AI-based belt foreign matter recognition system for automatic aggregate conveying

Abstract

The invention relates to the technical field of belt foreign matter recognition systems for aggregate conveying, in particular to an AI-based belt foreign matter recognition system for automatic aggregate conveying. According to the technical scheme, the AI-based belt foreign matter recognition system for automatic aggregate conveying comprises a monitoring module, a video analysis module, an early warning and feedback module, a user interaction module and a power supply and protection module; high-definition IP cameras are installed at key positions of the aggregate conveying belt, such as a starting point, an ending point, a turning position and an area where foreign matter is prone to being mixed, clear video images can be captured under various light conditions, in order to cope with the night or the environment with insufficient light, an infrared night vision camera is further arranged, and the system is convenient to use. The infrared night vision camera can capture clear video pictures in a completely dark environment, and 24-hour uninterrupted monitoring is ensured.

Description

Technical Field

[0001] This invention relates to the technical field of foreign object identification systems for aggregate conveyor belts, and more particularly to an AI-based automated foreign object identification system for aggregate conveyor belts. Background Technology

[0002] The foreign object identification system for aggregate conveyor belts is an automated monitoring system based on artificial intelligence technology. It is specifically designed for aggregate conveyor belt scenarios to monitor and identify foreign objects on the belt in real time, ensuring the safety and stability of the aggregate conveying process.

[0003] The existing foreign object recognition system for aggregate conveyor belts is greatly affected by the target resolution and light intensity. When the target is too small or the light is too weak, the recognition rate may decrease. In harsh environments such as mines, factors such as dust and uneven lighting may further affect the recognition accuracy.

[0004] The existing foreign object identification system for aggregate conveyor belts is significantly affected by target resolution and light intensity. When the target is too small or the light is too weak, the recognition rate may decrease. In harsh environments such as mines, dust and uneven lighting may further affect the recognition accuracy. This solution installs high-definition IP cameras at key locations on the aggregate conveyor belt, such as the start, end, turns, and areas prone to foreign object contamination. These cameras can capture clear video images under various lighting conditions. To cope with nighttime or low-light environments, infrared night vision cameras are also provided. These cameras can capture clear video images in complete darkness, ensuring 24-hour uninterrupted monitoring. By adjusting the installation angle and focal length of the cameras, it is ensured that the field of view of each camera covers the entire width of the belt, and there is some overlap between adjacent cameras, thereby eliminating blind spots. Summary of the Invention

[0005] To overcome the problems mentioned in the background art, the present invention proposes an AI-based automated aggregate conveyor belt foreign object identification system.

[0006] The technical solution of this invention is: an AI-based automated aggregate conveyor belt foreign object identification system, comprising: Monitoring module: Used for multi-directional real-time monitoring of aggregate conveyor belts, expanding the monitoring range; Video analytics module: used for real-time analysis of video streams and foreign object identification using AI algorithms; Early warning and feedback module: It is used to issue an early warning immediately when a foreign object is detected, and push the warning to relevant personnel through multiple channels to form a closed-loop management mechanism; User interaction module: Used by administrators to perform real-time monitoring, alarm handling, and data query operations; Power and protection module: Used to provide a stable power supply to the system and protect the equipment from damage caused by lightning strikes, dust and moisture.

[0007] Preferably, the monitoring module includes: A1001: Video acquisition unit, including a high-definition IP camera and an infrared night vision camera. The high-definition IP camera is deployed along the aggregate conveyor belt and has high resolution, wide dynamic range and low-light performance. The infrared night vision camera is used for 24-hour uninterrupted monitoring at night or in low-light environments. A1002: Video transmission unit, including network switch and fiber optic transmission equipment, is used to transmit the acquired video stream to the video analysis server.

[0008] Preferably, the monitoring module includes the following steps when it is in operation: S1001: Install high-definition IP cameras and infrared night vision cameras at key locations on the aggregate conveyor belt, including the starting point, ending point, turning points, and areas prone to foreign matter contamination. S1002: The camera's field of view is adjusted to cover the entire width of the belt, with some overlap between adjacent cameras, eliminating blind spots. The video coverage optimization formula is:

[0009] in The field of view coverage (%) measures the effective proportion of the belt covered by the camera. Let be the horizontal and vertical field of view range (°) of the i-th camera; Let be the horizontal and vertical overlap angle (°) of the j-th monitoring blind spot; The total length of the belt is (m). The belt width is in meters. S1003: Connect the camera to the fiber optic transmission equipment via a network switch to ensure stable and high-speed transmission of the video stream to the video analysis server; S1004: Configure network parameters, including IP address, subnet mask and gateway, and detect the network connectivity between the camera and the video analysis server; S1005: The camera collects the video stream of the aggregate conveyor belt in real time and transmits it to the video analysis server via the network; S1006: Detect that the fiber optic transmission equipment is not interfered with during video stream transmission.

