Laser powder bed melting quality analysis method, device, equipment and medium

Abstract

The invention discloses a laser powder bed melting quality analysis method, device and equipment and a medium, and relates to the technical field of laser powder bed melting quality analys.The method comprises the steps that a surface gray level image of a melting image is collected, and first-order statistic features of the image are extracted; the method comprises the following steps: acquiring feature data, performing multiple data dimensions and multiple attribute dimensions on the feature data, acquiring a potential relationship and correlation among the feature data, and establishing an association relationship between gray features and sample quality, thereby fully mining image information from a fused surface image of a sample without obvious defects as a whole, and obtaining a fused surface image of the sample without obvious defects. The data characteristics are visualized, so that the quality analysis of the sample without obvious defects is realized; meanwhile, the method can achieve automatic detection, reduces the manual participation degree, and enables the detection result to be more objective and accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of laser powder bed melting quality analysis, and in particular to a laser powder bed melting quality analysis method, device, equipment and medium. Background Art

[0002] Laser additive manufacturing technology is widely used in industrial manufacturing due to its high energy density and high molding precision. However, in the actual molding process, various factors may cause unpredictable quality problems in the molded parts. Therefore, like products manufactured by traditional parts processing, quality inspection and analysis of molded parts is an indispensable task. Currently, common quality inspection methods include density inspection method and roughness inspection method. The density inspection method generally requires the molded parts to be placed in the water of the test bench for multiple inspections to obtain the average value. When inspecting a large number of samples, this method is more complex, and excessive manual labor will lead to larger inspection errors. Roughness inspection requires the use of a surface roughness tester to take samples at different locations for multiple inspections. This method will increase the heavy manual workload when there are too many samples, and since the measurement accuracy of the inspection equipment will gradually decrease with the increase in the number of uses, the inspection and replacement of the equipment requires a certain amount of time and economic cost.

[0003] The laser additive manufacturing online monitoring system can collect a large amount of radiation intensity signal data generated during the printing process, and can analyze this type of data to detect the quality of the formed parts. Currently, analyzing the light signals collected during the printing process is the most widely used method in additive manufacturing data analysis. The molten surface image collected by a high-speed camera intuitively reflects the surface information of the printing process, and areas with obvious defects can be observed from the molten surface image; however, the collected molten surface image can only fully show whether there are defects or the location of the defective area. It is difficult to intuitively observe the quality analysis of samples without obvious defects from the collected molten surface image. Summary of the invention

[0004] The embodiments of the present invention provide a laser powder bed melting quality analysis method, device, equipment and medium, which can solve the problem in the prior art that the collected molten surface image can only fully show whether there are defects or the location of the defective area, and the quality analysis of samples without obvious defects is difficult to intuitively observe from the collected molten surface image.

[0005] The embodiment of the present invention provides a laser powder bed melting quality analysis method, comprising the following steps:

[0006] Obtain multiple fusion images during sample preparation with different process parameters;

[0007] Marking multiple fused images of samples with different process parameters with polygon vertices, segmenting the fused images according to the marked polygons and the marked shapes, to obtain multiple segmented images under different process parameters;

[0008] The image grayscale information of the segmented image is obtained, the grayscale pixel interval is set according to the grayscale information, and the grayscale of each pixel point of multiple segmented images under different process parameters and the number of pixels with the grayscale are obtained; the grayscale of each pixel point in the segmented image is used as the horizontal coordinate, and the number of pixels with the grayscale is used as the vertical coordinate to obtain multiple grayscale distribution histograms of the fused images with different process parameters;

[0009] According to multiple grayscale distribution histograms under the same process parameters, multiple grayscale eigenvalues ​​of the corresponding fused images are obtained; multiple grayscale eigenvalue statistics of multiple fused images under different process parameters are obtained respectively, the grayscale features of the fused images under the same process parameters are classified into one category, and a matrix scatter plot under different process parameters is established according to multiple grayscale eigenvalue statistics; the correspondence between process parameters and grayscale eigenvalues ​​is obtained according to the matrix scatter plot, and the distribution law of grayscale features when the process parameters change is obtained according to the corresponding relationship, and the quality of samples under different process parameters is judged by the distribution range of grayscale eigenvalues.

