A method and system for intelligent detection of defects for coating a core

Abstract

The application relates to a kind of intelligent detection method and system for coating core defects, and to the field of product quality detection technology, which comprises obtaining workpiece detection image;Determine the workpiece center point and workpiece feature point in the workpiece detection image, and draw a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius;Define the workpiece feature points on the virtual detection contour as online feature points, and determine the adjacent feature distance on the virtual detection contour according to the adjacent online feature points in the preset detection direction;Count according to different values of adjacent feature distance to determine the number of feature distance;If there is at least one virtual detection contour, the number of feature distances determined is greater than the number of permitted distances, then output the workpiece defect signal.The application has the function of improving the overall detection effect of the coated core.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of product quality detection technology, in particular to a defect intelligent detection method and system for a coated core. BACKGROUND

[0002] The coated core is a core component of a motor, a transformer and various inductive elements, which is usually composed of silicon steel sheets and coated with an insulating coating on the surface. The main function of the coating is to reduce the eddy current loss of the core, prevent short circuit between sheets, and play a role in rust prevention and corrosion resistance.

[0003] At present, in industrial production, the surface of the coated core is widely detected by using an automatic detection system based on machine vision. Since the defect type does not need to be determined under some working conditions, only whether the defect exists is needed, the most traditional "standard image comparison" technology is introduced. The basic principle is that: firstly, a standard coated core image without defects is obtained as a template; then, in the detection process, the shooting image of the workpiece to be detected is compared with the standard template image at the pixel level or the feature level, and whether the workpiece to be detected has defects is judged by analyzing the differences in gray scale, color or texture between the two.

[0004] In the related technology described above, since the defects that may exist on the coated core include small problems such as uneven coating thickness, it is difficult to compare the whole workpiece image, and in actual production, the process parameters of the coated core will be dynamically adjusted due to changes in product demand or external environment, so the standard image needs to be frequently replaced, resulting in poor overall detection effect, and there is still room for improvement. SUMMARY

[0005] In order to improve the overall detection effect of the coated core, the present application provides a defect intelligent detection method and system for a coated core.

[0006] In the first aspect, the present application provides a defect intelligent detection method for a coated core, which adopts the following technical scheme:

[0007] A defect intelligent detection method for a coated core, comprising:

[0008] obtaining a workpiece detection image;

[0009] determining a workpiece center point and a workpiece feature point in the workpiece detection image, and defining a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius;

[0010] defining the workpiece feature point on the virtual detection contour as an online feature point, and determining an adjacent feature distance on the virtual detection contour according to adjacent online feature points in a preset detection direction.

[0011] counting the adjacent feature distances according to different numerical values to determine the number of feature distances;

[0012] determining whether the number of feature distances determined on at least one virtual detection contour is greater than a preset permitted distance number;

[0013] outputting a workpiece qualified signal if the number of feature distances determined on at least one virtual detection contour is not greater than the permitted distance number;

[0014] outputting a workpiece defect signal if the number of feature distances determined on at least one virtual detection contour is greater than the permitted distance number.

[0015] Optionally, the intelligent defect detection method for coating the core further comprises the following steps after the adjacent feature distances are determined:

[0016] counting the adjacent feature distances according to the same numerical value to determine the number of single-point distances;

[0017] determining the maximum number of single-point distances according to a preset sorting rule, and defining the adjacent feature distance corresponding to the maximum number of single-point distances as a representative feature distance;

[0018] constructing a representative distance range according to the representative feature distance and a preset permitted error distance, and converting the adjacent feature distances within the representative distance range into current representative feature distances;

[0019] continuing to determine the representative feature distance according to the remaining adjacent feature distances after the adjacent feature distances are converted, until all the adjacent feature distances are within the constructed representative distance range.

