Coal conveyor belt deviation detection method, system and device based on YOLOv8 algorithm

Abstract

The present application discloses a method, system and device for detecting deviation of a coal conveyor belt based on a YOLOv8 algorithm. The method includes: real-time acquisition of an image of a coal conveyor belt and preprocessing; edge detection of a coal conveyor belt in an image of a coal conveyor belt using a YOLOv8 algorithm; performing deviation calculation according to a first deviation calculation step in combination with the edge coordinates of the coal conveyor belt to obtain a first deviation parameter; real-time acquisition of operating data and environmental data of the coal conveyor belt; using a first deviation parameter adaptive threshold adjustment mechanism, obtaining a first deviation parameter threshold adjustment value corresponding to the operating data and environmental data acquired in real time and adjusting a preset threshold of the first deviation parameter, determining whether the calculated first deviation parameter exceeds the preset threshold of the first deviation parameter after the threshold adjustment, so as to obtain a first result of deviation in real time and generate a corresponding prompt information of deviation of the coal conveyor belt. The present application can complete more accurate deviation detection of a coal conveyor belt.

Description

Technical Field

[0001] The present application relates to the technical field of deviation detection of coal conveyor belts, and specifically to a method, system and device for detecting deviation of coal conveyor belts based on the YOLOv8 algorithm. Background Art

[0002] Conveyor belts are an important part of modern mine automation production and are widely used in underground belt transportation systems in coal mines, material transmission in open-pit coal mines, and status monitoring of material conveyor lines in coal preparation plants. Traditional conveyor belt monitoring technology mainly relies on manual inspections and fixed sensors. Although manual inspections can find some obvious faults, they are inefficient and greatly affected by subjective factors, making it difficult to meet the needs of continuous operation in modern mines. Fixed sensors, due to their fixed installation positions, are difficult to fully cover complex working environments, especially in poor lighting conditions, where detection accuracy is limited, leading to frequent false alarms.

[0003] Existing solutions usually use infrared sensors or ultrasonic ranging to monitor the deviation of conveyor belts. These methods determine whether the conveyor belt deviates from the normal track by setting a fixed threshold. Infrared sensors use infrared beams to detect the distance of objects, while ultrasonic ranging determines the distance by emitting ultrasonic waves and measuring the time difference of echoes. Both methods can achieve preliminary detection of conveyor belt deviation to a certain extent. In addition, there are technologies that combine image detection technology to detect the deviation of coal conveyor belts, but the existing image detection technology has weak anti-interference ability and is easily affected by dust, moisture and lighting environmental factors in the environment, resulting in frequent false alarms. Moreover, because they are based on fixed threshold judgments, they cannot adapt to slight deviations caused by real-time working conditions such as changes in coal load on the conveyor belt, thereby reducing the reliability and applicability of the system. Therefore, a more accurate and reliable deviation detection method is urgently needed to solve these problems. Summary of the invention

[0004] In order to achieve more accurate deviation detection using image detection methods while considering the real-time working conditions and environmental factors of the coal conveyor belt, the present application provides a coal conveyor belt deviation detection method, system and device based on the YOLOv8 algorithm.

[0005] In a first aspect, the present application provides a method for detecting deviation of a coal conveyor belt based on a YOLOv8 algorithm, comprising:

[0006] Real-time acquisition and pre-processing of coal conveyor belt images;

[0007] The YOLOv8 algorithm is used to detect the edge of the coal conveyor belt in the coal conveyor belt image; the deviation calculation is performed according to the first deviation calculation step in combination with the edge coordinates of the coal conveyor belt obtained, and the first deviation parameter is obtained; the first deviation calculation step includes: fitting the edge line trajectories for the left and right edge coordinates of the coal conveyor belt obtained by detection, calculating the center line trajectory using the fitted left and right edge line trajectories and comparing it with the preset center line trajectory of the coal conveyor belt, and calculating the first deviation parameter, including the offset and the inclination angle;

[0008] Collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time; use the first deviation parameter adaptive threshold adjustment mechanism to obtain the first deviation parameter threshold adjustment value corresponding to the working condition data and environmental data collected in real time; the first deviation parameter adaptive threshold adjustment mechanism selects the first deviation parameter adaptive threshold adjustment function or the trained first deviation parameter adaptive threshold adjustment model fitted by the historical first deviation parameter threshold adjustment value corresponding to the historical first deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual first deviation parameter greater than a preset similarity under different historical working condition data and environmental data conditions;

[0009] The first deviation parameter threshold adjustment value obtained is used to adjust the first deviation parameter preset threshold, and it is determined whether the calculated first deviation parameter exceeds the first deviation parameter preset threshold after the threshold adjustment, so as to obtain the first result of whether the deviation occurs in real time and generate the coal conveyor belt deviation prompt information accordingly.

[0010] By adopting the above scheme, the YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image, and the left and right edge coordinates of the conveyor belt are accurately identified. In combination with the first deviation calculation step, the first deviation parameter is accurately obtained. At the same time, the operating condition data and environmental data are collected in real time, and the deviation parameter judgment threshold is dynamically adjusted through the adaptive threshold adjustment mechanism to avoid detection deviations caused by operating conditions and environmental factors, thereby improving the detection accuracy and enhancing the adaptability and reliability.

[0011] Preferably, it also includes:

[0012] The position range of the coal conveyor belt covered by the camera device is obtained in advance and the area is divided to obtain the straight area of ​​the coal conveyor belt and the curved area of ​​the coal conveyor belt; the preset length is matched corresponding to the divided area type of the coal conveyor belt, and the left and right edge coordinates of the coal conveyor belt are segmented using the matched preset length to obtain a plurality of segmented sets of the left and right edge coordinates of the coal conveyor belt;

[0013] According to each segmented set of left and right edge coordinates of the coal conveyor belt, line trajectories are fitted for the left and right edge coordinates of the coal conveyor belt obtained by detection, and the center line trajectory is calculated by segment using the fitted left and right edge line trajectories and compared with the preset center line trajectory of the coal conveyor belt, so as to calculate and obtain the first deviation parameters corresponding to several segments;

[0014] Obtain the working condition data and environmental data in the position range of the coal conveyor belt corresponding to the left and right edge line trajectories of the coal conveyor belt in sections, obtain the first deviation parameter threshold adjustment value accordingly, and generate the deviation parameter preset threshold value after the threshold adjustment in sections;

[0015] It is judged in segments whether the first deviation parameter calculated in real time exceeds the preset threshold value of the first deviation parameter after the threshold value of the corresponding segment is adjusted, and a coal conveyor belt deviation prompt information is generated for the first deviation parameter that exceeds the preset threshold value of the first deviation parameter after the threshold value is adjusted.

[0016] By adopting the above solution, different areas of the coal conveyor belt are processed in sections, and combined with the actual working conditions and environmental data of each section, more accurate deviation parameter calculation and threshold adjustment are achieved, thereby effectively reducing false alarms and missed alarms.

