State monitoring method based on cross-sea stockpiling line conveying belt

Abstract

The invention discloses a state monitoring method based on a conveying belt of a cross-sea stockpiling line, and relates to the technical field of state monitoring of the conveying belt of the cross-sea stockpiling line, comprehensive monitoring is realized through four steps: step 1, collecting belt no-load reference characteristics, and monitoring foreign matters and tearing through AI visual identification; the method comprises the following steps of: 1, cooperatively identifying abnormities by using vibration and temperature two-dimensional monitoring, 3, evaluating the risk of foreign matter evolution into tearing and the tearing cause, 4, integrating a multi-modal monitoring report to form closed-loop operation and maintenance, and carrying out visual, vibration and temperature multi-dimensional data fusion and reference feature comparison and dynamic verification to obtain a multi-dimensional data fusion model. Foreign matters such as rod-shaped objects and net-shaped objects and tearing are accurately recognized, the problem of visual failure caused by material shielding is effectively solved, the abnormity recognition accuracy is improved, closed-loop management from monitoring to disposal is achieved, safe and stable operation of a cross-sea stockpiling line conveying belt is guaranteed, the operation and maintenance cost is reduced, and the conveying efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of condition monitoring technology for conveyor belts in cross-sea stockpiling lines, and specifically to a condition monitoring method based on conveyor belts in cross-sea stockpiling lines. Background Technology

[0002] Conveyor belts on cross-sea stacking lines are key equipment in logistics scenarios such as ports. Traditional monitoring relies on manual inspection, which is inefficient and has a high rate of missed inspections. Complex working conditions such as material obstruction often lead to the failure of visual monitoring. They are also prone to malfunctions due to foreign objects getting stuck or tearing, affecting transportation. Therefore, there is an urgent need for accurate and multi-dimensional monitoring methods.

[0003] The existing or traditional methods for monitoring the condition of conveyor belts in cross-sea stockpiling lines have the following technical problems: 1. The existing methods for monitoring the condition of conveyor belts in cross-sea stockpiling lines lack a unified set of condition monitoring benchmark features. They do not collect benchmark images, average vibration signals and average temperatures under no-load conditions through the system. This leads to a lack of unified reference standards during the monitoring process. Monitoring data from different time periods and different equipment cannot be effectively compared. They are prone to misjudgment due to environmental interference or equipment errors, and it is difficult to distinguish between normal operation fluctuations and abnormal conditions.

[0004] 2. Traditional condition monitoring methods based on conveyor belts in cross-sea material stacking lines lack hierarchical processing of visual monitoring results. They do not distinguish between key and non-key foreign objects and tear monitoring locations in the dual-dimensional monitoring. This leads to the application of the same monitoring intensity to all areas when the vision is not in failure, resulting in a waste of resources. When the vision fails, it is impossible to accurately locate the areas that need to be monitored, which easily leads to missed detections. Especially under the condition of frequent material obstruction, the efficiency of anomaly identification drops significantly.

[0005] 3. Traditional methods for monitoring the condition of conveyor belts in cross-sea material stacking lines lack a dual-dimensional collaborative monitoring mechanism for vibration and temperature. Relying solely on visual monitoring or independent vibration or temperature monitoring can lead to an inability to identify anomalies by matching vibration signals with temperature changes when visual failure occurs due to material obstruction. This makes it difficult to determine the state of rod-shaped objects, mesh-like objects, or tears in the obstructed area, increasing safety hazards caused by obstruction.

[0006] 4. Traditional condition monitoring methods for conveyor belts in cross-sea stockpiling lines lack early warning mechanisms for the risk of foreign objects evolving into tearing. They do not track the dynamic changes in the vibration and temperature characteristics of foreign objects, which leads to the inability to identify the evolution process of "jamming - increased friction - tearing" in advance. They can only deal with it after the tearing occurs, which increases equipment downtime and maintenance costs, and may even cause more serious chain failures. Summary of the Invention

[0007] The purpose of this invention is to provide a condition monitoring method based on a cross-sea material stacking line conveyor belt, which solves the problems existing in the background art.

[0008] To solve the above technical problems, the present invention adopts the following technical solution: The present invention provides a condition monitoring method based on the conveyor belt of the cross-sea stockpiling line, including: Step 1, AI vision preliminary monitoring: collecting the condition monitoring benchmark features of each belt under no-load conditions, including fiber optic auscultation vibration benchmark, and using AI vision recognition combined with fiber optic vibration benchmark to monitor the transport of foreign objects and tearing of each belt.

[0009] Step 2, Dual-Dimensional Fusion Monitoring: When material obstruction causes visual failure, the belt is monitored for foreign objects and tears through a dual-dimensional approach of fiber optic auscultation vibration monitoring and temperature monitoring, and the corresponding condition of the belt is comprehensively evaluated.

[0010] Step 3: State Correlation Assessment: Based on the results of dual-dimensional monitoring and collaborative identification, and with fiber optic auscultation vibration data as the core, the risk of foreign objects evolving into tearing is warned within a set time period, and the cause of belt tearing is assessed.

[0011] Step 4: Multimodal fusion: Integrate visual, fiber optic auscultation vibration and temperature monitoring reports to form a closed-loop operation and maintenance process.

