameters; based on the energy consumption data, establish a mapping model between equipment operating parameters and energy consumption using a CART decision tree algorithm; determine the optimal energy consumption operating parameter combination through the mapping model; acquire the equipment wear status assessment result; and based on the wear status assessment result , a Bayesian network algorithm is used to establish an association model between the wear state and the grease filling amount and filling cycle; through the association model, the lubrication parameters are dynamically optimized, and the lubrication parameters include the grease filling amount and filling cycle; according to the equipment health status, energy consumption optimization results and wear prediction information, the weighted summation method is used to determine the optimal cleaning control strategy, and the optimal cleaning control strategy includes the optimized cantilever motion parameters and lubrication maintenance parameters; the optimal cleaning control strategy is sent to the cantilever control unit; the cantilever control unit adjusts the cantilever motion in real time through a closed-loop control method according to the received motion parameters; the cantilever control unit performs regular maintenance according to the optimized lubrication strategy based on the received lubrication maintenance parameters.

[0037] Exemplarily, according to the cantilever motion state and equipment health state, the motion control parameters are dynamically optimized. First, the state data of the cantilever vibration, temperature, etc. are collected in real time by sensors. The collected data is used to build an equipment health assessment model based on the support vector regression algorithm to determine the equipment health state level. If the equipment health state is lower than the preset threshold, the cleaning motion parameters are dynamically adjusted to reduce the operating speed and acceleration, reduce the cantilever amplitude, and ensure the safe operation of the equipment. When optimizing the motion control parameters, energy consumption and wear factors are comprehensively considered. By conducting energy consumption tests under different equipment operating parameters, the corresponding energy consumption data is obtained. Using the acquired data, the CART decision tree algorithm is used to establish a mapping model between equipment operating parameters and energy consumption, and to find the operating parameter combination with the best energy consumption. At the same time, according to the wear state evaluation results of the equipment, the Bayesian network algorithm is used to establish an association model between the wear state and the grease filling amount and filling cycle, and the lubrication parameters are dynamically optimized. After obtaining the equipment health state, energy consumption optimization results and wear prediction information, the weighted summation method is used to weigh the three optimization objectives to obtain the optimal cleaning control strategy that takes into account safety, energy saving and wear reduction. The optimal control strategy includes optimized boom motion parameters such as speed, acceleration, amplitude, and maintenance parameters such as grease filling amount and filling cycle. The optimized cleaning control parameters are sent to the boom control unit, including motion control parameters and lubrication maintenance parameters. The boom control unit adjusts the boom motion in real time through closed-loop control based on the received parameters, and performs regular maintenance according to the optimized lubrication strategy to ensure the continuous, efficient and safe operation of the equipment.

[0038] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method 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 in the scope of protection of the present invention.

Claims

1. An intelligent method for cleaning residual coal of a dumper, characterized in that: The method comprises: Obtain the internal space boundary information of the dumper and the cantilever motion trajectory planning results, and use the octree-based spatial collision detection algorithm to determine the feasibility of the fulcrum adjustment plan. If there is a risk of spatial interference, further optimize the adjustment plan until the spatial constraints are met; Obtain the information on the properties of the residual coal. If the residual coal is dry coal powder, select a high-speed pneumatic cleaning tool with a dust collection device to control the cantilever to move at a low and uniform speed. If the residual coal is a wet coal block, select a cleaning tool with vibration and scraping functions to control the cantilever to move at a low speed. Obtain the residual coal distribution and the current cantilever posture information, use the A-star algorithm to solve the optimal cleaning path under the obstacle avoidance constraint, set the cleaning completion threshold based on image processing, and judge whether the current area meets the predetermined cleaning standard. If not, return to the area for repeated cleaning until the requirements are met; The deep learning-based human detection and tracking algorithm is used to monitor the position and behavior of people in the working area in real time. If it is detected that a person enters a dangerous area or engages in unsafe behavior, an alarm mechanism is triggered to control the cantilever to stop moving or move to a safe position. Collect and analyze equipment operating parameters and vibration signal data, build an equipment health status assessment model based on support vector machines, realize early warning and diagnosis of equipment failures, and generate maintenance suggestions and plans if equipment abnormalities are found; Optimize cantilever motion planning and control strategies, reduce energy consumption and mechanical wear, monitor equipment status in real time, and dynamically adjust cleaning parameters based on health assessment results to ensure safe and efficient operation of the equipment.

