An image processing-based ring main unit production line quality management and control method

Abstract

The application relates to the technical field of image processing, in particular to a ring main unit production line quality control method based on image processing. The method comprises the following steps: obtaining a side gray image of a ring main unit to be detected, and determining a ring area with the outer contour of a target screw as an inner circle from the side gray image; determining a consistency index of the ring area of the target screw; obtaining a pixel probability distribution corresponding to the target screw, and determining a Jensen-Shannon divergence value between the pixel probability distribution to be detected and a standard pixel probability distribution; determining a loosening degree value of the target screw according to the consistency index of the ring area of the target screw and the Jensen-Shannon divergence value, and determining whether the target screw is loose according to the loosening degree value, so as to process the ring main unit to be detected with a screw on the side being loose. Through the above technical scheme, the fastening condition of the screw on the side of the ring main unit to be detected can be controlled.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of image processing, and particularly relates to a quality control method for ring main unit production line based on image processing. BACKGROUND

[0002] The ring main unit is a device used for power distribution and management in a power system, and can connect ring network lines, improve line power supply reliability, protect transformers and limit short-circuit current. For example, the ring main unit can quickly remove short-circuit faults inside the transformer within 10 milliseconds, thereby avoiding damage to the transformer due to sudden short-circuit fault current.

[0003] The ring main unit can be divided into a metal cabinet type ring main unit and a split interval type ring main unit. The high-voltage switch device in the metal cabinet type ring main unit is installed in a steel plate metal cabinet, has the characteristics of simple structure, small size and low price, and can adapt to various power supply scenarios. The split interval type ring main unit is composed of multiple interval units, is suitable for ring network power supply units, and has the characteristics of high flexibility and expandability.

[0004] The screws on the side of the ring main unit are mainly used for fastening connection, to ensure the structural stability and safety of the devices arranged in the ring main unit. The screws on the side of the ring main unit can effectively prevent the devices from loosening due to vibration or impact during operation, thereby ensuring the normal operation of the ring main unit.

[0005] In the production process of the ring main unit, there may be a situation that the screws are not tightened, or the screws are loosened due to vibration during transportation. The loosening of the screws will cause the connection between the parts of the ring main unit to be not firm, increase the vibration and impact during operation of the devices, and thus may cause short circuit or damage of the devices. Therefore, it is necessary to detect the fastening condition of the screws on the side of the ring main unit to be detected, to realize quality control of the ring main unit production line. SUMMARY

[0006] In order to control the quality of the ring main unit production line, the application provides a ring main unit production line quality control method based on image processing, which comprises the following steps: obtaining a side gray image of a ring main unit to be detected, determining a circular area where a target screw is located from the side gray image by using Hough circle transformation, and determining a ring area with the outer contour of the circular area of the target screw as the inner circle; dividing the ring area into a plurality of ring lines with the same center, and determining the consistency value of the pixel value of the pixel points on the same ring line relative to the adjacent ring line, taking the average value of the consistency values of different ring lines as the consistency index of the ring area of the target screw; obtaining the detection pixel probability distribution corresponding to the target screw according to the statistical result of the pixel value of the pixel points in the ring area, and determining the Jensen-Shannon divergence value between the detection pixel probability distribution and the standard pixel probability distribution; determining the loosening degree value of the target screw according to the consistency index of the ring area of the target screw and the Jensen-Shannon divergence value, and determining whether the target screw is loose according to the loosening degree value, so as to process the ring main unit to be detected whose side screw is loose.

[0007] In this way, the loose screw on the side of the ring main unit can be automatically detected, so as to realize the quality control of the production process of the ring main unit.

[0008] Optionally, the consistency value of the ring line is determined by the following method: for the target ring line in all ring lines of the ring area, the average value of the pixel value of the pixel points in the target ring line is taken as the feature value of the target ring line, and the consistency value of the target ring line is determined according to the difference between the feature values of the target ring line and the adjacent other ring lines.

[0009] In this way, the obtained consistency value can reflect the difference in gray feature between the ring line and the adjacent other ring lines.

[0010] Optionally, the Jensen-Shannon divergence value is determined by the following method: determining the average probability distribution of the detection pixel probability distribution and the standard pixel probability distribution; determining the first relative entropy between the detection pixel probability distribution and the average probability distribution, and determining the second relative entropy between the standard pixel probability distribution and the average probability distribution, and taking the average value of the first relative entropy and the second relative entropy as the Jensen-Shannon divergence value.

[0011] Optionally, the first relative entropy is determined by the following method: , wherein, is the first relative entropy, n is the number of gray interval in the detection pixel probability distribution, is the frequency ratio corresponding to the i-th pixel interval in the detection pixel probability distribution, ln is the logarithmic function with natural constant as the base, is the frequency ratio corresponding to the i-th pixel interval in the average probability distribution.