[0012] The beneficial effects of the present invention are as follows: 1. The condition monitoring method based on the cross-sea material stacking line conveyor belt provided by the embodiments of the present invention, by collecting the condition monitoring benchmark characteristics of each belt under no-load conditions, and by deploying equipment such as high frame rate industrial cameras, distributed fiber optic stethoscope sensors, and thermal imaging cameras, records benchmark images, average vibration signals and average temperatures without foreign objects and without tears, providing a unified reference standard for subsequent monitoring, which helps to eliminate identification deviations caused by the lack of benchmarks and ensures the accuracy of anomaly judgment.

[0013] 2. In the initial monitoring process of AI vision, this embodiment of the invention compares real-time images with a benchmark template to distinguish between key and non-key foreign objects and tear monitoring locations in the dual-dimensional monitoring. Manual monitoring is pushed to non-key locations, and dual-dimensional monitoring is initiated for key locations. This helps to achieve reasonable allocation of monitoring resources, avoid ineffective monitoring that consumes resources, and reduce the risk of missed detection when vision fails.

[0014] 3. In the dual-dimensional fusion monitoring process of this invention, when vision fails, vibration signal monitoring and temperature value monitoring work together to match and judge the characteristics of rod-shaped objects, mesh-like objects, and tears, which helps to make up for the limitations of single visual monitoring and can still accurately identify anomalies under material obstruction conditions.

[0015] 4. In the state correlation assessment process, this invention tracks the changes in vibration and temperature characteristics of foreign objects and traces back historical data before tearing to warn of the risk of foreign objects evolving into tearing and to clarify the cause of tearing. This is conducive to early intervention in potential risk points, reducing downtime losses caused by tearing, and providing a basis for targeted maintenance.

[0016] 5. In the multimodal fusion process, the embodiments of the present invention integrate visual, vibration and temperature monitoring reports, and eliminate false alarms of a single mode through cross-validation to form a comprehensive report that includes the anomaly type, location, evolution stage and cause. This promotes the formulation of operation and maintenance plans, effect tracking and baseline data updates, which is conducive to building a complete process of "monitoring-analysis-handling-closed loop", improving operation and maintenance efficiency and ensuring the continuous and stable operation of the conveyor belt of the cross-sea stockpile line. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the implementation steps of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please see Figure 1 As shown, the present invention provides a condition monitoring method for conveyor belts in cross-sea stockpiling lines. The method includes: Step 1, AI vision preliminary monitoring: collecting the condition monitoring benchmark features of each conveyor belt under no-load conditions, including fiber optic auscultation vibration benchmark, and using AI vision recognition combined with fiber optic vibration benchmark to monitor for transported foreign objects and tearing of each conveyor belt.

[0021] In a specific embodiment, the process of collecting the status monitoring benchmark features of each belt under no-load conditions is as follows: When the belt is in an no-load state where no materials are being transported and no operation is in progress, high frame rate, high pixel industrial cameras and dust-penetrating cameras are deployed above the head and tail of each belt and on the side plate of the material drop port sealing device. Distributed fiber optic stethoscope sensors are fixed along the belt steel frame, a four-channel stethoscope fiber optic host is placed in the SR102 head control room, and a thermal imaging camera is installed on the idler roller device bridge under the belt, aligned with the belt bearing surface.

[0022] The condition monitoring baseline features include visual baselines, vibration baselines, and temperature baselines. The vibration baseline includes the mechanical noise characteristics of the idler rollers, and the temperature baseline includes the normal operating temperature of the roller bearings. The normal operating temperature of the roller bearings is collected by temperature sensors installed in the roller bearing housings, and then the average value is calculated as the temperature baseline. High frame rate cameras and dust-penetrating cameras are activated to scan the entire surface of each belt and record baseline images without foreign objects or tears, including belt edge contours, idler roller positions, and surface flatness. These images are stored in the database as visual comparison templates.

[0023] Materials that have been screened and are free of foreign objects are transported via various types of conveyor belts for testing. Vibration signals at the feed inlets of each conveyor belt are collected at various time points using distributed fiber optic stethoscope sensors. The average vibration signal at the feed inlets of each conveyor belt at each time point is calculated by averaging. The average vibration signal includes the average vibration frequency of the idler rollers and the average vibration frequency of the conveyor belt. Based on this, the threshold ranges for the idler roller rotation frequency and the conveyor belt vibration frequency are set to ±30% of the average vibration frequency of the idler rollers and ±30% of the average vibration frequency of the conveyor belt, respectively. Temperature sensors and thermal imaging cameras are used to record the surface temperature of the conveyor belt and the temperature of the idler rollers at various time points during the transport test. The average surface temperature of the conveyor belt and the average temperature of the idler rollers during the transport test are then calculated by averaging.

[0024] It should be noted that in belt conveyor scenarios, "the feed inlet of each belt" refers to the entrance position where materials enter each belt. Specifically, the core function of a belt conveyor is to transport materials from one place to another, and the "feed inlet" is the starting point where materials enter the current belt. When materials enter the belt, they will directly interact with the belt and the idlers near the feed inlet, such as the material impacting the belt and causing changes in the force on the idlers. Therefore, the vibration signal at the feed inlet can reflect the operating status of the belt in the key link of receiving materials and starting to transport them.

