Belt state detection method based on AI detection and identification

Abstract

The invention relates to the technical field of belt state detection, and particularly discloses a belt state detection method based on AI detection identification, and the method comprises the steps: belt deviation detection, deviation reason confirmation and feedback, belt wear detection, wear degree confirmation and feedback, and belt tear detection early warning. Whether the target belt deviates or not is judged according to the running video of the target belt, if the target belt deviates, the deviation reason of the target belt is determined, the abrasion degree of each area of the target belt is determined by constructing the three-dimensional information model of each area, meanwhile, tearing severity early warning is conducted by analyzing the tearing severity coefficient of the target belt, and therefore the tearing severity of the target belt is improved. The method improves the troubleshooting efficiency of belt deviation reasons, improves the accuracy of belt wear degree judgment, accurately evaluates the expansion direction and speed of cracks, provides effective data support for tear prevention and control, and helps to find potential tear risks in time, thereby reducing the possibility of belt tear accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of belt state detection, and in particular to a belt state detection method based on AI detection and recognition. Background Art

[0002] In industries such as coal, mining, and electricity, conveyor belts are important material transportation equipment, and their operating status is directly related to production efficiency and safety. However, traditional belt status detection methods mostly rely on manual inspections and simple sensor monitoring, which have problems such as untimely monitoring, poor accuracy, and slow response speed. With the rapid development of artificial intelligence technology, AI-based belt status detection systems have gradually become a research hotspot and have broad application prospects.

[0003] The existing methods for monitoring energy consumption in engineering construction still have the following problems: 1. When detecting the belt deviation status, the current method is based only on manual observation to determine whether the belt is deviated, which results in a large error in confirming the deviation status. At the same time, when the belt deviates, manual investigation of the cause of the deviation is not only time-consuming, but also cumbersome, and the cause of the belt deviation cannot be accurately located, reducing the efficiency of investigating the cause of the belt deviation.

[0004] 2. When detecting the wear status of the belt, the wear degree of the belt is currently judged only by observing the overall pattern of the belt to see whether it is broken or flattened. The wear condition of each pattern in the belt in the vertical direction is not analyzed, which reduces the coverage of the belt wear analysis and the accuracy of the belt wear degree judgment, and cannot guarantee the good condition of the belt.

[0005] 3. When detecting the tearing status of the belt, the cracking posture of each crack in each area of ​​the belt is not deeply analyzed. Only the density of cracks in each area is considered without considering the cracking angle of each crack. The analysis dimension is relatively single, and it is impossible to accurately evaluate the expansion direction and speed of the cracks, thereby failing to provide effective data support for the prevention and control of tearing. At the same time, potential tearing risks may not be discovered in time, thereby increasing the possibility of belt tearing accidents. Summary of the invention

[0006] In view of this, in order to solve the problems raised in the above background technology, a belt status detection method based on AI detection and recognition is proposed.

[0007] The objective of the present invention can be achieved through the following technical solutions: The present invention provides a belt state detection method based on AI detection and recognition, comprising the following steps: S1, belt deviation detection: randomly select edge monitoring points on the edge of the target belt, use a high-definition camera to shoot the target belt in real time, obtain the running video of the target belt, and determine whether the target belt is deviated. If it is deviated, execute step S2, and if it is not deviated, execute step S3.

[0008] S2. Confirmation and feedback of the cause of deviation: Collect the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor, confirm the cause of the deviation of the target belt, and provide feedback.

[0009] S3. Belt wear detection: randomly arrange detection points in each component pattern of each area of ​​the target belt, construct a three-dimensional information model of each area, and obtain the y-axis coordinate value of each detection point in each component pattern of each area to analyze the wear degree of each area of ​​the target belt.

[0010] S4. Confirmation and feedback of wear degree: confirm the wear degree of each area of ​​the target belt and provide feedback of the wear degree of each area.

[0011] S5. Belt tear detection warning: collect images corresponding to each area of ​​the target belt, analyze the tear severity coefficient of the target belt, and compare it with the set value. If it is greater than or equal to the set value, a tear severity warning is issued.

[0012] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention determines whether the target belt is deviating based on the running video of the target belt. If it is deviating, the cause of the deviation of the target belt is confirmed according to the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor, thereby reducing the error existing in manual observation of whether the belt state is deviating, and avoiding the problems of long time-consuming cycle and cumbersome troubleshooting process in manual troubleshooting of the cause of the deviation, thereby improving the efficiency of troubleshooting the cause of the belt deviation.

