Conveyor belt deviation detection method and system

Abstract

The invention discloses a conveyor belt deviation detection method and system, and relates to the technical field of conveyor detection.The method comprises the steps that a belt working video shot by a camera is obtained; detecting the number of carrier rollers in a preset area in the belt working video through an image recognition algorithm; if it is detected that the number of the carrier rollers is smaller than or equal to a preset number threshold value, early warning information is sent out; if it is detected that the number of the carrier rollers is larger than the preset number threshold value, the offset distance of the edge of the belt is calculated; if it is detected that the offset distance is larger than a preset distance threshold value and the offset duration is larger than a preset time threshold value, early warning information is sent out. According to the method, automatic detection and early warning of belt deviation are achieved through the camera video and image recognition technology, the detection efficiency is greatly improved, the workload of manual inspection is reduced, problems can be found more timely, and potential safety hazards are reduced.

Description

Technical Field

[0001] The present application relates to the field of conveyor detection technology, and in particular to a conveyor belt deviation detection method and system. Background Art

[0002] Long-term operation of the conveyor belt will cause the conveyor belt centerline to deviate from the centerline of the frame, affecting production operations. Deviation can easily lead to conveyor belt scratches and wear problems.

[0003] Cargo transportation is mainly completed by belt conveyors (belt conveyors for short). Traditional belt conveyor inspections are mainly carried out through manual inspections of important parts such as belt conveyor rollers, drums, motors, reducers, etc. in each process, and abnormal phenomena such as conveyor belt deviation and blockage are judged by visual inspection.

[0004] However, the manual inspection method is inefficient and has a certain lag, making it difficult for relevant staff to promptly detect the safety hazards caused by belt deviation. Summary of the invention

[0005] The present application provides a conveyor belt deviation detection method and system, which performs image recognition on the video of the belt working process, determines the degree of belt deviation by detecting the number of rollers at different times, and then issues a belt deviation warning.

[0006] In a first aspect, the present application provides a conveyor belt deviation detection method, which is applied to a deviation detection system, and the method comprises: Get a video of the belt operation captured by a camera installed just above the belt; Detect the number of rollers in a preset area in the belt working video through an image recognition algorithm; If it is detected that the number of rollers is less than or equal to the preset number threshold, an early warning message is issued; If the detected number of rollers is greater than the preset number threshold, the offset distance of the belt edge is calculated; If it is detected that the offset distance is greater than the preset distance threshold and the offset duration is greater than the preset time threshold, an early warning message is issued. The preset distance threshold is less than the offset distance of the belt when the roller is completely offset and covered by the belt. The offset duration starts from the time the belt offset is detected.

[0007] Through the above-mentioned embodiments, the deviation detection system detects the number of rollers in the belt working video through an image recognition algorithm, and determines whether the belt has deviated based on the comparison result with the preset threshold. When the number of rollers is less than or equal to the preset threshold, it means that the belt has offset and covered part of the rollers, and an early warning is directly issued. When the number of rollers is greater than the preset threshold, the offset distance of the belt edge is further calculated. If the offset distance and duration both exceed the preset threshold, an early warning is issued. This method uses camera video and image recognition technology to realize automatic detection and early warning of belt deviation, greatly improving detection efficiency, reducing the workload of manual inspections, and being able to discover problems more promptly and reduce safety hazards.

[0008] In some embodiments, after the step of calculating the offset distance of the belt edge if the number of idlers detected is greater than the preset number threshold, the method further includes: If the detected offset distance is less than or equal to the preset distance threshold, a warning mark is issued; If the deviation duration of the belt is detected to be less than or equal to the preset time threshold, a warning mark is issued; Send warning marks to the central platform client at preset time intervals.

[0009] Through the above-mentioned embodiments, when it is detected that the belt deviation distance is small or the duration is short, although it is not enough to trigger an early warning, it has indicated that there may be a problem with the belt operation status. At this time, the deviation detection system gives a warning mark and sends it to the central platform at a regular interval, which can prompt the operation and maintenance personnel to pay attention in advance and closely monitor the belt operation status, further improving the timeliness and predictability of deviation detection.

[0010] In some embodiments, if it is detected that the offset distance is greater than a preset distance threshold and the offset duration is greater than a preset time threshold, the step of issuing a warning message specifically includes: After detecting that the offset distance is greater than a preset distance threshold, determining the offset trend of the belt, the offset trend including an offset distance increasing trend and an offset distance decreasing trend; After the migration duration is longer than the preset time threshold, if the migration trend is an increasing trend of the migration distance, a warning message is issued; After the migration duration is longer than the preset time threshold, if the migration trend is a trend of decreasing migration distance, a warning mark is issued.