[0025] It should also be noted that in industrial cameras, high frame rate refers to 30 frames per second or higher, which can capture details when the belt is running at high speed, and high pixel count refers to 2 million pixels or higher, which can clearly present the surface and edge features of the belt.

[0026] In a specific embodiment, the monitoring of foreign objects transported by each conveyor belt using AI visual recognition is carried out as follows: During the material transport process, industrial cameras and dust-penetrating cameras deployed above the head and tail of each conveyor belt and at the material drop-off port collect images of the conveyor belt surface at preset collection time intervals. The supplementary lighting device is activated, and the self-cleaning air device is activated. The AI ​​visual recognition algorithm first calls the visual comparison template in the database and compares the real-time collected images with the reference template. If an anomaly is found between the image corresponding to a certain collection time point and the image of the reference template, the image corresponding to that collection time point is recorded as an image of suspected foreign objects, and the location is recorded as a non-key foreign object monitoring location for dual-dimensional monitoring. At the same time, the location is marked on the cloud platform and pushed to the mobile APP to remind the maintenance personnel to conduct manual monitoring.

[0027] If no anomalies are found when comparing the image corresponding to a certain acquisition time point with the baseline template image, it is determined that the actual location of the image corresponding to that acquisition time point is obstructed by material, causing visual failure. The actual location of the image corresponding to that acquisition time point is recorded as the key foreign object monitoring location for dual-dimensional monitoring. Then, foreign object monitoring is carried out on the key monitoring location for dual-dimensional monitoring through dual-dimensional monitoring.

[0028] It should be noted that while monitoring the transport of foreign objects on each conveyor belt using AI visual recognition, the system also monitors belt deviation. By comparing the offset of the belt edge with the baseline contour in the real-time image, if the offset exceeds a set threshold, it is determined to be a deviation, thereby triggering an early warning notification.

[0029] It should also be noted that during the process of pushing notifications to the mobile app to remind maintenance personnel to conduct manual monitoring, the acoustic fingerprint data converted from the fiber optic vibration signal at that location is simultaneously synchronized to assist in manual annotation and judgment.

[0030] For example, when AI vision identifies an anomaly between an image and a baseline template in a certain area, such as an irregular shadow on the surface of a conveyor belt, and it is impossible to determine whether it is a foreign object or a material accumulation, the system marks this location as a non-key foreign object monitoring location for dual-dimensional monitoring and pushes it to the mobile APP. At the same time, it will synchronize the fiber optic vibration acoustic data of this location. If the acoustic data presents a regular high-frequency "clicking" impact sound, which is consistent with the collision characteristics of a rod-shaped object and the conveyor belt, the maintenance personnel can determine that it is a rod-shaped object and complete the labeling by combining the shape of the shadow in the image. If the acoustic data is only a messy low-frequency friction sound without obvious foreign object collision characteristics, it can be determined that it is a material accumulation rather than a foreign object. The acoustic data supplements the ambiguity of visual information, thereby improving the accuracy of manual labeling.

[0031] It should also be noted that, assuming the reference template is a surface image of the belt when it is unloaded, with no foreign objects, smooth edges, and clear idler positions, if a long strip-shaped object appears on the belt surface or a clear crack appears on the edge in the real-time image at a certain acquisition time point, it is considered "an anomaly exists". The system marks it as a non-key foreign object monitoring location in dual-dimensional monitoring and pushes it for manual confirmation. If the belt is completely covered by material in the real-time image, and the visual characteristics of the covered area are highly similar to the corresponding foreign object-free area in the reference template, making it impossible to identify whether there is a foreign object or tear, it is considered "no anomaly exists". It is determined that the visual failure is caused by material obstruction and is marked as a key foreign object monitoring location in dual-dimensional monitoring, and vibration and temperature monitoring are initiated.

[0032] In a specific embodiment, the tear detection of each belt using AI visual recognition is carried out as follows: When performing tear detection using AI visual recognition, an industrial camera captures images of the edges and surfaces of each belt in real time according to a preset acquisition time interval. The AI ​​algorithm calls the edge contour and surface flatness data of the tear-free state in the status monitoring benchmark features. The real-time captured images of the edges and surfaces of each belt are compared pixel-level with the edge contour and surface flatness data of the tear-free state in the status monitoring benchmark features to detect whether there are burrs that deviate from the benchmark contour by more than a set burr length threshold. If the detection result shows that there are burrs that deviate from the benchmark contour by more than the set burr length threshold, the actual position of the image corresponding to the deviation from the benchmark contour is recorded as a non-key tear detection position of the dual-dimensional monitoring. At the same time, the position is marked on the cloud platform and pushed to the mobile APP to remind the maintenance personnel to perform manual monitoring. If the detection result does not show that there are burrs that deviate from the benchmark contour by more than the set burr length threshold, it is determined that there is material occlusion causing visual failure. The actual position of the image corresponding to the image that does not deviate from the benchmark contour is recorded as a key tear detection position of the dual-dimensional monitoring. Then, tear detection is carried out on the key tear detection position of the dual-dimensional monitoring through dual-dimensional monitoring.

[0033] It should be noted that while using AI visual recognition to monitor for tearing of each belt, it also monitors for belt deviation. By comparing the edge offset distance at the pixel level, an early warning is triggered if the edge offset distance exceeds the benchmark range.