[0012] Optionally, the loosening degree value of the target screw is determined by the following method: obtaining a standard deviation of the consistency index corresponding to the predetermined qualified state of the screw, and determining a sum value of the standard deviation and a preset positive number, determining a ratio value of the consistency index of the annular region of the target screw and the sum value; processing the ratio value by using an exponential function with a natural constant as the base to obtain an exponential operation result, and taking a product of the reciprocal of the exponential operation result and the Jansen Shannon dispersion value as the loosening degree value of the target screw.

[0013] Optionally, the circular region where the target screw is located is determined by the following method: performing circular region detection on the side surface gray scale image by using Hough circle transformation, and screening the obtained multiple circular regions according to the size of the screw and the edge shape of the central region to determine different circular regions corresponding to different screws respectively.

[0014] Optionally, the processing of the ring main unit to be detected with loosened screws on the side surface includes: obtaining distance information of different operators to the ring main unit to be detected at the current time respectively, and assigning a target operator matched to the ring main unit to be detected according to the distance information; generating an operation task corresponding to the target operator; the operation task is used to instruct the target operator to tighten the screw with loosening on the side surface of the ring main unit to be detected.

[0015] In this way, the target operator matched to the ring main unit to be detected can be assigned according to the distance information, which helps to arrange the operator closer to the ring main unit to be detected to process the screw on the side surface of the ring main unit to be detected in priority.

[0016] Optionally, the method further includes: performing circular region detection on the side surface gray scale image by using Hough circle transformation, and screening the obtained multiple circular regions according to the size of the screw and the edge shape of the central region; determining the circular region after screening whose edge shape of the central region does not satisfy the characteristics of the screw as a circular region where the screw is missing in the ring main unit to be detected.

[0017] Optionally, whether the target screw is loosened is determined according to the loosening degree value, including: adding a vibration of a preset frequency to the side surface of the ring main unit to be detected, and determining a blurring degree value of the annular region of the target screw under the vibration; the blurring degree value is used to represent the blurring degree of the image region of the annular region under the influence of the vibration; taking a product of the normalized blurring degree value and the loosening degree value as an evaluation value, and determining whether the target screw is loosened according to the size relationship between the evaluation value and a preset evaluation threshold.

[0018] In this way, the vibration of the preset frequency is added to the side surface of the ring main unit to be detected, which can further expand the difference between the screw with loosening and the screw in fastening, so as to determine whether the target screw is loosened.

[0019] Optionally, the method further comprises: acquiring a to-be-detected assembly image of the inside of the ring main unit, and respectively matching a template image of different components required to be assembled with the to-be-detected assembly image to determine actual position distribution of the different components in the to-be-detected assembly image; in a case where the actual position distribution is different from the target position distribution, outputting prompt information; the prompt information is used to prompt that the assembly position of the components in the inside of the ring main unit is incorrect.

[0020] In this way, it is helpful to avoid that the assembly position of the components in the inside of the ring main unit is incorrect.

[0021] The technical scheme provided by the embodiment of the application can include the following beneficial effects: a side gray image of a to-be-detected ring main unit is acquired, and a ring area where a screw is located is determined from the side gray image; a Jansen-Shannon entropy value between a pixel probability distribution of the ring area and a standard pixel probability distribution corresponding to the fastening screw is determined; when the screw arranged on the side of the ring main unit is loose, the consistency of the ring area of the edge of the screw with the surface of the ring main unit becomes worse, and the pixel distribution probability of the ring area also changes accordingly, therefore, by using a looseness degree value determined according to the consistency index of the ring area of the target screw and the Jansen-Shannon entropy value, whether the target screw in the ring main unit is loose can be effectively determined, and the automatic detection of the screw looseness of the ring main unit is realized.

[0022] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the application. BRIEF DESCRIPTION OF DRAWINGS

[0023] Figure 1 is a flowchart of an image processing-based ring main unit production line quality control method according to an exemplary embodiment;

[0024] Figure 2 is a schematic diagram of a ring main unit in the embodiment of the application;

[0025] Figure 3 is a schematic diagram of a ring area of a certain screw of a ring main unit in the embodiment of the application;

[0026] Figure 4 is a schematic diagram of the change process of the gray value of different ring lines of the ring area with the radius in the embodiment of the application. DETAILED DESCRIPTION

[0027] First, a brief introduction to the application scenario of this application embodiment will be given. In the application scenario of this application, the screws provided on the side of the ring main unit can be used to fix the equipment or seals installed inside the ring main unit. If the screws are not tightened or become loose due to vibration during transportation, it will affect the use of the equipment installed inside the ring main unit. Therefore, it is necessary to detect whether the screws on the side of the ring main unit are loose in order to achieve quality control of the screw tightening operation of the ring main unit.