[0010] Preferably, the step of obtaining a plurality of fused images during sample preparation with different process parameters includes:

[0011] Set the process parameters according to the laser additive manufacturing equipment and forming requirements. Set one parameter as a variable in each group and set the remaining parameters as constants.

[0012] Keep other printing parameters unchanged, set the laser power as the variable of the first group, and then set the scanning speed, scanning spacing, and scanning direction as variables of the second, third, and fourth groups respectively, and print the sample;

[0013] The printed samples were tested for density and roughness using the drainage method and a roughness tester, and multiple fused images were collected under different process parameters.

[0014] Preferably, the step of obtaining a plurality of segmented images under different process parameters includes:

[0015] Use the labelme annotation tool to polygonally mark multiple fused images collected under different process parameters and generate a json file;

[0016] Parse the json file to obtain the name of each marker type and the coordinates of the polygon vertices corresponding to the type, fill the inside of the polygon with pixels with a grayscale value of 0, and fill the outside of the polygon with pixels with a grayscale value of 255 to form a mask image with the same size as the original image, and convert the mask image into an RGBA image;

[0017] The pixel values ​​of the four channels of the original image pixels corresponding to the 255 pixels in the mask image are transformed to (0, 0, 0, 0), that is, the pixel values ​​outside the marked polygon in the original image are transformed to transparent, and the pixel values ​​inside the marked polygon in the original image remain unchanged, so as to obtain multiple segmented images under different process parameters.

[0018] Preferably, the step of obtaining a plurality of grayscale distribution histograms of fusion images with different process parameters comprises:

[0019] Read the image grayscale information in the segmented image, set the grayscale pixel interval according to the grayscale information, obtain the grayscale of each pixel point and the number of pixels with the grayscale in multiple segmented images under different process parameters, and count the total number of pixels and the occurrence probability of each grayscale;

[0020] Taking the gray level of each pixel in the image as the horizontal coordinate and the frequency of the gray level as the vertical coordinate, multiple gray distribution histograms of the fused images with different process parameters are obtained; the method for obtaining the frequency of the gray level in the gray distribution histogram is:

[0021] P(r)=n r / (W*H)(r∈[0,L-1])

[0022] Where: W, H represent the width and height of the image respectively; n r It represents the number of pixels with value r in the image; P(r) represents the frequency of pixel value r in the image; L represents the number of gray levels of the image.

[0023] Preferably, the establishment of the matrix scatter plot under different process parameters includes:

[0024] The images of different layers under the same process parameters are set to the same name, multiple grayscale feature values ​​of multiple fused images are counted, and the statistics of each grayscale feature value under the multiple fused images are obtained according to the multiple grayscale feature values ​​of the multiple fused images;

[0025] The fused images are classified according to their serial numbers, and the data points under different process parameters are set to different colors. A matrix scatter plot under different process parameters is established according to the statistics of each grayscale feature value under different colors.

[0026] Preferably, the statistics of each grayscale feature value include:

[0027] The grayscale characteristic values ​​include: mean, median, standard deviation, 3 / 4 quantile, 1 / 4 quantile, minimum value, maximum value, entropy, peak value and mean absolute deviation;

[0028] The feature quantities represented by the taken eigenvalues ​​are different, and the statistics of the mean value, minimum value, entropy, and mean absolute deviation are determined as each grayscale feature.

[0029] Preferably, the determination of the sample quality includes:

[0030] By observing the grayscale feature matrix scatter plot of the fused image under different laser powers, scanning speeds, scanning intervals and scanning directions, the corresponding relationship between the process parameters and the grayscale feature values ​​is obtained according to the matrix scatter plot, and the grayscale feature distribution law when the process parameters change is obtained according to the corresponding relationship;

[0031] The quality of the sample is judged according to the distribution range of the grayscale feature value. When the average value, minimum value, entropy and average absolute deviation of the grayscale feature of the fused image are within the determined range, the sample is qualified, and the rest are unqualified.