[0020] Optionally, the method further comprises a step of determining the permitted error distance, which comprises the following steps:

[0021] constructing a historical interval with a current time point as a rear end point and a preset historical time length as a width on a preset time axis, and defining the representative distance range of the current detection distance as a reference distance range according to the historical interval;

[0022] determining internal interval distances according to the adjacent feature distances and the representative feature distances in the reference distance range, and defining the maximum internal interval distance as an upper limit interval distance;

[0023] randomly selecting one of all the upper limit interval distances as a main interval distance, and defining the remaining upper limit interval distances as secondary interval distances;

[0024] The main representative coefficient is determined according to the main distance and all the secondary distances, and the main distance corresponding to the maximum main representative coefficient is defined as the reasonable distance;

[0025] The permitted error distance is determined by summing the reasonable distance and the preset fixed adjustment distance.

[0026] Optionally, the method further comprises a random determination step of the detection distance, which comprises:

[0027] The detection distance corresponding to the output workpiece defect signal in the historical interval is defined as the abnormal distance;

[0028] The distance interval value is determined according to the abnormal distance and the detection distance, and the priority coefficient of each detection distance under the distance interval value is determined according to the preset priority matching relationship;

[0029] The single-point representative coefficient is determined by summing all the priority coefficients under the same detection distance;

[0030] The detection distances are sorted from front to back according to the single-point representative coefficients from large to small to determine the random generation order, and the detection distances are determined in sequence according to the random generation order.

[0031] Optionally, after the priority coefficient is determined, the intelligent defect detection method for coating the iron core further comprises:

[0032] The reference type set corresponding to the abnormal distance and the comparison type set corresponding to the detection distance are determined according to the preset type matching relationship;

[0033] The common type is determined according to the reference type set and the comparison type set, and the common quantity is determined by counting according to the common type;

[0034] The reference quantity is determined by counting according to the reference type set, and the type common proportion is determined by calculating according to the common quantity and the reference quantity;

[0035] The key adjustment coefficient corresponding to the type common proportion is determined according to the preset adjustment matching relationship, and the priority coefficient is updated by calculating according to the key adjustment coefficient and the priority coefficient.

[0036] Optionally, after the workpiece defect signal is output, the intelligent defect detection method for coating the iron core further comprises:

[0037] The virtual detection contour corresponding to the output workpiece defect signal is defined as the abnormal detection contour;

[0038] a discard profile with a preset detection distance is randomly drawn on the anomaly detection profile, and a region outside the discard profile on the anomaly detection profile is defined as a reserved profile;

[0039] analysis is performed according to the reserved profile to determine whether there is at least one reserved profile outputting a workpiece qualification signal;

[0040] If there is no at least one reserved profile outputting a workpiece qualification signal, a discard profile is added until there is at least one reserved profile outputting a workpiece qualification signal;

[0041] If there is at least one reserved profile outputting a workpiece qualification signal, the corresponding reserved profile is defined as a qualified profile, and the discard profile corresponding to the qualified profile is defined as a defect profile;

[0042] The defect profiles with common regions are fitted to construct a local profile, and the local defect region is determined according to the local profile.

[0043] Optionally, the step of determining the local defect region according to the local profile comprises:

[0044] A local center is determined in the local profile, and a local coverage region is determined with the local center as the center and a preset coverage distance as the radius;

[0045] The local coverage region containing the current local profile is defined as an effective coverage region, and the internal local number is determined according to the local profile in the effective coverage region;

[0046] The internal local number with the largest value is determined according to the sorting rule, and the effective coverage region corresponding to the internal local number is determined as the local defect region.