[0028] To address the aforementioned technical problems, this application provides a quality control method for ring main unit production lines based on image processing. Figure 1 This is a flowchart illustrating a quality control method for a ring main unit production line based on image processing, according to an exemplary embodiment. Figure 1 As shown, the method includes the following steps.

[0029] In step S101, a side grayscale image of the ring main unit to be inspected is obtained. The Hough circle transform is used to determine the circular region where the target screw is located from the side grayscale image, and an annular region with the outer contour of the circular region of the target screw as the inner circle is determined.

[0030] The side image of the ring main unit to be tested can be acquired through an image acquisition device. By processing the side image into grayscale, a grayscale image of the side of the ring main unit to be tested can be obtained. The image acquisition device can be positioned facing the side of the ring main unit to ensure that the features of the screws on the side of the ring main unit to be tested are presented more completely.

[0031] Figure 2 This is a schematic diagram of the ring main unit in the embodiments of this application, such as... Figure 2 As shown, the outer diameter of the screws on the side of the ring main unit is mainly circular, and the surface of the ring main unit is usually provided with corresponding recessed parts to avoid the surface of the screws protruding after they are screwed in. When the screws on the side of the ring main unit are in the tightened state, the screws are in close contact with the surface of the recessed parts of the ring main unit, so that the area around the edge of the screw shows the color characteristics of the ring main unit, and there is less shadow area at the edge of the screw.

[0032] When the screws on the side of the ring main unit are not tightened, there is a gap between the screw and the surface of the recessed part of the ring main unit, which makes the area around the edge of the screw appear as a shadowed ring area. The shadowed part of the screw edge is more. Therefore, by determining the grayscale image of the side of the ring main unit to be inspected, it is easy to detect the tightening status of the screws.

[0033] The Hough circle transform is an image processing algorithm based on parameter space mapping. The Hough circle transform realizes the recognition of a circular geometric feature by mapping a point in an image space to a three-dimensional parameter space. The Hough circle transform can be used in the fields of machine vision, industrial detection, medical imaging, and automatic driving.

[0034] Edge points in the image can be extracted by performing edge detection on the side gray image, the possible center and radius of each edge point in the three-dimensional parameter space can be calculated, and a vote can be performed in the accumulator of the Hough circle transform. By analyzing the local maximum value in the accumulator, the circular region existing in the side gray image can be determined. For details, refer to the existing art for the determination of the circular region by the Hough circle transform. The present embodiment will not be described here.

[0035] The difference between the radius of the outer circle of the annular region and the radius of the inner circle can be equal to a preset distance, and the preset distance can be between 20% and 50% of the diameter of the screw.

[0036] In an embodiment, the circular region where the target screw is located is determined by performing circular region detection on the side gray image by using the Hough circle transform, screening the obtained multiple circular regions according to the size of the screw and the edge shape of the central region, and determining different circular regions corresponding to different screws from the multiple circular regions.

[0037] For example, the Hough Circles function in OpenCV can be used to perform circular detection on the gray image to obtain the circular region in the side gray image. The diameter of the screw can be determined in advance. For all the detected circular regions, it can be judged whether the diameter is within the diameter range corresponding to the screw, so as to eliminate other circular regions that do not meet the screw to be detected outside the diameter range.

[0038] Since the obtained circular regions can include a circular region where the external shadow of the screw is located, and the circular region where the external shadow of the screw is located has the same center as the circular region of the outer boundary of the screw, for two or more circular regions with the same center determined, the circular region with the smallest diameter and within the diameter range corresponding to the screw can be retained.

[0039] The center of the screw usually has a T-shaped groove or a cross-shaped groove for rotating the screw. The T-shaped groove or the cross-shaped groove presents a relatively obvious edge feature at the center of the screw, and the center of other circular regions outside the circular region of the screw usually does not have a T-shaped groove or a cross-shaped groove. Therefore, the obtained multiple circular regions can also be screened in combination with the edge shape of the central region of the screw.

[0040] The obtained circular regions are filtered based on the screw size and the edge shape of the central area. The circular regions that are retained after filtering can be used as the outer boundary of the screw. Different screws correspond to different circular regions.

[0041] By performing circular region detection on the side grayscale image using Hough circular transform and then filtering the obtained circular regions to obtain the circular regions corresponding to screws, it is possible to automatically detect whether the screws on the side of the ring main unit are loose.

[0042] In step S102, the annular region is divided into multiple annular lines with the same center, and the consistency value of the pixel value of the pixel point on the same annular line is determined relative to the adjacent annular lines. The average value of the consistency values ​​of different annular lines is used as the consistency index of the annular region of the target screw.

