Image information generation device, image information generation method, image processing device, image processing method, defect prediction device, defect prediction method, and program

Abstract

An image information generation device generates image information indicating the shape of a molded object in which layers comprising a plurality of beads are stacked. The image information generation apparatus includes: a coordinate information acquisition unit that approximates a design shape for each layer of a model using a model that simulates a shape of a bead, and acquires coordinate information of a plurality of points of the approximated model; a contour extraction unit that finds a contour indicating the surface shape of the specified layer on the basis of the coordinate information; and an image information conversion unit that uses the contour to convert the uneven shape of the surface of the specified layer into image information.

Description

Technical Field

[0001] The present invention relates to an image information generating device and an image information generating method, an image processing device and an image processing method, a defect prediction device and a defect prediction method and a program. Background Art

[0002] In arc welding, there is a known technique for detecting defects generated in a structure being welded to determine whether the welding work has been performed properly. For example, Patent Document 1 discloses a technique for obtaining a plurality of information related to the behavior of the welder, the shape of the molten pool, and the length of the protruding wire in semi-automatic welding to determine the quality of the welding work (Patent Document 1).

[0003] Prior Art Literature

[0004] Patent Literature

[0005] Patent Document 1: Japanese Patent Application Publication No. 2008-110388 Summary of the invention

[0006] -Problems to be solved by the invention-

[0007] In the stacking molding of the molded object obtained by stacking weld beads by arc welding, a narrow portion may be generated between weld beads if the welding conditions or the weld bead forming path are inappropriate. If molding is continued in a state where a narrow portion is generated, a welding defect of poor fusion (non-fusion) may occur at the position of the narrow portion. Therefore, it is important to select welding conditions or weld bead forming paths that do not generate welding defects based on the possibility of the unevenness of the stacked shape and the generation of defects in advance.

[0008] However, in stacking molding, there is a tendency for the number of welds and the length of welds to increase compared to normal welding, and it is difficult to predict the shape of the molded object before molding and optimize the welding conditions. In addition, the more complex the shape of the molded object, the more complex the above prediction and optimization processes.

[0009] Therefore, the object of the present invention is to provide an image information generating device and an image information generating method, an image processing device and an image processing method, a defect prediction device and a defect prediction method, and a program, which can easily visually grasp information such as the concave and convex contours, excessive or insufficient stacking height, etc. even for objects with complex shapes that are difficult to grasp.

[0010] -Methods for solving problems-

[0011] The present invention is constituted by the following structure.

[0012] (1) An image information generating device for generating image information representing the shape of a shaped object in which layers composed of a plurality of weld beads are stacked, comprising:

[0013] A coordinate information acquisition unit that uses a model that simulates the shape of the weld bead to approximate the design shape of each layer of the shaped object, and acquires coordinate information of a plurality of points of the approximated model;

[0014] a contour extraction unit that obtains a contour representing a surface shape of the designated layer based on the coordinate information; and

[0015] The image information conversion unit converts the concavo-convex shape of the surface of the designated layer into image information using the contour.

[0016] (2) An image processing device comprising:

[0017] a narrow portion extracting unit that extracts a concave narrow portion formed by the welding bead by performing image processing based on the image information generated by the image information generating device according to (1); and

[0018] The feature quantity calculation unit calculates a feature quantity including any one of the position and the size of the extracted stenosis portion.

[0019] (3) A defect prediction device comprising:

[0020] (1) The image information generating device;

[0021] a narrow portion extracting unit for extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the image information generating device;

[0022] a feature quantity calculation unit that calculates a feature quantity including any one of the position and size of the extracted stenosis;

[0023] a stacking condition acquisition unit that acquires a stacking condition of the weld bead that forms the narrow portion; and

[0024] The defect determination unit predicts the presence or absence of a defect generated in the narrow portion of the object or the size of the defect based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

[0025] (4) An image information generating method for generating image information representing the shape of a molded object having a plurality of weld beads stacked on top of each other,

[0026] approximating the design shape of each layer of the object using a model that simulates the shape of the weld bead, and obtaining coordinate information of a plurality of points of the approximated model;

[0027] deriving a contour representing the surface shape of the designated layer based on the coordinate information,

[0028] The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

[0029] (5) An image processing method,

[0030] Based on the image information generated by the image information generation method described in (4), a concave narrow portion formed by the weld bead is extracted by performing image processing,

[0031] A feature quantity including either the position or the size of the extracted stenosis is calculated.