[0013] (2) The present invention constructs a three-dimensional information model of each area, and obtains the y-axis coordinate value of each detection point in each component pattern of each area, analyzes the wear degree of each area of ​​the target belt, improves the coverage of the belt wear analysis, and at the same time improves the accuracy of the belt wear degree judgment, thereby ensuring the good condition of the belt.

[0014] (3) The present invention calculates the crack density and crack opening angle hazard coefficient of the target belt based on the images corresponding to each area of ​​the target belt, thereby analyzing the tear severity coefficient of the target belt. It can accurately evaluate the expansion direction and speed of the crack, thereby providing effective data support for the prevention and control of tearing, and helping to timely discover potential tearing risks, thereby reducing the possibility of belt tearing accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for describing the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0016] Figure 1 The figure is a schematic flow chart of the steps of the method of the present invention. DETAILED DESCRIPTION

[0017] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0018] See also Figure 1 As shown, the present invention provides a belt state detection method based on AI detection and recognition, including: S1, belt deviation detection: randomly select edge monitoring points on the edge of the target belt, use a high-definition camera to shoot the target belt in real time, obtain the running video of the target belt, and judge whether the target belt is deviated. If it is deviated, execute step S2, if it is not deviated, execute step S3.

[0019] In a specific embodiment of the present invention, the target belt has two edges. This embodiment takes any one of the edges as an example for analysis. The analysis of the two edges is similar to the analysis of one edge, which will not be repeated here.

[0020] In a specific embodiment of the present invention, the specific process of determining whether the target belt is deviated is as follows: A1. Dividing the running video of the target belt into running images in frames, and locating the positions of the edge monitoring points from the running images of the target belt.

[0021] A2. Overlap and compare each running image of the target belt with the preset boundary image of the target belt stored in the database, so as to locate the distance from each edge monitoring point in each running image to the preset boundary line, and record it as Lij , where i represents the number of the running image, i=1,2,...,n, and j represents the number of the edge monitoring point, j=1,2,...,m.

[0022] A3. Calculate the deviation β of the target belt.

[0023] In a specific embodiment of the present invention, the specific process of calculating the deviation of the target belt is as follows: B1, subtracting the distance from each edge monitoring point in each running image of the target belt to the preset boundary line from the set reference distance to obtain the deviation distance of each edge monitoring point in each running image of the target belt, and performing mean calculation to obtain the average deviation distance of the edge monitoring points in each running image of the target belt, and recording it as

[0024] B2. Extract the maximum value of the deviation distance of each edge monitoring point in each running image of the target belt and record it as

[0025] B3. Calculate the deviation β of the target belt. Among them, L′ represents the distance of the reference setting, a 1 and a 2 They represent the weights of the average deviation distance and the maximum deviation distance corresponding to the deviation degree assessment, respectively. 1 +a 2 =1, e represents a natural constant, and n represents the number of running images.

[0026] In a specific embodiment of the present invention, a 1 The setting value is 0.5, a 2 The setting value is 0.5. The average deviation distance deviation can reflect the average deviation degree of the belt during the overall operation process, helping to understand the overall trend of belt deviation. The maximum deviation distance deviation can reveal the degree of belt deviation under extreme conditions, which is of great significance for evaluating the safety risks that may be caused by deviation. It is usually necessary to comprehensively consider the average deviation distance deviation and the maximum deviation distance deviation. The two can complement each other and provide more comprehensive belt deviation information.

[0027] A4. Compare the deviation of the target belt with the deviation of the set reference. If the deviation of the target belt is greater than or equal to the deviation of the set reference, it indicates that the target belt is deviated. Otherwise, it indicates that the target belt is not deviated.

[0028] S2. Confirmation and feedback of the cause of deviation: Collect the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor, confirm the cause of the deviation of the target belt, and provide feedback.

[0029] It should be noted that the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor are collected by the belt tension sensor, the roller rotation speed sensor and the vibration detector respectively.

[0030] In a specific embodiment of the present invention, the specific process of confirming the cause of the deviation of the target belt is: C1. Based on the rotation speed of the two rollers of the target belt during each operation, the roller rotation synchronization degree ω of the target belt is calculated.

[0031] In a specific embodiment of the present invention, the specific process of calculating the roller rotation synchronization of the target belt is as follows: D1, the speeds of the two rollers of the target belt during each operation are subtracted from each other to obtain the speed deviation of the two rollers of the target belt during each operation, and recorded as Δv d , where d represents the number of each run, d = 1, 2, ..., k.