[0043] For screws on the side of a ring main unit, screws that are tightened will have a uniform or gradually changing texture or light reflection on their outer side in the radial direction, resulting in a higher consistency value. For screws that are loose or not tightened to the end, the grayscale distribution between the ring lines in the ring area where the screw is located will change abruptly, resulting in a lower consistency value.

[0044] For example, for a screw that is tightened on the side of a ring main unit, the annular area of ​​the screw may only include the surface of the ring main unit; for a screw that is not tightened on the side of a ring main unit, the annular area of ​​the screw may include both the surface of the ring main unit and the shadow formed by the gap between the loose screw and the ring main unit. Therefore, using consistency indicators can make it easy to determine whether the target screw is tightened.

[0045] Figure 3 This is a schematic diagram of the annular region of a screw in a ring main unit in an embodiment of this application. Figure 3 An example of the annular region of a screw in a tightened state is shown; to illustrate the annular region more intuitively, in Figure 3 The pixels outside the circular area are displayed as black.

[0046] When the screws are tightened, they fit more closely to the surface of the ring main unit, so that the annular area outside the outer contour of the screw includes the surface area of ​​the ring main unit. Figure 3 The white outline of the annular area corresponds to the area on the surface of the ring main unit; the annular area includes multiple ring lines with the same center. By utilizing the characteristics of the different ring lines in the annular area, the tightening status of the screws can be reflected.

[0047] In one embodiment, the consistency value of the annular line is determined by: for a target annular line in all annular lines of the annular region, taking the average value of the pixel values of the pixel points in the target annular line as the characteristic value of the target annular line, and determining the consistency value of the target annular line according to the difference in the characteristic values between the target annular line and the adjacent other annular lines.

[0048] The target annular line can be any one of all annular lines of the annular region, and all the annular lines collectively form the annular region corresponding to the screw; the average value of the pixel values of the pixel points in the target annular line as the characteristic value of the target annular line can reflect the overall gray value of the pixel points on the target annular line.

[0049] The difference in the characteristic values between the target annular line and the adjacent other annular lines may, for example, be equal to the absolute value of the difference between the characteristic value of the target annular line and the characteristic value of the adjacent other annular lines; the greater the difference in the characteristic values between the target annular line and the adjacent other annular lines, the greater the difference in the overall gray value between the target annular line and the adjacent other annular lines, and the more likely the target annular line is to be in the annular line where the screw is not tightened.

[0050] The difference in the characteristic values between the target annular line and the adjacent other annular lines is negatively correlated with the consistency value of the target annular line; for example, the consistency value of the target annular line can be equal to the inverse of the difference in the characteristic values between the target annular line and the adjacent other annular lines.

[0051] When the consistency value of the target annular line is determined according to the difference in the characteristic values between the target annular line and the adjacent other annular lines, the comparison object is the annular line adjacent to the target annular line, and the consistency value can describe the more subtle features of the annular region where the screw is located.

[0052] Figure 4 A schematic diagram of the change process of the gray values of different annular lines of the annular region in the embodiments of the present application is shown in FIG. 1, where the average gray values corresponding to the annular lines with different radii of the annular region are different, but the average gray values of the pixel points in the annular lines are mainly concentrated in the gray value range of 196 to 208, which is consistent with the gray value features of the surface of the ring net cabinet. Figure 4

[0053] In step S103, the target screw corresponding to the to-be-detected pixel probability distribution is obtained according to the statistical result of the pixel values of the pixel points in the annular region, and the Jensen-Shannon divergence value between the to-be-detected pixel probability distribution and the standard pixel probability distribution is determined.

[0054] ​All pixel points in the annular region can be traversed, and the gray value of each pixel is recorded. The gray values of all pixel points in the annular region are histogram counted, and the gray values are divided into several intervals. For example, 256 gray levels from 0 to 255 can be divided into 16 or 32 gray intervals. According to the proportion of the number of pixel points falling into the gray interval in the total number of all pixel points in the annular region, the probability value corresponding to the gray interval is determined.

[0055] The probability distribution of the to-be-detected pixels includes probability values respectively corresponding to different gray intervals of the annular region of the target screw to be detected. The probability distribution of the to-be-detected pixels can be used to describe the gray distribution of the pixel points in the annular region.

[0056] Referring to the determination manner of the probability distribution of the to-be-detected pixels of the target screw to be detected, the standard probability distribution of the annular region of the screw in the fastened state can be determined in advance. The standard probability distribution can be used to describe the gray feature of the annular region where the outer side of the screw in the fastened state is located.