[0032] (6) A defect prediction method,

[0033] Based on the image information generated by the image information generation method described in (4), a concave narrow portion formed by the weld bead is extracted by performing image processing,

[0034] calculating a feature quantity including any one of the position and size of the extracted stenosis,

[0035] obtaining the lamination conditions of the weld bead forming the narrow portion,

[0036] The presence or absence of a defect generated in the narrow portion of the object or the size of the defect is predicted based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

[0037] (7) A program for executing the step of generating image information representing the shape of a molded object in which layers composed of a plurality of weld beads are stacked,

[0038] Used to make the computer perform the following steps:

[0039] Using a model that mimics the shape of the weld bead to approximate the design shape of each layer of the object, and obtaining coordinate information of a plurality of points of the approximated model;

[0040] deriving a contour representing a surface shape of the designated layer based on the coordinate information; and

[0041] The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

[0042] -Effects of the Invention-

[0043] According to the present invention, even for a complex and difficult-to-grasp object, image information can be obtained that allows easy visual grasp of information such as the contour of the concave and convex, the excess or deficiency of the stacking height, etc. Thus, the defect size can be predicted with high accuracy, which helps to suppress the occurrence of defects. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is the overall structural diagram of the welding system.

[0045] Figure 2 It is a structural block diagram of a defect prediction device.

[0046] Figure 3 This is a flowchart showing the steps of a defect prediction method performed by a defect prediction device.

[0047] Figure 4 This is an explanatory diagram showing a case where the shape of a modeled object is reproduced by a plurality of unit models.

[0048] Figure 5 This is an explanatory diagram showing a dot matrix of vertices of a unit model.

[0049] Figure 6 This is an explanatory diagram showing a row of unit models extending in the depth direction.

[0050] Figure 7 This is an explanatory diagram showing a case where the shape of a modeled object is reproduced by an elliptical unit model.

[0051] Fig. 8A This is an explanatory diagram showing a state where contours are extracted from a plurality of unit models that approximate the shape of a modeled object.

[0052] Figure 8B This is an explanatory diagram showing a state where contours are extracted from a plurality of unit models that approximate the shape of a modeled object.

[0053] Fig. 9A This is a schematic diagram of an image based on image information expressed in the form of contour lines.

[0054] Fig. 9B Yes means Fig. 9A A graph of the pixel values ​​on the IX-IX line.

[0055] Fig.10 This is an explanatory diagram showing a situation where modeling is performed on an inclined base.

[0056] Fig.11 This is an explanatory diagram showing an example of characteristic quantities of a stenosis portion.

[0057] Fig.12 This is a functional block diagram of a defect determination unit included in a defect prediction device.

[0058] Fig.13 This is a model diagram showing the contents of a stacking plan for a modeled object based on multiple weld beads.

[0059] Fig.14 It means along Fig.13 A graph of the height distribution of the model in the cross section of line XIV-XIV.

[0060] Fig.15 It means from Fig.13 The information of the dot matrix extracted from the model diagram is visualized as a result of the image.

[0061] Fig.16 Yes Fig.15 The image shown is a result of image processing in which the stenosis is emphasized. DETAILED DESCRIPTION

[0062] Here, an example of a stacking molding device for manufacturing a molded object by stacking weld beads with arc welding is described, but the manufacturing method of the molded object and the device structure for stacking molding are not limited to this.

[0063] <Structure of stacking molding device>

[0064] Figure 1 1 is an overall structural diagram of a stacking molding apparatus. The stacking molding apparatus 100 includes a molding unit 11 and a control device 13. The molding unit 11 includes a robot 15, a robot control unit 17, a filling material supply unit 19, and a heat source control unit 21.

[0065] The robot control unit 17 controls the robot 15 and the heat source control unit 21. A controller (not shown) is connected to the robot control unit 17, and an operator can instruct arbitrary operations to the robot control unit 17 via the controller.