[0032] D2. Calculate the roller rotation synchronization of the target belt, ω. Wherein, Δv′ represents the speed deviation of the set reference, and k represents the number of operations.

[0033] C2. Based on the tension of each monitoring point of the target belt during each operation, the tension uniformity δ of the target belt during the operation is calculated.

[0034] It should be noted that the specific process of calculating the tension uniformity of the target belt during operation is as follows: the tension of each monitoring point of the target belt during each operation is recorded as τ dt , where t represents the number of the monitoring point, t = 1, 2, ..., h.

[0035] Calculate the tension uniformity δ of the target belt during operation, Among them, τ d ( t+1 ) represents the tension of the target belt at the t+1th monitoring point during the dth operation, and Δτ represents the tension deviation of the set reference.

[0036] C3. Based on the vibration frequency of the motor of the target belt during each operation, the vibration abnormality index ξ of the target belt during the operation is calculated.

[0037] It should be noted that the specific process of calculating the vibration abnormality index of the target belt during operation is as follows: the vibration frequency of the motor of the target belt during each operation is recorded as μ d .

[0038] Calculate the vibration abnormality index ξ of the target belt during operation, Here, μ′ and Δμ represent the vibration frequency and vibration frequency deviation of the set reference, respectively.

[0039] C4. When δ<δ′, the cause of the deviation of the target belt is recorded as uneven tension; when ω<ω′, the cause of the deviation of the target belt is recorded as asynchronous roller rotation; when ξ≥ξ′, the cause of the deviation of the target belt is recorded as abnormal motor vibration, wherein δ′, ω′ and ξ′ represent the tension uniformity, roller rotation synchronization and vibration abnormality index of the set reference, respectively.

[0040] The embodiment of the present invention determines whether the target belt is deviating based on the running video of the target belt. If deviating, the cause of the deviation of the target belt is confirmed according to the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor, thereby reducing the error existing in manual observation of whether the belt state is deviating, and at the same time avoiding the problems of long time-consuming cycle and cumbersome troubleshooting process in manual troubleshooting of the cause of the deviation, thereby improving the efficiency of troubleshooting the cause of the belt deviation.

[0041] S3. Belt wear detection: randomly arrange detection points in each component pattern of each area of ​​the target belt, construct a three-dimensional information model of each area, and obtain the y-axis coordinate value of each detection point in each component pattern of each area to analyze the wear degree of each area of ​​the target belt.

[0042] It should be noted that the specific method of constructing the three-dimensional information model of each area is: by collecting images of each area of ​​the target belt, importing the images of each area into the three-dimensional modeling software, and identifying and constructing the three-dimensional information model of each area through the three-dimensional modeling software.

[0043] In a specific embodiment of the present invention, the specific process of analyzing the wear degree of each area of ​​the target belt is as follows: E1. Based on the y-axis coordinate value of each detection point in each component texture of each area, the wear degree of each component texture of each area of ​​the target belt is calculated. Among them, r represents the number of the region, r = 1, 2, ..., z, g represents the number of the constituent texture, g = 1, 2, ..., p.

[0044] In a specific embodiment of the present invention, the specific process of calculating the wear degree of each component pattern of each area of ​​the target belt is as follows: F1, the y-axis coordinate value of each detection point in each component pattern of each area is recorded as y rgf , where f represents the number of the detection point, f = 1, 2, ..., q.

[0045] F2. Extract the initial position of each detection point in each component pattern of each area of ​​the target belt from the database, and bring it into the three-dimensional information model of the corresponding area to obtain the initial y-axis coordinate value of each detection point in each component pattern of each area, recorded as

[0046] F3. Calculate the wear degree of each component pattern in each area of ​​the target belt Wherein, Δy′ represents the allowable deviation value of the y-axis setting, and q represents the number of detection points.

[0047] E2. Compare the wear degree of each component pattern of each area of ​​the target belt with the wear degree of the set reference. If the wear degree of a component pattern is greater than the wear degree of the set reference, then the component pattern is recorded as the wear pattern. Count the number of wear patterns in each area of ​​the target belt, recorded as ε r .

[0048] E3. Calculate the wear degree of each area of ​​the target belt Among them, σ represents the proportion of the number of wear lines set as reference, and p represents the number of component lines.

[0049] S4. Confirmation and feedback of wear degree: confirm the wear degree of each area of ​​the target belt and provide feedback of the wear degree of each area.