[0047] In a second aspect, the application provides a defect intelligent detection system for coating an iron core, which adopts the following technical scheme:

[0048] A defect intelligent detection system for coating an iron core, comprising:

[0049] An acquisition module for acquiring a workpiece detection image;

[0050] A processing module connected with the acquisition module and the judgment module for storing and processing information;

[0051] A judgment module connected with the acquisition module and the processing module for information judgment;

[0052] The processing module determines a workpiece center point and a workpiece feature point in the workpiece detection image, and draws a virtual detection profile with the workpiece center point as the center and a random detection distance as the radius;

[0053] The processing module defines the feature points of the workpiece on the virtual detection contour as online feature points, and determines the adjacent feature distance according to adjacent online feature points on the virtual detection contour in a preset detection direction;

[0054] The processing module counts the adjacent feature distances according to different numerical values to determine the feature distance quantity;

[0055] The judgment module judges whether the feature distance quantity determined on at least one virtual detection contour is greater than the preset permitted distance quantity;

[0056] If the judgment module judges that the feature distance quantity determined on at least one virtual detection contour is not greater than the permitted distance quantity, the processing module outputs a workpiece qualified signal;

[0057] If the judgment module judges that the feature distance quantity determined on at least one virtual detection contour is greater than the permitted distance quantity, the processing module outputs a workpiece defect signal.

[0058] In summary, the present application includes at least one of the following beneficial technical effects:

[0059] In the process of detecting defects of the coated iron core, the defects can be determined by the multi-point mutual comparison monitoring method based on the shape of the coated iron core, thereby improving the overall detection effect of the coated iron core;

[0060] In the process of detection, the area with a high probability of defects can be detected preferentially, so that the workpiece with defects can be quickly determined, and the overall detection efficiency is improved. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 is a flowchart of the intelligent defect detection method for the coated iron core.

[0062] Figure 2 is a schematic diagram of a workpiece detection image of the coated iron core.

[0063] Figure 3 is a module flowchart of the intelligent defect detection method for the coated iron core. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical scheme and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. Figures 1-3 The specific embodiments described herein are only used to explain the present application and do not limit the present application.

[0065] The embodiments of the present application will be further described in detail below with reference to the accompanying drawings.

[0066] The embodiment of the application discloses a method for intelligent detection of defects of a coated iron core Figure 1 The method flow of the method for intelligent detection of defects of a coated iron core comprises the following steps:

[0067] Step S100: acquire a workpiece detection image.

[0068] The workpiece detection image is an image acquired when the coated iron core is placed on a fixed detection area by an image shooting device, which can be referred to as Figure 2 .

[0069] Step S101: determine a workpiece center point and a workpiece feature point in the workpiece detection image, and draw a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius.

[0070] The workpiece center point is a center point of the iron core when the workpiece is placed, which can be referred to as Figure 2 , and the workpiece center point can be fixed because the position of the iron core in the detection area is fixed; the workpiece feature point is a position point of the region where the workpiece is located, which can be referred to as Figure 2 ; the detection distance is a randomly generated value, and the value can be generated in a range set by a worker in advance, and the virtual detection contour is a contour line of a circle drawn with the workpiece center point as the center and the detection distance as the radius.

[0071] Step S102: define the workpiece feature point on the virtual detection contour as an online feature point, and determine an adjacent feature distance on the virtual detection contour according to adjacent online feature points in a preset detection direction.

[0072] The online feature point is defined to distinguish different workpiece feature points, which is convenient for subsequent analysis; the detection direction is a counterclockwise direction or a clockwise direction set by the worker, and the adjacent feature distance is a distance value of adjacent online feature points on the virtual detection contour in the detection direction.

[0073] Step S103: count the adjacent feature distances of different values to determine a feature distance quantity.

[0074] The feature distance quantity is the quantity of the adjacent feature distances of different values.

[0075] Step S104: determine whether there is at least one feature distance quantity determined on the virtual detection contour greater than a preset permitted distance quantity.

[0076] The maximum number of allowable distances is the maximum number of allowable distances allowed to appear when the arrangement of feature points on the identified virtual detection contour is normal, and the corresponding number of allowable distances is 2 in this application. The purpose of the judgment is to know whether there is a defect that leads to different adjacent feature distances, that is, to judge whether the current core has a defect.

[0077] Step S1041: If the number of feature distances determined on at least one virtual detection contour is greater than the number of allowable distances, output a workpiece qualified signal.

[0078] When the number of feature distances determined on at least one virtual detection contour is greater than the number of allowable distances, it means that the workpiece in all regions is normal, that is, there is no defect, at this time, the workpiece qualified signal is output to identify this situation.