[0057] In one embodiment, the Jensen-Shannon divergence value is determined by determining the average probability distribution of the probability distribution of the to-be-detected pixels and the standard probability distribution; determining the first relative entropy between the probability distribution of the to-be-detected pixels and the average probability distribution, and determining the second relative entropy between the standard probability distribution and the average probability distribution, and taking the average value of the first relative entropy and the second relative entropy as the Jensen-Shannon divergence value.

[0058] The average probability distribution of the probability distribution of the to-be-detected pixels and the standard probability distribution can measure the symmetric difference between the probability distribution of the to-be-detected pixels and the standard probability distribution. In the obtained average probability distribution, the probability value corresponding to the gray interval is equal to the average value of the probability values of the same gray interval in the probability distribution of the to-be-detected pixels and the standard probability distribution.

[0059] The first relative entropy between the probability distribution of the to-be-detected pixels and the average probability distribution can reflect the difference between the probability distribution of the to-be-detected pixels and the average probability distribution in the overall distribution shape.

[0060] The first relative entropy between the standard probability distribution and the average probability distribution can reflect the difference between the standard probability distribution and the average probability distribution in the overall distribution shape.

[0061] The average value of the first relative entropy and the second relative entropy can reflect the difference between the probability distribution of the to-be-detected pixels and the standard probability distribution in the overall.

[0062] When the target screw of the ring main unit is in the fastened state, the target screw and the surface of the ring main unit are in the adhering state, so that the outside of the screw in the obtained image exhibits a relatively consistent gray value, so that the pixel probability distribution of the annular area of the screw in the fastened state is mainly concentrated in the gray level corresponding to the smaller number of categories of the surface of the ring main unit.

[0063] When the target screw of the ring main unit is not in the fastened state, the target screw and the surface of the ring main unit are in the separated state, so that the outside of the screw in the obtained image exhibits a relatively diverse gray value, so that the pixel probability distribution of the annular area of the screw is more dispersedly distributed in the gray level corresponding to the larger number of categories of the surface of the ring main unit and the possible shadow.

[0064] In one embodiment, the first relative entropy is determined by: wherein, is the first relative entropy, n is the number of gray interval in the pixel probability distribution to be detected, is the frequency ratio corresponding to the i-th pixel interval in the pixel probability distribution to be detected, and ln is the logarithmic function with the natural constant as the base, is the frequency ratio corresponding to the i-th pixel interval in the average probability distribution.

[0065] In the calculation formula of the first relative entropy, the greater the value of the same gray interval in the pixel probability distribution to be detected than in the average probability distribution, the more concentrated the gray interval in the pixel probability distribution to be detected, the greater the difference between the pixel probability distribution to be detected and the average probability distribution at the gray interval, and The greater the value of the same gray interval in the pixel probability distribution to be detected than in the average probability distribution, the greater the weight value of occupies in the summation result, which can improve the contribution of the gray interval with greater difference to the first relative entropy.

[0066] For example, if the pixel probability distribution to be detected is (0.1, 0.1, 0.2, 0.2, 0.2, 0.2), the standard pixel probability distribution is (0.02, 0.03, 0.05, 0.15, 0.35, 0.4), and the average probability distribution of the pixel probability distribution to be detected and the standard pixel probability distribution is (0.06, 0.065, 0.125, 0.175, 0.275, 0.3).

[0067] The first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution is 0.1*ln(0.1 / 0.06)+0.1*ln(0.1 / 0.065)+0.2*ln(0.2 / 0.125)+0.2*ln(0.2 / 0.175)+0.2*ln(0.2 / 0.275)+0.2*ln(0.1 / 0.3)=0.0511+0.0431+0.00940-0.00637-0.0811=0.00701.

[0068] When the to-be-detected pixel probability distribution and the average probability distribution are not completely the same, the first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution is greater than 0; when the to-be-detected pixel probability distribution and the average probability distribution are completely the same, the first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution is equal to 0.

[0069] The to-be-detected pixel probability distribution corresponding to the annular region of the target screw to be detected is usually not completely the same as the standard pixel probability distribution corresponding to the annular region of the screw in the fastened state, so that the to-be-detected pixel probability distribution and the average probability distribution are not completely the same, and thus the first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution is greater than 0.

[0070] The greater the first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution, the greater the difference between the to-be-detected pixel probability distribution and the average probability distribution, and the greater the probability that the to-be-detected pixel probability distribution and the standard pixel probability distribution are different.

[0071] With reference to the obtaining step of the first relative entropy, the second relative entropy is determined in the following manner: , wherein is the second relative entropy, n is the number of gray interval in the standard pixel probability distribution, is the frequency ratio corresponding to the i-th pixel interval in the standard pixel probability distribution, ln is a logarithmic function with a natural constant as the base number, is the frequency ratio corresponding to the i-th pixel interval in the average probability distribution.