[0050] In a specific embodiment of the present invention, the method for confirming the degree of wear of each area of ​​the target belt is: comparing the degree of wear of each area of ​​the target belt with the wear degree set corresponding to each degree of wear stored in the database; if the degree of wear of a certain area is within the wear degree set corresponding to a certain degree of wear, then the degree of wear is used as the degree of wear of the area, thereby obtaining the degree of wear of each area of ​​the target belt.

[0051] The embodiment of the present invention constructs a three-dimensional information model of each area, and obtains the y-axis coordinate value of each detection point in each component texture of each area, analyzes the wear of each area of ​​the target belt, thereby improving the coverage of the belt wear analysis, and at the same time improves the accuracy of judging the degree of belt wear, thereby ensuring the good condition of the belt.

[0052] S5. Belt tear detection warning: collect images corresponding to each area of ​​the target belt, analyze the tear severity coefficient of the target belt, and compare it with the set value. If it is greater than or equal to the set value, a tear severity warning is issued.

[0053] It should be noted that the images corresponding to each area of ​​the target belt are acquired by a high-definition camera.

[0054] In a specific embodiment of the present invention, the specific process of analyzing the tear severity coefficient of the target belt is as follows: G1, locating the number of cracks, the length of each crack, the length of the region and the area of ​​the region from the image corresponding to each region of the target belt, and calculating the crack density of the target belt

[0055] It should be noted that the specific process of calculating the crack density of the target belt is as follows: the number of cracks, the length of the region and the area of ​​the region corresponding to each region of the target belt are respectively recorded as M r , l r and S r .

[0056] The maximum value of each crack length corresponding to each area of ​​the target belt is extracted and recorded as

[0057] Calculate the crack density of the target belt Among them, σ 1 and σ 2 They respectively represent the percentage of crack number and crack length per unit area of ​​the set reference, and z represents the number of regions.

[0058] G2. Calculate the crack angle hazard coefficient ψ of the target belt.

[0059] In a specific embodiment of the present invention, the specific process of calculating the crack opening angle hazard coefficient of the target belt is: H1. Establish a rectangular coordinate system with the center point of one side edge of the target belt as the origin, the direction of the belt conveying the goods as the y-axis, and the direction perpendicular to the belt conveying the goods as the x-axis.

[0060] H2. Locate the coordinates of the two endpoints of each crack corresponding to each region in the rectangular coordinate system and record them as and Wherein, w represents the crack number, w=1,2,...,u.

[0061] H3. Calculate the angle θ of each crack in each area rw ,

[0062] H4, the angle of each crack in each area is 0 The difference between each crack in each area and 0 0 The deviation angle between them is recorded as the first deviation angle. If the first deviation angle of a crack in a certain area is less than the set first deviation angle, the crack is determined to be a first deviation crack. The number of first deviation cracks in each area is counted and recorded as

[0063] H5. Compare the angles of each crack in each region with 90 0 The difference between each crack in each area and 90 0 The deviation angle between the cracks in each area and the 0The absolute value of the deviation angle between is recorded as the second deviation angle. If the second deviation angle of a crack in a certain area is less than the set second deviation angle, the crack is determined to be a second deviation crack. The number of second deviation cracks in each area is counted and recorded as

[0064] H6. Calculate the crack angle hazard coefficient ψ of the target belt. Where ζ′ represents the number of deviation cracks set as a reference.

[0065] It should be noted that the crack angles of the target belt in the direction of the goods conveyed by the belt and perpendicular to the direction of the goods conveyed by the belt are more likely to lead to rapid crack expansion and cause more serious structural damage to the target belt. Therefore, when calculating the crack angle hazard factor of the target belt, the crack angles in the direction of the goods conveyed by the belt and perpendicular to the direction of the goods conveyed by the belt are mainly considered.

[0066] G3. Calculate the tear severity coefficient θ of the target belt. Among them, b 1 and b 2 They represent the weights of the tear severity assessment corresponding to the crack density and crack angle hazard factor of the reference, b 1 +b 2 =1.

[0067] In a specific embodiment of the present invention, b 1 The setting value of is 0.5, b 2 The setting value is 0.5. The crack density reflects the number and distribution of cracks on the belt surface. A high density of cracks means that the structural integrity of the belt is more seriously damaged, which may cause the belt to lose its load-bearing capacity more quickly or break suddenly. Therefore, the crack density is an important indicator for measuring the risk of belt tearing. The crack opening angle can affect the crack propagation speed and direction, and thus affect the overall strength and stability of the belt. Therefore, when calculating the tear severity coefficient of the target belt, the crack density and crack opening angle hazard coefficient should be comprehensively considered.