[0032] The embodiment of the present invention further provides a laser powder bed melting mass analysis device, comprising:

[0033] An image module is used to obtain multiple fusion images during the preparation of samples with different process parameters;

[0034] A segmentation module is used to mark the polygon vertices of multiple fused images of samples with different process parameters, and segment the fused images according to the marked polygons according to the marked shapes to obtain multiple segmented images under different process parameters;

[0035] The feature solution module is used to obtain the image grayscale information of the segmented image, set the grayscale pixel interval according to the grayscale information, obtain the grayscale of each pixel point of multiple segmented images under different process parameters and the number of pixels with the grayscale level; take the grayscale level of each pixel point in the segmented image as the horizontal coordinate and the number of pixels with the grayscale level as the vertical coordinate to obtain multiple grayscale distribution histograms of the fused images with different process parameters;

[0036] According to multiple grayscale distribution histograms under the same process parameters, multiple grayscale characteristic values ​​of the corresponding fused images are obtained; multiple grayscale characteristic value statistics of multiple fused images under different process parameters are obtained respectively, the grayscale features of the fused images under the same process parameters are classified into one category, and a matrix scatter plot under different process parameters is established according to multiple grayscale characteristic value statistics; the corresponding relationship between the process parameters and the grayscale characteristic values ​​is obtained according to the matrix scatter plot, and the grayscale characteristic distribution law when the process parameters change is obtained according to the corresponding relationship;

[0037] The analysis module determines the quality of samples under different process parameters through the distribution range of grayscale eigenvalues.

[0038] An embodiment of the present invention further provides an electronic device, including a memory and a processor;

[0039] The memory is used to store computer programs;

[0040] The processor is used to implement the steps of the laser powder bed melting quality analysis method as described above when executing the computer program stored in the memory.

[0041] An embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, wherein when the computer program is executed by a processor, the steps of the laser powder bed melting quality analysis method as described above are implemented.

[0042] The embodiments of the present invention provide a laser powder bed melting quality analysis method, device, equipment and medium. Compared with the prior art, the advantages thereof are as follows:

[0043] The present invention collects the surface grayscale image of the fused image, extracts the first-order statistical features of the image, and performs multivariate analysis on the feature data in multiple data dimensions and multiple attribute dimensions to obtain the potential relationship and correlation between the feature data, and thereby establishes a correlation between the grayscale features and the quality of the sample. On the whole, the image information is fully mined from the fused surface image of the sample without obvious defects, and the data features are visualized to achieve quality analysis of the sample without obvious defects. At the same time, the method of the present invention can realize automated detection, reduce manual participation, and make the detection results more objective and accurate. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 An overall flow diagram of a laser powder bed melting quality analysis method, device, equipment and medium provided by an embodiment of the present invention;

[0045] Figure 2 A schematic diagram of an image segmentation idea of ​​a laser powder bed melting quality analysis method, device, equipment and medium provided by an embodiment of the present invention;

[0046] Figure 3 A schematic diagram of a method, device, equipment and medium grayscale histogram drawing idea for laser powder bed melting quality analysis provided by an embodiment of the present invention;

[0047] Figure 4 A schematic diagram of a method, device, equipment and medium grayscale feature distribution scatter plot drawing idea for a laser powder bed melting quality analysis method, device, and medium provided in an embodiment of the present invention;

[0048] Figure 5A schematic diagram of image segmentation in a specific image of a laser powder bed melting quality analysis method, device, equipment and medium provided by an embodiment of the present invention;

[0049] Figure 6 A schematic diagram of characteristic processing of a laser powder bed melting quality analysis method, device, equipment and medium when the power changes provided in an embodiment of the present invention; wherein: (a) is a grayscale image of the sample; (b) is a grayscale distribution histogram; (c) is a line graph of process parameters and density roughness measurement values ​​of the corresponding sample. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present invention, so the present invention is not limited by the specific embodiments disclosed below.