[0017] Preferably, it also includes:

[0018] The YOLOv8 algorithm is used to perform edge detection on the coal carried on the coal conveyor belt in the coal conveyor belt image to obtain the edge coordinates and the second confidence of the coal; the deviation calculation is performed according to the second deviation calculation step in combination with the obtained edge coordinates of the coal to obtain the second deviation parameter; the second deviation calculation step includes: performing contour fitting on the edge coordinates of the coal obtained by the detection, calculating the centroid position of the coal area, and comparing it with the centroid position of the coal carried by the preset coal conveyor belt, and calculating the second deviation parameter, including: offset;

[0019] Collect basic information data of coal carried on the coal conveyor belt in real time, including: type and weight; use the second deviation parameter adaptive threshold adjustment mechanism to obtain the second deviation parameter threshold adjustment value corresponding to the real-time collected working condition data, environmental data and basic information data of coal; the second deviation parameter adaptive threshold adjustment mechanism selects a second deviation parameter adaptive threshold adjustment function or a trained second deviation parameter adaptive threshold adjustment model that is fitted with a historical second deviation parameter threshold adjustment value corresponding to a historical second deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual second deviation parameter greater than a preset similarity under different historical working condition data, environmental data and basic information data of coal;

[0020] The second deviation parameter preset threshold is adjusted by obtaining the second deviation parameter threshold adjustment value, and it is determined whether the calculated second deviation parameter exceeds the second deviation parameter preset threshold after the threshold adjustment, so as to obtain a second result of whether the deviation occurs in real time;

[0021] A first confidence is obtained when the YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image; a comprehensive confidence is calculated according to the weighted first confidence and the second confidence, and a confidence threshold is set; the comprehensive confidence is compared with the confidence threshold, and if the comprehensive confidence is greater than the confidence threshold, and any of the first result and the second result is that there is deviation, then a corresponding coal conveyor belt deviation prompt information is generated instead of directly generating the coal conveyor belt deviation prompt information according to the first result.

[0022] By adopting the above scheme, the YOLOv8 algorithm is used to perform edge detection on the coal carried on the coal conveyor belt in the coal conveyor belt image, and the second deviation parameter is accurately obtained in combination with the second deviation calculation step; at the same time, the collected working condition data, environmental data and coal basic data are combined, and the preset threshold of the deviation parameter is dynamically adjusted through the adaptive threshold adjustment mechanism, so as to accurately obtain the deviation detection result; and from the perspective of multiple target detection such as the coal conveyor belt itself and the coal carried on the coal conveyor belt, a comprehensive confidence level is set and a collaborative analysis is performed to comprehensively judge the accuracy of the deviation detection result, thereby avoiding the uncertainty brought by a single detection result and improving the accuracy of the alarm.

[0023] Preferably, the weight ratio corresponding to the first confidence level and the second confidence level in the weighted calculation of the comprehensive confidence level according to the first confidence level and the second confidence level is dynamically adjusted according to the working condition data, the environmental data and the basic information data of the coal, and the weight ratio corresponding to the first confidence level and the second confidence level is obtained including:

[0024] Constructing and training a first neural network model, wherein the first neural network model is generated by training a data set consisting of a weight ratio corresponding to a first confidence level and a second confidence level when the accuracy of generating coal conveyor belt deviation prompt information is greater than a preset accuracy rate using historical different operating condition data, environmental data, basic information of coal, and the first neural network model, wherein the input of the first neural network model is real-time operating condition data, environmental data and basic information of coal, and the output is the weight ratio corresponding to the first confidence level and the second confidence level;

[0025] The currently collected real-time operating data, environmental data and basic information of coal are input into the first neural network model to obtain the weight ratio corresponding to the first confidence level and the second confidence level.

[0026] By adopting the above scheme, the weight ratio of the two confidence levels is automatically adjusted according to the real-time collected operating data, environmental data and basic information of coal, so as to more accurately reflect the reliability of the current detection results, realize the dynamic adjustment of the first confidence level and the second confidence level, and improve the adaptability of the system to different operating conditions and environmental conditions.

[0027] Preferably, it also includes:

[0028] According to the probability of the actual deviation under the corresponding conditions of different historical working condition data and environmental data, the corresponding matching continuous judgment time is preset for the working condition data and environmental data combinations within different preset ranges of the deviation probability; wherein, the larger the preset range of the occurrence probability is, the shorter the preset matching continuous judgment time is;

[0029] The continuous judgment duration is obtained by correspondingly matching the working condition data and the environmental data obtained in real time; according to the obtained continuous judgment duration, the YOLOv8 algorithm is continuously used to detect the edge of the coal conveyor belt in each frame image of the coal conveyor belt and the first deviation parameter is calculated accordingly;

[0030] According to the time sequence of the image frames, if the first deviation parameter obtained and calculated exceeds the first result of the deviation parameter preset threshold after the threshold adjustment within the continuous judgment time, a corresponding coal conveyor belt deviation prompt information is generated, replacing the original first result of obtaining whether the deviation is obtained in real time and generating the corresponding coal conveyor belt deviation prompt information.

[0031] By adopting the above scheme, taking into account the possible real-time detection errors that may lead to false alarms of deviation, and combining the actual probability of deviation under different working conditions and environments, the continuous judgment time is flexibly adjusted, and the real-time reporting results are replaced by the continuous detection results within the continuous judgment time to finally determine whether the coal conveyor belt has deviated.

[0032] Preferably, it also includes:

[0033] The YOLOv8 algorithm is used to detect the torn area of ​​the coal conveyor belt in the coal conveyor belt image, obtain the edge coordinates of the torn area of ​​the coal conveyor belt, and calculate the area of ​​the torn area of ​​the coal conveyor belt;

[0034] Continuously obtain the area of ​​the tearing area; input the area of ​​the tearing area, the operating condition data and the environmental data obtained at the current moment and the previous preset time period into the second neural network, and output whether there is a deviation result of the coal conveyor belt at the current moment; the second neural network model is trained and generated by marking the tearing area area of ​​the continuous time period with the deviation result of the coal conveyor belt under different historical operating condition data and environmental data conditions;

[0035] While obtaining the result of whether the coal conveyor belt has deviated, verify whether the first result of obtaining whether the deviation exists in real time corresponds to the generated coal conveyor belt deviation prompt information; if not, it is determined that there is an error in the first result judgment, and the YOLOv8 algorithm is optimized and the optimized YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image at the next moment.

[0036] By adopting the above scheme, the YOLOv8 algorithm is further used to detect the torn area of ​​the coal conveyor belt in the coal conveyor belt image, and the area change of the torn area is accurately identified and calculated to determine whether the current coal conveyor belt is offset. The deviation detection results obtained in real time are automatically verified by the tearing condition of the coal conveyor belt, and the YOLOv8 algorithm is further optimized through the verification results to improve the accuracy of coal conveyor belt deviation detection.