[0034] It should also be noted that when pushing notifications to the mobile app to remind maintenance personnel to perform manual monitoring, it is necessary to combine the vibration sound pattern characteristics collected by the fiber optic stethoscope sensor to improve the accuracy of manual labeling. For example, when AI vision identifies a blurred abnormal outline on the edge of the belt, such as a small tear or attached blocky material, and pushes a notification to the mobile app to remind manual monitoring, if the vibration sound pattern collected by the fiber optic stethoscope sensor is combined simultaneously, if the sound pattern shows a high-frequency, intermittent "banging" sound, which matches the friction characteristics of blocky material and belt, the maintenance personnel can determine that it is foreign object attachment. If the sound pattern shows a low-frequency, continuous "scratching" sound, which matches the friction characteristics of tear and idler roller, it can be determined as an early tear. By complementing the sound pattern characteristics with the visual image, misjudgments caused by image blurring are avoided, thereby improving the accuracy of manual labeling.

[0035] In the initial monitoring process of AI vision, this invention compares real-time images with a benchmark template to distinguish between key and non-key foreign objects and tear monitoring locations in the dual-dimensional monitoring. Manual monitoring is pushed to non-key locations, and dual-dimensional monitoring is initiated for key locations. This helps to achieve a reasonable allocation of monitoring resources, avoids ineffective monitoring that consumes resources, and reduces the risk of missed detection when vision fails.

[0036] Step 2, Dual-Dimensional Fusion Monitoring: When material obstruction causes visual failure, the belt is monitored for foreign objects and tears through a dual-dimensional approach of fiber optic auscultation vibration monitoring and temperature monitoring, and the corresponding condition of the belt is comprehensively evaluated.

[0037] In a specific embodiment, the process of using fiber optic auscultation vibration monitoring to detect foreign objects and tears in the conveyor belt is as follows: The dual-dimensional monitoring method consists of vibration signal monitoring and temperature value monitoring. When material obstruction causes visual failure, the distributed fiber optic auscultation sensor first starts operating. Based on the conveyor belt segment corresponding to the non-key foreign object monitoring location in the dual-dimensional monitoring, the sensor collects the idler rotation frequency and belt vibration frequency at the feed inlet of the conveyor belt segment corresponding to the non-key foreign object monitoring location in the dual-dimensional monitoring, and compares them with the idler rotation frequency threshold and belt vibration frequency recorded when transporting materials without foreign objects in the previous period. If high-frequency pulse vibration is detected, and the idler rotation frequency and belt vibration frequency are not within the idler rotation frequency threshold range and belt vibration frequency range, then it is determined to be an obstructed rod-shaped object.

[0038] If low-frequency continuous vibration occurs, and the rotation frequency of the idler roller and the vibration frequency of the belt are both outside the threshold range of the idler roller rotation frequency and the belt vibration frequency range, and the amplitude is higher than the set percentage of the average signal, and the spectrum shows the characteristics of multi-frequency superposition, then it is determined to be a blocked mesh. If low-frequency strong vibration is captured, and the rotation frequency of the idler roller and the vibration frequency of the belt are both outside the threshold range of the idler roller rotation frequency and the belt vibration frequency range, and the amplitude is higher than the set percentage of the average signal, and the duration exceeds the set duration, then it is determined to be a blocked tear. Then, the monitoring of transported foreign objects and tear detection is carried out by temperature value monitoring.

[0039] In a specific embodiment, the conveyor belt is monitored for foreign object transport and tearing by temperature monitoring. The specific process is as follows: The temperature monitoring method uses a thermal imaging camera installed on the idler roller bridge under the belt to record the temperature of the obscured area in real time according to a preset acquisition time interval. The temperature is compared with the average belt surface temperature and average idler roller temperature obtained from the transportation test. When monitoring the key foreign object monitoring position of the dual-dimensional monitoring, the temperature at the corresponding position changes synchronously with the vibration pulse, showing an instantaneous rise and fall. The temperature difference fluctuation meets the preset pulsed temperature difference threshold based on the average belt surface temperature and average idler roller temperature, that is, it matches the vibration characteristics of the rod-shaped object. Then it is determined that there is a rod-shaped object obscured by material at the key tear monitoring position of the dual-dimensional monitoring. If the temperature increases stepwise, the cumulative temperature difference per unit time is within the preset stepwise temperature difference range with reference to the reference temperature, and the degree of overall temperature deviation from the average belt surface temperature and average idler roller temperature is consistent with the friction characteristics of the mesh, then it is determined that there is a rod-shaped object obscured by material at the key tear monitoring position of the dual-dimensional monitoring.

[0040] If, when monitoring the key tear monitoring location in the dual-dimensional monitoring system, the temperature rises linearly, the high-temperature area forms a continuous strip shape with the belt running and there is no pulse fluctuation, and the temperature difference change range meets the preset tear characteristic temperature difference range based on the reference temperature, then it is determined that there is a tear that is blocked by material at the key tear monitoring location in the dual-dimensional monitoring system.

[0041] It should be noted that "instantaneous rise and fall" refers to the temperature changing rapidly in sync with the vibration pulse. When the rod-shaped object collides intermittently with the idler roller and belt, heat is generated due to friction at the moment of each collision, and the temperature will suddenly rise. After the collision ends, the heat is gradually dissipated, and the temperature slowly drops, forming a pulse-like change from a sudden rise to a slow fall. The "preset pulse-like temperature difference threshold" is a reasonable fluctuation range set based on the average belt surface temperature and the average idler roller temperature, used to define whether this instantaneous change conforms to the characteristics of the rod-shaped object.

