Method for detecting surface defects of polygonal-section material

Abstract

The invention relates to the field of automatic detection of material surface defects, in particular to a method for detecting surface defects of a polygonal-section material. The method comprises the following steps: aiming at a polygonal section material, calculating an error between a polygonal section appearance and a standard section appearance of the material in operation, and finally connecting the errors of all sections to obtain defect parameters of the whole material surface. According to the method, errors caused by rotation and shaking in the material running process are corrected, meanwhile, the method for solving the minimum distance under the XY coordinates is used for replacing errors formed by solving errors at the same angle of polar coordinates, the detection precision is improved, meanwhile, the time period is short, the steps are simple, the analysis timeliness is accelerated, the detection efficiency and stability are improved, and the method is suitable for popularization and application. And the overall accuracy is good, so that the quality parameters of the product have traceability.

Description

Technical Field

[0001] The invention relates to the field of automatic detection of defects on the surface of a material, and in particular to a method for detecting surface defects of a material with a polygonal cross section. Background Art

[0002] With the continuous improvement of industrial automation technology, the deepening of industrial upgrading, the continuous improvement of material production quality and quality requirements, the data detection and inspection methods in the material production and processing process are becoming more and more important. The original material processing process does not have the function of online dimension collection and surface defect detection. It is necessary to complete the error measurement of each point manually or offline after the material processing is completed or during the inspection process. The measurement process is long, the steps are cumbersome, the efficiency is low, and the measurement accuracy and real-time performance are poor.

[0003] The prior art also proposes an online detection method. For example, the document with application number "CN200980139089" discloses a method for determining shape parameters. A known method is used to generate and measure at least three projected edges that are attached to the elongated product to be measured, surround the elongated product and form a polygon by means of a measuring device with at least two laser scanners, and the corresponding tangents are calculated from them. The laser scanners each have a photosensitive sensor and a laser. A contour is calculated from the polygon, and multiple contour segments are determined on the contour. The desired shape parameters can be calculated from these data. However, there are the following problems: 1. It cannot be quickly targeted at materials with different polygonal cross-sectional shapes, the processing time cycle is long, and the steps are cumbersome; 2. Due to the different data acquisition methods, the data processing methods are also different, and the data density and speed are low, and the overall accuracy is poor. Summary of the invention

[0004] The present invention provides a method for detecting surface defects of a material with a polygonal cross section, so as to solve the problems of the prior art, such as long time period, complicated steps and poor overall accuracy.

[0005] In order to achieve the above object, the technical solution of the present invention is as follows: a method for detecting shape defects of a polygonal cross section, characterized in that it comprises the following steps:

[0006] Step 1: Use multiple line laser or binocular vision sensor data evenly distributed in the circumferential direction of the same cross-section of the material to stitch the data of multiple sensors on the cross-section;

[0007] Step 2: Obtain the spliced ​​cross-sectional shape in real time and calculate the error between it and the standard polygonal cross-sectional shape learned in advance;

[0008] Step 3, obtain a cross-section shape error curve: for the two matched two-dimensional arrays, extract the XY coordinate data of the real-time cross-section shape in sequence through a for loop, calculate the error when the point is the minimum value of all the points of the standard shape, record the error, and after a loop, obtain the cross-section shape error curve between the real-time cross-section shape data and the standard shape data;

[0009] Step 4: Form multiple cross-sectional shape error curves on the time axis, and splice the error curves to obtain a three-dimensional defect array of the material surface.

[0010] Furthermore, the specific steps of the above step 2 include:

[0011] Step 2.1: Convert the two-dimensional array in polar coordinates obtained in step 1 into a two-dimensional array in XY coordinates;

[0012] Step 2.2: Calculate the center of gravity of the two-dimensional array obtained in step 2.1 to obtain the center of gravity coordinates, perform center of gravity calibration on the acquired real-time shape data, and obtain the real-time cross-sectional shape data after center of gravity calibration;

[0013] Step 2.3: Convert the two-dimensional array in XY coordinates obtained in step 2.2 into a two-dimensional array in polar coordinates;

[0014] Step 2.4: Extract the one-dimensional radius array from the two-dimensional array obtained in step 2.3, obtain its salient point positions, and form a new position array;

[0015] Step 2.5: According to the values ​​of two adjacent points in the new position array, cut the corresponding two-dimensional array obtained in step 6;

[0016] Step 2.6: Extract the position values ​​of the first and last two points of each of the above four array segments;

[0017] Step 2.7: Calculate the rotation angle of the corresponding array and the X-axis according to the above values, and thus calculate the rotation angle of the real-time section relative to the standard section in step 2.6;

[0018] Step 2.8: Perform angle correction on the standard cross-section shape data learned earlier;

[0019] Step 2.9: Use the modified standard polygonal cross-section shape data obtained in step 2.8 to unify and match the data length points with the real-time cross-section polygonal shape data obtained in step 2.6 in polar coordinates.