[0011] Through the above embodiments, when the deviation detection system detects that the belt deviation exceeds the distance and time thresholds and shows a trend of gradually increasing, it determines that the belt deviation is worsening and immediately issues an early warning. When the deviation exceeds the threshold but shows a trend of gradually decreasing, it is determined that the belt may be automatically resetting and the risk is relatively controllable. At this time, only an early warning mark is required, avoiding unnecessary early warnings, making early warning judgments more intelligent, and being able to identify serious deviations that are truly risky, reducing false alarms.

[0012] In some embodiments, the step of detecting the number of rollers in a preset area in the belt working video by an image recognition algorithm specifically includes: Extracting picture frames of the belt working video under various lighting environments at different time periods, the picture frames including picture frames when the belt is moving and picture frames when the belt is stationary; Use a preset data calibration tool to calibrate the rollers appearing in the picture frame to obtain a training data set; The deviation detection model is trained using the training data set to detect the number of rollers in the preset area in the belt working video.

[0013] Through the above embodiments, the deviation detection system collects video frame images of the belt in static and moving states under different environments, and marks the roller positions therein to produce a training data set for the deviation detection model. The diversity of the training set samples improves the environmental adaptability of the deviation detection model and the accuracy of roller identification, thereby improving the accuracy of roller quantity detection.

[0014] In some embodiments, the step of obtaining the belt working video captured by the camera specifically includes: The pressure data on both sides of the roller are obtained through the sensor and the pressure difference is calculated; When the belt offset position triggers a warning message, if the pressure difference detected exceeds the preset pressure threshold, a prompt message that the belt is unevenly stressed is added to the warning message.

[0015] Through the above embodiments, the deviation detection system obtains the force data on both sides of the belt by installing pressure sensors on both sides of the roller. When a deviation warning is detected, if the pressure data shows that the difference between the two sides is too large, it is determined that the cause of the belt deviation is uneven force on the belt, so corresponding prompt information is added to the warning information. This provides a direction for subsequent judgment of the cause of deviation and troubleshooting of belt faults.

[0016] In some embodiments, if it is detected that the number of rollers is greater than a preset number threshold, the step of calculating the offset distance of the belt edge specifically includes: Use edge detection algorithm to determine the edge position of the belt in the belt working video; Calculate the offset distance of the edge position from the image center line of the belt working video.

[0017] Through the above embodiments, the deviation detection system can use the edge detection algorithm to identify the edge position of the belt, and obtain the offset distance by calculating the offset from the edge to the reference center line, thereby providing a reliable basis for belt deviation judgment and warning.

[0018] In some embodiments, after the step of obtaining the belt working video captured by the camera, the method further includes: Obtaining inspection information of the belt conveyor through a variety of detection devices, the inspection information includes the sound waveform of the belt conveyor, temperature information and nearby pedestrian information; The inspection information and belt working video are sent to the central platform client for display.

[0019] Through the above embodiments, the deviation detection system can collect more comprehensive conveyor operation status data by installing multiple sensors such as audio sensors, temperature sensors, human infrared sensors, etc. on the belt conveyor, and upload these data together with the video data to the monitoring platform to achieve all-round monitoring of the conveyor.

[0020] In a second aspect, the present application provides a deviation detection system, the deviation detection system comprising: one or more processors and a memory; The memory is coupled to the one or more processors, and the memory is used to store computer program code, which includes computer instructions. The one or more processors call the computer instructions so that the deviation detection system can implement a conveyor belt deviation detection method provided in the above embodiment, which will not be repeated here.

[0021] In a third aspect, the present application provides a computer-readable storage medium, including instructions. When the instructions are executed on a deviation detection system, the deviation detection system can implement a conveyor belt deviation detection method provided in the above-mentioned embodiment, which will not be repeated here.

[0022] In a fourth aspect, the present application provides a computer program product. When the computer program product runs on a deviation detection system, the deviation detection system can implement a conveyor belt deviation detection method provided in the above-mentioned embodiment, which will not be repeated here.

[0023] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. Utilizing camera video and image recognition technology, the automatic detection and early warning of belt deviation is realized by detecting the number of rollers in the belt working video and calculating the belt edge offset distance. By setting the offset distance threshold and duration threshold, it is possible to detect and warn in time when the belt offset is small and the problem is not serious, while reducing false alarms. Compared with manual inspections, this method greatly improves detection efficiency and timeliness, and reduces the safety hazards of belt deviation.

[0024] 2. When a long-term belt deviation is detected, the belt deviation trend is further judged to determine whether to issue an early warning. When the deviation shows a trend of gradual increase, an early warning is issued immediately, and when the deviation shows a trend of decrease, only a mark is made, reducing unnecessary early warnings. By dynamically evaluating the deviation trend, the early warning judgment is made more intelligent, and the real risk of deviation is identified, reducing false alarms.