[0010] Preferably, the video analytics module includes: A2001: The video stream receiving unit includes: a video decoding card, used to receive video streams from the monitoring module and convert the video streams from an encoded format to a processable format; A2002: The AI ​​algorithm processing unit includes a high-performance GPU server, an AI accelerator card, and a YOLOX optimized algorithm, which is used for real-time monitoring of aggregate conveying scenarios, accurately identifying foreign objects of different sizes, shapes, and colors, and performing real-time analysis. A2003: Result Output Unit: Data Interface Card, used to output the analysis results of the AI ​​algorithm processing unit to the early warning and feedback module, and provide a data interface with the early warning and feedback module.

[0011] Preferably, the video analytics module includes the following steps when it is in operation: S2001: Start the high-performance GPU server and load the AI ​​accelerator card driver; S2002: Configure server network parameters, detect and monitor network connectivity of the early warning and feedback modules, and analyze the latency formula in real time: in Total delay (ms); The camera acquisition time (ms); The video stream transmission time (ms); AI algorithm processing time (ms); S2003: Load the optimized version of the YOLOX algorithm on the video analytics server and configure the algorithm parameters; S2004: The detection algorithm is adapted to aggregate conveying scenarios and has a high recognition rate for foreign objects of different sizes, shapes and colors; S2005: The video analysis server receives video streams from the monitoring module and performs real-time analysis using the YOLOX algorithm; S2006: The algorithm extracts features from video frames through network layers such as convolutional layers and pooling layers, and combines them with inter-layer fusion operations to output foreign object recognition results, including the location, size, and type of the foreign object. The formula for the foreign object recognition accuracy is: in, The recognition accuracy rate (%). The number of foreign objects correctly identified; This represents the number of falsely reported foreign objects.

[0012] As a preferred option, the early warning and feedback module includes: A3001: Alarm Information Processing Unit: Alarm Server, equipped with a high-performance processor and sufficient storage space, responsible for real-time processing and long-term storage of alarm information. It is used to receive analysis results from the video analysis module, generate alarm information, and perform storage and statistical analysis. A3002: Alarm push unit, including SMS push module and APP push module, is used to integrate SMS gateway to push alarm information to relevant personnel via SMS. It also integrates real-time monitoring and alarm processing functions, so that managers can keep track of the belt conveyor's operating status in real time. A3003: On-site alarm unit, including an audible and visual alarm, is deployed on the aggregate conveyor belt site. When a foreign object is detected, it emits a strong audible and visual alarm signal to remind on-site personnel to handle the situation in a timely manner and prevent the foreign object from damaging the conveyor belt or causing a safety accident.

[0013] Preferably, the early warning and feedback module includes the following steps when it is in operation: S3001: Start the alarm server, configure the connection parameters with the video analysis server, and check that it can receive alarm information from the video analysis server; S3002: Set the alarm information storage path and statistical analysis rules. The alarm response time formula is: in, Total response time (s); Foreign object detection time (s); Alarm push time (s); Personnel confirmation time (s); S3003: When the video analysis server detects a foreign object, it immediately generates an alarm message and sends it to the alarm server; S3004: The alarm server pushes alarm information to relevant personnel through SMS push module, APP push module and audible and visual alarms according to preset rules. When the alarm server's alarm fatigue level approaches the threshold, it notifies relevant personnel. The alarm fatigue suppression formula is: in, To alert for fatigue levels; Total number of alarms; This represents the actual number of foreign object alarms. This represents the number of false alarms. This is the false alarm weighting coefficient; S3005: After receiving the alarm information, relevant personnel can handle the alarm through the web or mobile APP. S3006: The alarm server stores alarm processing records and performs statistical analysis according to preset rules; S3007: Administrators can view alarm handling status through the web or mobile app and further optimize and adjust the system.