[0020] Furthermore, the above step 2.9 is to use a for loop in polar coordinates to sequentially extract the angle values ​​of the two-dimensional array of the standard polygonal section shape, and for a corresponding angle value, find the minimum absolute value of the error of the one-dimensional array of angles in the real-time section polygon shape data; extract the angle and radius of the position as output, and after the for loop, extract the real-time section shape data with the same length as the standard section polygon array and arranged at almost the same angle to form a new two-dimensional array of real-time section shape.

[0021] Furthermore, the specific steps of the above step three are:

[0022] Step 3.1: Convert the real-time cross-sectional shape data in polar coordinates obtained in step 2 to XY coordinates;

[0023] Step 3.2: Use the for loop to sequentially extract the corresponding data coordinate x and y values ​​from the two-dimensional array of real-time XY coordinates obtained in step 3.1, calculate the minimum absolute value of the difference between the coordinate point value and each point under the XY coordinates of the standard section, and record the position with the minimum absolute value;

[0024] Step 3.3: Extract the xy values ​​of the corresponding real-time data and standard data according to the position with the smallest absolute value obtained in step 3.2, convert them into polar coordinates, and calculate the radius difference between the two positions;

[0025] Step 3.4: According to the radius difference in step 3.3, determine the positive and negative direction of the error. If the real-time data point is greater than the standard data point, the error value is positive; if the real-time data point is less than the standard data point, the error value is negative; if the real-time data point is equal to the standard data point, its error value is 0;

[0026] Step 3.5: After the For loop, a set of error waveforms with the same number and angle order in polar coordinates as the standard cross-section array is formed.

[0027] Furthermore, the specific steps of the above step 1 are:

[0028] Step 1.1: Use multiple line laser sensors placed in the circumferential direction to obtain two-dimensional data corresponding to the respective depths in the Z and X directions, and filter the Z direction data;

[0029] Step 1.2: Perform differential calculations on the arrays in the Z direction, use the differential value as the segmentation point, and find the longest data segment within the differential value with the data value within the correct range as the parameter for subsequent splicing;

[0030] Step 1.3: For the reference coordinate values ​​and angles of multiple sensors, perform the same XY coordinate conversion and connection: convert the array from XY coordinates to polar coordinates, sort them according to the angle size, and thus obtain a two-dimensional array of the real-time material cross-sectional shape under polar coordinates.

[0031] Furthermore, the above step 4 specifically utilizes the above steps 1 to 3 to be cycled multiple times on the time axis to obtain the surface defect parameters of the entire material.

[0032] Compared with the prior art, the present invention has the following beneficial effects:

[0033] 1. The method for detecting surface defects of materials with polygonal cross-sections proposed by the present invention takes into account the errors in machining and installation of supporting structures, mounting structures, etc. in practical applications, and the non-stationary errors in electrical control operation. During the detection process, the errors caused by the rotation and shaking of the materials during operation are corrected. At the same time, during the research and development process, when the calculation of circles and the data analysis of material cross-sections with line segment structures such as polygons are performed, it is found that the use of the polar coordinate method for material cross-sections with line segment structures will cause large errors. Therefore, in step 2 and step 3, the present invention takes into account the difference between the data density of the polygon at the intersection of the data analysis and the density of the line segment area data, and the error in the calculation of the actual value and the ideal value when the line segment area polar coordinates are used as a unified standard. The method of finding the minimum distance under XY coordinates is used instead of the error formed when finding the error at the same angle of polar coordinates. Therefore, the method provided by the present invention can greatly reduce the error when comparing at a unified angle under polar coordinates, thereby improving the accuracy of detection.