[0025] 3. Various sensors such as audio, temperature, and infrared are installed on the belt conveyor to collect data such as the sound, temperature, and surrounding personnel of the belt conveyor, and upload them to the monitoring platform together with the video data, thus realizing all-round monitoring of the conveyor. This multi-sensor data fusion can timely detect and locate various faults of the belt conveyor, such as abnormal operation, overheating, and illegal intrusion, greatly improving the operating safety of the conveyor. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a flow chart of a conveyor belt deviation detection method in an embodiment of the present application; Figure 2 This is another flow chart of a conveyor belt deviation detection method in an embodiment of the present application; Figure 3 is a schematic diagram of an exemplary structure of a roller device in an embodiment of the present application; Figure 4 It is a schematic diagram of the structure of a physical device of the deviation detection system in the embodiment of the present application. DETAILED DESCRIPTION

[0027] The terms used in the following embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to be used as limitations to the present application. As used in the specification and appended claims of the present application, the singular expressions "one", "a kind of", "said", "above", "the" and "this" are intended to also include plural expressions, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in the present application refers to any or all possible combinations comprising one or more listed items.

[0028] In the following, the terms "first" and "second" are used for descriptive purposes only and are not to be understood as suggesting or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features, and in the description of the embodiments of the present application, unless otherwise specified, "plurality" means two or more.

[0029] For ease of understanding, the following describes the process of the method provided by this implementation in combination with the above scenario. Figure 1 The figure is a flow chart of a conveyor belt deviation detection method in an embodiment of the present application.

[0030] S101, obtaining a belt operation video captured by a camera.

[0031] Specifically, this step is performed when it is necessary to detect the deviation of the belt conveyor. In the embodiment of the present application, the camera is installed directly above the center of the belt conveyor, and each camera is responsible for continuously shooting the running area of ​​different sections of the belt. When it is necessary to detect the deviation of the belt, the camera uploads the belt working video within the specified time period to the system platform for belt deviation detection.

[0032] In addition, the camera in the embodiment of the present application is a high-definition camera. Figure 3 As shown, in order to meet the recognition needs of the captured video, the installation of the high-definition camera must at least meet the following requirements: the camera should be kept vertical to the belt and directly above the center line of the belt; it should ensure that the entire belt is captured while taking up as much of the screen as possible, with less interference on the belt edge; the light brightness on the belt should be consistent at different positions to avoid the influence of environmental factors such as shadows; the camera should be at an appropriate height, and a pair of rollers on both sides of the belt should be exposed in the captured video. Of course, the installation conditions can also be adjusted according to actual needs, and there is no limitation here.

[0033] S102, detecting the number of rollers in a preset area in the belt working video through an image recognition algorithm.

[0034] Among them, the image recognition algorithm refers to the algorithm that uses computers to process and analyze images, extract interesting target features from them, and detect, classify and identify targets. Rollers refer to roller-shaped components used to support and carry conveyor belts (belts) in belt conveyors, usually arranged at equal intervals along the conveyor frame. The preset area refers to a specific area pre-defined in the belt working video screen, which needs to cover the full belt width and is used to limit the range of roller detection.

[0035] Specifically, the basic framework of the deviation detection model is first built on the deviation detection system based on the image recognition algorithm, and a large number of picture frames are obtained from the video taken by the camera. Then, the picture frames are annotated using a specific annotation tool to determine the rollers that appear in each picture frame, and the annotated picture frames are used to build a training set for the deviation detection model. After the system inputs the belt working video into the deviation detection model, the model marks the rollers that appear in the video and counts the number of rollers at different times.

[0036] S103: If it is detected that the number of rollers is less than or equal to a preset number threshold, a warning message is issued.

[0037] The preset number threshold represents a set critical value of the number of rollers. When the number of rollers actually detected is less than or equal to the threshold, it is determined that serious belt deviation has occurred and a warning needs to be issued. Warning information refers to information prompts used to indicate abnormal conditions in equipment or systems and to prompt relevant personnel to handle them in a timely manner. Its form includes but is not limited to text descriptions, indicator lights, sound alarms, etc., which are not limited here.

[0038] Specifically, this step is performed after the detection of the number of rollers is completed. Relevant technicians set a preset number threshold in the system based on the design parameters of the conveyor, such as belt width, roller diameter and spacing, that is, the number of rollers that should be observed in the picture when the belt is in normal working condition. When it is detected at a certain moment that the number of rollers appearing in the belt working video is less than or equal to the preset number threshold, it indicates that the belt has deviated from the normal position at this time, and the corresponding rollers are blocked by the belt, making it impossible for the camera to capture them, that is, it is determined that deviation has occurred. At this time, the system immediately issues a warning message to remind relevant personnel that the belt has deviated and needs to be stopped for maintenance in time.

[0039] S104: If it is detected that the number of rollers is greater than a preset number threshold, the offset distance of the belt edge is calculated.

[0040] The offset distance indicates the deviation of the belt edge position from the design reference position. The belt edge indicates the boundary line on both sides of the conveyor belt along the width direction. The offset distance refers to the vertical distance from the belt edge to the reference position. The unit can be pixels or actual length, which is not limited here.