[0014] Preferably, the user interaction module includes: A4001: Web Management Unit, which includes a Web server and a database server. The Web server is used to deploy the Web management platform and provides user login, permission management and data query functions. The database server is used to store system operation data and alarm information and supports data query and statistical functions. A4002: The mobile APP unit includes a mobile APP development framework and a push service. The mobile APP development framework is used for cross-platform development, and the push service is used to integrate third-party push services to push alarm information in real time. A4003: Field operation unit, including a touch screen operating interface, is used to deploy on the field control cabinet of the aggregate conveyor belt for easy operation and monitoring by field personnel.

[0015] As a preferred option, the user interaction module is used to develop a web-based management platform, allowing safety inspectors or system administrators to log in via a browser to perform production supervision, alarm handling, and data query operations; a mobile APP is also developed to support real-time monitoring and alarm handling on mobile devices. Administrators can view the operating status and alarm information of the aggregate conveyor belt in real time through the mobile APP and perform corresponding processing operations. At the same time, a touch screen operating interface is deployed on the aggregate conveyor belt field control cabinet, allowing on-site personnel to operate and monitor the system.

[0016] Preferably, the power supply and protection module includes: A5001: Power supply unit, including USB uninterruptible power supply, used to provide short-term power supply during power outages to ensure normal system operation and secure data storage; A5002: Equipment protection unit, including a lightning protection module and a dustproof and waterproof housing. The lightning protection module is deployed at the power input and signal output terminals to protect the equipment from lightning damage. The dustproof and waterproof housing is for the stable operation of the equipment in harsh environments.

[0017] The beneficial effects of this invention are: 1. Compared with existing foreign object identification systems for aggregate conveyor belts, the algorithm is greatly affected by target resolution and light intensity. When the target is too small or the light is too weak, the recognition rate may decrease. In harsh environments such as mines, dust, uneven lighting and other factors may further affect the recognition accuracy. This solution installs high-definition IP cameras at key locations on the aggregate conveyor belt, such as the starting point, ending point, turning points and areas where foreign objects are prone to enter. These cameras can capture clear video images under various lighting conditions. In order to cope with nighttime or low-light environments, infrared night vision cameras are also equipped. Infrared night vision cameras can capture clear video images in completely dark environments, ensuring 24-hour uninterrupted monitoring. At the same time, by adjusting the installation angle and focal length of the cameras, it is ensured that the field of view of each camera can cover the entire width of the belt, and there is a partial overlap between adjacent cameras, thereby eliminating monitoring blind spots. 2. Compared with existing foreign object recognition systems for aggregate conveyor belts, which require a large amount of labeled data for model training to improve recognition accuracy, this solution uses a high-performance GPU server as the core processing unit for video analysis. Specifically, the YOLOX algorithm is optimized for the characteristics of aggregate conveying scenarios. By adjusting algorithm parameters and increasing training data, the algorithm can more accurately identify foreign objects of different sizes, shapes, and colors. Furthermore, the algorithm optimization reduces the false alarm rate, enabling it to more accurately distinguish between foreign objects and normal materials, thus reducing production interruptions and manpower waste caused by false alarms. 3. Compared to existing foreign object detection systems for aggregate conveyor belts, which are prone to false alarms or missed alarms in complex scenarios such as mixed materials on the belt or foreign objects that are similar in color to the materials, this solution addresses this issue. When the system detects a foreign object, it immediately sends an alarm message to relevant personnel via SMS. The SMS message includes key information such as the location, size, and type of the foreign object, facilitating quick location and handling. In addition to SMS, the system also supports sending alarm messages to relevant personnel via an app. The app allows for real-time viewing of alarm details and alarm task processing, improving alarm handling efficiency. Furthermore, an audible and visual alarm is deployed at the aggregate conveyor belt site. When the system detects a foreign object, the alarm emits a strong audible and visual signal, alerting on-site personnel to handle the situation promptly and prevent damage to the conveyor belt or safety accidents. Attached Figure Description

[0018] Figure 1 The diagram shown is a schematic flowchart of an AI-based automated aggregate conveying belt foreign object identification system according to the present invention. Figure 2The diagram shows the workflow of the monitoring module of the AI-based automated aggregate conveying belt foreign object identification system according to the present invention. Figure 3 The diagram shows the workflow of the video analysis module of an AI-based automated aggregate conveying belt foreign object recognition system according to the present invention. Figure 4 The diagram shows the workflow of the early warning and feedback module of the AI-based automated aggregate conveyor belt foreign object identification system according to the present invention. Detailed Implementation