[0072] In the obtained first relative entropy or second relative entropy, the probability values of the gray intervals with different contribution degrees are weighted, that is, the probability values of the gray intervals with different degrees of difference in the same interval in the average probability distribution have different contributions to the first relative entropy, so that the obtained Jensen-Shannon divergence can amplify the key difference between the to-be-detected pixel probability distribution and the standard pixel probability distribution, so as to determine whether the target screw to be detected is loose.

[0073] In step S104, the loosening degree value of the target screw is determined according to the consistency index of the annular region of the target screw and the Jensen-Shannon divergence value, and whether the target screw is loose is determined according to the loosening degree value, so as to process the ring main unit whose side surface has a screw in a loose state.

[0074] The consistency index of the annular region of the target screw can represent the color difference of different annular lines of the target screw, and the screw in the fastened state has a higher consistency index. The Jensen-Shannon divergence value of the annular region of the target screw can reflect the difference in gray scale distribution between the annular region of the target screw and the screw in the fastened state. The combination of the consistency index and the Jensen-Shannon divergence value can facilitate the judgment of the fastened state of the target screw.

[0075] In one embodiment, the loosening degree value of the target screw is determined by the following method: obtaining the standard deviation of the consistency index corresponding to the screw in the qualified state, determining the sum of the standard deviation and a preset positive number, determining the ratio of the consistency index of the annular region of the target screw to the sum, using an exponential function with a natural constant as the base to process the ratio to obtain an exponential operation result, and taking the product of the reciprocal of the exponential operation result and the Jensen-Shannon divergence value as the loosening degree value of the target screw.

[0076] The standard deviation of the consistency index corresponding to the screw in the qualified state can reflect the fluctuation range of the screw in the fastened state in the consistency index. Determining the sum of the standard deviation and a preset positive number can ensure that the obtained sum is at least greater than 0. Determining the ratio of the consistency index of the annular region of the target screw to the sum can realize the normalization processing of the consistency index, so as to compare and calculate between variables.

[0077] The value of the reciprocal of the exponential operation result is greater than 0 and less than 1, and the reciprocal of the exponential operation result is negatively correlated with the consistency index. The smaller the consistency index corresponding to the target screw is, the greater the reciprocal of the exponential operation result is, the higher the degree of abnormality of the fastened state of the target screw is, and the greater the probability of the target screw in the loose state is.

[0078] By determining the loosening degree value of the target screw, the automatic judgment of the loosening degree of the target screw can be realized, so as to process the loose screw that may exist on the side surface of the ring main unit, and ensure the stability of the installation of the equipment in the ring main unit.

[0079] The following takes a formula as an example to more intuitively introduce the determination process of the loosening degree value in the embodiment of the application: , S is the loosening degree value of the target screw, T is the Jensen-Shannon divergence value, e is a natural constant, and C is the consistency index of the annular region of the target screw. The standard deviation of the consistency index corresponding to the screw in the qualified state.

[0080] With reference to the step of determining whether the target screw is loose, it can be determined whether different screws in the ring main unit are loose to determine the screws on the side of the ring main unit that are loose.

[0081] In one embodiment, the determination of whether the target screw is loose according to the looseness degree value includes: adding a vibration of a preset frequency to the side of the ring main unit to be detected, and determining a blur degree value of the annular region of the target screw under the vibration; the blur degree value is used to represent the blur degree of the image region of the annular region under the influence of the vibration; the product of the normalized blur degree value and the looseness degree value is taken as an evaluation value, and whether the target screw is loose is determined according to the size relationship between the evaluation value and a preset evaluation threshold.

[0082] For the screw on the side of the ring main unit in the fastened state, the screw is closely attached to the surface of the ring main unit, and when the side of the ring main unit is vibrated, the screw in the fastened state will vibrate together with the side of the ring main unit, or the vibration amplitude generated by the screw in the fastened state is smaller, so that the determined annular region is clearer in the image.

[0083] For the screw on the side of the ring main unit in the loose state, there is a gap between the screw and the surface of the ring main unit, and when the side of the ring main unit is vibrated, the vibration amplitude generated by the screw in the fastened state is greater than the vibration of the side of the ring main unit, so that the determined annular region is more blurred in the image.

[0084] In order to avoid the influence of the added vibration on the equipment installed in the interior of the ring main unit, the vibration of the preset frequency can be smaller than the minimum amplitude that causes the equipment installed in the interior of the ring main unit to vibrate, and the preset frequency can be a frequency other than the frequency that can cause resonance.

[0085] Adding a vibration of a preset frequency to the side of the ring main unit to be detected can further expand the difference between the fastened screw and the loose screw, and can facilitate more accurate determination of the screw that is loose.