[0041] Specifically, when the number of rollers is greater than the threshold, it means that the belt is still within the normal working range, but may have slightly deviated. At this time, it is necessary to quantify and calculate the actual offset distance to determine whether an offset warning is needed. First, determine the reference position for deviation judgment in the video image, that is, the midpoint of the line connecting the centers of the rollers. Then, the edge line of the belt is extracted from the image through the edge detection algorithm. Finally, calculate the deviation between the edge line and the reference position to obtain the offset distance value.

[0042] It should be noted that in some embodiments, the offset distance can be calculated in a variety of ways: optionally, if the camera has calibration parameters with known accuracy, the pixel coordinates of the reference position and the edge line can be converted into actual physical coordinates and then the difference can be made; optionally, the pixel distance can be directly used instead of the actual distance. It is understandable that with the development of machine vision technology, more advanced positioning and measurement methods can also be used to obtain offset distance information, which is not limited here.

[0043] S105: If it is detected that the offset distance is greater than a preset distance threshold and the offset duration is greater than a preset time threshold, a warning message is issued.

[0044] The preset distance threshold refers to a set offset distance critical value. When the actual measured offset distance exceeds the threshold and the duration exceeds the preset time threshold, the system determines that a belt deviation requiring an early warning has occurred. In addition, the preset time threshold refers to the critical value of the duration of the offset state, which is used to filter out instantaneous offsets caused by accidental factors.

[0045] This step combines the two factors of the belt's offset distance and duration to make a further judgment on the belt deviation. Specifically, when the system detects that the belt's offset distance at a certain moment exceeds the preset distance threshold, a timer is started to record the start time of the offset state. If the offset distance in subsequent image frames remains above the preset distance threshold, the timer reading is continuously updated until the offset distance drops below the threshold, and the total duration of the offset is calculated. The duration is then compared with the preset time threshold. If it is greater than the time threshold, it is determined to be a valid belt deviation event, and an early warning message is issued to prompt relevant personnel to handle it. If the duration does not exceed the threshold, it is determined to be a short-term offset caused by accidental factors such as belt vibration, and no early warning processing is performed.

[0046] In the above embodiment, the deviation detection system detects the number of rollers in the belt working video through an image recognition algorithm, and determines whether the belt has deviated based on the comparison result with the preset threshold. When the number of rollers is less than or equal to the preset threshold, it means that the belt has offset and covered part of the rollers, and an early warning is directly issued. When the number of rollers is greater than the preset threshold, the offset distance of the belt edge is further calculated. If both the offset distance and duration exceed the preset threshold, an early warning is issued. This method uses camera video and image recognition technology to realize automatic detection and early warning of belt deviation, greatly improving detection efficiency, reducing the workload of manual inspections, and being able to discover problems more promptly and reduce safety hazards.

[0047] After combining the above scenarios, the following further describes the method provided by this implementation in a more specific process. Figure 2 As shown, it is another flow chart of a conveyor belt deviation detection method in an embodiment of the present application.

[0048] S201, extracting picture frames of the belt working video under various lighting environments at different time periods.

[0049] The time period refers to the different time intervals in a 24-hour day, such as daytime, dusk, and night. The lighting environment refers to the lighting conditions that affect the camera's imaging effect, including the intensity, angle, color temperature, and other parameters of light sources such as natural light and artificial lighting. The picture frame refers to a single-frame still image captured from a video sequence, representing the picture content of the video at a certain moment.

[0050] Specifically, this step is performed when preparing training data for the deviation detection model. Since conveyors usually work continuously in open-air environments, it is necessary to consider the impact of light changes during the day and night on video quality. In order to enable the trained deviation detection model to adapt to different lighting conditions, the system collects belt working video samples from different time periods and extracts representative video frames as training data. For example, choose to record a video during the day, dusk, night and other time periods with large differences in lighting, and then capture a certain number of frame images from the video sequence at equal intervals to cover the different working states of the conveyor as much as possible, and include the picture difference characteristics caused by lighting changes.

[0051] S202: Use a preset data calibration tool to calibrate the rollers appearing in the picture frame to obtain a training data set.

[0052] It should be noted that the preset data calibration tool refers to a software tool used to mark the target of interest on the image, and the target object to be identified is framed and outlined in the image through manual operation or automatic algorithm, and the corresponding category label is affixed. These annotation information together with the original image constitute a training data set, which is used to supervise the training process of the learning model. In an embodiment of the present application, roller calibration refers to the process of marking the area contour belonging to the roller on the extracted video frame image and giving the roller category label.