[0019] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0020] Please see Figure 1-4 This invention provides an embodiment: an AI-based automated aggregate conveyor belt foreign object identification system, comprising: Monitoring module: Used for multi-directional real-time monitoring of aggregate conveyor belts, expanding the monitoring range; Video analytics module: used for real-time analysis of video streams and foreign object identification using AI algorithms; Early warning and feedback module: It is used to issue an early warning immediately when a foreign object is detected, and push the warning to relevant personnel through multiple channels to form a closed-loop management mechanism; User interaction module: Used by administrators to perform real-time monitoring, alarm handling, and data query operations; Power and protection module: Used to provide a stable power supply to the system and protect the equipment from damage caused by lightning strikes, dust and moisture.

[0021] Preferably, the monitoring module includes: A1001: Video acquisition unit, including a high-definition IP camera and an infrared night vision camera. The high-definition IP camera is deployed along the aggregate conveyor belt and has high resolution, wide dynamic range and low-light performance. The infrared night vision camera is used for 24-hour uninterrupted monitoring at night or in low-light environments. A1002: Video transmission unit, including network switch and fiber optic transmission equipment, is used to transmit the acquired video stream to the video analysis server.

[0022] Preferably, the monitoring module includes the following steps when it is in operation: S1001: Install high-definition IP cameras and infrared night vision cameras at key locations on the aggregate conveyor belt, including the starting point, ending point, turning points, and areas prone to foreign matter contamination. S1002: The camera's field of view is adjusted to cover the entire width of the belt, with some overlap between adjacent cameras, eliminating blind spots. The video coverage optimization formula is:

[0023] in The field of view coverage (%) measures the effective proportion of the belt covered by the camera. Let be the horizontal and vertical field of view range (°) of the i-th camera; Let be the horizontal and vertical overlap angle (°) of the j-th monitoring blind spot; The total length of the belt is (m). The belt width is in meters. S1003: Connect the camera to the fiber optic transmission equipment via a network switch to ensure stable and high-speed transmission of the video stream to the video analysis server; S1004: Configure network parameters, including IP address, subnet mask and gateway, and detect the network connectivity between the camera and the video analysis server; S1005: The camera collects the video stream of the aggregate conveyor belt in real time and transmits it to the video analysis server via the network; S1006: Detect that the fiber optic transmission equipment is not interfered with during video stream transmission.

[0024] As described above, the monitoring module, through the combined use of high-definition IP cameras and infrared night vision cameras, achieves all-weather, all-round monitoring of the aggregate conveyor belt, ensuring no blind spots. The infrared night vision cameras maintain monitoring capabilities even at night or in low-light conditions, improving the system's stability under different environmental conditions. Utilizing network switches and fiber optic transmission equipment ensures stable and high-speed transmission of video streams to the video analysis server, providing a reliable data foundation for real-time analysis. By precisely adjusting the camera's field of view to cover the entire belt width and eliminate blind spots, the comprehensiveness and accuracy of monitoring are improved.

[0025] Preferably, the video analytics module includes: A2001: The video stream receiving unit includes: a video decoding card, used to receive video streams from the monitoring module and convert the video streams from an encoded format to a processable format; A2002: The AI ​​algorithm processing unit includes a high-performance GPU server, an AI accelerator card, and a YOLOX optimized algorithm, which is used for real-time monitoring of aggregate conveying scenarios, accurately identifying foreign objects of different sizes, shapes, and colors, and performing real-time analysis. A2003: Result Output Unit: Data Interface Card, used to output the analysis results of the AI ​​algorithm processing unit to the early warning and feedback module, and provide a data interface with the early warning and feedback module.

[0026] Preferably, the video analytics module includes the following steps when it is in operation: S2001: Start the high-performance GPU server and load the AI ​​accelerator card driver; S2002: Configure server network parameters, detect and monitor network connectivity of the early warning and feedback modules, and analyze the latency formula in real time: in Total delay (ms); The camera acquisition time (ms); The video stream transmission time (ms); AI algorithm processing time (ms); S2003: Load the optimized version of the YOLOX algorithm on the video analytics server and configure the algorithm parameters; S2004: The detection algorithm is adapted to aggregate conveying scenarios and has a high recognition rate for foreign objects of different sizes, shapes and colors; S2005: The video analysis server receives video streams from the monitoring module and performs real-time analysis using the YOLOX algorithm; S2006: The algorithm extracts features from video frames through network layers such as convolutional layers and pooling layers, and combines them with inter-layer fusion operations to output foreign object recognition results, including the location, size, and type of the foreign object. The formula for the foreign object recognition accuracy is: in, The recognition accuracy rate (%). The number of foreign objects correctly identified; This represents the number of falsely reported foreign objects.