[0079] Step S1042: If the number of feature distances determined on at least one virtual detection contour is greater than the number of allowable distances, output a workpiece defect signal.

[0080] When the number of feature distances determined on at least one virtual detection contour is greater than the number of allowable distances, it means that the current core has at least one region with a defect, so the workpiece defect signal is output to identify this situation, which is convenient for subsequent processing of the workpiece.

[0081] After the adjacent feature distance is determined, the intelligent defect detection method for coating the core further comprises:

[0082] Step S200: Counting the adjacent feature distances with the same value to determine the number of single-point distances.

[0083] The number of single-point distances is the number of adjacent feature distances with the same value.

[0084] Step S201: Determine the maximum number of single-point distances according to the preset sorting rule, and define the adjacent feature distance corresponding to the number of single-point distances as the representative feature distance.

[0085] The sorting rule is a method set by the staff to sort the values, such as bubble method. Through the sorting rule, the maximum number of single-point distances can be determined, that is, the adjacent feature distance corresponding to the number of single-point distances appears the most times and is most consistent with the normal feature interval requirements of the current workpiece. Therefore, it is defined as the representative feature distance to distinguish different adjacent feature distances for subsequent analysis.

[0086] Step S202: Construct a representative distance range according to the representative feature distance and the preset allowable error distance, and convert the adjacent feature distance within the representative distance range to the current representative feature distance.

[0087] The permitted error distance is an error value of the distance determined by the staff under the same condition, and the two end points of the representative distance range can be determined by adding and subtracting the permitted error distance to the representative characteristic distance respectively; at this time, the adjacent characteristic distance within the representative distance range is converted into the current representative characteristic distance to uniformly summarize the data under the same condition, thereby improving the accuracy of data analysis.

[0088] Step S203: After the adjacent characteristic distance is converted, the representative characteristic distance is determined according to the remaining adjacent characteristic distance until all the adjacent characteristic distances are within the constructed representative distance range.

[0089] By continuously converting the representative characteristic distance, all the adjacent characteristic distances can be uniformly summarized to the numerical value with analysis significance, thereby reducing the influence of small errors on data analysis and improving the stability of data analysis.

[0090] The method further comprises a step of determining the permitted error distance, and the step comprises:

[0091] Step S300: A history interval with the current time point as the rear end point and the width of the preset history length is constructed on the preset time axis, and the representative distance range of the current detection distance under the history interval is defined as the reference distance range.

[0092] The time axis is a coordinate axis formed by combining each time point, and the coordinate axis points from the time point that has passed to the time point that has not arrived, wherein the time point that has passed is on the left side of the coordinate axis, and the left side of the coordinate axis is defined as the front side of the time axis; the history length is the interval length from the time point when the current batch of workpieces starts to be detected to the current time point, and the history interval is constructed to facilitate the acquisition and analysis of the data within the history length; the reference distance range is defined to identify the representative distance range under the same condition in the history, which is convenient for subsequent analysis.

[0093] Step S301: Determine the internal separation distance according to the adjacent characteristic distance and the representative characteristic distance in the reference distance range, and define the maximum internal separation distance as the upper limit separation distance.

[0094] The internal separation distance is the difference between the adjacent characteristic distance and the representative characteristic distance, and the difference is an absolute value; the upper limit separation distance is defined to determine the condition farthest from the representative characteristic distance, which is convenient for subsequent analysis.

[0095] Step S302: Randomly select one of all the upper limit separation distances as the main separation distance, and define the remaining upper limit separation distances as the secondary separation distance.

[0096] The main distance and the secondary distance are defined to distinguish the upper limit distance in different cases, which is convenient for subsequent analysis.

[0097] Step S303: Analyze according to the main distance and all secondary distances to determine the main representative coefficient, and define the main distance corresponding to the maximum main representative coefficient as the reasonable distance.