[0034] 2. The present invention obtains the material shape data in the production process or the inspection process in an online and real-time manner. When splicing, the shape data of multiple groups of sensors are combined in spatial positions to form the shape of the polygonal material in the spatial position. When matching, the cross-sectional shape data of each polygon is compared with the cross-sectional shape data of the standard polygon, and its defect error parameters are calculated. Finally, multiple cross-sectional shape defect parameters are spliced ​​using the time axis to form the corresponding material surface defect parameters. During the process, the rotation angle difference and center of gravity between the real-time section and the standard section are continuously calculated and corrected, effectively avoiding the calculation error caused by the jumping and rotation during the material movement. Since the method for detecting errors is adjusted, the real-time rotation angle difference and center of gravity are paid attention to at the same time, and the parameters are corrected in real time, the accuracy of the detection can be effectively improved. Through the acquisition method of the present invention, a large amount of data can be obtained, the time for data acquisition is reduced, the time period is short, the steps are simple, the timeliness of the analysis is accelerated, the efficiency and stability of the detection are improved, and the overall accuracy is good, so that the quality parameters of the product have traceability. DETAILED DESCRIPTION

[0035] To facilitate understanding of the present invention, the present invention will be described more fully below with reference to embodiments.

[0036] The design idea of ​​the present invention is: to use several groups of sensor data evenly distributed in the circumferential direction to form cross-sectional shape data, compare the cross-sectional shape data measured in real time with the standard cross-sectional parameters of the polygonal cross-sectional material, obtain the error parameters on the cross-sectional shape, and splice multiple cross-sectional shape errors on the time axis to form a three-dimensional defect array on the material surface. In the process, the angular rotation and center of gravity offset of the material during operation are considered, so as to finally obtain the defect parameters of the entire material surface.

[0037] Based on the above design ideas, the present invention provides a method for detecting surface defects of polygonal cross-section materials, comprising the following steps:

[0038] Step 1: Use multiple line laser or binocular vision sensor data evenly distributed in the circumferential direction of the same cross-section of the material to stitch the data of multiple sensors on the cross-section;

[0039] Step 2. Obtain the spliced ​​cross-sectional shape in real time and calculate the error with the standard polygonal cross-sectional shape learned in advance: calculate the rotation angle of each real-time cross-sectional figure and the standard shape, and at the same time, sort the cross-sectional data points acquired in real time according to the points of the standard shape in the array position of the nearest angle in polar coordinates, and match the real-time cross-sectional shape data with the standard shape data in length and angle position.

[0040] Step 3. Get a cross-section shape error curve: For the two matched two-dimensional arrays, extract the XY coordinate data of the real-time cross-section shape in sequence through a for loop, calculate the error from its point to the minimum value of all points of the standard shape, and record the error. After the loop, you can get the cross-section shape error curve between the real-time cross-section shape data and the standard shape data in sequence.

[0041] Step 4: On the time axis, as the material moves, it forms multiple cross-sectional shape error curves. By splicing the various error curves, a three-dimensional defect array of the material surface can be obtained.

[0042] Embodiment: Taking a quadrilateral material as the implementation object, the error calculation between the real-time cross-sectional shape data and the standard shape data is performed.

[0043] The sensor system evenly distributed in the circumferential direction learns the standard circular and quadrilateral sample rods and calibrates the reference coordinate values ​​(x 0 ,y 0 )、(x 1 ,y 1 )、(x 2 ,y 2 )、(x3 ,y 3 ) and four angles β 0 , β 1 , β 2 , β 3 .

[0044] A method for detecting shape defects of a polygonal cross section is provided, which specifically comprises the following steps:

[0045] Step 1: Splicing:

[0046] Step 1.1: Use multiple line laser sensors placed in the circumferential direction to obtain the two-dimensional data corresponding to the respective depths in the Z and X directions. In this embodiment, four sensors are evenly placed at four positions, and the Z direction data of the four sensors are filtered separately;

[0047] Step 1.2: Perform differential calculations on each of the four Z-direction arrays above, use the set differential value 1 as the split point, and find the longest data segment with a differential value less than 1 and a data value within the correct range. In this embodiment, the longest segment with a differential value less than 1 and a Z-direction value between 150-350 is intercepted as the parameter for subsequent splicing.

[0048] Step 1.3: Combine the four segments of data from the four sensors obtained in step 1.2 with the reference coordinate values ​​(x 0 ,y 0 )、(x 1 ,y 1 )、(x 2 ,y 2 )、(x 3 ,y 3 ) and four angles β 0 , β 1 , β 2 , β 3 , perform the same XY coordinate conversion, and connect the four converted data; convert the array from XY coordinates to polar coordinates, and sort them according to the angle size under polar coordinates, so as to obtain a two-dimensional array of the real-time material cross-section shape under polar coordinates.