[0037] Preferably, it also includes:

[0038] The edge computing method is collected and different edge computing nodes are used to realize distributed computing of edge detection and deviation calculation of the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm and edge detection and deviation calculation of the coal carried on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm.

[0039] By adopting the above solution, different edge computing nodes are used to perform distributed computing on coal conveyor belt detection based on different detection target objects, thereby improving the detection computing capability and response speed.

[0040] In the second aspect, the present application provides a coal conveyor belt deviation detection system based on the YOLOv8 algorithm, comprising:

[0041] The coal conveyor belt image acquisition and processing module is used to use a camera device to collect images of the coal conveyor belt in real time and perform preprocessing;

[0042] A first deviation parameter acquisition module is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm; perform deviation calculation according to the first deviation calculation step in combination with the acquired edge coordinates of the coal conveyor belt to obtain a first deviation parameter; the first deviation calculation step includes: performing edge line trajectory fitting for the left and right edge coordinates of the coal conveyor belt obtained by the detection, calculating the center line trajectory using the fitted left and right edge line trajectories and comparing it with the preset center line trajectory of the coal conveyor belt, and calculating the first deviation parameter, including an offset and an inclination angle;

[0043] The coal conveyor belt working condition and environment collection module is used to collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time;

[0044] The first deviation parameter adjustment module is used to obtain a first deviation parameter threshold adjustment value corresponding to the working condition data and environmental data collected in real time by using a first deviation parameter adaptive threshold adjustment mechanism; the first deviation parameter adaptive threshold adjustment mechanism selects a first deviation parameter adaptive threshold adjustment function or a trained first deviation parameter adaptive threshold adjustment model that is fitted by a historical first deviation parameter threshold adjustment value corresponding to a historical first deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual first deviation parameter greater than a preset similarity under conditions of different historical working condition data and environmental data; and adjusts the preset threshold of the first deviation parameter by using the obtained first deviation parameter threshold adjustment value;

[0045] The coal conveyor belt deviation result acquisition module is used to determine whether the calculated first deviation parameter exceeds the preset threshold of the first deviation parameter after the threshold adjustment, so as to obtain the first result of deviation and generate corresponding coal conveyor belt deviation prompt information.

[0046] By adopting the above scheme and combining it with the YOLOv8 algorithm for target edge detection, the deviation parameters of the coal conveyor belt can be accurately determined. The real-time working condition and environmental data of the coal conveyor belt are collected, and the adaptive deviation parameter adaptive threshold adjustment mechanism is combined to realize the adaptive deviation parameter threshold adjustment, so as to further accurately obtain the detection results of whether the coal conveyor belt has deviated, and issue early warning in time.

[0047] In a third aspect, the present application provides a coal conveyor belt device, which includes an image acquisition device and an image analysis device. The image analysis device includes a memory, a processor, and a program stored and executable on the memory. When the program is executed by the processor, the steps of the above method are implemented.

[0048] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute the method as described above.

[0049] In summary, this application has the following beneficial effects:

[0050] 1. Based on the YOLOv8 algorithm, the edge position and deviation parameters of the coal conveyor belt can be accurately identified; the adaptive threshold adjustment mechanism can dynamically adjust the threshold of the deviation parameter according to the real-time collected working condition data and environmental data; by comparing the accurately calculated deviation parameters with the deviation parameter threshold set by dynamic adaptive setting, the deviation detection results of the coal conveyor belt can be obtained, the deviation detection accuracy of the coal conveyor belt can be improved, and the false alarm rate can be reduced;

[0051] 2. By segmenting different areas of the coal conveyor belt and combining the actual working conditions and environmental data of each segment, more accurate deviation parameter calculation and threshold adjustment are achieved, thereby improving detection accuracy and effectively reducing false alarms and missed alarms;

[0052] 3. In addition to the detection of the coal conveyor belt itself, the coal conveyor belt deviation detection is carried out from the angle of the coal carried on the coal conveyor belt, and the deviation parameters corresponding to the multi-angle detection are obtained to provide a data basis for collaborative analysis; the comprehensive confidence is introduced, and the calculation of the comprehensive confidence is combined with the deviation results judged by different deviation parameters to comprehensively judge the coal conveyor belt deviation detection results, thereby improving the accuracy of the coal conveyor belt deviation detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] Figure 1 It is a flow chart of the coal conveyor belt deviation detection method based on the YOLOv8 algorithm described in the specific embodiment;

[0054] Figure 2 It is a structural schematic diagram of the coal conveyor belt deviation detection system based on the YOLOv8 algorithm described in the specific embodiment. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0056] like Figure 1 As shown, the embodiment of the present application discloses a method for detecting deviation of a coal conveyor belt based on the YOLOv8 algorithm, and the specific steps include:

[0057] S1. Use a camera device to collect images of the coal conveyor belt in real time and perform preprocessing.

[0058] Specifically, a video stream of the coal conveyor belt is continuously collected using a camera device, which may be an industrial-grade camera connected to a back-end processing system via a network interface (such as Ethernet) to transmit the video stream in real time.

[0059] Frame extraction is performed on the received video stream, which can be achieved by extracting frames at time intervals, such as capturing a frame of image every certain time (1 second). Preprocessing is performed on the extracted image frames, including grayscale processing and image enhancement processing. Specifically, the grayscale processing includes converting the color image into a grayscale image to reduce the computational complexity. The specific grayscale formula is as follows:

[0060]

[0061] The image enhancement process includes using adaptive histogram equalization (AHE) to improve image contrast and enhance image details in local areas.

[0062] S2. Use the YOLOv8 algorithm to perform edge detection on the coal conveyor belt in the coal conveyor belt image, combine the acquired edge coordinates of the coal conveyor belt, perform deviation calculation according to the first deviation calculation step, and obtain the first deviation parameter.

[0063] Specifically, the camera device equipped with the YOLOv8 algorithm model can be directly used to perform edge detection on the coal conveyor belt in each frame of the preprocessed coal conveyor belt image, and the edge coordinates of each frame of the coal conveyor belt can be obtained in real time; or the preprocessed coal conveyor belt image can be transmitted to an edge computing node with a built-in YOLOv8 algorithm, and the edge detection can be performed on the coal conveyor belt in each frame of the preprocessed coal conveyor belt image, and the edge coordinates of each frame of the coal conveyor belt can be obtained in real time.

[0064] Combined with the edge coordinates of each frame of the coal conveyor belt obtained in real time, the first deviation parameter is calculated according to the first deviation calculation step. Specifically, the first deviation calculation step includes: performing edge line trajectory fitting for the left and right edge coordinates of the coal conveyor belt obtained by detection, such as: obtaining the coordinate set of the left and right edges of the conveyor belt from the detection results of the YOLOv8 model and , use the least squares method to fit the straight line equations of the left and right edges respectively: For the left edge:

[0065]

[0066] For the right edge:

[0067]

[0068] in: ,

[0069] Calculate centerline trajectory For comparison, , ; Compare it with the preset coal conveyor belt centerline trajectory and calculate the first deviation parameter, including the offset and ;

[0070] The formula for calculating the offset is:

[0071] in, The center line trajectory is at a fixed vertical coordinate in the image The horizontal coordinate at To preset the centerline trajectory of the coal conveyor belt The horizontal coordinate at

[0072] Calculate the tilt angle:

[0073] in, is the slope of the predetermined trajectory.