[0098] The main representative coefficient is the parameter value of the feasibility of the current main distance representing all upper limit distances. The larger the value is, the more representative it is. It is determined by summing the absolute value of the difference between the main distance and the secondary distance and taking the reciprocal. The reasonable distance is defined to determine the most common error distance under the current detection distance, which is convenient for subsequent analysis.

[0099] Step S304: Sum according to the reasonable distance and the preset fixed adjustment distance to determine the permitted error distance.

[0100] The fixed adjustment distance is a fixed value set by the staff. The reasonable distance plus the fixed adjustment distance can determine the permitted error distance, so that the allowed error of each area is determined according to the specific processing conditions of each area, and the overall effect is better. The method of steps S300-S304 is suitable for the case where the number of reference distance ranges is greater than the preset reference value. For example, steps S300-S304 are executed only after 10 workpieces are detected. When the number is less than the reference value, the corresponding permitted error distance is a fixed value set by the staff according to the actual situation.

[0101] It also includes a random determination step of the detection distance, which includes:

[0102] Step S400: Define the detection distance corresponding to the output workpiece defect signal in the history interval as the abnormal distance.

[0103] The abnormal distance is defined to distinguish the detection distance in different cases, which is convenient for subsequent analysis.

[0104] Step S401: Determine the distance separation value according to each abnormal distance and the detection distance, and determine the priority coefficient of each detection distance under the distance separation value according to the preset priority matching relationship.

[0105] The distance separation value is the difference between the abnormal distance and the detection distance, which is an absolute value. The smaller the value, the higher the possibility that the defect corresponding to the current abnormal distance affects the range of the corresponding detection distance. The priority coefficient is the coefficient of the detection distance that needs to be detected preferentially. Different distance separation values indicate different defect conditions, and the corresponding priority coefficients are also different. The priority matching relationship between them is determined by the staff in advance, and it is necessary to ensure that the larger the distance separation value, the smaller the corresponding priority coefficient.

[0106] Step S402: Sum all the priority coefficients at the same detection distance to determine the single-point representative coefficient.

[0107] The single-point representative coefficient is the sum of the priority coefficients determined for each workpiece in the historical interval at the same detection distance.

[0108] Step S403: Sort the corresponding detection distances from front to back according to the single-point representative coefficient from large to small to determine the random generation order, and determine the detection distance in turn according to the random generation order.

[0109] The larger the single-point representative coefficient, the higher the possibility of defects at the corresponding detection distance. Therefore, it is arranged in the front to generate the random generation order. At this time, the current detection distance is determined in turn according to the random generation order to ensure that the area with a higher possibility of defects is detected, so that the workpiece with defects does not need to be detected as a whole, thereby improving the overall detection efficiency.

[0110] After the priority coefficient is determined, the intelligent defect detection method for coating the iron core further comprises:

[0111] Step S500: Determine the reference type set corresponding to the abnormal distance and the comparison type set corresponding to the detection distance according to the preset type matching relationship.

[0112] The reference type set is a set of defect types that may occur in the area detected at the abnormal distance, such as concave, over-thick coating, etc. The comparison type set is a set of defect types that may occur in the area detected at the detection distance. Different distances correspond to different areas, so different defect types may occur. The type matching relationship between them is constructed by the staff by statistically analyzing each defect in advance.

[0113] Step S501: Determine the common type according to the reference type set and the comparison type set, and count the common type to determine the common number.

[0114] The common type is a type that exists in both the reference type set and the comparison type set, and the common number is the number of the determined common type.

[0115] Step S502: Counting according to the reference type set to determine the reference quantity, and calculating according to the common quantity and the reference quantity to determine the type common proportion.

[0116] The reference quantity is the total number of types recorded in the reference type set, and the type common proportion is the proportion value of the common type in all types in the reference type set, which is determined by dividing the common quantity by the reference quantity.

[0117] Step S503: According to the preset adjustment matching relationship, the key adjustment coefficient corresponding to the type common proportion is determined, and the priority coefficient is updated according to the key adjustment coefficient and the priority coefficient.