[0027] As described above, the video analysis module employs a high-performance GPU server, AI accelerator card, and the optimized YOLOX algorithm to accurately identify foreign objects of different sizes, shapes, and colors in aggregate conveying scenarios. By optimizing the algorithm and hardware configuration, the analysis speed of the video stream is improved, ensuring real-time performance. The optimized YOLOX algorithm is specifically optimized for aggregate conveying scenarios, improving adaptability to complex scenes and recognition rate. The algorithm can output the location, size, and type information of foreign objects, providing strong support for subsequent early warning and feedback.

[0028] As a preferred option, the early warning and feedback module includes: A3001: Alarm Information Processing Unit: Alarm Server, equipped with a high-performance processor and sufficient storage space, responsible for real-time processing and long-term storage of alarm information. It is used to receive analysis results from the video analysis module, generate alarm information, and perform storage and statistical analysis. A3002: Alarm push unit, including SMS push module and APP push module, is used to integrate SMS gateway to push alarm information to relevant personnel via SMS. It also integrates real-time monitoring and alarm processing functions, allowing managers to monitor the operating status of belt conveyor in real time. A3003: On-site alarm unit, including an audible and visual alarm, is deployed on the aggregate conveyor belt site. When a foreign object is detected, it emits a strong audible and visual alarm signal to remind on-site personnel to handle the situation in a timely manner and prevent the foreign object from damaging the conveyor belt or causing a safety accident.

[0029] Preferably, the early warning and feedback module includes the following steps when it is in operation: S3001: Start the alarm server, configure the connection parameters with the video analysis server, and check that it can receive alarm information from the video analysis server; S3002: Set the alarm information storage path and statistical analysis rules. The alarm response time formula is: in, Total response time (s); Foreign object detection time (s); Alarm push time (s); Personnel confirmation time (s); S3003: When the video analysis server detects a foreign object, it immediately generates an alarm message and sends it to the alarm server; S3004: The alarm server pushes alarm information to relevant personnel through SMS push module, APP push module and audible and visual alarms according to preset rules. When the alarm server's alarm fatigue level approaches the threshold, it notifies relevant personnel. The alarm fatigue suppression formula is: in, To alert for fatigue levels; Total number of alarms; This represents the actual number of foreign object alarms. This represents the number of false alarms. This is the false alarm weighting coefficient; S3005: After receiving the alarm information, relevant personnel can handle the alarm through the web or mobile APP. S3006: The alarm server stores alarm processing records and performs statistical analysis according to preset rules; S3007: Administrators can view alarm handling status through the web or mobile app and further optimize and adjust the system.

[0030] As described above, the combined use of an alarm server, SMS push module, APP push module, and audible and visual alarms ensures that an immediate warning is issued upon detection of a foreign object, and the warning is pushed to relevant personnel through multiple channels. An alarm fatigue suppression formula is introduced to prevent personnel from ignoring important alarms due to frequent alarm notifications. Relevant personnel can handle alarms via the web or mobile APP, improving processing efficiency and response speed. The alarm server stores alarm handling records and performs statistical analysis according to preset rules, providing support for system optimization and decision-making.

[0031] Preferably, the user interaction module includes: A4001: Web Management Unit, which includes a Web server and a database server. The Web server is used to deploy the Web management platform and provides user login, permission management and data query functions. The database server is used to store system operation data and alarm information and supports data query and statistical functions. A4002: Mobile APP unit, including mobile APP development framework and push service. The mobile APP development framework is used for cross-platform development, and the push service is used to integrate third-party push services to push alarm information in real time. A4003: Field operation unit, including a touch screen operating interface, is used to deploy on the field control cabinet of the aggregate conveyor belt for easy operation and monitoring by field personnel.

[0032] As a preferred option, the user interaction module is used to develop a web-based management platform, allowing safety inspectors or system administrators to log in via a browser to perform production supervision, alarm handling, and data query operations; a mobile APP is also developed to support real-time monitoring and alarm handling on mobile devices. Administrators can view the operating status and alarm information of the aggregate conveyor belt in real time through the mobile APP and perform corresponding processing operations. At the same time, a touch screen operating interface is deployed on the aggregate conveyor belt field control cabinet, allowing on-site personnel to operate and monitor the system.