[0074] S3. Collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time.

[0075] Specifically, the sensor is used to synchronize with each frame of the coal conveyor belt image, and the operating data of the coal conveyor belt is collected in real time, including: the transmission speed of the coal conveyor belt, the load of the coal conveyor belt, etc., and the environmental data around the coal conveyor belt, including: light, temperature, humidity, etc.

[0076] S4. Using the first deviation parameter adaptive threshold adjustment mechanism, obtain a first deviation parameter threshold adjustment value corresponding to the operating condition data and environmental data collected in real time.

[0077] Specifically, considering the influence of different working conditions and environmental data conditions on deviation detection, a first deviation parameter adaptive threshold adjustment mechanism is introduced to dynamically adjust a preset first deviation parameter threshold, that is, a first deviation parameter preset threshold.

[0078] The first deviation parameter adaptive threshold adjustment mechanism can be obtained by using a first deviation parameter adaptive threshold adjustment function or a first deviation parameter adaptive threshold adjustment model. , The threshold adjustment value of the historical first deviation parameter corresponding to the historical first deviation parameter after threshold adjustment and whose similarity with the historical actual first deviation parameter is greater than the preset similarity under the conditions of different historical working condition data and environmental data is formed by numerical fitting, that is, for each combination of working condition data and environmental data, the historical first deviation parameter after threshold adjustment and whose similarity with the historical first deviation parameter and the actual deviation parameter is greater than the preset similarity is obtained, and the first deviation parameter threshold adjustment mean is determined according to the obtained first deviation parameter and the corresponding historical actual first deviation parameter, and then each working condition and environmental data is used as an independent variable and the corresponding determined first deviation parameter threshold adjustment mean is used as a dependent variable to fit and generate a first deviation parameter adaptive threshold adjustment function.

[0079] The first deviation parameter adaptive threshold adjustment model is generated by training the historical first deviation parameter threshold adjustment value corresponding to the historical first deviation parameter that has been threshold adjusted and has a similarity with the historical actual first deviation parameter greater than a preset similarity under different historical operating data and environmental data conditions; wherein the first deviation parameter adaptive threshold adjustment model adopts a neural network, the model input is the operating data and environmental data, and the model output is the first deviation parameter threshold value.

[0080] The first deviation parameter threshold adjustment value obtained in real time is used to adjust the first deviation parameter preset threshold, such as:

[0081]

[0082] in, is a preset offset threshold and a preset tilt angle threshold, The offset threshold adjustment function and the preset tilt angle threshold adjustment function can be obtained according to data fitting.

[0083] S5. According to the calculated first deviation parameter and the preset threshold of the first deviation parameter after threshold adjustment, a first result of whether the first deviation is real-time or not is obtained and correspondingly a coal conveyor belt deviation prompt information is generated.

[0084] Specifically, the first deviation parameter calculated in real time is compared with the preset threshold value of the first deviation parameter after the threshold adjustment. If the calculated first deviation parameter exceeds the preset threshold value of the first deviation parameter after the threshold adjustment, such as: or The first result of obtaining whether the conveyor belt is deviating in real time is that there is a coal conveyor belt deviation in the current camera shooting area, and corresponding coal conveyor belt deviation prompt information is generated.

[0085] By adopting the above solution, high-precision deviation detection is achieved in complex environments (especially low-light conditions) based on the YOLOv8 algorithm, and the threshold is dynamically adjusted to adapt to different working conditions and environmental data, effectively reducing the false alarm rate of deviation detection.

[0086] Example 2

[0087] The difference from Example 1 is that, in order to take into account the situation where the shape of the coal conveyor belt is relatively complex and the surrounding environment is relatively complex, in order to further obtain more accurate detection results and determine the specific location where the specific coal conveyor belt is deviated, the method also includes:

[0088] The position range of the coal conveyor belt covered by the camera device is obtained in advance and the area is divided to obtain the straight area and the curved area of ​​the coal conveyor belt; the preset length is matched corresponding to the divided area type of the coal conveyor belt, such as: the first preset length is matched for the straight area of ​​the coal conveyor belt, and the second preset length is matched for the curved area of ​​the coal conveyor belt, and the first preset length is greater than the second preset length; and the left and right edge coordinates of the coal conveyor belt are segmented using the matched preset length to obtain several segment sets of the left and right edge coordinates of the coal conveyor belt; such as: the preset length is 1 cm, and for the edge coordinates (1.21, 1.80), (1.256, 1.80), (1.289, 1.81), (1.32, 1.809), the segment sets (1.21, 1.80), (1.256, 1.80), (1.289, 1.81) and (1.32, 1.809) are obtained.

[0089] According to each segmented set of left and right edge coordinates of the coal conveyor belt, line trajectories are fitted for the left and right edge coordinates of the coal conveyor belt obtained by detection, and the center line trajectory is calculated using the fitted left and right edge line trajectories and compared with the preset center line trajectory of the coal conveyor belt, and the first deviation parameters corresponding to several segments are calculated; such as: .

[0090] For the operating condition data of the coal conveyor belt and the environmental data around the coal conveyor belt collected in real time, the operating condition data and environmental data within the position range of the coal conveyor belt can be obtained in segments according to the line trajectories of the left and right edges of the coal conveyor belt, and then the first deviation parameter threshold adjustment value can be obtained accordingly and the deviation parameter preset threshold after the threshold adjustment can be generated in segments.

[0091] The first deviation parameter calculated in real time is judged in segments whether it exceeds the preset threshold of the first deviation parameter after the threshold adjustment of the corresponding segment, and a coal conveyor belt deviation prompt information is generated for the first deviation parameter that exceeds the preset threshold of the first deviation parameter after the threshold adjustment, including specific segment information, so as to obtain the deviation situation of the specific segment area.

[0092] By adopting the above solution, the deviation detection of the coal conveyor belt can be further refined.

[0093] Example 3

[0094] The difference from the above-mentioned embodiment 1 is that in order to further obtain accurate detection results, the YOLOv8 model is extended to perform multi-target detection, that is, in addition to detecting the edge of the conveyor belt, the coal carried on the conveyor belt can be further detected, and whether deviation occurs based on the distribution detection of the coal is determined. Combining multiple detection results, the running state of the conveyor belt is comprehensively judged to provide more accurate detection results. The method also includes:

[0095] The YOLOv8 algorithm is used to perform edge detection on the coal carried on the coal conveyor belt in the coal conveyor belt image to obtain the edge coordinates and the second confidence level of the coal.