[0118] The key adjustment coefficient is a parameter value used to adjust the priority coefficient. The larger the type common proportion, the greater the possibility that the same defect will occur in two regions under the same defect cause, so the corresponding key adjustment coefficient is also larger. The adjustment matching relationship between the two is determined by the staff in advance. The priority coefficient can be updated by adding the key adjustment coefficient to the priority coefficient, thereby improving the accuracy of the determination of the detection distance and further improving the detection efficiency.

[0119] After the workpiece defect signal is output, the intelligent defect detection method for coating the iron core further comprises:

[0120] Step S600: Defining the virtual detection contour corresponding to the output workpiece defect signal as an abnormal detection contour.

[0121] Defining the abnormal detection contour distinguishes different virtual detection contours for subsequent analysis.

[0122] Step S601: Randomly defining a discard contour with a preset detection distance on the abnormal detection contour, and defining the area on the abnormal detection contour outside the discard contour as a reserved contour.

[0123] The detection distance is a fixed distance set by the staff. By randomly defining the discard contour, the defect position can be simulated and determined for subsequent analysis. At this time, the reserved contour is defined to distinguish different contours for subsequent analysis.

[0124] Step S602: Analyzing the reserved contour to determine whether at least one reserved contour outputs a workpiece qualified signal.

[0125] The purpose of the judgment is to know whether the detected defect is on the determined discard contour.

[0126] Step S6021: If there is no reserved profile output workpiece qualified signal, a segment of discarded profile is added until there is at least one reserved profile output workpiece qualified signal.

[0127] When there is no reserved profile output workpiece qualified signal, it means that all defects existing on the current abnormal detection profile cannot be determined by using the current discarded profile, so a segment of discarded profile is added again to expand the defect determination range. By continuously determining the discarded profile and analyzing, the defect condition can be determined. When there is at least one reserved profile output workpiece qualified signal, it means that the corresponding defect at this time is in the determined discarded profile, which is convenient for subsequent analysis.

[0128] Step S6022: If there is at least one reserved profile output workpiece qualified signal, the corresponding reserved profile is defined as a qualified profile, and the discarded profile corresponding to the qualified profile is defined as a defect profile.

[0129] When there is at least one reserved profile output workpiece qualified signal, it means that the current defect is included in the discarded profile, so the qualified profile and the defect profile are defined to distinguish different profiles, which is convenient for subsequent analysis.

[0130] Step S603: The defect profiles with common areas are fitted to construct a local profile, and the local defect area is demarcated according to the local profile.

[0131] The local profile is a profile composed of all defect profiles with common areas, that is, the profile passes through the area where the defect is located, so the local defect area can be demarcated according to the local profile defect curve. The specific demarcation method of the area can refer to steps S700-S702.

[0132] The step of demarcating the local defect area according to the local profile includes:

[0133] Step S700: Determine the local center point in the local profile, and demarcate the local coverage area with the local center as the center and the preset coverage distance as the radius.

[0134] The local center point is the center point of the local profile, and the coverage distance is the set value distance of the worker. The local coverage area is demarcated to effectively demarcate the area where the defect is located.

[0135] Step S701: Define the local coverage area containing the current local profile as an effective coverage area, and count the internal local number in the effective coverage area according to the local profile.

[0136] The effective coverage area is defined to effectively identify the local profile caused by the defect corresponding to the current area, and the internal local quantity is the total number of the determined local profile in the effective coverage area.

[0137] Step S702: According to the sorting rule, the internal local quantity with the maximum value is determined, and the effective coverage area corresponding to the internal local quantity is defined as the local defect area.

[0138] The internal local quantity with the maximum value is determined through the sorting rule, that is, it is indicated that the condition of the current defect in the corresponding effective coverage area is the most accurate, and therefore it is defined as the local defect area, thereby facilitating the subsequent re-inspection by the staff.