[0068] The embodiment of the present invention calculates the crack density and crack opening angle hazard coefficient of the target belt according to the images corresponding to each area of ​​the target belt, thereby analyzing the tear severity coefficient of the target belt, and can accurately evaluate the expansion direction and speed of the crack, thereby providing effective data support for preventing and controlling tearing, and helping to timely discover potential tearing risks, thereby reducing the possibility of belt tearing accidents.

[0069] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, they shall all fall within the protection scope of the present invention.

Claims

1. A belt state detection method based on AI detection and recognition, characterized in that: The steps include: S1, belt deviation detection: randomly select edge monitoring points on the edge of the target belt, use a high-definition camera to take real-time photos of the target belt, obtain the running video of the target belt, and determine whether the target belt is deviated. If it is deviated, execute step S2, if not, execute step S3; S2. Confirmation and feedback of the cause of deviation: collect the tension of each monitoring point of the target belt during each operation, the rotation speed of the two rollers and the vibration frequency of the motor, confirm the cause of the deviation of the target belt, and provide feedback; S3. Belt wear detection: randomly arrange detection points in each component pattern of each area of ​​the target belt, construct a three-dimensional information model of each area, and obtain the y-axis coordinate value of each detection point in each component pattern of each area, and analyze the wear degree of each area of ​​the target belt; S4, wear degree confirmation and feedback: confirm the wear degree of each area of ​​the target belt and provide feedback on the wear degree of each area; S5. Belt tear detection warning: collect images corresponding to each area of ​​the target belt, analyze the tear severity coefficient of the target belt, and compare it with the set value. If it is greater than or equal to the set value, a tear severity warning is issued.

2. The belt state detection method based on AI detection and recognition according to claim 1 is characterized in that: The specific process of judging whether the target belt is deviated is as follows: A1. Divide the running video of the target belt into running images in frames, and locate the position of each edge monitoring point from each running image of the target belt; A2. Overlap and compare each running image of the target belt with the preset boundary image of the target belt stored in the database, so as to locate the distance from each edge monitoring point in each running image to the preset boundary line, and record it as L ij , where i represents the number of the running image, i=1,2,...,n, and j represents the number of the edge monitoring point, j=1,2,...,m; A3. Calculate the deviation β of the target belt; A4. Compare the deviation of the target belt with the deviation of the set reference. If the deviation of the target belt is greater than or equal to the deviation of the set reference, it indicates that the target belt is deviated. Otherwise, it indicates that the target belt is not deviated.

3. The belt state detection method based on AI detection and recognition according to claim 2 is characterized in that: The specific process of calculating the deviation of the target belt is as follows: B1. Subtract the distance from each edge monitoring point to the preset boundary line in each running image of the target belt from the distance of the set reference to obtain the deviation distance of each edge monitoring point in each running image of the target belt, and calculate the average value to obtain the average deviation distance of the edge monitoring point in each running image of the target belt, and record it as B2. Extract the maximum value of the deviation distance of each edge monitoring point in each running image of the target belt and record it as B3. Calculate the deviation β of the target belt. Wherein, L′ represents the set reference distance, a1 and a2 represent the set average deviation distance deviation and maximum deviation distance deviation corresponding to the deviation degree evaluation weights, a1+a2=1, e represents a natural constant, and n represents the number of running images.

4. The belt state detection method based on AI detection and recognition according to claim 3 is characterized in that: The specific process of confirming the cause of the deviation of the target belt is as follows: C1. Calculate the roller rotation synchronization ω of the target belt based on the rotation speed of the two rollers of the target belt during each operation; C2. Calculate the tension uniformity δ of the target belt during operation based on the tension of each monitoring point during each operation of the target belt; C3. Calculate the vibration abnormality index ξ of the target belt during operation based on the vibration frequency of the motor during each operation of the target belt; C4. When δ<δ′, the cause of the deviation of the target belt is recorded as uneven tension; when ω<ω′, the cause of the deviation of the target belt is recorded as asynchronous roller rotation; when ξ≥ξ′, the cause of the deviation of the target belt is recorded as abnormal motor vibration, wherein δ′, ω′ and ξ′ represent the tension uniformity, roller rotation synchronization and vibration abnormality index of the set reference, respectively.