[0096] Combined with the edge coordinates of the coal obtained, the deviation calculation is performed according to the second deviation calculation step to obtain the second deviation parameter. The second deviation calculation step includes: performing contour fitting on the edge coordinates of the coal obtained by detection, and calculating the centroid position of the coal area; the specific formula includes: assuming that the coal with contour fitting is an object of arbitrary shape and uneven mass distribution, the centroid The coordinates are:

[0097]

[0098] In the formula, it is assumed that the mass density function of the loaded coal is Represents the coal area element.

[0099] Compare the calculated coal area centroid position with the coal centroid position carried by the preset coal conveyor belt to calculate the second deviation parameter, including: offset .

[0100] Among them, in order to complete the detection in time, edge computing was chosen to transmit the collected images of the coal conveyor belt to different edge nodes, and these different edge computing nodes with built-in YOLOv8 algorithms were used to realize distributed computing of using the YOLOv8 algorithm to perform edge detection and deviation calculation on the coal conveyor belt in the coal conveyor belt image, and using the YOLOv8 algorithm to perform edge detection and deviation calculation on the coal carried on the coal conveyor belt in the coal conveyor belt image.

[0101] Taking into account different working conditions and environmental conditions, coal itself is also an influencing factor of the final deviation detection. The basic information data of the coal carried on the coal conveyor belt is collected in real time, including: coal type (such as lignite, bituminous coal and anthracite, etc.), weight, etc.

[0102] The second deviation parameter adaptive threshold adjustment mechanism is used to obtain the second deviation parameter threshold adjustment value corresponding to the real-time collected working condition data, environmental data and basic information data of coal. Specifically, the second deviation parameter adaptive threshold adjustment mechanism adopts a second deviation parameter adaptive threshold adjustment function or a second deviation parameter adaptive threshold adjustment model; wherein the second deviation parameter adaptive threshold adjustment function is generated by fitting the historical second deviation parameter threshold adjustment value corresponding to the historical second deviation parameter after threshold adjustment and whose similarity with the historical actual second deviation parameter is greater than the preset similarity under the conditions of different historical working condition data, environmental data and basic information data of coal; the second deviation parameter adaptive threshold adjustment model is generated by training the historical second deviation parameter threshold adjustment value corresponding to the historical second deviation parameter after threshold adjustment and whose similarity with the historical actual second deviation parameter is greater than the preset similarity under the conditions of different historical working condition data, environmental data and basic information data of coal, the input of the model is the working condition data, environmental data and basic information data of coal, and the output of the model is the second deviation parameter threshold adjustment value.

[0103] The second deviation parameter threshold adjustment value is obtained to adjust the second deviation parameter preset threshold, and it is determined whether the calculated second deviation parameter exceeds the second deviation parameter preset threshold after the threshold adjustment, so as to obtain the second result of deviation in real time; specifically, if it exceeds, , then the second result is that there is deviation.

[0104] In order to make a further comprehensive judgment, a comprehensive confidence calculation method is introduced. When the YOLOv8 algorithm is used to detect the edge of the coal conveyor belt in the coal conveyor belt image, the first confidence is obtained; the comprehensive confidence is calculated by weighting the first confidence and the second confidence, and the comprehensive confidence = (first confidence * weight 1) + (second comprehensive confidence * weight 2) / (weight 1 + weight 2); and a confidence threshold is set, and the comprehensive confidence is compared with the confidence threshold. If the comprehensive confidence is greater than the confidence threshold, and any of the first result and the second result is that there is deviation, then the corresponding coal conveyor belt deviation prompt information is generated instead of directly generating the coal conveyor belt deviation prompt information according to the first result; if the comprehensive confidence is lower than the confidence threshold, even if both the first result and the second result are deviations, it may be caused by detection errors, and it cannot be directly determined that the coal conveyor belt deviation prompt information is generated, and further verification is required.

[0105] Among them, the setting of weight 1 and weight 2 is to distribute the credibility of the two methods in proportion, and the sum is 1. Specifically, considering that the credibility of the results of different target detections under the working condition data, environmental data and basic information data of coal is different, the weight ratio corresponding to the first confidence and the second confidence in the weighted calculation of the comprehensive confidence according to the first confidence and the second confidence is dynamically adjusted according to the working condition data, environmental data and basic information data of coal, and the weight ratio corresponding to the first confidence and the second confidence is obtained including:

[0106] Construct and train a first neural network model, wherein the first neural network model is generated by training a data set consisting of historical data of different operating conditions, environmental data, basic information of coal, and a weight ratio corresponding to a first confidence level and a second confidence level when the accuracy of generating coal conveyor belt deviation warning information is greater than a preset accuracy rate. The input of the first neural network model is real-time operating condition data, environmental data and basic information of coal, and the output is a weight ratio corresponding to the first confidence level and the second confidence level; the currently collected real-time operating condition data, environmental data and basic information of coal are input into the first neural network model to obtain the weight ratio corresponding to the first confidence level and the second confidence level.

[0107] Similarly, for other targets, the coal conveyor belt deviation detection is carried out, and the corresponding detection confidence is obtained. The comprehensive confidence is calculated by integrating the corresponding confidences obtained by all detection methods, and then the analysis and judgment are combined with all the detection results. For example, suppose the different detection target sets are , the corresponding confidence level is , the comprehensive confidence calculation formula is: In the formula, For the target risk weight.

[0108] Set different target detection result sets as With confidence threshold ,At Exceeding the set threshold When , it indicates that the current detection error is small. If any of the detection results is deviation, a coal conveyor belt deviation prompt message will be generated.

[0109] Example 4

[0110] The difference from the first embodiment is that, considering that the sensor or camera device at a single moment is affected by external factors, resulting in incorrect data collected, and thus misjudgment occurs, in order to achieve timely and accurate deviation detection and prompting while trying to avoid misjudgment at a single moment, the method further includes:

[0111] According to the probability of actual deviation under the corresponding conditions of different historical working condition data and environmental data, the corresponding matching continuous judgment time is preset for the working condition data and environmental data combinations within different deviation probability preset ranges; for example, the deviation probability preset range is set to the first deviation probability , Second deviation probability And the third deviation probability , corresponding to the first continuous judgment time, the second continuous judgment time and the third continuous judgment time; wherein, the combination of working condition data and environmental data that is in a larger preset range of occurrence probability has a shorter corresponding matching continuous judgment time.

[0112] The continuous judgment duration is obtained by correspondingly matching the working condition data and the environmental data obtained in real time; according to the obtained continuous judgment duration, the YOLOv8 algorithm is continuously used to detect the edge of the coal conveyor belt in each frame image of the coal conveyor belt and the first deviation parameter is calculated accordingly;

[0113] According to the time sequence of the image frames, if the first deviation parameter obtained and calculated exceeds the first result of the deviation parameter preset threshold after the threshold adjustment within the continuous judgment time, a corresponding coal conveyor belt deviation prompt information is generated, replacing the original first result of obtaining whether the deviation is obtained in real time and generating the corresponding coal conveyor belt deviation prompt information.