[0066] The manipulator 15 is, for example, a multi-joint robot, and is supported so as to be able to continuously supply a filler material M to a welding torch 23 provided on a front end axis. The welding torch 23 holds the filler material (also referred to as welding wire) M in a state of protruding from the front end. The position and posture of the welding torch 23 can be arbitrarily set three-dimensionally within the range of the degrees of freedom of the robot arm constituting the manipulator 15. The manipulator 15 preferably has more than six degrees of freedom, and preferably allows the axial direction of the heat source at the front end to be arbitrarily changed. In addition to Figure 1 In addition to the multi-joint robot with four or more axes shown above, various types of robots such as a robot having angle adjustment mechanisms on two or more orthogonal axes may be used.

[0067] The welding torch 23 has a shielding nozzle not shown in the figure, and a shielding gas is supplied from the shielding nozzle. The shielding gas blocks the atmosphere, prevents oxidation and nitridation of the molten metal during welding, and suppresses poor welding. The arc welding method used in this structure can be any of a consumable electrode type such as coated arc welding or carbon dioxide arc welding, TIG (Tungsten Inert Gas: tungsten inert gas shielded welding) welding, or plasma arc welding non-consumable electrode type, which can be appropriately selected according to the modeling object. Here, metal arc welding is taken as an example for explanation. In the case of the consumable electrode type, a conductive nozzle is arranged inside the shielding nozzle, and the filler material M to which the current is supplied is held by the conductive nozzle. The welding torch 23 holds the filler material M, and generates an arc from the front end of the filler material M under a shielding gas atmosphere.

[0068] The filling material supply unit 19 supplies the filling material M toward the welding torch 23. The filling material supply unit 19 includes a reel 19a on which the filling material M is wound and an extraction mechanism 19b for extracting the filling material M from the reel 19a. The filling material M is fed to the welding torch 23 while being conveyed in the forward direction or the reverse direction as required by the extraction mechanism 19b. The extraction mechanism 19b is not limited to a push-out type that is arranged on the side of the filling material supply unit 19 and squeezes out the filling material M, but may also be a pull-out type or a push-pull type that is arranged on a robot arm or the like.

[0069] The heat source control unit 21 is a welding power source that supplies power required for welding of the robot 15. The heat source control unit 21 adjusts the welding current and welding voltage supplied when forming a weld bead that melts and solidifies the filler material M. In addition, the filler material supply speed of the filler material supply unit 19 is adjusted in conjunction with the welding conditions such as the welding current and welding voltage set by the heat source control unit 21.

[0070] The heat source used to melt the filler material M is not limited to the above-mentioned electric arc. For example, a heating method using a combination of electric arc and laser, a heating method using plasma, a heating method using electron beam or laser, and other heat sources may also be used. In the case of heating using an electron beam or laser, the amount of heat can be more finely controlled, and the state of the formed weld bead can be more appropriately maintained, which helps to further improve the quality of the stacked structure. In addition, there is no particular limitation on the material of the filler material M, such as low-carbon steel, high-strength steel, aluminum, aluminum alloy, nickel, nickel-based alloy, etc. The type of filler material M used may vary depending on the characteristics of the molded object Wk.

[0071] The control device 13 controls the above-mentioned components in a centralized manner. The control device 13 is constituted by hardware using an information processing device such as a PC (Personal Computer).

[0072] The stacking molding device 100 of the above structure operates according to the molding program created based on the stacking plan of the molded object Wk. The molding program is composed of a plurality of command codes and is created based on an appropriate algorithm according to various conditions such as the shape, material, and heat input of the molded object Wk. If the welding torch 23 is moved according to the molding program, the fed filler material M is melted and solidified, and a linear weld bead B as a molten solidified body of the filler material M is formed on the base 25. That is, the robot control unit 17 drives the robot 15 and the heat source control unit 21 based on the given molding program provided from the control device 13. The robot 15 forms the weld bead B while melting the filler material M using an arc according to the instruction from the robot control unit 17. By sequentially forming and stacking the weld bead B in this way, the molded object Wk of the target shape can be obtained.

[0073] Next, a stacking plan of an object to be formed by the stacking forming apparatus 100 will be described.

[0074] The stacking plan is determined based on a given algorithm according to the shape of the object to be molded and the specifications of each part constituting the stacking molding device 100. Specifically, the stacking plan includes information on the trajectory (path) for moving the welding torch 23, information on welding conditions for forming a weld bead, etc. The specific steps for determining the stacking plan are well known, so the description thereof is omitted here.