[0139] Reference Figure 3 Based on the same inventive concept, the embodiment of the present application provides a defect intelligent detection system for coating an iron core, comprising:

[0140] An acquisition module is configured to acquire a workpiece detection image.

[0141] A processing module is connected with the acquisition module and the judgment module, and is configured to store and process information.

[0142] A judgment module is connected with the acquisition module and the processing module, and is configured to judge information.

[0143] The processing module determines a workpiece center point and a workpiece feature point in the workpiece detection image, and defines a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius.

[0144] The processing module defines the workpiece feature point on the virtual detection contour as an online feature point, and determines an adjacent feature distance on the virtual detection contour according to adjacent online feature points in a preset detection direction.

[0145] The processing module counts the adjacent feature distances with different values to determine a feature distance quantity.

[0146] The judgment module judges whether there is at least one virtual detection contour with a feature distance quantity greater than a preset permitted distance quantity.

[0147] If the judgment module judges that there is no at least one virtual detection contour with a feature distance quantity greater than the permitted distance quantity, the processing module outputs a workpiece qualified signal.

[0148] If the judgment module judges that there is at least one virtual detection contour with a feature distance quantity greater than the permitted distance quantity, the processing module outputs a workpiece defect signal.

[0149] The adjacent feature distance induction module is used for inducing the adjacent feature distances with close numerical values;

[0150] The license error distance determination module is used for determining the appropriate license error distance according to different positions;

[0151] The detection distance random determination module is used for determining the random determination sequence of the detection distance to improve the overall detection efficiency;

[0152] The priority coefficient update module is used for updating the priority coefficients of the defect types that may occur in each region;

[0153] The defect region positioning module is used for positioning the region with actual defects;

[0154] The local defect region determination module is used for determining the local defect region according to the local contour distribution.

[0155] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional modules is taken as an example for illustration, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the device is divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, device and unit described above can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

Claims

1. A method for intelligent detection of defects for coating a core, characterized by, The method comprises the following steps: acquiring a workpiece detection image; determining a workpiece center point and a workpiece feature point in the workpiece detection image, and defining a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius; defining the workpiece feature point on the virtual detection contour as an online feature point, and determining an adjacent feature distance on the virtual detection contour according to adjacent online feature points in a preset detection direction; counting the adjacent feature distances of different values to determine a feature distance number; judging whether there is at least one virtual detection contour with a feature distance number greater than a preset permitted distance number; outputting a workpiece pass signal if there is no virtual detection contour with a feature distance number greater than the permitted distance number; outputting a workpiece defect signal if there is at least one virtual detection contour with a feature distance number greater than the permitted distance number; the step of counting the adjacent feature distances of different values to determine a feature distance number comprises: counting the adjacent feature distances of the same value to determine a single-point distance number; determining the maximum single-point distance number according to a preset sorting rule, and defining the adjacent feature distance corresponding to the single-point distance number as a representative feature distance; constructing a representative distance range according to the representative feature distance and a preset permitted error distance, and converting the adjacent feature distances within the representative distance range into current representative feature distances; continuing to determine the representative feature distance according to the remaining adjacent feature distances after the conversion of the adjacent feature distances, until all the adjacent feature distances are within the constructed representative distance range, wherein the feature distance number is determined according to the representative feature distances.

2. The method for defect intelligent inspection of a coated core according to claim 1, wherein The method further comprises a step of determining the permitted error distance, which comprises: constructing a history interval with a current time point as the back end point and a preset history length as the width on a preset time axis, and defining the representative distance range of the current detection distance as a reference distance range according to the history interval; determining an internal interval distance according to the adjacent feature distances and the representative feature distance in the reference distance range, and defining the maximum internal interval distance as an upper limit interval distance; randomly selecting one of all the upper limit interval distances as a main interval distance, and defining the remaining upper limit interval distances as secondary interval distances; analyzing the main interval distance and all the secondary interval distances to determine a main representative coefficient, and defining the main interval distance corresponding to the maximum main representative coefficient as a reasonable interval distance; summing the reasonable interval distance and a preset fixed adjustment distance to determine the permitted error distance.