[0114] Example 5

[0115] The difference from Example 1 is that, in order to further ensure the accuracy of the detection result, different target detection results can be used for verification, and the method further includes:

[0116] The YOLOv8 algorithm is used to detect the torn area of ​​the coal conveyor belt in the coal conveyor belt image, obtain the edge coordinates of the torn area of ​​the coal conveyor belt, and calculate the area of ​​the torn area of ​​the coal conveyor belt;

[0117] Continuously obtain the area of ​​the tearing region; input the area of ​​the tearing region, the operating condition data and the environmental data obtained at the current moment and the previous preset time period into the second neural network, and output whether there is a deviation result of the coal conveyor belt at the current moment; the second neural network model is trained and generated by marking the area of ​​the tearing region of continuous time periods with the result of coal conveyor belt deviation under different historical operating condition data and environmental data conditions; the input of the model is the operating condition data and environmental data at the current moment and the area of ​​the tearing region at the current moment and the previous preset time period.

[0118] While obtaining the result of whether the coal conveyor belt has deviated, verify whether the first result of obtaining whether the deviation exists in real time corresponds to the generated coal conveyor belt deviation prompt information; if not, it is determined that there is an error in the first result judgment, and the YOLOv8 algorithm is optimized and the optimized YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image at the next moment.

[0119] like Figure 2 As shown, the embodiment of the present application provides a coal conveyor belt deviation detection system based on the YOLOv8 algorithm, which specifically includes:

[0120] The coal conveyor belt image acquisition and processing module 101 is used to use a camera device to collect images of the coal conveyor belt in real time and perform preprocessing;

[0121] The first deviation parameter acquisition module 102 is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm; perform deviation calculation according to the first deviation calculation step in combination with the acquired edge coordinates of the coal conveyor belt to obtain the first deviation parameter; the first deviation calculation step includes: performing edge line trajectory fitting for the left and right edge coordinates of the coal conveyor belt obtained by the detection, calculating the center line trajectory using the fitted left and right edge line trajectories and comparing it with the preset center line trajectory of the coal conveyor belt, and calculating the first deviation parameter, including the offset and the tilt angle;

[0122] The coal conveyor belt working condition and environment acquisition module 103 is used to collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time;

[0123] The first deviation parameter adjustment module 104 is used to obtain a first deviation parameter threshold adjustment value corresponding to the working condition data and environmental data collected in real time by using a first deviation parameter adaptive threshold adjustment mechanism; the first deviation parameter adaptive threshold adjustment mechanism selects a first deviation parameter adaptive threshold adjustment function or a trained first deviation parameter adaptive threshold adjustment model that is fitted with a historical first deviation parameter threshold adjustment value corresponding to a historical first deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual first deviation parameter greater than a preset similarity under conditions of different historical working condition data and environmental data; and adjusts the preset threshold of the first deviation parameter by using the obtained first deviation parameter threshold adjustment value;

[0124] The coal conveyor belt deviation result acquisition module 105 is used to determine whether the calculated first deviation parameter exceeds the first deviation parameter preset threshold after the threshold adjustment, so as to obtain the first result of deviation in real time and generate corresponding coal conveyor belt deviation prompt information.

[0125] In a specific embodiment, the system further comprises:

[0126] The second deviation parameter acquisition module 106 is used to perform edge detection on the coal carried on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm to obtain the edge coordinates and second confidence of the coal; perform deviation calculation according to the second deviation calculation step based on the obtained edge coordinates of the coal to obtain the second deviation parameter; the second deviation calculation step includes: performing contour fitting on the edge coordinates of the coal obtained by the detection, calculating the centroid position of the coal area, and comparing it with the centroid position of the coal carried by the preset coal conveyor belt, and calculating the second deviation parameter, including: offset;

[0127] The coal basic information collection module 107 is used to collect basic information data of the coal carried on the coal conveyor belt in real time, including: type and weight;

[0128] The second deviation parameter adjustment module 108 is used to obtain the second deviation parameter threshold adjustment value corresponding to the real-time collected working condition data, environmental data and basic information data of coal by using the second deviation parameter adaptive threshold adjustment mechanism; the second deviation parameter adaptive threshold adjustment mechanism selects the second deviation parameter adaptive threshold adjustment function or the trained second deviation parameter adaptive threshold adjustment model fitted by the historical second deviation parameter threshold adjustment value corresponding to the historical second deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual second deviation parameter greater than a preset similarity under the conditions of different historical working condition data, environmental data and basic information data of coal;

[0129] The coal conveyor belt deviation result acquisition module 105 is also used to adjust the second deviation parameter preset threshold by obtaining the second deviation parameter threshold adjustment value, determine whether the calculated second deviation parameter exceeds the second deviation parameter preset threshold after the threshold adjustment, so as to obtain the second result of deviation in real time; obtain the first confidence when performing edge detection on the coal conveyor belt in the coal conveyor belt image by using the YOLOv8 algorithm; calculate the comprehensive confidence by weighting the first confidence and the second confidence, and set the confidence threshold; compare the comprehensive confidence with the confidence threshold, if the comprehensive confidence is greater than the confidence threshold, and any of the first result and the second result is that there is deviation, then generate the coal conveyor belt deviation prompt information instead of directly generating the coal conveyor belt deviation prompt information according to the first result.

[0130] In a specific embodiment, the system further comprises:

[0131] The first deviation parameter 102 is also used to pre-acquire the position range of the coal conveyor belt covered by the camera device and perform area division to obtain the straight area of ​​the coal conveyor belt and the curved area of ​​the coal conveyor belt; match the preset length corresponding to the divided area type of the coal conveyor belt, and use the matched preset length to segment the left and right edge coordinates of the coal conveyor belt to obtain a plurality of segment sets of the left and right edge coordinates of the coal conveyor belt; according to each segment set of the left and right edge coordinates of the coal conveyor belt, respectively perform line trajectory fitting on the left and right edge coordinates of the coal conveyor belt obtained by detection, calculate the center line trajectory of the segmented left and right edge line trajectories using the fitted left and right edge line trajectories, and compare it with the preset center line trajectory of the coal conveyor belt, and calculate and obtain the first deviation parameters corresponding to the plurality of segments;

[0132] The first deviation parameter adjustment module 104 is further used to obtain the working condition data and environmental data within the position range of the coal conveyor belt corresponding to the left and right edge line trajectories of the coal conveyor belt in sections, obtain the first deviation parameter threshold adjustment value accordingly, and generate the deviation parameter preset threshold value after the threshold adjustment in sections;

[0133] The coal conveyor belt deviation result acquisition module 105 is also used to determine in segments whether the first deviation parameter calculated in real time exceeds the preset threshold of the first deviation parameter after the threshold adjustment of the corresponding segment, and generate coal conveyor belt deviation prompt information for the first deviation parameter that exceeds the preset threshold of the first deviation parameter after the threshold adjustment.