[0042] "Step-by-step increase" refers to the fact that the mesh generates heat due to continuous friction. The temperature does not rise continuously and uniformly, but increases by a certain amount at intervals, showing a step-like trend from stability to a small jump to stability again. This is because the friction intensity between the mesh and the belt is relatively stable but heat continues to accumulate. The "preset step-by-step temperature difference range" is the temperature rise per unit time and the upper limit of the overall temperature difference set based on the reference temperature, which is used to match the friction characteristics of the mesh.

[0043] It should also be noted that, assuming a certain area is obscured by material during belt conveyor transport, and the baseline average belt surface temperature is 25°C, if thermal imaging shows that the temperature in that area rises sharply from 25°C to 30°C with vibration and then falls back, meeting the pulsed temperature difference threshold, then rod-like objects are identified. If the temperature rises by 3°C every 30 seconds, reaching an overall temperature of 37°C, meeting the stepped temperature pattern, then mesh-like objects are identified. If the temperature rises by 10°C per minute, forming a continuous strip of high-temperature material, then tearing is identified. These examples are merely illustrative and are not the only possible explanations.

[0044] In a specific embodiment, the comprehensive evaluation of the belt's corresponding state is carried out as follows: If the idler rotation frequency and belt vibration frequency are both outside the idler rotation frequency threshold range and belt vibration frequency range, and the corresponding temperature change differs from the average temperature in accordance with the temperature matching relationship, the abnormality type is confirmed. Then, the location of the abnormality on the belt is determined by combining the fiber optic sensor position. If the idler rotation frequency and belt vibration frequency are both outside the idler rotation frequency threshold range and belt vibration frequency range, but the temperature does not differ from the average temperature, it is determined to be material unevenness interference, and no alarm is triggered. If the temperature deviates from the average temperature, but the vibration signal has no corresponding characteristics, it is determined to be material temperature difference, and no alarm is triggered.

[0045] In the dual-dimensional fusion monitoring process of this invention, when visual monitoring fails, vibration signal monitoring and temperature value monitoring work together to match and judge the characteristics of rod-shaped objects, mesh-like objects, and tears, which helps to overcome the limitations of single visual monitoring and can still accurately identify anomalies under material obstruction conditions.

[0046] Step 3: State Correlation Assessment: Based on the results of dual-dimensional monitoring and collaborative identification, and with fiber optic auscultation vibration data as the core, the risk of foreign objects evolving into tearing is warned within a set time period, and the cause of belt tearing is assessed.

[0047] In a specific embodiment, the process of providing early warning of the risk of foreign object tearing within a set time period is as follows: When providing early warning of the risk of foreign object tearing within a set time period, the vibration and temperature characteristics of the identified foreign object are tracked and compared with the status monitoring benchmark characteristics of the transportation test. The vibration acoustic signal of the foreign object captured by the fiber optic auscultation sensor is monitored in real time to determine whether the frequency deviation has further expanded from the idler roller rotation frequency threshold range and the belt vibration frequency range, whether the amplitude has continued to increase, and whether the temperature rise rate has changed from pulsed and step-like to linear growth. If the above characteristics appear within the set time period, and the vibration spectrum begins to show the low-frequency strong vibration trend characteristic of tearing, it is determined that the foreign object has entered the evolution stage from jamming to increased friction and then to imminent tearing. An upgraded early warning is immediately triggered to indicate that the belt is at risk of tearing.

[0048] It should be noted that, assuming the transport test sets the idler rotation frequency threshold range to 70-130Hz and the belt vibration frequency threshold range to 35-65Hz, if the initial detected vibration of the rod-shaped object shows an idler rotation frequency of 135Hz and a belt vibration frequency of 70Hz, and within the next 30 seconds the idler rotation frequency rises to 150Hz and the belt vibration frequency rises to 85Hz, this is considered a "further expansion" of the frequency deviation. At the same time, the amplitude increases from the initial 0.1mm to 0.3mm, the temperature change changes from a sudden pulse to a linear increase, and low-frequency strong vibrations of 500-1.5kHz appear in the vibration spectrum, it is determined that the rod-shaped object has entered the stage of increased friction from jamming, which is about to lead to tearing, and an upgrade warning is immediately triggered.

[0049] It should be noted that "the temperature rise rate changes from pulsed and stepwise to linear growth" means that the initial temperature change of the foreign object conforms to its own characteristics. For rod-shaped objects, it is pulsed, meaning that the temperature rises and falls instantaneously with the vibration pulse and then falls back; for mesh-shaped objects, it is stepwise, meaning that the temperature rises by a certain amount every once in a while and then stabilizes briefly. However, when the foreign object evolves towards tearing, the temperature rise pattern changes from the original intermittent and staged temperature rise to continuous and uniform temperature rise, meaning that the temperature rise per unit time is basically the same, without obvious drops or pauses. For example, if it is initially identified as a rod-shaped object, and the temperature change is a sudden rise of 5°C during vibration and then falls back to the baseline after the vibration stops, then it is pulsed. When the rod-shaped object becomes more stuck, the friction changes from intermittent to continuous, and the temperature changes from a sudden rise and fall to a stable rise of 8°C per minute, and then no longer falls back, then it is linear growth. Similarly, if the mesh-shaped object changes from a stable stepwise temperature rise of 3°C every 30 seconds to a linear growth of a uniform temperature rise of 6°C per minute, it indicates that the friction has entered a state of continuous intensification, which is a signal of evolution towards tearing.