2. The method according to claim 1, characterized in that The method of obtaining the internal space boundary information of the dumper and the cantilever motion trajectory planning result, and using the octree-based space collision detection algorithm to determine the feasibility of the fulcrum adjustment plan, and if there is a risk of space interference, further optimizing the adjustment plan until the space constraint conditions are met, includes: After obtaining the internal space boundary information of the dumper and the cantilever motion trajectory planning results, the boundary information is preprocessed; Gaussian filtering algorithm is used to smooth and denoise the boundary point cloud data to improve the quality of boundary information; Then, an octree model of the internal space of the dumper is constructed for subsequent spatial collision detection; Discretize the cantilever motion trajectory to obtain a series of cantilever position points; For each cantilever posture point, obtain the corresponding cantilever geometric model; The cantilever geometric model can be modeled by CAD software and converted into a triangular patch mesh model; Perform spatial collision detection in the octree model to determine whether there is interference between the cantilever mesh model and the internal space of the dumper; Spatial collision detection is implemented based on the axis-aligned bounding box tree (AABBTree); If there is spatial interference, obtain the interference position information; According to the interference position and type, the A heuristic search algorithm is used to optimize the fulcrum adjustment scheme; Algorithm A takes the current fulcrum position as the starting point, aims to eliminate spatial interference, and searches for the optimal fulcrum position within the preset fulcrum feasible region; The heuristic function is designed as the Euclidean distance from the current pivot position to the target position, which is used to estimate the cost; Apply the optimized fulcrum adjustment scheme to the cantilever motion trajectory planning to regenerate the motion trajectory; According to the new fulcrum position, the numerical iterative algorithm that cyclically utilizes the Jacobian matrix is ​​used to solve the cantilever inverse kinematics, calculate the motion parameters of each cantilever joint, and generate a new joint space trajectory; Then, through forward kinematics calculation, the joint space trajectory is converted into Cartesian space to form a new cantilever motion trajectory; Repeat the above collision detection and trajectory optimization process until there is no spatial interference on the entire motion trajectory, or the preset maximum number of iterations is reached; Output the fulcrum adjustment plan and the optimized cantilever motion trajectory that meet the space constraints.

3. The method according to claim 1, characterized in that The method of obtaining the residual coal property information includes: if the residual coal is dry coal powder, selecting a high-speed pneumatic cleaning tool with a dust collecting device to control the cantilever to move at a low and uniform speed; if the residual coal is a wet coal block, selecting a cleaning tool with vibration and scraping functions to control the cantilever to move at a low speed, including: Obtain information on the moisture and particle size properties of residual coal; Determining the type of the residual coal according to the property information; If the residual coal is dry coal powder, a high-speed pneumatic cleaning tool with a dust collection device is selected from a preset cleaning tool database, and the maximum speed threshold of the cantilever's low-speed uniform motion is determined through finite element analysis according to the physical parameters of the cantilever, and the high-speed pneumatic cleaning tool is installed at the end of the cantilever, and the cantilever is controlled to move at a low-speed uniform speed at the maximum speed threshold, and the high-speed pneumatic cleaning tool is started to perform cleaning operations; If the residual coal is a wet coal block, a cleaning tool with vibration and scraping functions is selected from the cleaning tool database, the cleaning tool is installed at the end of the cantilever, the cantilever is controlled to move at the lowest speed, and the vibration and scraping device of the cleaning tool are started to loosen and remove the wet coal block; During the cleaning process, the industrial camera is used to collect real-time images of the residual coal surface, and the image processing algorithm is used to identify the degree of residual coal cleaning. According to the image recognition results, the cantilever movement speed and the vibration frequency parameters of the cleaning tool are dynamically adjusted through the control algorithm until the residual coal is cleaned.

4. The method according to claim 1, characterized in that: The method of obtaining the residual coal distribution and the current cantilever posture information, using the A-star algorithm to solve the optimal cleaning path under the obstacle avoidance constraint, setting the cleaning completion threshold based on image processing, and judging whether the current area meets the predetermined cleaning standard, if not, returning to the area for repeated cleaning until the requirements are met, includes: After obtaining the remaining coal distribution information and the current cantilever posture information, the obtained information is preprocessed to extract effective information that can be used for the A algorithm; The pre-processed information is input into the A algorithm, and the obstacle avoidance constraints are considered to plan and generate the optimal cleaning path; According to the preset cleaning completion threshold, the image of the cleaned area is processed using image processing techniques such as image segmentation and binarization in OpenCV to extract the remaining coal area after cleaning; Calculate the area ratio of the remaining coal area after cleaning and use it as the current cleaning completion degree; Determine whether the current cleaning completion degree reaches the preset threshold. If so, the area is considered to be cleaned and the next area is entered; If it is not reached, the A algorithm is used to generate a path to return to the uncleaned area and control the cantilever to return to the area; According to the return path, the cantilever is controlled to return to the unreached area and the cleaning operation is repeated; Obtain the cleaned image again, recalculate the cleaning completion degree, and determine whether it meets the standard; Repeat the above cleaning, returning and judging process until the cleaning completion degree of the current area meets the preset threshold requirement, or the number of repetitions reaches the set upper limit; Marking the area whose cleaning completion degree reaches the preset threshold as a cleaned area, removing it from the remaining coal distribution information, and updating the remaining coal distribution information; The image of the cleaned area is replaced with a coal-free image, and the residual coal distribution map is updated; According to the updated residual coal distribution information, proceed to the cleaning operation of the next area until all areas are cleaned.