[0134] In a specific embodiment, the system further comprises:

[0135] The coal conveyor belt deviation result acquisition module 105 is also used to preset the corresponding matching continuous judgment time for the working condition data and environmental data combination within different preset ranges of deviation probability according to the probability of actual deviation under the corresponding conditions of different historical working condition data and environmental data; wherein, the working condition data and environmental data combination within the larger preset range of occurrence probability has a shorter preset corresponding matching continuous judgment time; the continuous judgment time is obtained according to the corresponding matching of the working condition data and environmental data acquired in real time; according to the acquired continuous judgment time, the YOLOv8 algorithm is continuously used to perform edge detection on the coal conveyor belt in each frame image of the coal conveyor belt and the first deviation parameter is calculated accordingly; according to the time sequence of the image frames, when the first deviation parameter obtained and calculated exceeds the first result of the deviation parameter preset threshold value after the threshold value is adjusted within the continuous judgment time, the coal conveyor belt deviation prompt information is generated accordingly, replacing the original first result of obtaining whether the deviation is in real time and generating the coal conveyor belt deviation prompt information accordingly.

[0136] In a specific embodiment, the system further comprises:

[0137] The coal conveyor belt deviation result verification module 109 is used to use the YOLOv8 algorithm to detect the tear area of ​​the coal conveyor belt in the coal conveyor belt image, obtain the edge coordinates of the tear area of ​​the coal conveyor belt, and calculate the area of ​​the tear area of ​​the coal conveyor belt; continuously obtain the area of ​​the tear area; input the area of ​​the tear area, the operating data and the environmental data obtained at the current moment and the previous preset time period into the second neural network, and output whether there is a coal conveyor belt deviation result at the current moment; the second neural network model is trained and generated by marking the tear area of ​​continuous time periods with coal conveyor belt deviation results under different historical operating data and environmental data conditions; while obtaining the coal conveyor belt deviation result, verify whether the first result of obtaining whether the deviation is obtained in real time corresponds to the generation of coal conveyor belt deviation prompt information; if not, it is determined that the first result judgment is wrong, the YOLOv8 algorithm is optimized, and the optimized YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image at the next moment.

[0138] The embodiments of the present application also disclose a coal conveyor belt device.

[0139] Specifically, the coal conveyor belt device includes a conveying device, an image acquisition device and an image analysis device. The image analysis device includes a memory, a processor and a program stored and executable on the memory. When the program is executed by the processor, the steps of the coal conveyor belt deviation detection method based on the YOLOv8 algorithm are implemented as described above.

[0140] The embodiment of the present application also discloses a computer-readable storage medium.

[0141] Specifically, the computer-readable storage medium stores a computer program that can be loaded by a processor and executes the above-mentioned coal conveyor belt deviation detection method based on the YOLOv8 algorithm. The computer-readable storage medium includes, for example: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0142] The above are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Any feature disclosed in this specification (including the abstract and drawings), unless otherwise stated, can be replaced by other equivalent or alternative features with similar purposes. That is, unless otherwise stated, each feature is only an example of a series of equivalent or similar features.

Claims

1. A coal conveyor belt deviation detection method based on YOLOv8 algorithm, characterized in that: include: The image of the coal conveyor belt is collected in real time by using a camera device and pre-processed; The YOLOv8 algorithm is used to detect the edge of the coal conveyor belt in the coal conveyor belt image; the deviation calculation is performed according to the first deviation calculation step based on the acquired edge coordinates of the coal conveyor belt to obtain the first deviation parameter; The first deviation calculation step includes: fitting edge line trajectories for the acquired left and right edge coordinates of the coal conveyor belt, calculating the center line trajectory using the fitted left and right edge line trajectories and comparing it with the preset center line trajectory of the coal conveyor belt, and calculating the first deviation parameter, including the offset and the inclination angle; Collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time; The first deviation parameter adaptive threshold adjustment mechanism is used to obtain a first deviation parameter threshold adjustment value corresponding to the working condition data and environmental data collected in real time; the first deviation parameter adaptive threshold adjustment mechanism selects a first deviation parameter adaptive threshold adjustment function fitted with a historical first deviation parameter threshold adjustment value corresponding to a historical first deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual first deviation parameter greater than a preset similarity under different historical working condition data and environmental data conditions, or a trained first deviation parameter adaptive threshold adjustment model; The first deviation parameter threshold adjustment value obtained is used to adjust the first deviation parameter preset threshold, and it is determined whether the calculated first deviation parameter exceeds the first deviation parameter preset threshold after the threshold adjustment, so as to obtain the first result of whether the deviation occurs in real time and generate the coal conveyor belt deviation prompt information accordingly.

2. The coal conveyor belt deviation detection method based on the YOLOv8 algorithm according to claim 1 is characterized in that: Also includes: Pre-acquire the position range of the coal conveyor belt covered by the camera device and divide the area to obtain the straight area of ​​the coal conveyor belt and the curved area of ​​the coal conveyor belt; Matching the preset length corresponding to the divided area type of the coal conveyor belt, and using the matched preset length to segment the left and right edge coordinates of the coal conveyor belt to obtain a plurality of segment sets of the left and right edge coordinates of the coal conveyor belt; According to each segmented set of left and right edge coordinates of the coal conveyor belt, line trajectories are fitted for the left and right edge coordinates of the coal conveyor belt obtained by detection, and the center line trajectory is calculated by segment using the fitted left and right edge line trajectories and compared with the preset center line trajectory of the coal conveyor belt, so as to calculate and obtain the first deviation parameters corresponding to several segments; Obtain the working condition data and environmental data in the position range of the coal conveyor belt corresponding to the left and right edge line trajectories of the coal conveyor belt in sections, obtain the first deviation parameter threshold adjustment value accordingly, and generate the deviation parameter preset threshold value after the threshold adjustment in sections; It is judged in segments whether the first deviation parameter calculated in real time exceeds the preset threshold value of the first deviation parameter after the threshold value of the corresponding segment is adjusted, and a coal conveyor belt deviation prompt information is generated for the first deviation parameter that exceeds the preset threshold value of the first deviation parameter after the threshold value is adjusted.