[0051] See also Figure 1 The embodiment of the present invention provides a laser powder bed melting quality analysis method, device, equipment and medium, including the following steps:

[0052] Step 1: Collect sample quality and fused image information under different printing parameters.

[0053] Step 2: Segment the fused grayscale image under different printing parameters.

[0054] Step 3: Draw the grayscale distribution histogram of the fused image and extract the grayscale features.

[0055] Step 4: Conduct correlation analysis on printing parameters, sample quality and grayscale features.

[0056] Specifically:

[0057] Step 1: Collect sample quality and fused image information under different process parameters.

[0058] ① Set the process parameters in groups according to the laser additive manufacturing equipment and forming requirements. In each group, set one parameter as a variable and the other parameters as constants. At the same time, build an online monitoring system to prepare samples.

[0059] ② First, keep other printing parameters unchanged and set the laser power as the variable of the first group. Then set the scanning speed, scanning spacing and scanning direction as variables of the second, third and fourth groups respectively. After the parameter setting is completed, print the sample. Use the drainage method and roughness detector to detect the density and roughness of the printed samples, and collect the fused surface images under different process parameters through the online monitoring system during the printing process. Since there are too many fused image data during the printing process, uniform sampling is performed in the collected multi-layer fused surface images to obtain multiple fused images under each parameter, and the extracted fused images are subsequently processed.

[0060] Step 2: Segment the fused grayscale image under different process parameters.

[0061] ① Use the labelme annotation tool to polygonally mark the collected images and generate a json file.

[0062] ②Use the image segmentation algorithm to classify and annotate the grayscale images in the json file, and segment the fused images under different parameters from the collected images to extract the grayscale features under each parameter. In order to achieve fast segmentation of labeled images of arbitrary shapes, the segmentation algorithm of the labeled images is designed as follows:

[0063] Parse the json file to get the name of each marker type and the corresponding polygon vertex coordinates, fill the inside of the polygon with a grayscale value of 0, and fill the outside with a grayscale value of 255, forming a polygon with the same size as the original. Figure 1 The mask image is created and the original image and the mask image are converted into RGBA images; the four channel pixel values ​​of the original image pixels corresponding to the 255 pixels in the mask image are changed to (0, 0, 0, 0), that is, the pixel values ​​outside the marked polygon in the original image are changed to transparent (the α channel is 0), and the pixel values ​​inside the polygon remain unchanged. In this way, the original image is segmented into different categories according to the marked shape. The algorithm idea is as follows Figure 2 shown.

[0064] Step 3: Draw the fused grayscale distribution histogram and extract grayscale features.

[0065] ① The grayscale histogram drawing algorithm is used to draw the grayscale distribution histogram of the sample molten surface image under different process parameters, which are the grayscale distribution histograms when the laser power changes, the scanning speed changes, the scanning spacing changes, and the scanning direction changes. The process parameter variables are matched with the sample quality to obtain the change of the sample quality when the process parameters change; the grayscale histogram is calculated by the following formula:

[0066] P(r)=n r / (W*H)(r∈[0,L-1])

[0067] Where: W, H represent the width and height of the image respectively; n r represents the number of pixels with value r in the image; P(r) represents the frequency of pixels with value r in the image; L represents the grayscale level of the image. The idea of ​​grayscale histogram drawing algorithm is as follows:

[0068] Read the grayscale information of the segmented image, set the grayscale pixel interval according to the grayscale information, calculate the grayscale of the pixel and the number of pixels with this grayscale, count the total number of pixels and the probability of occurrence of each grayscale; use the grayscale of each pixel in the image as the horizontal axis and the frequency of occurrence of the grayscale as the vertical axis to draw a grayscale distribution histogram. The drawing idea is as follows Figure 3 shown.

[0069] ② The grayscale feature distribution map drawing algorithm is used to extract the grayscale features of the grayscale histogram of the fused image under all parameters, and then the eigenvalues ​​of the grayscale histogram are calculated and counted to calculate the grayscale features of each fused image. Finally, it is drawn as a grayscale feature distribution map of the fused image to visualize the image grayscale features.