[0086] The blur degree value of the annular region of the target screw under the vibration can be determined by the structural similarity between the two image regions before and after the annular region is vibrated, and the smaller the structural similarity, the higher the degree of influence of the vibration on the annular region, and the greater the blur degree value of the annular region under the vibration.

[0087] Alternatively, the blur degree value can be determined by the mean square error of the annular region between the two image regions before and after being vibrated, the greater the mean square error between the two image regions, the higher the degree of influence on the annular region, and the greater the blur degree value of the annular region under vibration.

[0088] Here, the annular region of the fastened screw is mainly the more single texture of the surface of the ring main unit, and the annular region of the fastened screw still shows the surface texture of the ring main unit when the fastened screw is vibrated, so that the blur degree value obtained by the fastened screw is smaller.

[0089] The annular region of the loose screw is a more complex texture including the surface of the ring main unit and the shadow region, and the shadow of the annular region of the loose screw will change when the loose screw is vibrated, and the original texture will show more diversity under vibration, so that the blur degree value obtained by the loose screw is greater. Therefore, the difference between the fastened screw and the loose screw can be further expanded by the blur degree value to identify the loose screw on the side surface of the ring main unit.

[0090] When the evaluation value is greater than the preset evaluation threshold, it indicates that the characteristics of the target screw are more consistent with the characteristics of the loose screw, and it can be determined that the target screw is loose. On the contrary, when the evaluation value is less than or equal to the preset evaluation threshold, it indicates that the characteristics of the target screw are more consistent with the characteristics of the fastened screw, and it can be determined that the target screw is not loose. The preset evaluation threshold can be set according to actual needs. For example, when the evaluation value is normalized, the value of the preset evaluation threshold can be between 0.2 and 0.5.

[0091] In one embodiment, the ring main unit to be detected with a loose screw on the side surface is processed, including: respectively acquiring distance information of different operators to the ring main unit at the current time, and assigning a target operator matched to the ring main unit to be detected according to the distance information; generating an operation task corresponding to the target operator; the operation task is used to instruct the target operator to tighten the screw that is loose on the side surface of the ring main unit to be detected.

[0092] Respectively acquiring distance information of different operators to the ring main unit at the current time can preferentially assign operators closer to the ring main unit to be detected, and preferentially assign operators closer to the ring main unit to be detected among multiple operators as target operators, and generate operation tasks corresponding to the target operators, which can avoid interference with other operators, and can also help to process the ring main unit to be detected in time.

[0093] The different operators can be, for example, a device or a person capable of tightening a loose screw on the side of the ring main unit to be detected; for example, the generated operation task can be sent to a terminal device bound to the operator, so as to prompt the operator to perform the corresponding operation task through the terminal device.

[0094] The operation task can include position information, model information and other identification information of the ring main unit to be detected, so that the operator can process the loose screw in the ring main unit to be detected according to the indication of the operation task.

[0095] In an embodiment, the side gray-scale image can also be subjected to circular region detection by using Hough circle transformation, and the obtained multiple circular regions are screened according to the size of the screw and the edge shape of the central region; the circular region whose edge shape of the screened central region does not satisfy the characteristics of the screw is determined as a circular region in which the screw in the ring main unit to be detected is missing.

[0096] According to the fact that the central region of the screw usually has a cross, a single or an internal hexagonal edge shape, and the size of the screw is usually within a predetermined range; for the circular region whose edge shape of the screened central region does not satisfy the characteristics of the screw, the circular region has a greater probability of being a mounting hole reserved for screw installation, and the corresponding position has not been installed with a screw.

[0097] In this way, by determining the circular region in which the screw in the ring main unit to be detected is missing, the user can conveniently supplement the screw in the region in which the screw is missing.

[0098] In an embodiment, an inside detection assembly image of the ring main unit to be detected can also be obtained, and the template image of each different component to be assembled is matched with the inside detection assembly image to determine the actual position distribution of each different component in the inside detection assembly image; if there is a difference between the actual position distribution and a target position distribution, a prompt information is output; the prompt information is used to prompt that the assembly position of the component in the inside of the ring main unit is wrong.

[0099] Different components have different appearance characteristics, and corresponding template images can be made in advance for different components, different template images are matched with the inside detection assembly image, if there is an image region in the inside detection assembly image that matches the template image, it can be determined that the corresponding component is installed at the corresponding position in the inside of the ring main unit to be detected.

[0100] When the operator places or installs the components inside the ring net cabinet, the components can be placed or installed to the wrong position. By determining the actual position distribution of different components in the detection assembly image and comparing the actual position distribution with the target position distribution, the position installation error of the components can be avoided, so that the safety of the ring net cabinet and the components arranged inside the ring net cabinet is ensured.

[0101] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope of the application being indicated by the following claims.