3. The coal conveyor belt deviation detection method based on the YOLOv8 algorithm according to claim 1 is characterized in that: Also includes: The YOLOv8 algorithm is used to detect the edge of the coal carried on the coal conveyor belt in the coal conveyor belt image, and the edge coordinates and the second confidence level of the coal are obtained; Based on the acquired edge coordinates of the coal, the deviation calculation is performed according to the second deviation calculation step to obtain a second deviation parameter; The second deviation calculation step includes: performing contour fitting on the coal edge coordinates obtained by detection, calculating the centroid position of the coal area, and comparing it with the centroid position of the coal carried by the preset coal conveyor belt, and calculating the second deviation parameters, including: offset; Collect basic information data of coal carried on the coal conveyor belt in real time, including: type and weight; use the second deviation parameter adaptive threshold adjustment mechanism to obtain the second deviation parameter threshold adjustment value corresponding to the real-time collected working condition data, environmental data and basic information data of coal; the second deviation parameter adaptive threshold adjustment mechanism selects a second deviation parameter adaptive threshold adjustment function or a trained second deviation parameter adaptive threshold adjustment model that is fitted with a historical second deviation parameter threshold adjustment value corresponding to a historical second deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual second deviation parameter greater than a preset similarity under different historical working condition data, environmental data and basic information data of coal; The second deviation parameter preset threshold is adjusted by obtaining the second deviation parameter threshold adjustment value, and it is determined whether the calculated second deviation parameter exceeds the second deviation parameter preset threshold after the threshold adjustment, so as to obtain a second result of whether the deviation occurs in real time; A first confidence is obtained when the YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image; a comprehensive confidence is calculated according to the weighted first confidence and the second confidence, and a confidence threshold is set; the comprehensive confidence is compared with the confidence threshold, and if the comprehensive confidence is greater than the confidence threshold, and any of the first result and the second result is that there is deviation, then a corresponding coal conveyor belt deviation prompt information is generated instead of directly generating the coal conveyor belt deviation prompt information according to the first result.

4. The coal conveyor belt deviation detection method based on the YOLOv8 algorithm according to claim 3 is characterized in that: The weight ratio corresponding to the first confidence level and the second confidence level in the weighted calculation of the comprehensive confidence level according to the first confidence level and the second confidence level is dynamically adjusted according to the working condition data, the environmental data and the basic information data of the coal, and the weight ratio corresponding to the first confidence level and the second confidence level is obtained, including: Constructing and training a first neural network model, wherein the first neural network model is generated by training a data set consisting of a weight ratio corresponding to a first confidence level and a second confidence level when the accuracy of generating coal conveyor belt deviation prompt information is greater than a preset accuracy rate using historical different operating condition data, environmental data, basic information of coal, and the first neural network model, wherein the input of the first neural network model is real-time operating condition data, environmental data and basic information of coal, and the output is the weight ratio corresponding to the first confidence level and the second confidence level; The currently collected real-time operating data, environmental data and basic information of coal are input into the first neural network model to obtain the weight ratio corresponding to the first confidence level and the second confidence level.

5. The coal conveyor belt deviation detection method based on YOLOv8 algorithm according to claim 1 is characterized in that: Also includes: According to the probability of the actual deviation under the corresponding conditions of different historical working condition data and environmental data, the corresponding matching continuous judgment time is preset for the working condition data and environmental data combinations within different preset ranges of the deviation probability; wherein, the larger the preset range of the occurrence probability is, the shorter the preset matching continuous judgment time is; The continuous judgment duration is obtained by correspondingly matching the working condition data and the environmental data obtained in real time; according to the obtained continuous judgment duration, the YOLOv8 algorithm is continuously used to detect the edge of the coal conveyor belt in each frame image of the coal conveyor belt and the first deviation parameter is calculated accordingly; According to the time sequence of the image frames, if the first deviation parameter obtained and calculated exceeds the first result of the deviation parameter preset threshold after the threshold adjustment within the continuous judgment time, a corresponding coal conveyor belt deviation prompt information is generated, replacing the original first result of obtaining whether the deviation is obtained in real time and generating the corresponding coal conveyor belt deviation prompt information.

6. The method for detecting deviation of a coal conveyor belt based on the YOLOv8 algorithm according to claim 1, characterized in that: Also includes: The YOLOv8 algorithm is used to detect the torn area of ​​the coal conveyor belt in the coal conveyor belt image, obtain the edge coordinates of the torn area of ​​the coal conveyor belt, and calculate the area of ​​the torn area of ​​the coal conveyor belt; Continuously obtain the area of ​​the tearing area; input the area of ​​the tearing area, the operating condition data and the environmental data obtained at the current moment and the previous preset time period into the second neural network, and output whether there is a deviation result of the coal conveyor belt at the current moment; the second neural network model is trained and generated by marking the tearing area area of ​​the continuous time period with the deviation result of the coal conveyor belt under different historical operating condition data and environmental data conditions; While obtaining the result of whether the coal conveyor belt is deviating, verify whether the first result of obtaining whether the coal conveyor belt is deviating in real time corresponds to the generated coal conveyor belt deviating prompt information; If not generated, it is determined that the first result is wrong, the YOLOv8 algorithm is optimized, and the optimized YOLOv8 algorithm is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image at the next moment.

7. The method for detecting deviation of a coal conveyor belt based on the YOLOv8 algorithm according to claim 3, characterized in that: Also includes: The edge computing method is collected and different edge computing nodes are used to realize distributed computing of edge detection and deviation calculation of the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm and edge detection and deviation calculation of the coal carried on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm.

8. A coal conveyor belt deviation detection system based on YOLOv8 algorithm, characterized in that: include: The coal conveyor belt image acquisition and processing module is used to use a camera device to collect images of the coal conveyor belt in real time and perform preprocessing; A first deviation parameter acquisition module is used to perform edge detection on the coal conveyor belt in the coal conveyor belt image using the YOLOv8 algorithm; based on the acquired edge coordinates of the coal conveyor belt, perform deviation calculation according to the first deviation calculation step to obtain a first deviation parameter; The first deviation calculation step includes: fitting edge line trajectories for the left and right edge coordinates of the coal conveyor belt obtained by detection, calculating the center line trajectory using the fitted left and right edge line trajectories and comparing it with the preset center line trajectory of the coal conveyor belt, and calculating the first deviation parameter, including the offset and the inclination angle; The coal conveyor belt working condition and environment collection module is used to collect the working condition data of the coal conveyor belt and the environmental data around the coal conveyor belt in real time; The first deviation parameter adjustment module is used to obtain a first deviation parameter threshold adjustment value corresponding to the working condition data and environmental data collected in real time by using a first deviation parameter adaptive threshold adjustment mechanism; the first deviation parameter adaptive threshold adjustment mechanism selects a first deviation parameter adaptive threshold adjustment function or a trained first deviation parameter adaptive threshold adjustment model that is fitted by a historical first deviation parameter threshold adjustment value corresponding to a historical first deviation parameter that has undergone threshold adjustment and has a similarity with the historical actual first deviation parameter greater than a preset similarity under conditions of different historical working condition data and environmental data; and adjusts the preset threshold of the first deviation parameter by using the obtained first deviation parameter threshold adjustment value; The coal conveyor belt deviation result acquisition module is used to determine whether the calculated first deviation parameter exceeds the preset threshold of the first deviation parameter after the threshold adjustment, so as to obtain the first result of deviation in real time and generate corresponding coal conveyor belt deviation prompt information.

9. A coal conveyor belt device, characterized in that: The coal conveyor belt device includes an image acquisition device and an image analysis device. The image analysis device includes a memory, a processor, and a program stored and executable on the memory. When the program is executed by the processor, the steps of the method as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the method according to any one of claims 1 to 7.

Images