[0075] The more complex the shape of the object, the more difficult it is to grasp the content of the stacking plan. Therefore, it is preferred to represent the stacking plan with image information, and the image information can be displayed on a display unit such as a monitor. In this case, the operator can easily visually confirm the content of the stacking plan, extract the characteristic parts and evaluate them. That is, if the stacking plan can be visualized, it is easy to extract the factors that cause various defects such as unwelded defects, and it is also possible to predict defects based on the information of the extracted defect factors. As a defect factor, for example, a narrow portion formed by a weld bead can be listed. It is known that unwelded welds are easily generated in narrow portions. Here, the presence or absence of defects or the size of defects is predicted based on the information of the narrow portion. In the following description, the narrow portion is described as a defect factor, but it is not limited to this, and other factors can also be extracted to predict defects.

[0076] Next, a defect prediction method and a defect prediction device for predicting defect information included in a narrow portion in a molded object based on a stacking plan will be described in detail below.

[0077] Figure 2 It is a structural block diagram of the defect prediction device 200.

[0078] The defect prediction device 200 includes a stacking condition acquisition unit 210 that acquires stacking conditions from a prepared stacking plan, an image information generation device 220 , an image processing device 230 , and a defect determination unit 250 .

[0079] The image information generating device 220 uses a model that simulates the shape of a weld bead to represent the design shape of each layer of the modeling object based on the stacking plan, and generates image information representing the surface shape of the designated layer based on the coordinate information of multiple points of the model. The designation of the layer can be arbitrarily set by the user of the defect prediction device. The image information generating device 220 includes a coordinate information acquisition unit 31, a contour extraction unit 33, and an image information conversion unit 35, which will be described in detail later. The coordinate information acquisition unit 31 uses a model that simulates the shape of a weld bead to approximate the design shape of each layer of the modeling object, and acquires the coordinate information of multiple points of the approximated model. The contour extraction unit 33 obtains the contour representing the surface shape of the designated layer based on the obtained coordinate information. The image information conversion unit 35 uses the obtained contour to convert the concave and convex shape of the surface of the designated layer into image information.

[0080] In addition, the image processing device 230 includes: a narrow portion extraction unit 37, which performs image processing on the image information generated by the image information generation device 220 to extract the concave narrow portion formed by the weld bead; and a feature quantity calculation unit 39, which calculates a feature quantity including any of the position and size of the extracted narrow portion based on the image information.

[0081] The defect prediction device 200 may include a display unit 240 for displaying the image information output from the image information generating device 220 and the feature amount information output from the image processing device 230. By displaying the output information on the display unit 240, the set stacking conditions can be easily understood visually.

[0082] The defect prediction device 200 further includes a defect determination unit 250 that predicts the presence or size of a defect such as an unmelted defect included in the narrow portion of the molded object based on the feature value of the narrow portion calculated by the feature value calculation unit 39 and the stacking conditions of the narrow portion.

[0083] The defect prediction device 200 is similar to the control device 13 and is composed of hardware using an information processing device such as a PC (Personal Computer). The control function of the defect prediction device 200 is realized by a control device (not shown) reading a program with a specific function stored in a storage device and executing the program. As a control device, a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processor Unit) or a dedicated circuit can be cited. As a storage device, a memory such as a RAM (Random Access Memory) or a ROM (Read Only Memory), a storage device such as a HDD (Hard Disk Drive) or a SSD (Solid State Drive) can be exemplified. In addition to being configured to include various structural elements such as the above-mentioned image information generating device 220 and the image processing device 230, the defect prediction device 200 can also be a method of remotely connecting various structural elements via a network or the like. In addition, the image processing device 230 can also be configured to be suitable for high-speed processing including a GPU (Graphics Processing Unit) in a processor.

[0084] Figure 3 200 is a flowchart showing the steps of a defect prediction method of the defect prediction device.