3. The method for defect intelligent inspection of a coated core according to claim 2, wherein The method further comprises a step of randomly determining the detection distance, which comprises: defining the detection distance corresponding to the output workpiece defect signal in the history interval as an abnormal distance; determining a distance interval value according to each abnormal distance and the detection distance, and determining a priority coefficient of each detection distance according to a preset priority matching relationship according to the distance interval value; summing all the priority coefficients of the same detection distance to determine a single-point representative coefficient. The single points are sorted from front to back according to the corresponding detection distances represented by the single points from large to small to determine a random generation order, and the detection distances are sequentially determined according to the random generation order.

4. The method for defect intelligent inspection of a coated core according to claim 3, wherein After the priority coefficient is determined, the intelligent defect detection method for the coated iron core further includes: According to the preset type matching relationship, a reference type set corresponding to the abnormal distance and a comparison type set corresponding to the detection distance are determined; According to the reference type set and the comparison type set, a common type is determined, and counting is performed according to the common type to determine a common quantity; According to the reference type set, counting is performed to determine a reference quantity, and the common quantity and the reference quantity are calculated to determine a type common proportion; According to the preset adjustment matching relationship, a key adjustment coefficient corresponding to the type common proportion is determined, and the priority coefficient is updated by calculating according to the key adjustment coefficient and the priority coefficient.

5. The method for defect intelligent inspection of a coated core according to claim 1, wherein After the workpiece defect signal is output, the intelligent defect detection method for the coated iron core further includes: The virtual detection contour corresponding to the output workpiece defect signal is defined as an abnormal detection contour; A discard contour with a preset detection distance is randomly drawn on the abnormal detection contour, and a region outside the discard contour on the abnormal detection contour is defined as a reserved contour; According to the reserved contour, it is judged whether there is at least one reserved contour output workpiece qualified signal; If there is no at least one reserved contour output workpiece qualified signal, a discard contour is added until there is at least one reserved contour output workpiece qualified signal; If there is at least one reserved contour output workpiece qualified signal, the corresponding reserved contour is defined as a qualified contour, and the discard contour corresponding to the qualified contour is defined as a defect contour; The defect contours with common regions are fitted to construct a local contour, and the local defect region is determined according to the local contour.

6. The method for defect intelligent inspection of a coated core according to claim 5, wherein The step of determining the local defect region according to the local contour includes: Determine the local center point in the local contour, and draw a local coverage region with the local center as the center and a preset coverage distance as the radius; The local coverage region containing the current local contour is defined as an effective coverage region, and the internal local quantity is determined by counting the local contour in the effective coverage region; According to the sorting rule, the internal local quantity with the maximum value is determined, and the effective coverage region corresponding to the internal local quantity is determined as the local defect region.

7. A system for intelligent detection of defects in a coated core for implementing a method for intelligent detection of defects in a coated core as claimed in any one of claims 1 to 6, characterized in that It includes: An acquisition module for acquiring a workpiece detection image; A processing module connected with the acquisition module and the judgment module for information storage and processing; A judgment module connected with the acquisition module and the processing module for information judgment; The processing module determines the workpiece center point and the workpiece feature point in the workpiece detection image, and draws a virtual detection contour with the workpiece center point as the center and a random detection distance as the radius; The processing module defines the workpiece feature points on the virtual detection contour as online feature points, and determines the adjacent feature distance in the preset detection direction according to the adjacent online feature points on the virtual detection contour; The processing module counts the adjacent feature distances with different values to determine the feature distance quantity; The judging module judges whether the number of feature distances determined on at least one virtual detection contour is greater than a preset permitted distance number; If the judging module judges that the number of feature distances determined on at least one virtual detection contour is not greater than the permitted distance number, the processing module outputs a workpiece qualified signal; If the judging module judges that the number of feature distances determined on at least one virtual detection contour is greater than the permitted distance number, the processing module outputs a workpiece defect signal.

Images