[0070] There are 10 types of grayscale features extracted from the grayscale feature distribution histogram, including: mean, median, standard deviation, 3 / 4 quantile, 1 / 4 quantile, minimum, maximum, entropy, peak, and mean absolute deviation. Since there are too many features to analyze, four of them are taken for calculation and statistics, namely mean, minimum, entropy, and mean absolute deviation. Since a multi-layer image is extracted, there are multiple grayscale distribution feature values ​​under each parameter. The mean, minimum, entropy, and mean absolute deviation of the grayscale distribution feature values ​​of multiple grayscale images under each parameter are calculated, and the grayscale feature values ​​of multiple grayscale histograms under the same parameter are calculated. Then, the grayscale feature values ​​of multiple grayscale histograms under different parameters in the same group are calculated. The grayscale feature values ​​of the same group are plotted as a grayscale feature value distribution scatter plot. The final grayscale feature distribution graph visualizes the grayscale features and more intuitively reflects the distribution of grayscale feature values ​​under parameter changes. The relationship between parameters and grayscale characteristics is obtained by analyzing the graph. Figure 4 As shown in the figure, the idea of ​​the grayscale feature distribution scatter plot drawing algorithm is as follows:

[0071] Set the images of different layers under the same parameters to the same name, count the grayscale eigenvalues ​​of each image, and obtain the feature parameters such as the mean and minimum value, that is, the grayscale eigenvalues ​​of each image, and then store these eigenvalue data in a table to obtain the grayscale eigenvalue data tables of multiple groups of images under different parameters.

[0072] Taking the processing of eigenvalue data under power change as an example, the characteristic data in the table is read, the image grayscale eigenvalue data under power parameter change is taken, the category of the image drawn at the upper triangle position is set to a scatter plot, and the category of the image drawn at the lower triangle position and the diagonal line is set to a kernel density map; the image is classified according to the number name, and the data points under different powers are set to different colors, so that a matrix scatter plot of the grayscale eigenvalues ​​selected in the horizontal and vertical tables can be obtained; according to this step, the matrix scatter plots when other parameters change are drawn respectively, and the matrix scatter plots under each parameter change can be obtained, and the relationship between the parameters and the grayscale eigenvalues ​​can be analyzed through the drawn matrix scatter plots.

[0073] Step 4: Conduct correlation analysis on printing parameters, sample quality and grayscale features.

[0074] The obtained grayscale feature distribution diagram is analyzed, and the grayscale feature distribution of the molten image under different laser powers, scanning speeds, scanning intervals, and scanning directions is observed. The grayscale feature distribution law when the process parameters change is obtained, thereby establishing a correlation between the process parameters and the grayscale features. Based on the relationship between the process parameters and the sample quality obtained in step three, the correlation between the process parameters-sample quality=grayscale features is obtained.

[0075] According to the correlation between sample quality and printing parameters, the optimal printing parameters are corrected to determine the optimal printing parameters; according to the correlation between sample quality and grayscale characteristics, the sample quality is preliminarily judged by the distribution range of grayscale feature values. The parts with the mean, minimum value, entropy and average absolute deviation of the grayscale characteristics of the fused image within a certain range are qualified, and the rest are unqualified; by setting the value range, the judgment of unqualified parts can be made more accurate, and then the qualified parts can be tested, thereby improving the detection effect.

[0076] The specific tests are as follows:

[0077] like Figure 5 As shown in the figure, it is a schematic diagram of the image segmentation in the experiment; each layer of the fused image contains 4 groups (4 process parameters as variables) of 16 sample images, and the outline of each sample in the image is marked using the labelme program of the python software, and the coordinate information of the sample outline is saved as a json file; the coordinate information of the single sample outline is read from the json file, and the image is filled according to the coordinate information, so that the internal pixels of the single sample outline are 0 and the external pixels are 255, and then the filled image is used as a mask image and superimposed on the original image to obtain an image of a single sample with the original pixel value maintained, and this step is repeated to segment the grayscale images of the 16 samples in turn.