[0053] This step is performed after obtaining representative video frame samples (frames of the belt working video). Since rollers are the core objects of deviation detection, it is necessary to construct a high-quality roller image dataset for training the deviation detection model. Specifically, the deviation detection system displays the video frame image through a data calibration tool, and polygons or rectangular frames are manually drawn on the image to select the area of ​​interest of each roller and mark its category with a unified label. The annotation results are repeatedly verified by multiple people to ensure that the annotation box fits the roller contour as accurately as possible, so that the label corresponds to the roller area one by one. The system then organizes and packages all calibrated images, annotation information, and category labels in a unified format to obtain a roller dataset for training the deviation detection model.

[0054] It should be noted that in some embodiments, the calibration of rollers can be achieved in a variety of ways: optionally, use general labeling tools such as LabelImg and LabelMe to manually draw a rectangular box to select each roller, then enter the category label, and export the labeling file in XML or JSON format; optionally, develop customized labeling software, design a contour point function for the circular feature of the roller, and fit the elliptical contour of the roller through a small number of key points to improve the labeling accuracy and efficiency. Other calibration methods can also be selected according to actual needs, which are not limited here.

[0055] S203: Use the training data set to train the deviation detection model to detect the number of rollers in a preset area in the belt working video.

[0056] It is understandable that the deviation detection model is a machine learning model used to detect roller targets from input images. The EfficientNet series models (such as EfficientNet-B0) can be used to build the basic framework of the deviation detection model. Its training process mainly uses the calibrated training data set as the supervision signal and optimizes the model parameters through repeated iterations so that it can learn to identify roller targets from images.

[0057] This step is performed after the system generates a roller detection data set. Specifically, the system builds the network structure of the model and initializes the model parameters. The calibrated images and roller annotation information are divided into training sets and validation sets according to a certain ratio, and sent to the model for training. By setting hyperparameters such as the number of training rounds and learning rate, the model is trained and optimized with the training set, and the model performance is evaluated with the validation set after each round of training, and the model performance is continuously iterated until the model performance is stable. After the training is completed, the detection accuracy of the model is tested with the validation set, and roller detection is performed on new belt working video frames to evaluate the generalization ability of the model. After the model training and optimization is completed, it can be used for actual online deviation detection tasks.

[0058] It should be noted that EfficientNet-B0 is the smallest version in the EfficientNet model series, which is a new neural network architecture proposed by Google researchers. The EfficientNet model series uses the Compound Scaling Method to balance the width, depth, and resolution of the network to achieve better efficiency and accuracy.

[0059] In the above embodiment, the deviation detection system collects video frame images of the belt in static and moving states under different environments, and marks the roller positions therein to produce a training data set for the deviation detection model. The diversity of the training set samples improves the environmental adaptability of the deviation detection model and the accuracy of roller identification, thereby improving the accuracy of roller quantity detection.

[0060] S204: The number of rollers is less than or equal to a preset number threshold.

[0061] The output detection results of the deviation detection model include the position coordinates and confidence of each roller identified in the image. The detection results with confidence higher than the set threshold are filtered out, and their number is counted, which is the number of rollers observed in the frame image. The system then compares the number with the preset number threshold. If it is less than or equal to the threshold, it is determined that serious deviation may have occurred at this time, and the belt has deviated to the point that the rollers are not completely visible, which is an abnormal situation. When the number of rollers in several consecutive frames of images is less than the threshold, step S212 will be entered, triggering the next step of early warning processing.

[0062] S205: Use an edge detection algorithm to determine the edge position of the belt in the belt working video.

[0063] Among them, the edge detection algorithm refers to a computer vision algorithm used to extract the edge of the target contour from the image, and finds the boundary line with obvious pixel value mutation by detecting the changes in the gradient, contrast and other characteristics of the local area of ​​the image. Commonly used edge detection algorithms include Canny, Sobel, Laplacian, etc., which are not limited here. In addition, in the embodiment of the present application, the edge position of the belt refers to the pixel coordinate position of the side edge line of the conveyor belt along the width direction projected on the image.

[0064] S206, calculating the offset distance between the edge position and the image center line of the belt working video.

[0065] It can be understood that the center line refers to the straight line that divides the image into two equal parts in the width direction, and its position coordinate is half of the image width. The offset distance refers to the pixel distance that the edge line of the belt deviates from the center line in the lateral direction, which is used to quantitatively evaluate the degree of belt offset.

[0066] The method for calculating the offset distance has been introduced in step S104 and will not be repeated here.

[0067] It should be noted that, in some embodiments, the calculation of the offset distance can be achieved in a variety of ways: optionally, for a belt parallel to the horizontal axis of the image, the difference between the horizontal coordinate of the edge line and the horizontal coordinate of the center line can be directly calculated to obtain the offset distance; for an inclined belt, it is necessary to calculate the hypotenuse projection length through trigonometric functions, etc. For different belt slopes, camera viewing angles and other working conditions, the calculation of the offset distance needs to be adjusted specifically, and the method given here is not limited to the above examples.

[0068] S207: The offset distance is greater than a preset distance threshold.