[0033] As mentioned above, through the web-based management platform and mobile app, managers can monitor the operating status and alarm information of the aggregate conveyor belt in real time and perform corresponding processing operations, thereby improving management efficiency. The deployment of a touch screen operating interface on the aggregate conveyor belt field control cabinet facilitates operation and monitoring by on-site personnel and reduces the difficulty of operation. The mobile app development framework supports cross-platform development, enabling managers to monitor and manage anytime and anywhere on different devices.

[0034] Preferably, the power supply and protection module includes: A5001: The power supply unit includes a USB uninterruptible power supply (UPS) to provide short-term power supply during power outages, ensuring normal system operation and secure data storage. A5002: The equipment protection unit includes a lightning protection module and a dustproof and waterproof housing. The lightning protection module is deployed at the power input and signal output terminals to protect the equipment from lightning damage, and the dustproof and waterproof housing is used for the stable operation of the equipment in harsh environments.

[0035] As described above, the power and protection module provides short-term power supply during power outages via USB uninterruptible power supply, ensuring normal system operation and secure data storage; the application of the lightning protection module and dustproof and waterproof housing protects the equipment from damage by lightning strikes, dust and moisture, improving the stability and reliability of the equipment in harsh environments.

[0036] Example 1: Upgrade Project of Foreign Object Identification System for Aggregate Conveyor Belt at Maanshan Iron & Steel Mining Co., Ltd. Implementation Background: Maanshan Iron & Steel Mining's existing foreign object identification system for aggregate conveyor belts is based on the YOLOX algorithm. However, in actual operation, due to the complex mining environment, dust, uneven lighting, and other factors, the identification accuracy is affected, and the false alarm rate is high in complex scenarios such as mixed material transportation. In order to improve the system performance, Maanshan Iron & Steel Mining decided to introduce an AI-based automated foreign object identification system for aggregate conveyor belts.

[0037] Solution Application: 1. Monitoring module: High-definition IP cameras and infrared night vision cameras are installed at key locations on the aggregate conveyor belt, such as the starting point, ending point, turning points, and areas prone to foreign matter contamination, to ensure 24-hour uninterrupted monitoring.

[0038] The camera's field of view was adjusted to cover the entire width of the belt, with some overlap between adjacent cameras, eliminating blind spots and increasing the field of view coverage to 98%.

[0039] 2. Video Analysis Module: High-performance GPU servers and AI accelerator cards are used, and the optimized YOLOX algorithm is loaded to improve the accuracy and real-time performance of foreign object recognition.

[0040] The algorithm extracts features from video frames through network layers such as convolutional layers and pooling layers, and combines inter-layer fusion operations to output foreign object recognition results, including information on the location, size, and type of the foreign object.

[0041] 3. Early Warning and Feedback Module: The alarm server receives the analysis results from the video analysis module, generates alarm information, and pushes it to relevant personnel through the SMS push module, APP push module, and audible and visual alarm.

[0042] An alarm fatigue suppression formula is introduced to prevent personnel from ignoring alarms, and the alarm fatigue level is controlled within 1.2.

[0043] 4. User interaction module: Develop a web-based management platform and a mobile app to support administrators in real-time monitoring, alarm handling, and data querying.

[0044] A touch screen operating interface is deployed on the field control cabinet of the aggregate conveyor belt to facilitate operation and monitoring by on-site personnel.

[0045] 5. Power supply and protection module: A USB uninterruptible power supply is used to ensure that the system can continue to operate normally when power is interrupted.

[0046] The equipment protection unit includes a lightning protection module and a dustproof and waterproof housing, protecting the equipment from harsh environments and achieving a system stability of 99.9%.

[0047] Improvement comparison

[0048] Example 2: New Project of Foreign Object Identification System for Aggregate Conveyor Belt in a Large Mine Background: A large mine plans to build a foreign object identification system for aggregate conveyor belts to improve production safety and efficiency. Considering the complexity of the mining environment and the requirement for high system reliability, the mine decided to adopt an AI-based automated aggregate conveyor belt foreign object identification system.

[0049] Solution Application: 1. Monitoring module: High-definition IP cameras and infrared night vision cameras are deployed along the aggregate conveyor belt to ensure comprehensive and blind-spot-free monitoring.