[0078] Read the grayscale image information of the single sample after segmentation, set the grayscale pixel interval according to the grayscale information, calculate the grayscale value of the pixel point and the number of pixels with this grayscale, and count the total number of pixels and the probability of occurrence of each grayscale. Draw a grayscale distribution histogram with the grayscale of each pixel in the image as the horizontal axis and the frequency of occurrence of the grayscale as the vertical axis; Figure 6 (a) shows the fused image without segmentation. Figure 6 (b) shows the grayscale distribution histogram of the fused image. Figure 6 (c) in the figure shows a line diagram of the process parameters and density roughness measurement values ​​of the samples corresponding to each fused image.

[0079] The mean, minimum, entropy and mean absolute deviation of the grayscale distribution eigenvalues ​​of multiple grayscale images under each parameter are calculated respectively, and the grayscale eigenvalues ​​of multiple grayscale histograms under the same parameter are statistically calculated. Then, the grayscale eigenvalues ​​of each grayscale histogram under different parameters in the same group are statistically calculated, and the grayscale eigenvalues ​​of the same group are plotted as a grayscale eigenvalue distribution scatter plot matrix; the final scatter plot matrix visualizes the changing rules of the grayscale features and clearly reflects the changing trend of the image grayscale distribution eigenvalues ​​when the process parameters change; the overall distribution of the data can be seen by visualizing the data through the kernel density map. The grayscale features of the sample images under different power parameters are obviously distributed in different areas. The data points with the same parameters have obvious clustering and have an obvious center point. The data points of the same type are approximately normally distributed around the center point.

[0080] The present invention collects grayscale images of the fused surface during the printing process, processes the first-order statistical features of the image, extracts the information contained in the data, visualizes the data information and establishes a relationship between the data and the powder bed melting quality; it provides a new idea for laser powder bed melting quality analysis. Compared with traditional physical detection methods, the quality detection achieved by the method of the present invention has a short detection time, is easy to realize automatic detection, is simple to operate, reduces manual labor, and the detection results are more objective and accurate.

[0081] The method of the present invention establishes a correlation between process parameters, sample quality and grayscale features through image processing and other algorithms. Therefore, the printing parameters can be optimized by analyzing the grayscale features of the fused image. The sample quality can be initially determined by analyzing the grayscale features of the fused image. The automatic selection of the optimal parameters and the automatic determination of the sample quality can be achieved by analyzing a large amount of data using machine learning.

[0082] The present invention selects multiple data dimensions or attributes for multivariate analysis of the fused image information data collected during the printing process, and obtains the potential relationship and correlation between the data through variable analysis. The method fully mines the image information, improves data utilization while obtaining deep correlation information of the data, and the analysis method visualizes the data features, making the results more intuitive and reliable.

[0083] The embodiment of the present invention further provides a laser powder bed melting mass analysis device, comprising:

[0084] The image module is used to obtain multiple fusion images during the sample preparation process with different process parameters.

[0085] The segmentation module is used to mark the polygon vertices of multiple fused images of samples with different process parameters, and segment the fused images according to the marked polygons according to the marked shapes to obtain multiple segmented images under different process parameters.

[0086] The feature solution module is used to obtain the image grayscale information of the segmented image, set the grayscale pixel interval according to the grayscale information, obtain the grayscale of each pixel point of multiple segmented images under different process parameters and the number of pixels with this grayscale; take the grayscale level of each pixel point in the segmented image as the horizontal coordinate and the number of pixels with this grayscale level as the vertical coordinate to obtain multiple grayscale distribution histograms of fused images with different process parameters.

[0087] According to multiple grayscale distribution histograms under the same process parameters, multiple grayscale eigenvalues ​​of the corresponding fused images are obtained; multiple grayscale eigenvalue statistics of multiple fused images under different process parameters are obtained respectively, the grayscale features of the fused images under the same process parameters are classified into one category, and a matrix scatter plot under different process parameters is established according to multiple grayscale eigenvalue statistics; the correspondence between the process parameters and the grayscale eigenvalues ​​is obtained according to the matrix scatter plot, and the grayscale feature distribution law when the process parameters change is obtained according to the corresponding relationship.