[0069] This step is executed after the system obtains the quantitative value of the belt offset distance, and is used to determine whether the belt deviation degree reaches the warning level. Specifically, the system sets the offset distance threshold and divides the deviation into slight, medium, severe and other levels. When the belt offset distance exceeds the preset medium deviation threshold, it means that there are certain safety hazards in the belt operation state at this time, which requires attention, and a low warning level warning mark can be made. When the offset distance further exceeds the severe deviation threshold, the belt is likely to cause lateral cutting, breakage and other accidents. At this time, it is necessary to further detect whether the offset duration is too long, that is, enter step S208.

[0070] S208: The offset duration is greater than a preset time threshold.

[0071] The deviation duration refers to the duration of time that the belt deviation distance continuously exceeds the preset threshold, which is used to evaluate the stability and severity of the deviation. The preset time threshold refers to the maximum time allowed for instantaneous deviation, which is pre-set according to the design requirements and material properties of the conveyor. If this time is exceeded, it is considered that the deviation will cause damage to the equipment. The threshold is usually determined by experienced technicians based on previous data analysis, such as continuous deviation of 30 seconds will cause belt wear, etc., which is not limited here.

[0072] It is understandable that belt deviation is divided into two situations: accidental instantaneous deviation and continuous long-term deviation, which have different degrees of impact on the equipment. Instantaneous deviation is often caused by factors such as sudden accumulation of materials and short-term jamming. It is random and unpredictable, but it lasts for a short time and causes less wear on the belt. Long-term deviation is usually caused by factors such as misalignment of belt joints and imbalance of rollers. Once formed, it is difficult to correct automatically. The lateral shear stress on the belt acts for a longer time and causes more serious wear. Therefore, when the system detects that the belt has a large deviation in step S207 (that is, the deviation distance is greater than the severe deviation threshold), it will further record the duration of the belt being in a large deviation state (that is, the deviation duration). If it is detected that the deviation duration is greater than the preset time threshold, it is determined that the deviation has a large impact, and step S209 is entered to detect the belt deviation trend; conversely, if it is detected that the deviation duration is less than or equal to the preset time threshold, it is determined to be a short-term deviation, and step S213 is entered to only perform a warning mark.

[0073] In the above embodiment, when it is detected that the belt deviation distance is small or the duration is short, although it is not enough to trigger an early warning, it has indicated that there may be a problem with the belt operation status. At this time, the deviation detection system will give a warning mark and send it to the central platform at a regular intervals, which can prompt the operation and maintenance personnel to pay attention in advance and closely monitor the belt operation status, further improving the timeliness and predictability of deviation detection.

[0074] S209: Whether the offset distance has an increasing trend.

[0075] The offset distance refers to the lateral offset pixel distance between the belt edge line and the image center line calculated in step S206, which is used to quantitatively evaluate the degree of belt offset. The increasing trend refers to the trend of continuous increase in the offset distance in multiple consecutive frames of images, reflecting that the belt offset is increasing and there is a risk of loss of control.

[0076] Specifically, this step is performed after obtaining the offset distance of the belt in multiple consecutive frames, and is used to predict the development trend of the deviation. It is understandable that since the deviation has a certain inertia, the current offset state will often continue for a period of time and show a gradual characteristic. If the offset distance of the belt continues to increase in several consecutive frames, it means that the degree of deviation is constantly increasing, and it is very likely to exceed the safety threshold in the future. Therefore, after the system detects that the belt is in a trend of increasing offset distance, it can enter step S210 for further judgment, or it can directly issue an early warning message. In addition, if the offset distance fluctuates randomly within a small range, it can be regarded as the vibration of the belt itself, and there is no need to pay too much attention. Therefore, after the system detects that the belt is in a trend of decreasing offset distance, it can enter step S213 for early warning marking to reduce the occurrence of false alarms.

[0077] In the above embodiment, when the deviation detection system detects that the belt deviation exceeds the distance and time thresholds and shows a trend of gradually increasing, it is determined that the belt deviation is worsening and an early warning is issued immediately. When the deviation exceeds the threshold but shows a trend of gradually decreasing, it is determined that the belt may be automatically resetting and the risk is relatively controllable. At this time, only an early warning mark is required, avoiding unnecessary early warnings, making the early warning judgment more intelligent, and being able to identify serious deviations that are really risky, reducing false alarms.

[0078] S210, obtaining pressure data at both sides of the roller through a sensor and calculating a pressure difference.

[0079] Among them, the sensor refers to the pressure sensor installed on both sides of the roller bearing seat, which is used to collect the belt pressure borne by the roller in real time and reflect the force distribution condition of the belt.

[0080] This step is performed after the risk of belt deviation is detected (i.e. the belt is in a trend of increasing deviation distance) and is an optional step used to assist in assessing the severity of the deviation.