[0050] In a specific embodiment, the process of assessing the cause of belt tearing is as follows: Based on the dual-dimensional monitoring data at the time of tearing, which includes vibration signal data and temperature value data, the historical records of the tearing set time period are then traced back. If there is a vibration signal that matches the characteristics of a foreign object in the historical records of the tearing set time period, and the temperature anomaly at the corresponding location matches the characteristics of the foreign object, then it is determined that the tearing is caused by foreign object obstruction.

[0051] If there are no abnormal vibrations or temperatures related to foreign objects in the historical records of the tear within the set time period, and only the vibration and temperature characteristics unique to the tear are present, then the tear is determined to be caused by non-foreign object factors.

[0052] It should be noted that the dual-dimensional monitoring data at the time of the tear is based on the vibration acoustic data recorded by the fiber optic auscultation sensor to assess the cause of the belt tear.

[0053] In the state correlation assessment process, this invention tracks changes in the vibration and temperature characteristics of foreign objects and reviews historical data before tearing to warn of the risk of foreign objects evolving into tearing and to clarify the cause of tearing. This is beneficial for early intervention in potential risk points, reducing downtime losses caused by tearing, and providing a basis for targeted maintenance.

[0054] Step 4: Multimodal fusion: Integrate visual, fiber optic auscultation vibration and temperature monitoring reports to form a closed-loop operation and maintenance process.

[0055] In a specific embodiment, the integration of visual, vibration, and temperature monitoring reports is carried out as follows: Based on the baseline characteristics and real-time identification results of visual monitoring, the abnormal spectrum and frequency deviation data of vibration monitoring, and the abnormal temperature rise trend and baseline temperature difference information of temperature monitoring, combined with the risk warning of foreign object evolution and the assessment conclusion of tearing cause, false alarms of a single mode are eliminated through cross-validation, and a comprehensive report containing the anomaly type, precise location, evolution stage, and cause is formed. Based on this, the operation and maintenance process is promoted to formulate handling plans, track the effects, and dynamically update the baseline data, thus completing the closed-loop process from monitoring to resolution.

[0056] In the multimodal fusion process, this invention integrates visual, vibration, and temperature monitoring reports, eliminates false alarms from a single mode through cross-validation, and generates a comprehensive report that includes the anomaly type, location, evolution stage, and cause. This promotes the formulation of operation and maintenance plans, effect tracking, and baseline data updates, which is conducive to building a complete "monitoring-analysis-handling-closed-loop" process, improving operation and maintenance efficiency, and ensuring the continuous and stable operation of the conveyor belt of the cross-sea stockpile line.

[0057] The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line provided by this invention collects the condition monitoring benchmark characteristics of each belt under no-load conditions. By deploying equipment such as high frame rate industrial cameras, distributed fiber optic stethoscope sensors, and thermal imaging cameras, it records benchmark images, average vibration signals, and average temperatures without foreign objects or tears. This provides a unified reference standard for subsequent monitoring, which helps to eliminate identification bias caused by the lack of benchmarks and ensures the accuracy of anomaly judgment.

[0058] The above description is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they should all fall within the protection scope of the present invention.

Claims

1. A condition monitoring method based on conveyor belts in a cross-sea stockpiling line, characterized in that, include: Step 1: Preliminary AI Vision Monitoring: Collect the status monitoring benchmark characteristics of each belt under no-load conditions, including fiber optic auscultation vibration benchmark, and use AI vision recognition combined with fiber optic vibration benchmark to monitor the transport of foreign objects and tearing of each belt. Step 2, Dual-dimensional fusion monitoring: When material obstruction causes visual failure, the belt is monitored for transporting foreign objects and for tearing through a dual-dimensional approach of fiber optic auscultation vibration monitoring and temperature monitoring, and the corresponding condition of the belt is comprehensively evaluated. Step 3, State Correlation Assessment: Based on the results of dual-dimensional monitoring and collaborative identification, and taking fiber optic auscultation vibration data as the core, the risk of foreign objects evolving into tearing is warned within a set time period, and the cause of belt tearing is assessed. Step 4: Multimodal fusion: Integrate visual, fiber optic auscultation vibration and temperature monitoring reports to form a closed-loop operation and maintenance process.

2. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 1, characterized in that, The specific process for collecting the status monitoring benchmark characteristics of each conveyor belt under no-load conditions is as follows: When the belt is in an unloaded state without transporting materials or running, first fix the distributed fiber optic auscultation sensor along the belt steel frame, place the four-channel auscultation fiber optic host in the SR102 head control room, then deploy the high frame rate, high pixel industrial camera and dust-penetrating camera on the head and tail of each belt and the side plate of the material drop sealing device, and install the thermal imaging camera on the idler roller device bridge under the belt, aiming at the belt bearing surface; The condition monitoring baseline features include visual baselines, vibration baselines, and temperature baselines. The vibration baseline includes the mechanical noise characteristics of the idler rollers, and the temperature baseline includes the normal operating temperature of the roller bearings. The normal operating temperature of the roller bearings is collected by temperature sensors installed in the roller bearing housings, and then the average value is calculated as the temperature baseline. High frame rate cameras and dust-penetrating cameras are activated to scan the entire surface of each belt and record baseline images without foreign objects or tears, including belt edge contours, idler roller positions, and surface flatness. These images are stored in the database as visual comparison templates. Materials that have been screened and are free of foreign objects are transported via various types of conveyor belts for testing. Vibration signals at the feed inlets of each conveyor belt are collected at various time points using distributed fiber optic stethoscope sensors. The average vibration signal at the feed inlets of each conveyor belt at each time point is calculated by averaging. The average vibration signal includes the average vibration frequency of the idler rollers and the average vibration frequency of the conveyor belt. Based on this, the threshold ranges for the idler roller rotation frequency and the conveyor belt vibration frequency are set to ±30% of the average vibration frequency of the idler rollers and ±30% of the average vibration frequency of the conveyor belt, respectively. Temperature sensors and thermal imaging cameras are used to record the surface temperature of the conveyor belt and the temperature of the idler rollers at various time points during the transport test. The average surface temperature of the conveyor belt and the average temperature of the idler rollers during the transport test are then calculated by averaging.

3. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 2, characterized in that, The process of using AI visual recognition to monitor for foreign objects transported on each conveyor belt is as follows: During the conveyor belt transportation process, industrial cameras and dust-penetrating cameras deployed above the head and tail of each conveyor belt and at the material drop-off point collect images of the conveyor belt surface at preset collection time intervals. The supplementary lighting device is activated, and the self-cleaning air device is activated. The AI ​​visual recognition algorithm first calls the visual comparison template in the database and compares the real-time collected images with the reference template. If the image corresponding to a certain collection time point is abnormal after comparison with the reference template image, the image corresponding to that collection time point is recorded as an image of suspected foreign objects, and the location is recorded as a non-key foreign object monitoring location in the dual-dimensional monitoring. At the same time, the location is marked on the cloud platform and pushed to the mobile APP to remind the operation and maintenance personnel to conduct manual monitoring. If no anomalies are found when comparing the image corresponding to a certain acquisition time point with the baseline template image, it is determined that the actual location of the image corresponding to that acquisition time point is obstructed by material, causing visual failure. The actual location of the image corresponding to that acquisition time point is recorded as the key foreign object monitoring location for dual-dimensional monitoring. Then, foreign object monitoring is carried out on the key monitoring location for dual-dimensional monitoring through dual-dimensional monitoring.

4. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 3, characterized in that, The process of using AI visual recognition to detect tearing in each belt is as follows: When using AI visual recognition for tear detection, industrial cameras capture images of the edges and surfaces of each belt in real time according to a preset acquisition time interval. The AI ​​algorithm calls upon the edge contours and surface flatness data of the tear-free state in the condition monitoring benchmark features. It performs pixel-level comparison between the real-time captured images of the edges and surfaces of each belt and the edge contours and surface flatness data of the tear-free state in the condition monitoring benchmark features to detect whether there are burrs that deviate from the benchmark contour by more than a set burr length threshold. If the detection result shows that there are burrs that deviate from the benchmark contour by more than the set burr length threshold, the actual location of the image corresponding to the deviation from the benchmark contour is recorded as a non-key tear detection location in the dual-dimensional monitoring. At the same time, this location is marked on the cloud platform and pushed to the mobile APP to remind maintenance personnel to conduct manual monitoring. If the detection result does not show that there are burrs that deviate from the benchmark contour by more than the set burr length threshold, it is determined that there is material occlusion causing visual failure. The actual location of the image corresponding to the image that does not deviate from the benchmark contour is recorded as a key tear detection location in the dual-dimensional monitoring. Then, tear detection is performed on the key tear detection location in the dual-dimensional monitoring through dual-dimensional monitoring.

5. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 4, characterized in that, The process of using fiber optic auscultation vibration monitoring to detect foreign objects and tears in the conveyor belt is as follows: The dual-dimensional monitoring method consists of vibration signal monitoring and temperature value monitoring. When material obstruction causes visual failure, the distributed fiber optic auscultation sensor first starts to operate. Based on the belt segment corresponding to the non-key foreign object monitoring location in the dual-dimensional monitoring, the sensor collects the roller rotation frequency and belt vibration frequency at the feed inlet of the belt segment corresponding to the non-key foreign object monitoring location in the dual-dimensional monitoring. It is then compared with the roller rotation frequency threshold and belt vibration frequency recorded when transporting materials without foreign objects in the previous period. If high-frequency pulse vibration is detected, and the roller rotation frequency and belt vibration frequency are not within the roller rotation frequency threshold range and belt vibration frequency range, it is determined to be an obstructed rod-shaped object. If low-frequency continuous vibration occurs, and the rotation frequency of the idler roller and the vibration frequency of the belt are both outside the threshold range of the idler roller rotation frequency and the belt vibration frequency range, and the amplitude is higher than the set percentage of the average signal, and the spectrum shows the characteristics of multi-frequency superposition, then it is determined to be a blocked mesh. If low-frequency strong vibration is captured, and the rotation frequency of the idler roller and the vibration frequency of the belt are both outside the threshold range of the idler roller rotation frequency and the belt vibration frequency range, and the amplitude is higher than the set percentage of the average signal, and the duration exceeds the set duration, then it is determined to be a blocked tear. Then, the monitoring of transported foreign objects and tear detection is carried out by temperature value monitoring.

6. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 5, characterized in that, The conveyor belt is monitored for foreign objects and tears using temperature monitoring methods. The specific process is as follows: Temperature monitoring is achieved by a thermal imaging camera installed on the idler roller bridge under the belt. The camera records the temperature of the obscured area in real time according to a preset acquisition time interval. The temperature is compared with the average belt surface temperature and average idler roller temperature obtained from the transportation test. When monitoring the key foreign object monitoring location in the dual-dimensional monitoring, the temperature at the corresponding location changes synchronously with the vibration pulse, showing an instantaneous rise and fall. If the temperature difference fluctuation meets the preset pulsed temperature difference threshold based on the average belt surface temperature and average idler roller temperature, that is, it matches the vibration characteristics of the rod-shaped object, then it is determined that there is a rod-shaped object obscured by material at the key tear monitoring location in the dual-dimensional monitoring. If the temperature increases in a stepwise manner, the cumulative temperature difference per unit time is within the preset stepwise temperature difference range based on the reference temperature, and the degree to which the overall temperature deviates from the average belt surface temperature and average idler roller temperature is consistent with the friction characteristics of the mesh, then it is determined that there is a rod-shaped object obscured by material at the key tear monitoring location in the dual-dimensional monitoring. If, when monitoring the key tear monitoring location in the dual-dimensional monitoring system, the temperature rises linearly, the high-temperature area forms a continuous strip shape with the belt running and there is no pulse fluctuation, and the temperature difference change range meets the preset tear characteristic temperature difference range based on the reference temperature, then it is determined that there is a tear that is blocked by material at the key tear monitoring location in the dual-dimensional monitoring system.

7. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 6, characterized in that, The specific process for comprehensively evaluating the condition of the belt is as follows: If the idler rotation frequency and belt vibration frequency are both outside the idler rotation frequency threshold range and belt vibration frequency range, and the corresponding temperature change differs from the average temperature in accordance with the temperature matching relationship, the abnormality type is confirmed. Then, the location of the abnormality on the belt is determined by the fiber optic sensor. If the idler rotation frequency and belt vibration frequency are both outside the idler rotation frequency threshold range and belt vibration frequency range, but the temperature does not differ from the average temperature, it is determined to be material unevenness interference, and no alarm is triggered. If the temperature deviates from the average temperature, but the vibration signal has no corresponding characteristics, it is determined to be the material's own temperature difference, and no alarm is triggered.

8. The condition monitoring method based on the conveyor belt of the cross-sea stockpiling line according to claim 7, characterized in that, The specific process for issuing an early warning of the risk of foreign objects evolving into tearing within a set time period is as follows: When warning of the risk of foreign object tearing within a set time period, the vibration and temperature characteristics of the identified foreign object are tracked and compared with the status monitoring benchmark characteristics of the transport test. The vibration sound signal of the foreign object captured by the fiber optic auscultation sensor is monitored in real time to determine whether the frequency deviation has further expanded from the idler roller rotation frequency threshold range and the belt vibration frequency range, whether the amplitude continues to increase, and whether the temperature rise rate has changed from pulsed and step-like to linear growth. If the above characteristics appear within the set time period, and the vibration spectrum begins to show the low-frequency strong vibration trend unique to tearing, it is determined that the foreign object has entered the evolution stage from jamming to increased friction and then to imminent tearing. An upgraded warning is immediately triggered to indicate that the belt is at risk of tearing.

9. The condition monitoring method based on a cross-sea stockpiling line conveyor belt according to claim 8, characterized in that, The specific process for assessing the cause of belt tearing is as follows: Based on the dual-dimensional monitoring data at the time of the tear, which includes vibration signal data and temperature value data, and by tracing back the historical records of the tear within a set time period, if there are vibration signals that match the characteristics of a foreign object in the historical records of the tear within the set time period, and the temperature anomaly at the corresponding location matches the characteristics of the foreign object, then it is determined that the tear was caused by foreign object obstruction. If there are no abnormal vibrations or temperatures related to foreign objects in the historical records of the tear within the set time period, and only the vibration and temperature characteristics unique to the tear are present, then the tear is determined to be caused by non-foreign object factors.

10. The condition monitoring method based on a cross-sea stockpiling line conveyor belt according to claim 9, characterized in that, The process for integrating visual, vibration, and temperature monitoring reports is as follows: Based on the baseline features and real-time identification results of visual monitoring, the abnormal spectrum and frequency deviation data of vibration monitoring, the abnormal temperature rise trend and baseline temperature difference information of temperature monitoring, combined with the risk warning of foreign object evolution and the assessment conclusion of tearing cause, the system eliminates false alarms of a single mode through cross-validation, and summarizes a comprehensive report containing the anomaly type, precise location, evolution stage and cause. Based on this, the system promotes the formulation of handling plans, effect tracking and dynamic updating of baseline data in the operation and maintenance process, and completes the closed-loop process from monitoring to resolution.

Images