[0088] The analysis module determines the quality of samples under different process parameters through the distribution range of grayscale eigenvalues.

[0089] An embodiment of the present invention further provides an electronic device, including a memory and a processor.

[0090] The memory is used to store computer programs.

[0091] When the processor is used to execute the computer program stored in the memory, the steps of the above-mentioned laser powder bed melting quality analysis method are implemented.

[0092] An embodiment of the present invention further provides a computer-readable storage medium for storing a computer program, which implements the steps of the above-mentioned laser powder bed melting quality analysis method when executed by a processor.

[0093] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A laser powder bed melting quality analysis method, characterized in that: The following steps are involved: Obtain multiple fusion images during sample preparation with different process parameters; Marking multiple fused images of samples with different process parameters with polygon vertices, segmenting the fused images according to the marked polygons and the marked shapes, to obtain multiple segmented images under different process parameters; Obtaining image grayscale information of the segmented image, setting the grayscale pixel interval according to the grayscale information, and obtaining the grayscale of each pixel point of multiple segmented images under different process parameters and the number of pixels with the grayscale; Taking the gray level of each pixel in the segmented image as the horizontal coordinate and the number of pixels with the gray level as the vertical coordinate, multiple gray distribution histograms of the fused images with different process parameters are obtained; According to multiple grayscale distribution histograms under the same process parameters, multiple grayscale characteristic values ​​of the corresponding fused image are obtained; A plurality of grayscale eigenvalue statistics of a plurality of fused images under different process parameters are respectively obtained, the grayscale features of the fused images under the same process parameters are classified into one category, and a matrix scatter plot under different process parameters is established according to the plurality of grayscale eigenvalue statistics; The correspondence between process parameters and grayscale eigenvalues ​​is obtained according to the matrix scatter plot, and the grayscale feature distribution law when the process parameters change is obtained according to the correspondence, and the quality of samples under different process parameters is judged by the distribution range of grayscale eigenvalues.

2. A laser powder bed melting quality analysis method according to claim 1, characterized in that: The method of obtaining multiple fused images during the process of preparing samples with different process parameters includes: Set the process parameters according to the laser additive manufacturing equipment and forming requirements. Set one parameter as a variable in each group and set the remaining parameters as constants. Keep other printing parameters unchanged, set the laser power as the variable of the first group, and then set the scanning speed, scanning spacing, and scanning direction as variables of the second, third, and fourth groups respectively, and print the sample; The printed samples were tested for density and roughness using the drainage method and a roughness tester, and multiple fused images were collected under different process parameters.

3. A laser powder bed melting quality analysis method according to claim 1, characterized in that: The step of obtaining a plurality of segmented images under different process parameters includes: Use the labelme annotation tool to polygonally mark multiple fused images collected under different process parameters and generate a json file; Parse the json file to obtain the name of each marker type and the coordinates of the polygon vertices corresponding to the type, fill the inside of the polygon with pixels with a grayscale value of 0, and fill the outside of the polygon with pixels with a grayscale value of 255 to form a mask image with the same size as the original image, and convert the mask image into an RGBA image; The pixel values ​​of the four channels of the original image pixels corresponding to the 255 pixels in the mask image are transformed to (0, 0, 0, 0), that is, the pixel values ​​outside the marked polygon in the original image are transformed to transparent, and the pixel values ​​inside the marked polygon in the original image remain unchanged, so as to obtain multiple segmented images under different process parameters.