[0081] It is understandable that a normally operating belt is evenly stressed in the width direction, and the pressure on both sides is basically symmetrical and balanced. When the belt deviates, it deviates to one side, causing the force on the roller on that side to increase significantly, the pressure sensor value to increase, and the force on the other side to decrease accordingly. The greater the pressure difference on both sides, the more serious the belt deviation is, and the greater the wear on the roller and the belt itself. Therefore, the introduction of pressure detection can double confirm the degree of deviation, and cross-verify the information in two dimensions, image and pressure, to reduce the false alarm rate of deviation judgment.

[0082] Specific as Figure 3 The figure is a schematic diagram of an exemplary structure of the roller device in the embodiment of the present application. When the belt is working normally, it is directly above the roller, wherein the left and right rollers are in contact with the edge of the belt, and the belt shields part of the left and right rollers respectively. When the belt is working normally, the left and right rollers evenly bear the pressure applied downward by the belt. When the belt deviates, the contact position between the left and right rollers and the edge of the belt changes, so that the shielding area of ​​the left and right rollers by the belt changes respectively, and at the same time, the pressure applied downward by the belt to the left and right rollers changes, and the pressure on the rollers in the direction of belt deviation increases, while the pressure on the rollers in the opposite direction of belt deviation decreases.

[0083] S211. If it is detected that the pressure difference exceeds a preset pressure threshold, a prompt message is added to the warning message.

[0084] The pressure difference refers to the percentage difference between the pressure sensor data on both sides of the roller calculated in step S210, which is used to quantitatively evaluate the degree of deviation of the belt force distribution. The preset pressure threshold refers to the upper limit of the allowable pressure deviation that is preset according to the material properties of the roller and the belt and the design safety factor. It is usually calculated by experienced engineers. If the pressure difference exceeds 30%, it may cause accelerated wear of the bearing, etc., which is not limited here.

[0085] This step is a follow-up to pressure detection and provides a supplement to image detection in terms of early warning. Since the response of pressure sensors is generally faster than image processing, when the pressure difference reaches a dangerous level, it is likely that the offset distance has not exceeded the preset threshold. Therefore, relying solely on image warnings may result in late warnings, leaving insufficient time for on-site operators to react. Adding abnormal pressure prompts to the early warning information can draw attention earlier and provide more time for the formulation of emergency plans. In addition, in actual applications, this step is equivalent to realizing the fusion judgment of image and pressure information, which can effectively improve the real-time and accuracy of deviation warnings.

[0086] S212. Issue a warning message.

[0087] When the system determines that the belt is seriously deviated (the determination conditions are shown in the above steps S204 to S209), the system implements an early warning mechanism to send the position information of the belt deviation and the corresponding picture frame data to the client monitoring platform to realize the early warning function.

[0088] In addition, the system can also obtain inspection information of the belt conveyor through a variety of detection devices, such as installing a bone conduction industrial stethoscope on the belt conveyor rollers, drums, motors, and reducer boxes to detect abnormal noise information in key parts of the belt conveyor through learning sound waveforms; installing a thermal imaging camera monitoring device on the drive transmission part of the belt conveyor to check the temperature of the drive transmission part; deploying high-definition smart cameras with intelligent algorithms for personnel area intrusion detection along the belt conveyor to detect whether personnel have intruded, etc. The system integrates the above inspection information, belt working videos, and early warning information into the client monitoring platform to facilitate the processing, analysis, storage, display, query, linkage and other functions of data information.

[0089] S213. Make a warning mark.

[0090] When the system determines that the belt deviation is not serious (the determination conditions are shown in the above steps S204 to S209), it will mark the time, location and duration of the abnormality. And send it to the client monitoring platform for filing at a preset time interval.

[0091] The deviation detection system of the embodiment of the present invention is an electronic device. Figure 4 A schematic diagram of the architecture of an electronic device suitable for implementing an embodiment of the present invention is shown.

[0092] It should be noted that Figure 4 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present invention.

[0093] Those skilled in the art will appreciate that all or part of the steps in the various methods of the above embodiments can be completed by instructions (computer programs), or by controlling related hardware through instructions (computer programs), and the instructions can be stored in a computer-readable storage medium and loaded and executed by a processor. The electronic device of this embodiment includes a storage medium and a processor, wherein a plurality of instructions are stored in the storage medium, and the instructions can be loaded by the processor to execute any step of the method provided in the embodiment of the present invention.

[0094] Specifically, the storage medium and the processor are electrically connected directly or indirectly to realize data transmission or interaction. For example, these elements can be electrically connected to each other through one or more signal lines. The storage medium stores computer execution instructions for implementing the data access control method, including at least one software function module that can be stored in the storage medium in the form of software or firmware. The processor executes various functional applications and data processing by running the software program and module stored in the storage medium. The storage medium can be, but is not limited to, random access storage medium (Random Access Memory, referred to as: RAM), read-only storage medium (Read Only Memory, referred to as: ROM), programmable read-only storage medium (Programmable Read-Only Memory, referred to as: PROM), erasable read-only storage medium (Erasable Programmable Read-Only Memory, referred to as: EPROM), electrically erasable read-only storage medium (Electric Erasable Programmable Read-Only Memory, referred to as: EEPROM), etc. Among them, the storage medium is used to store programs, and the processor executes the program after receiving the execution instruction.