[0049] Step 2: Obtain the spliced ​​cross-sectional shape in real time and calculate the error between it and the standard polygonal cross-sectional shape learned in advance:

[0050] Step 2.1: Convert the two-dimensional array in polar coordinates obtained in step 1 into a two-dimensional array in XY coordinates;

[0051] Step 2.2: Calculate the center of gravity of the two-dimensional array of the shape data obtained in step 2.1 to obtain the center of gravity coordinates, and perform center of gravity calibration on the acquired real-time shape data, that is, subtract the center of gravity parameters from the array to obtain the real-time cross-sectional shape data after center of gravity calibration.

[0052] Step 2.3: Convert the two-dimensional array in XY coordinates obtained in step 2.2 into a two-dimensional array in polar coordinates;

[0053] Step 2.4: Extract the one-dimensional radius array from the two-dimensional array obtained in step 2.3, obtain its salient point positions, and form a new position array.

[0054] Step 2.5: According to the values ​​of two adjacent points in the new position array, cut the corresponding two-dimensional array obtained in step 2.6. In this embodiment, the position array value is 4. According to the values ​​of 1 and 2, cut the first segment, the values ​​of 2 and 3 cut the second segment, and the values ​​of 3 and 4 cut the third segment. The values ​​of 4 and 1 represent the array formed by connecting the position value of 4 to the end of the array and the array formed by connecting the starting point of the array to the position of 1.

[0055] Step 2.6: Extract the position values ​​of the first and last two points of each of the above four array segments.

[0056] Step 2.7: Calculate the rotation angle of the corresponding array and the X-axis based on the above values, thereby calculating the rotation angle of the real-time section in step 2.6 relative to the standard section.

[0057] Step 2.8: Make angle corrections to the standard cross-section shape data learned earlier.

[0058] Step 2.9: Use the modified standard polygonal cross-section shape data obtained in step 2.8 to unify and match the data length points with the real-time cross-section polygonal shape data obtained in step 2.6 in polar coordinates.

[0059] That is, in polar coordinates, a for loop is used to sequentially extract the angle values ​​of the two-dimensional array of the standard polygonal cross-section shape. For a corresponding angle value, the minimum absolute value of the error of the one-dimensional array of angles in the real-time cross-section polygonal shape data, that is, the closest angle position, is found, and the angle and radius of the position are extracted as output. After the for loop, the real-time cross-section shape data of the same length as the standard cross-section polygon array and arranged at almost the same angle can be extracted, thereby achieving the unification of the length of the real-time cross-section shape data and the length of the standard cross-section shape data and the unification of the angle sorting, that is, forming a new two-dimensional array of real-time cross-section shape.

[0060] Step 3: Get the shape error curve of a section:

[0061] Step 3.1: Convert the real-time cross-sectional shape data in polar coordinates obtained in step 2 to XY coordinates;

[0062] Step 3.2: Use the for loop to extract the corresponding data coordinate x and y values ​​in sequence from the two-dimensional array of real-time XY coordinates obtained in step 3.1, calculate the minimum absolute value of the difference between the coordinate point value and each point under the XY coordinates of the standard section, and record the position with the minimum absolute value. The minimum absolute value of the difference is the closest distance, which is basically the vertical distance from the point to the line.

[0063] Step 3.3: Extract the xy values ​​of the corresponding real-time data and standard data according to the position with the smallest absolute value obtained in step 3.2, convert them into polar coordinates, and calculate the radius difference between the two positions.

[0064] Step 3.4: Based on the radius difference in step 3.3, determine the positive or negative value of the minimum absolute value of the difference obtained in step 3.2, that is, the positive or negative direction of the error. If the real-time data point is larger than the standard data point, the error value is positive; if the real-time data point is smaller than the standard data point, the error value is negative; if the real-time data point is equal to the standard data point, its error value is 0.

[0065] Step 3.5: After the For loop, a set of error waveforms with the same number and angle order in polar coordinates as the standard cross-section array is formed.

[0066] Step 4. Obtain the three-dimensional defect array of the material surface: By using the above steps 1 to 3 on the time axis, multiple cycles can be used to obtain the error waveforms of multiple cross sections of the entire material. By splicing them, the surface defect parameters of the entire material can be obtained.

[0067] The above-mentioned embodiments only express the preferred implementation of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be based on the attached claims.

Claims

1. A method for detecting shape defects of polygonal cross sections, characterized in that: The following steps are involved: Step 1: Use multiple line laser or binocular vision sensor data evenly distributed in the circumferential direction of the same cross-section of the material to stitch the data of multiple sensors on the cross-section; Step 2: Obtain the spliced ​​cross-sectional shape in real time and calculate the error between it and the standard polygonal cross-sectional shape learned in advance; Step 3, obtain a cross-section shape error curve: for the two matched two-dimensional arrays, extract the XY coordinate data of the real-time cross-section shape in sequence through a for loop, calculate the error when the point is the minimum value of all the points of the standard shape, record the error, and after a loop, obtain the cross-section shape error curve between the real-time cross-section shape data and the standard shape data; Step 4: Form multiple cross-sectional shape error curves on the time axis, and splice the error curves to obtain a three-dimensional defect array of the material surface.