4. The laser powder bed melting quality analysis method according to claim 1, characterized in that: The method of obtaining a plurality of grayscale distribution histograms of the fusion images with different process parameters comprises: Read the image grayscale information in the segmented image, set the grayscale pixel interval according to the grayscale information, obtain the grayscale of each pixel point and the number of pixels with the grayscale in multiple segmented images under different process parameters, and count the total number of pixels and the occurrence probability of each grayscale; Taking the gray level of each pixel in the image as the horizontal coordinate and the frequency of the gray level as the vertical coordinate, multiple gray distribution histograms of the fused images with different process parameters are obtained; the method for obtaining the frequency of the gray level in the gray distribution histogram is: P(r)=n r / (W*H) (r∈[0,L-1]) Where: W, H represent the width and height of the image respectively; n r It represents the number of pixels with value r in the image; P(r) represents the frequency of pixel value r in the image; L represents the number of gray levels of the image.

5. The laser powder bed melting quality analysis method according to claim 1, characterized in that: The establishment of the matrix scatter plot under different process parameters includes: The images of different layers under the same process parameters are set to the same name, multiple grayscale feature values ​​of multiple fused images are counted, and the statistics of each grayscale feature value under the multiple fused images are obtained according to the multiple grayscale feature values ​​of the multiple fused images; The fused images are classified according to their serial numbers, and the data points under different process parameters are set to different colors. A matrix scatter plot under different process parameters is established according to the statistics of each grayscale feature value under different colors.

6. A laser powder bed melting quality analysis method according to claim 5, characterized in that: The statistics of each grayscale eigenvalue include: The grayscale characteristic values ​​include: mean, median, standard deviation, 3 / 4 quantile, 1 / 4 quantile, minimum value, maximum value, entropy, peak value and mean absolute deviation; The feature quantities represented by the taken eigenvalues ​​are different, and the statistics of the mean value, minimum value, entropy, and mean absolute deviation are determined as each grayscale feature.

7. The laser powder bed melting quality analysis method according to claim 1, characterized in that: The judgment of the sample quality includes: By observing the grayscale feature matrix scatter plot of the fused image under different laser powers, scanning speeds, scanning intervals and scanning directions, the corresponding relationship between the process parameters and the grayscale feature values ​​is obtained according to the matrix scatter plot, and the grayscale feature distribution law when the process parameters change is obtained according to the corresponding relationship; The quality of the sample is judged according to the distribution range of the grayscale feature value. When the average value, minimum value, entropy and average absolute deviation of the grayscale feature of the fused image are within the determined range, the sample is qualified, and the rest are unqualified.

8. A laser powder bed melting mass analysis device, characterized in that: include: An image module is used to obtain multiple fusion images during the preparation of samples with different process parameters; A segmentation module is used to mark the polygon vertices of multiple fused images of samples with different process parameters, and segment the fused images according to the marked polygons according to the marked shapes to obtain multiple segmented images under different process parameters; The feature solving module is used to obtain the image grayscale information of the segmented image, set the grayscale pixel interval according to the grayscale information, and obtain the grayscale of each pixel point and the number of pixels with the grayscale of multiple segmented images under different process parameters; Taking the gray level of each pixel in the segmented image as the horizontal coordinate and the number of pixels with the gray level as the vertical coordinate, multiple gray distribution histograms of the fused images with different process parameters are obtained; According to multiple grayscale distribution histograms under the same process parameters, multiple grayscale characteristic values ​​of the corresponding fused image are obtained; A plurality of grayscale eigenvalue statistics of a plurality of fused images under different process parameters are respectively obtained, the grayscale features of the fused images under the same process parameters are classified into one category, and a matrix scatter plot under different process parameters is established according to the plurality of grayscale eigenvalue statistics; Obtain the corresponding relationship between the process parameters and the grayscale eigenvalues ​​according to the matrix scatter plot, and obtain the grayscale feature distribution law when the process parameters change according to the corresponding relationship; The analysis module determines the quality of samples under different process parameters through the distribution range of grayscale eigenvalues.

9. An electronic device, characterized in that: include: Memory and processor; The memory is used to store computer programs; The processor is used to implement the steps of a laser powder bed melting quality analysis method as described in any one of claims 1 to 7 when executing the computer program stored in the memory.

10. A computer-readable storage medium, characterized in that: Used to store a computer program, which, when executed by a processor, implements the steps of a laser powder bed melting quality analysis method as described in any one of claims 1 to 7.

Images