5. The method according to claim 1, characterized in that The human body detection and tracking algorithm based on deep learning is used to monitor the position and behavior of people in the working area in real time. If it is detected that a person enters a dangerous area or has unsafe behavior, an alarm mechanism is triggered to control the cantilever to stop moving or move to a safe position, including: Obtain video images in the work area, use the pre-trained human detection model to detect and locate people in the image, and obtain the location coordinate information of the people; Compare the position coordinates of the person with the preset dangerous area coordinate range to determine whether the person has entered the dangerous area; If the personnel position coordinates enter the danger zone, the alarm mechanism will be triggered; A human posture estimation model based on key point detection is used to detect key body parts of a person and obtain the coordinate information of key points of the human skeleton; Inputting the key point coordinates into a pre-trained behavior recognition model to classify and recognize the behavior of the person; If unsafe behavior is identified, an alarm mechanism is triggered; The alarm mechanism includes emitting an audible and visual alarm signal and sending the alarm information to the cantilever control system; After receiving the alarm information, the cantilever control system controls the cantilever to stop the current movement and records the position coordinates when it stops; Select the nearest safety coordinate from the preset safety position coordinate list, and control the cantilever to move to the safety coordinate position; When it is detected that a person has left the danger zone and has not re-entered for a certain period of time, the boom is controlled to return to the position coordinates recorded when it stopped and continue the previous operation process.

6. The method according to claim 1, characterized in that The device operating parameters and vibration signal data are collected and analyzed, and a device health status assessment model based on a support vector machine is constructed to achieve early warning and diagnosis of device failures. If an abnormality is found in the device, maintenance suggestions and plans are generated, including: Acquire equipment operating parameters and vibration signal data, and preprocess the data to obtain a preprocessed data set; Dividing the preprocessed data set into a training set and a test set; A support vector machine algorithm is used to construct an equipment health status assessment model on the training set, and hyperparameters of the assessment model are optimized through grid search; Inputting the operating parameters and vibration signal data collected by the equipment in real time into the evaluation model to obtain the health status score of the equipment; If the health status score is lower than a preset threshold, it is determined that the device is abnormal; According to the abnormal parameters and signal characteristics of the equipment, combined with the pre-established expert knowledge base and fault case library, the expert system is used for rule reasoning, and the optimal maintenance plan is determined in combination with the maintenance cost and equipment downtime; Send the optimal maintenance plan to relevant personnel to guide them in equipment maintenance and repair, and feed back the maintenance process and results to the expert knowledge base and fault case library for updating; Continuously monitor the operating status of the equipment, regularly input the equipment data into the evaluation model, and dynamically evaluate the health status of the equipment; Adjust the equipment maintenance strategy based on the evaluation results, and trigger the fault diagnosis and maintenance decision-making process when the equipment health status deteriorates.

7. The method according to claim 1, characterized in that The optimized cantilever motion planning and control strategy reduces energy consumption and mechanical wear, monitors the equipment status in real time, and dynamically adjusts the cleaning parameters based on the health assessment results to ensure safe and efficient operation of the equipment, including: Acquire cantilever vibration and temperature state data, wherein the state data is collected in real time by a sensor; Based on the status data, a device health assessment model is constructed using a support vector regression algorithm; Determine the health status level of the equipment through the equipment health assessment model; If the health status level of the device is lower than a preset threshold, the cleaning motion parameters are dynamically adjusted, and the cleaning motion parameters include running speed, acceleration and amplitude; Obtain energy consumption data under different equipment operating parameters; Based on the energy consumption data, a mapping model between equipment operating parameters and energy consumption is established using a CART decision tree algorithm; Determining the optimal energy consumption operation parameter combination through the mapping model; Obtain equipment wear status assessment results; According to the wear state evaluation results, a correlation model between the wear state and the grease filling amount and filling cycle is established using a Bayesian network algorithm; Dynamically optimizing lubrication parameters through the association model, wherein the lubrication parameters include grease filling amount and filling cycle; According to the equipment health status, energy consumption optimization results and wear prediction information, the optimal cleaning control strategy is determined by using the weighted summation method, and the optimal cleaning control strategy includes the optimized cantilever motion parameters and lubrication maintenance parameters; Sending the optimal cleaning control strategy to the cantilever control unit; The cantilever control unit adjusts the cantilever motion in real time through a closed-loop control method according to the received motion parameters; The cantilever control unit performs regular maintenance according to the received lubrication maintenance parameters and the optimized lubrication strategy.

Images