[0102] It should be understood that the application is not limited to the precise construction that has been described above and shown in the accompanying drawings, and that various modifications and changes can be made by those skilled in the art without departing from the scope of the application.

Claims

1. An image processing-based quality management method for a ring main unit production line, characterized by, The method comprises the following steps: acquiring a side gray image of a ring main unit to be detected, determining a circular region where a target screw is located from the side gray image by using a Hough circle transform, and determining a ring region with an outer contour of the circular region of the target screw as an inner circle; dividing the ring region into a plurality of ring lines with the same center, and determining a consistency value of pixel values of pixel points on the same ring line relative to adjacent ring lines, taking an average value of the consistency values of different ring lines as a consistency index of the ring region of the target screw; obtaining a to-be-detected pixel probability distribution corresponding to the target screw according to a statistical result of the pixel values of the pixel points in the ring region, and determining a Jensen-Shannon divergence value between the to-be-detected pixel probability distribution and a standard pixel probability distribution; determining a looseness degree value of the target screw according to the consistency index of the ring region of the target screw and the Jensen-Shannon divergence value, comprising: acquiring a standard deviation of the consistency index corresponding to a screw in a qualified state determined in advance, determining a sum value of the standard deviation and a preset positive number, determining a ratio of the consistency index of the ring region of the target screw to the sum value, processing the ratio by using an exponential function with a natural constant as a base to obtain an exponential operation result, and taking a product of an inverse of the exponential operation result and the Jensen-Shannon divergence value as the looseness degree value of the target screw, and determining whether the target screw is loose according to the looseness degree value, so as to process the ring main unit to be detected in which the screw on the side is loose.

2. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, The consistency value of the ring line is determined in the following manner: for a target ring line in all ring lines of the ring region, taking an average value of the pixel values of the pixel points in the target ring line as a characteristic value of the target ring line, and determining the consistency value of the target ring line according to the difference between the characteristic values of the target ring line and other adjacent ring lines.

3. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, The Jensen-Shannon divergence value is determined in the following manner: determining an average probability distribution of the to-be-detected pixel probability distribution and the standard pixel probability distribution; determining a first relative entropy between the to-be-detected pixel probability distribution and the average probability distribution, and a second relative entropy between the standard pixel probability distribution and the average probability distribution, and taking an average value of the first relative entropy and the second relative entropy as the Jensen-Shannon divergence value.

4. The image processing-based quality management method for ring main unit production lines according to claim 3, characterized in that, The first relative entropy is determined in the following manner: wherein, is the first relative entropy, n is the number of gray interval in the pixel probability distribution to be detected, is the frequency ratio corresponding to the i-th pixel interval in the pixel probability distribution to be detected, and ln is the logarithmic function with the natural constant as the base number, is the frequency ratio corresponding to the i-th pixel interval in the average probability distribution.

5. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, The circular region where the target screw is located is determined in the following manner: performing circular region detection on the side gray image by using a Hough circle transform, screening a plurality of circular regions obtained according to the size of the screw and the edge shape of the central region, and determining different circular regions corresponding to different screws from the plurality of circular regions.

6. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, Processing the ring main unit to be detected in which the screw on the side is loose comprises: acquiring distance information of different operators to the ring main unit to be detected at a current time, and assigning a target operator matched to the ring main unit to be detected according to the distance information; generating an operation task corresponding to the target operator; the operation task is used to instruct the target operator to tighten the screw loose on the side of the ring main unit to be detected.

7. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, The method further comprises: The side gray image is detected by using the Hough circle transformation, and the multiple circular regions obtained are screened according to the size of the screw and the edge shape of the central region; The circular region whose edge shape of the central region after screening does not satisfy the characteristics of the screw is determined as a circular region where the screw in the ring main unit to be detected is missing.

8. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, According to the loosening degree value, it is determined whether the target screw is loose, including: Adding vibration of a preset frequency to the side of the ring main unit to be detected, and determining the blur degree value of the annular region of the target screw under vibration; the blur degree value is used to represent the blur degree of the image region of the annular region under the influence of vibration; The product of the normalized blur degree value and the loosening degree value is used as the evaluation value, and according to the size relationship between the evaluation value and the preset evaluation threshold, it is determined whether the target screw is loose.

9. The image processing-based quality management method for ring main unit production lines according to claim 1, characterized in that, The method further includes: Obtaining a to-be-detected assembly image of the inside of the ring main unit to be detected, and matching the template image of each different component required to be assembled with the to-be-detected assembly image to determine the actual position distribution of each different component in the to-be-detected assembly image; In the case where there is a difference between the actual position distribution and the target position distribution, output a prompt information; the prompt information is used to prompt that the assembly position of the component in the inside of the ring main unit is wrong.

Images