2. A method for detecting shape defects of a polygonal cross section according to claim 1, characterized in that: The specific steps of step 2 include: Step 2.1: Convert the two-dimensional array in polar coordinates obtained in step 1 into a two-dimensional array in XY coordinates; Step 2.2: Calculate the center of gravity of the two-dimensional array obtained in step 2.1 to obtain the center of gravity coordinates, perform center of gravity calibration on the acquired real-time shape data, and obtain the real-time cross-sectional shape data after center of gravity calibration; Step 2.3: Convert the two-dimensional array in XY coordinates obtained in step 2.2 into a two-dimensional array in polar coordinates; Step 2.4: Extract the one-dimensional radius array from the two-dimensional array obtained in step 2.3, obtain its salient point positions, and form a new position array; Step 2.5: According to the values ​​of two adjacent points in the new position array, cut the corresponding two-dimensional array obtained in step 6; Step 2.6: Extract the position values ​​of the first and last two points of each of the above four array segments; Step 2.7: Calculate the rotation angle of the corresponding array and the X-axis according to the above values, and thus calculate the rotation angle of the real-time section relative to the standard section in step 2.6; Step 2.8: Perform angle correction on the standard cross-section shape data learned earlier; Step 2.9: Use the modified standard polygonal cross-section shape data obtained in step 2.8 to unify and match the data length points with the real-time cross-section polygonal shape data obtained in step 2.6 in polar coordinates.

3. A method for detecting shape defects of a polygonal cross section according to claim 2, characterized in that: The step 2.9 is to sequentially extract the angle values ​​of the two-dimensional array of the standard polygonal cross-section shape by using a for loop in polar coordinates, and to find the minimum absolute value of the error of the one-dimensional array of the angle in the real-time cross-section polygonal shape data for a corresponding angle value; The angle and radius of the position are extracted as output. After the for loop, the real-time cross-section shape data with the same length as the standard cross-section polygon array and arranged at almost the same angle are extracted to form a new two-dimensional array of real-time cross-section shape.

4. A method for detecting shape defects of a polygonal cross section according to claim 2 or 3, characterized in that: The specific steps of step three are: Step 3.1: Convert the real-time cross-sectional shape data in polar coordinates obtained in step 2 to XY coordinates; Step 3.2: Use the for loop to sequentially extract the corresponding data coordinate x and y values ​​from the two-dimensional array of real-time XY coordinates obtained in step 3.1, calculate the minimum absolute value of the difference between the coordinate point value and each point under the XY coordinates of the standard section, and record the position with the minimum absolute value; Step 3.3: Extract the xy values ​​of the corresponding real-time data and standard data according to the position with the smallest absolute value obtained in step 3.2, convert them into polar coordinates, and calculate the radius difference between the two positions; Step 3.4: According to the radius difference in step 3.3, determine the positive and negative direction of the error. If the real-time data point is greater than the standard data point, the error value is positive; if the real-time data point is less than the standard data point, the error value is negative; If the real-time data point is equal to the standard data point, its error value is 0; Step 3.5: After the For loop, a set of error waveforms with the same number and angle order in polar coordinates as the standard cross-section array is formed.

5. A method for detecting shape defects of a polygonal cross section according to claim 4, characterized in that: The specific steps of step one are: Step 1.1: Use multiple line laser sensors placed in the circumferential direction to obtain two-dimensional data corresponding to the respective depths in the Z and X directions, and filter the Z direction data; Step 1.2: Perform differential calculations on the arrays in the Z direction, use the differential value as the segmentation point, and find the longest data segment within the differential value with the data value within the correct range as the parameter for subsequent splicing; Step 1.3: For the reference coordinate values ​​and angles of multiple sensors, perform the same XY coordinate conversion and connection: convert the array from XY coordinates to polar coordinates, sort them according to the angle size, and thus obtain a two-dimensional array of the real-time material cross-sectional shape under polar coordinates.

6. A method for detecting shape defects of a polygonal cross section according to claim 5, characterized in that: Step 4 specifically utilizes the above steps 1 to 3 to be cycled multiple times on the time axis to obtain the surface defect parameters of the entire material.