[0085] First, the stacking condition acquisition unit 210 acquires control information related to the stacking of weld beads during the stacking modeling of the modeling object, that is, information on the stacking plan (S1). For example, various stacking conditions required for prediction are extracted from the prepared stacking plan of the modeling object. The required stacking conditions include information related to the wire movement trajectory of the welding torch, information related to the welding torch posture during wire movement, etc. The information related to the wire movement trajectory includes, for example, the three-dimensional coordinates (x, y, z) of a specific point P located on the trajectory (modeling path). In addition, the information related to the welding torch posture includes, for example, the angles (α, β, γ) around each coordinate axis. In addition, the coordinate information and the information on the welding torch posture are not limited to the above examples. For example, it can also be information on a cylindrical coordinate system (r, θ, z) represented by polar coordinates. Regarding the welding torch posture, it can also be determined according to the joint angles or joint positions of the manipulator holding the welding torch. In addition, it can also include welding current, welding voltage, welding wire feed speed, wire movement trajectory pattern (whether there is swing), and the path time from the end of a specific modeling path to the beginning of the next path. The coordinate information acquisition unit 31 of the image information generating device 220 extracts the above-mentioned coordinate information from the input stacking conditions.

[0086] Next, the shape of the shaped object is predicted based on the acquired stacking conditions ( S2 ). Here, an outline is obtained by expressing the stacking shape when a plurality of weld beads are stacked using a model that simulates the weld bead shape of the weld bead.

[0087] Figure 4 This is an explanatory diagram showing the situation where the shape of a modeled object is reproduced by multiple unit models BM. The contour Prf of the stacked shape is preferably predicted using a unit model BM that simulates the shape of the weld bead. By using the unit model BM, the shape can be simplified and the calculation of coordinate values ​​and the like can be reduced. Specifically, multiple rows of unit models BM representing the shape of the weld bead are arranged in multiple layers along the modeling path so as to approximate the shape of the modeled object to be modeled. The shape of the unit model BM itself changes according to the welding conditions (welding current, welding voltage, wire feed speed, welding speed, etc.). Furthermore, each unit model BM preferably makes the overlap of adjacent weld beads, the droop of the weld beads, etc. reflected in the model shape according to its configuration state. The shape of the unit model BM can be various shapes other than a trapezoid or a shape close to a trapezoid, such as a polygonal shape, an elliptical shape, etc., without any particular limitation.

[0088] The multiple unit models BM configured in this way can reproduce the stacked shape of the modeled object by applying information such as welding conditions and target positions of weld bead formation. Based on the stacked shape reproduced by the multiple unit models BM, the shape of the outer surface of the model formed by overlapping the unit models BM can be grasped. Figure 4 As shown in FIG. 1 , this shape is represented by a contour Prf formed by connecting the outermost edges of the unit model BM. The contour Prf is extracted by Figure 2 The contour extraction unit 33 shown in the figure is used for the modeling. The contour Prf of the outer surface of the model can be extracted for each modeling path or for each layer. That is, as long as the contour Prf representing the stacked shape in any layer is extracted at any time during the modeling, the stacked shape in the layer at that time can be freely predicted.

[0089] Next, based on the information of the contour Prf representing the shape of the model outer surface of the designated layer, the information of the three-dimensional dot matrix is ​​generated (S3). Figure 4 When the straight lines are bent as shown, a plurality of points P that are corners of each straight line can be used as a lattice. In addition, each vertex in the shape of each unit model BM can be used as a lattice.

[0090] Figure 5 is an illustration of a dot matrix representing the vertices of a unit model. Figure 5As shown, the points P of the vertices of the unit model BM with a triangle attached to the base of the trapezoid or the trapezoid are used as the information of the lattice. For example, the vertices of the unit model BM of the first layer can be defined as the lattice of the first layer. In addition, in the case of the unit model BM of the second layer, the vertices exposed outside the unit model of the second layer and the vertices of the unit model of the first layer exposed outside from the unit model of the second layer can be defined as the lattice of the second layer.

[0091] In addition, the information of the dot matrix is ​​not only from Figure 5 The points on the cross section shown are extracted from Figure 5 Extract multiple unit models that extend in the depth direction (weld bead formation direction).

[0092] Figure 6 This is an explanatory diagram showing a row of unit models extending in the depth direction. The above-mentioned unit model BM is described with a cross-sectional shape in the zx plane, but the actual unit model BM, for example, is a hexahedron with a total of eight vertices when the cross-sectional shape is a trapezoid. In this way, multiple points of the three-dimensional unit model BM are extracted as information on the dot matrix.

[0093] The above-mentioned point matrix may be points other than the vertices of the unit model BM.