[0050] The camera is connected to the fiber optic transmission equipment via a network switch to ensure stable and high-speed transmission of the video stream to the video analysis server, with a video transmission bandwidth of 100Mbps.

[0051] 2. Video Analysis Module: It employs a high-performance GPU server cluster and AI accelerator card, and loads a deeply optimized YOLOX algorithm to achieve real-time foreign object recognition.

[0052] The algorithm extracts features from video frames through network layers such as convolutional layers and pooling layers, and combines inter-layer fusion operations to improve recognition accuracy, achieving a foreign object recognition accuracy of 99%.

[0053] 3. Early Warning and Feedback Module: The alarm server receives the analysis results from the video analysis module, generates alarm information, and pushes it to relevant personnel through various channels.

[0054] Establish an alarm response time formula to ensure that the system can respond to and process alarms quickly, with the total response time controlled within 10 seconds.

[0055] 4. User interaction module: Develop an advanced web-based management platform and mobile app to support simultaneous online operation and monitoring by multiple users.

[0056] A high-sensitivity touch screen operating interface is deployed on the field control cabinet of the aggregate conveyor belt to facilitate rapid response and alarm handling by field personnel, with an operation response time of ≤500ms.

[0057] 5. Power supply and protection module: It adopts a high-capacity USB uninterruptible power supply and an intelligent power management system to ensure stable operation of the system for a long time, with a UPS runtime of up to 60 minutes.

[0058] The equipment protection unit adopts military-grade lightning protection modules and an IP68-level dustproof and waterproof shell to ensure that the equipment can work normally in harsh environments, and the lightning protection voltage reaches 6kV.

[0059] Improvement Comparison:

[0060] Conclusion: Based on the above two embodiments, the AI-based automated aggregate conveyor belt foreign object identification system has achieved significant improvements in multiple aspects such as identification accuracy, response time, and system stability, providing a stronger guarantee for mine production safety.

[0061] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. An AI-based automated aggregate conveyor belt foreign object identification system; characterized in that: Including: Monitoring module: Used for multi-directional real-time monitoring of aggregate conveyor belts, expanding the monitoring range; Video analytics module: used for real-time analysis of video streams and foreign object identification using AI algorithms; Early warning and feedback module: It is used to issue an early warning immediately when a foreign object is detected, and push the warning to relevant personnel through multiple channels to form a closed-loop management mechanism; User interaction module: Used by administrators to perform real-time monitoring, alarm handling, and data query operations; Power and protection module: Used to provide a stable power supply to the system and protect the equipment from damage caused by lightning strikes, dust and moisture.

2. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 1, characterized in that: The monitoring module includes: A1001: Video acquisition unit, including a high-definition IP camera and an infrared night vision camera. The high-definition IP camera is deployed along the aggregate conveyor belt and has high resolution, wide dynamic range and low-light performance. The infrared night vision camera is used for 24-hour uninterrupted monitoring at night or in low-light environments. A1002: Video transmission unit, including network switch and fiber optic transmission equipment, is used to transmit the acquired video stream to the video analysis server.

3. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 2, characterized in that: When the monitoring module is working, it includes the following steps: S1001: Install high-definition IP cameras and infrared night vision cameras at key locations on the aggregate conveyor belt, including the starting point, ending point, turning points, and areas prone to foreign matter contamination. S1002: The camera's field of view is adjusted to cover the entire width of the belt, and there is some overlap between adjacent cameras, eliminating blind spots in monitoring; S1003: Connect the camera to the fiber optic transmission equipment via a network switch to ensure stable and high-speed transmission of the video stream to the video analysis server; S1004: Configure network parameters, including IP address, subnet mask and gateway, and detect the network connectivity between the camera and the video analysis server; S1005: The camera collects the video stream of the aggregate conveyor belt in real time and transmits it to the video analysis server via the network; S1006: Detect that the fiber optic transmission equipment is not interfered with during video stream transmission.

4. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 1, characterized in that: The video analytics module includes: A2001: The video stream receiving unit includes: a video decoding card, used to receive video streams from the monitoring module and convert the video streams from an encoded format to a processable format; A2002: The AI ​​algorithm processing unit includes a high-performance GPU server, an AI accelerator card, and a YOLOX optimized algorithm, which is used for real-time monitoring of aggregate conveying scenarios, accurately identifying foreign objects of different sizes, shapes, and colors, and performing real-time analysis. A2003: Result Output Unit: Data Interface Card, used to output the analysis results of the AI ​​algorithm processing unit to the early warning and feedback module, and provide a data interface with the early warning and feedback module.

5. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 4, characterized in that: When the video analytics module is working, Includes the following steps: S2001: Start the high-performance GPU server and load the AI ​​accelerator card driver; S2002: Configure server network parameters and detect and monitor the network connectivity of the early warning and feedback modules; S2003: Load the optimized version of the YOLOX algorithm on the video analytics server and configure the algorithm parameters; S2004: The detection algorithm is adapted to aggregate conveying scenarios and has a high recognition rate for foreign objects of different sizes, shapes and colors; S2005: The video analysis server receives video streams from the monitoring module and performs real-time analysis using the YOLOX algorithm; S2006: The algorithm extracts features from video frames through network layers such as convolutional layers and pooling layers, and combines inter-layer fusion operations to output foreign object recognition results, including information on the location, size, and type of the foreign object.

6. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 1, characterized in that: The early warning and feedback module includes: A3001: Alarm Information Processing Unit: Alarm Server, equipped with a high-performance processor and sufficient storage space, responsible for real-time processing and long-term storage of alarm information. It is used to receive analysis results from the video analysis module, generate alarm information, and perform storage and statistical analysis. A3002: Alarm push unit, including SMS push module and APP push module, is used to integrate SMS gateway to push alarm information to relevant personnel via SMS. It also integrates real-time monitoring and alarm processing functions, allowing managers to monitor the operating status of belt conveyor in real time. A3003: On-site alarm unit, including an audible and visual alarm, is deployed on the aggregate conveyor belt site. When a foreign object is detected, it emits a strong audible and visual alarm signal to remind on-site personnel to handle the situation in a timely manner and prevent the foreign object from damaging the conveyor belt or causing a safety accident.

7. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 6, characterized in that: The early warning and feedback module operates by including the following steps: S3001: Start the alarm server, configure the connection parameters with the video analysis server, and check that it can receive alarm information from the video analysis server; S3002: Set the alarm information storage path and statistical analysis rules; S3003: When the video analysis server detects a foreign object, it immediately generates an alarm message and sends it to the alarm server; S3004: The alarm server pushes alarm information to relevant personnel through SMS push module, APP push module and sound and light alarm according to preset rules. When the alarm server's alarm fatigue level approaches the threshold, it notifies relevant personnel. S3005: After receiving the alarm information, relevant personnel can handle the alarm through the web or mobile APP. S3006: The alarm server stores alarm processing records and performs statistical analysis according to preset rules; S3007: Administrators can view alarm handling status through the web or mobile app and further optimize and adjust the system.

8. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 1, characterized in that: The user interaction module includes: A4001: Web Management Unit, which includes a Web server and a database server. The Web server is used to deploy the Web management platform and provides user login, permission management and data query functions. The database server is used to store system operation data and alarm information and supports data query and statistical functions. A4002: Mobile APP unit, including mobile APP development framework and push service. The mobile APP development framework is used for cross-platform development, and the push service is used to integrate third-party push services to push alarm information in real time. A4003: Field operation unit, including a touch screen operating interface, is used to deploy on the field control cabinet of the aggregate conveyor belt for easy operation and monitoring by field personnel.

9. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 8, characterized in that: The user interaction module is used to develop a web-based management platform, allowing safety inspectors or system administrators to log in via a browser to perform production supervision, alarm handling, and data query operations. A mobile app is also developed to support real-time monitoring and alarm handling on mobile devices. Administrators can view the operating status and alarm information of the aggregate conveyor belt in real time through the mobile app and perform corresponding operations. At the same time, a touch screen operating interface is deployed on the aggregate conveyor belt field control cabinet, allowing on-site personnel to operate and monitor the system.

10. The AI-based automated aggregate conveyor belt foreign object identification system according to claim 1, characterized in that: The power supply and protection module includes: A5001: Power supply unit, including USB uninterruptible power supply, used to provide short-term power supply during power outages to ensure normal system operation and secure data storage; A5002: Equipment protection unit, including a lightning protection module and a dustproof and waterproof housing. The lightning protection module is deployed at the power input and signal output terminals to protect the equipment from lightning damage. The dustproof and waterproof housing is for the stable operation of the equipment in harsh environments.

Images