[0095] Furthermore, the software programs and modules in the above-mentioned storage medium may also include an operating system, which may include various software components and / or drivers for managing system tasks (such as memory management, storage device control, power management, etc.), and may communicate with various hardware or software components to provide an operating environment for other software components. The processor may be an integrated circuit chip having signal processing capabilities. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc., which may implement or execute the various methods, steps, and logic flow diagrams disclosed in this embodiment. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0096] Since the instructions stored in the storage medium can execute the steps in any method provided in the embodiments of the present invention, the beneficial effects of any method provided in the embodiments of the present invention can be achieved. Please refer to the previous embodiments for details and will not be repeated here.

[0097] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A conveyor belt deviation detection method, applied to a deviation detection system, characterized in that: The method comprises: Obtaining a belt working video captured by a camera, wherein the camera is installed directly above the belt; Detecting the number of rollers in a preset area in the belt working video by an image recognition algorithm; If it is detected that the number of the rollers is less than or equal to a preset number threshold, a warning message is issued; If it is detected that the number of rollers is greater than the preset number threshold, the offset distance of the belt edge is calculated; If it is detected that the offset distance is greater than the preset distance threshold and the offset duration is greater than the preset time threshold, an early warning message is issued. The preset distance threshold is less than the offset distance of the belt when the roller is completely offset and covered by the belt, and the offset duration starts from the time when the belt offset is detected.

2. The method according to claim 1, characterized in that After the step of calculating the offset distance of the belt edge if it is detected that the number of rollers is greater than the preset number threshold, the method further includes: If it is detected that the offset distance is less than or equal to a preset distance threshold, a warning mark is issued; If the deviation duration of the belt is detected to be less than or equal to the preset time threshold, a warning mark is issued; The warning mark is sent to the central platform client at preset time intervals.

3. The method according to claim 1, characterized in that The step of issuing a warning message if it is detected that the offset distance is greater than a preset distance threshold and the offset duration is greater than a preset time threshold specifically includes: After detecting that the offset distance is greater than a preset distance threshold, determining the offset trend of the belt, wherein the offset trend includes an offset distance increasing trend and an offset distance decreasing trend; After the offset duration is longer than a preset time threshold, if the offset trend is an increasing offset distance trend, issuing a warning message; After the offset duration is greater than a preset time threshold, if the offset trend is a trend of decreasing offset distance, a warning mark is performed.

4. The method according to claim 1, characterized in that: The step of detecting the number of rollers in a preset area in the belt working video by using an image recognition algorithm specifically includes: Extracting picture frames of the belt working video under various lighting environments at different time periods, wherein the picture frames include picture frames when the belt is moving and picture frames when the belt is stationary; Using a preset data calibration tool to calibrate the rollers appearing in the picture frame to obtain a training data set; The deviation detection model is trained using the training data set to detect the number of rollers in a preset area in the belt working video.

5. The method according to claim 1, characterized in that The step of obtaining the belt working video captured by the camera specifically includes: The pressure data on both sides of the roller are obtained through the sensor and the pressure difference is calculated; When the offset position of the belt triggers a warning message, if it is detected that the pressure difference exceeds a preset pressure threshold, a prompt message that the belt is unevenly stressed is added to the warning message.

6. The method according to claim 1, characterized in that If it is detected that the number of rollers is greater than the preset number threshold, the step of calculating the offset distance of the belt edge specifically includes: Using an edge detection algorithm to determine the edge position of the belt in the belt working video; The offset distance between the edge position and the image center line of the belt working video is calculated.

7. The method according to claim 1, characterized in that After the step of obtaining the belt working video shot by the camera, the method further includes: Obtaining inspection information of the belt conveyor through a variety of detection devices, the inspection information including the sound waveform of the belt conveyor, temperature information and nearby pedestrian information; The inspection information and belt working video are sent to the central platform client for display.

8. A deviation detection system, characterized in that: The deviation detection system includes: one or more processors and a memory; The memory is coupled to the one or more processors, and the memory is used to store computer program codes, wherein the computer program codes include computer instructions, and the one or more processors call the computer instructions to enable the deviation detection system to execute the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that: When the instruction is executed on the deviation detection system, the deviation detection system is caused to execute the method as described in any one of claims 1 to 7.

10. A computer program product, characterized in that When the computer program product runs on a deviation detection system, the deviation detection system is enabled to execute the method according to any one of claims 1 to 7.

Images