[0094] Figure 7 This is an explanatory diagram showing the case where the shape of a modeled object is reproduced by an elliptical unit model. In the case where the unit model reproduces the stacked shape with a smooth curve such as an ellipse, the collection of points P obtained by sampling the outer surface shape of the unit model at a certain interval (pitch Pt) can also be defined as the above-mentioned dot matrix. The information possessed by the dot matrix is ​​not limited to the orthogonal coordinate system (x, y, z), and can also be converted into a preferred coordinate system including the polar coordinate system (r, θ, z). In addition, in this case, as described above, the points P sampled at a given certain interval are also obtained in the depth direction, and these points P are defined as a dot matrix.

[0095] Fig. 8A , Figure 8B 1 is an explanatory diagram showing a case where a contour Prf is extracted from a plurality of unit models having a shape similar to that of a modeled object. The modeled object illustrated here has a frame-shaped wall portion and a filling portion that fills a region surrounded by the wall portion. Fig. 8A , Figure 8B A cross section of a model that represents a portion of it.

[0096] exist Fig. 8A4-layer unit models BMw constituting the wall portion and two-layer unit models BM1 and BM2 constituting the filling portion whose bead width and bead height are set larger than the wall portion in the rib portion of the unit model BMw are shown. The unit model BM1 is a model of the first layer, and the unit model BM2 is a model of the second layer. Each of the unit models BMw, BM1, and BM2 is a shape based on a trapezoid, and it is assumed that weld beads are formed in sequence in the filling portion from the side close to the unit model BMw that becomes the wall portion.

[0097] Fig. 8A The stacked shape of the object shown can be set to Figure 8B The contour Prf of the outer surface of the outermost unit model is shown. The contour Prf is also shown Figure 6 As shown, it is Figure 8B The outline of the three-dimensional shape that is also connected in the depth direction represents the height distribution on the xy plane.

[0098] Next, the information of the dot matrix obtained as described above is converted into a contour map showing the shape of the object ( S4 ). Specifically, image information (contour map) showing the concavities and convexities on the surface of the object as contour lines with respect to a plane in a specific direction is generated. Figure 2 The image information conversion unit 35 shown in the figure generates information of a three-dimensional dot matrix based on the information of the above-mentioned profile Prf and converts it into a contour map.

[0099] Fig. 9A This is a schematic diagram of an image based on image information expressed in the form of contour lines. Fig. 9B Yes means Fig. 9A The image 51 represents the height information as pixel values ​​(shades) on a two-dimensional plane, as Fig. 9B The pixel value of the height information is represented by Figure 8B The height distribution of the profile Prf is shown.

[0100] The operator passes Figure 2 By observing the image 51 on a display unit 240 such as a monitor, the state of the concave and convex of the dot matrix information that is difficult to understand intuitively can be easily captured visually. In addition, by converting the dot matrix information into image information, the amount of information to be stored can be reduced. Furthermore, if image processing techniques such as binarization are applied, the desired feature information can be easily extracted.

[0101] In addition, the pixel value of the image information for contour display may be the stacking height z, or the difference between the planned height z0 based on the stacking plan and the stacking height z obtained by modeling, or the height difference in a specific small unit interval, etc. That is, in addition to the absolute value of the height formed by the bumps and depressions of the surface shape, it may also be a value indicating the degree of inclination of the bumps and depressions. When the pixel value of the image information is set to the difference between the planned height z0 and the stacking height z, it becomes appropriate information when determining the excess or insufficient weld area of ​​the weld bead. Furthermore, the inclination angle of the bumps and depressions relative to a specific axial direction may also be set as the value for contour display. For example, Figure 8B If the above-mentioned inclination angle is defined as inclination from the vertical direction (z-axis direction), the inclined part is highlighted. Therefore, it is easy to evaluate the flatness of the molding surface, and it is easy to find the situation when there is a large concave part 53.

[0102] In addition, Fig. 9A In the figure, the height in the z-axis direction relative to the xy plane is displayed as a contour line, but the present invention is not limited to this. For example, the value in the x-direction relative to the yz plane may be displayed as a contour line, or other values ​​may be displayed.