5. The belt state detection method based on AI detection and recognition according to claim 4 is characterized in that: The specific process of calculating the roller rotation synchronization of the target belt is as follows: D1. Subtract the speeds of the two rollers of the target belt during each operation to obtain the speed deviation of the two rollers of the target belt during each operation, and record it as Δv d , where d represents the number of each run, d = 1, 2, ..., k; D2. Calculate the roller rotation synchronization of the target belt, ω. Wherein, Δv′ represents the speed deviation of the set reference, and k represents the number of operations.

6. The belt state detection method based on AI detection and recognition according to claim 3 is characterized in that: The specific process of analyzing the wear degree of each area of ​​the target belt is as follows: E1. Calculate the wear degree of each component pattern of each area of ​​the target belt based on the y-axis coordinate value of each detection point in each component pattern of each area Where r represents the number of the region, r = 1, 2, ..., z, g represents the number of the component texture, g = 1, 2, ..., p; E2. Compare the wear degree of each component pattern of each area of ​​the target belt with the wear degree of the set reference. If the wear degree of a component pattern is greater than the wear degree of the set reference, then the component pattern is recorded as the wear pattern. Count the number of wear patterns in each area of ​​the target belt, recorded as ε r ; E3. Calculate the wear degree of each area of ​​the target belt Among them, σ represents the proportion of the number of wear lines set as reference, and p represents the number of component lines.

7. The belt state detection method based on AI detection and recognition according to claim 6 is characterized in that: The specific process of calculating the wear degree of each component pattern of each area of ​​the target belt is as follows: F1. The y-axis coordinate value of each detection point in each component pattern of each area is recorded as y rgf , where f represents the number of the detection point, f = 1, 2, ..., q; F2. Extract the initial position of each detection point in each component pattern of each area of ​​the target belt from the database, and bring it into the three-dimensional information model of the corresponding area to obtain the initial y-axis coordinate value of each detection point in each component pattern of each area, recorded as F3. Calculate the wear degree of each component pattern in each area of ​​the target belt Wherein, Δy′ represents the allowable deviation value of the y-axis setting, and q represents the number of detection points.

8. The belt state detection method based on AI detection and recognition according to claim 6 is characterized in that: The method for confirming the degree of wear of each area of ​​the target belt is: comparing the degree of wear of each area of ​​the target belt with the wear degree set corresponding to each degree of wear stored in the database; if the degree of wear of a certain area is within the wear degree set corresponding to a certain degree of wear, then the degree of wear is used as the degree of wear of the area, thereby obtaining the degree of wear of each area of ​​the target belt.

9. The belt state detection method based on AI detection and recognition according to claim 6 is characterized in that: The specific process of analyzing the tear severity coefficient of the target belt is as follows: G1. Locate the number of cracks, the length of each crack, the length of the region and the area of ​​the region from the images corresponding to each region of the target belt, and calculate the crack density of the target belt G2. Calculate the crack angle hazard coefficient ψ of the target belt; G3. Calculate the tear severity coefficient θ of the target belt. Among them, b1 and b2 represent the weights of the tear severity assessment corresponding to the set reference crack density and crack opening angle hazard factor, respectively, b1+b2=1.

10. The belt state detection method based on AI detection and recognition according to claim 9, characterized in that: The specific process of calculating the crack angle hazard coefficient of the target belt is as follows: H1. Establish a rectangular coordinate system with the center point of one side edge of the target belt as the origin, the direction of the goods conveyed by the belt as the y-axis, and the direction perpendicular to the direction of the goods conveyed by the belt as the x-axis; H2. Locate the coordinates of the two endpoints of each crack corresponding to each region in the rectangular coordinate system and record them as and Wherein, w represents the crack number, w = 1, 2, ..., u; H3. Calculate the angle θ of each crack in each area rw , H4. Subtract the angle of each crack in each region from 0° to obtain the deviation angle between each crack in each region and 0°, which is recorded as the first deviation angle. If the first deviation angle of a crack in a certain region is less than the set first deviation angle, the crack is determined to be a first deviation crack. The number of first deviation cracks in each region is counted and recorded as H5. Subtract the angle of each crack in each region from 90° to obtain the deviation angle between each crack in each region and 90°, and record the absolute value of the deviation angle between each crack in each region and 90° as the second deviation angle. If the second deviation angle of a crack in a certain region is less than the set second deviation angle, the crack is determined to be a second deviation crack. The number of second deviation cracks in each region is counted and recorded as H6. Calculate the crack angle hazard coefficient ψ of the target belt. Where ζ′ represents the number of deviation cracks set as a reference.

Images