[0103] Fig.10 This is an explanatory diagram showing the case of modeling on an inclined base 25. When modeling a modeled object in layers, the base 25 is sometimes tilted from the horizontal plane (xy plane) and a weld bead B is formed on the inclined base surface for layering. In this case, it is preferred that the object displayed in the above-mentioned contour line form is not the height z of the coordinate system (x, y, z), but the value of the zL axis in the local coordinate system (xL, yL, zL) in which the inclined base surface is set as the xL-yL plane and the normal direction of the base surface is set as the zL direction. This makes it easy to grasp the shape change caused by the stacking height of the weld bead of the modeled object.

[0104] Next, the narrow portion of the weld bead is extracted from the obtained contour map ( S5 ). Figure 2 The narrow portion extraction unit 37 of the image information generating device 220 shown extracts the narrow portion generated between the weld beads formed based on the stacking plan (between the unit models BM or between the rows of the unit models) from the image information converted by the image information conversion unit 35. The narrow portion extraction unit 37 outputs the information such as the position and size of the extracted narrow portion to the feature quantity calculation unit 39. The feature quantity calculation unit 39 calculates the feature quantity of the narrow portion based on the input information of the narrow portion and the coordinate information of the narrow portion and the periphery of the narrow portion.

[0105] The stenosis extraction unit 37 extracts the stenosis by performing appropriate image processing on the input image information. For example, a stenosis with a sharp depression can also be extracted by binarization processing or the like. Specifically, for an image of the image information, a depressed area extracted by binarization processing with a specific pixel value as a threshold is captured as a stenosis. In addition, a reference mask representing a specific concave shape can be prepared, and a portion with a high degree of similarity to the concave shape can be obtained by mask processing, or the bottom position can be obtained by applying the steepest descent method to the image information. The extraction method is not particularly limited. The feature quantity calculation unit 39 obtains various feature quantities such as the size and center position of the stenosis extracted by the stenosis extraction unit 37. In addition, the outer surface shape near the stenosis can also be obtained as the feature quantity of the stenosis.

[0106] Fig.11 is an explanatory diagram showing an example of characteristic quantities of a stenosis. Fig.11 As schematically shown, a pair of adjacent weld beads are formed on the base surface FL representing the surface of the existing weld bead (not shown) of the base 25 or the lower layer, and a narrow portion is formed between the pair of weld beads. In this case, the direction perpendicular to the direction of weld bead formation ( Fig.11 In the left and right directions, the interval between the model lower ends Pb1 and Pb2 at the bottom of the valley formed by the pair of models BMa and BMb representing the weld bead is set as the bottom interval U, and the interval between the top Pt1 of the model BMa on one side and the top Pt2 of the model BMb on the other side is set as the weld bead interval W. The bottom interval U may also be the interval in the long side direction of the pair of models BMa and BMb (the interval in the direction from the starting end toward the ending end of the weld bead). In addition, the average value of the height from the base surface FL to the top Pt1 of the model BMa and the height from the base surface FL to the top Pt2 of the model BMb is set as the average height H. The average height H is equivalent to the valley depth of the valley formed by the pair of models BMa and BMb in the stacking direction. These bottom interval U, weld bead interval W, and average height H may also be used as characteristic quantities of the narrow portion.

[0107] The calculation of the above-mentioned U, W, and H can also be obtained by preparing a model function that simulates the shape of the stenosis in advance, and fitting the model function to the shape of the stenosis. For example, a specific reference profile function that can approximate the shape of the stenosis of the profile Prf can be selected by using the above-mentioned profile Prf, and the selected reference profile function can be fitted to set the feature value corresponding to the parameter value of the function. As the reference profile function, for example, it can be a quadratic curve, a hyperbola, a Gaussian curve, an arc of a specific curvature radius, a V-shaped line of a specific angle, etc.

[0108] Next, based on the characteristic amount of the narrow portion calculated as described above and information on the lamination conditions of the weld bead forming the narrow portion, prediction of defect occurrence or prediction of defect size is performed ( S6 ). Figure 2 The stacking condition acquisition unit 210 refers to the stacking plan to obtain the stacking condition of the weld bead forming the narrow portion extracted by the narrow portion extraction unit 37. The obtained stacking condition information and the information of the feature quantity of the narrow portion output from the feature quantity calculation unit 39 are input to the defect determination unit 250.

[0109] Fig.12 This is a functional block diagram of the defect determination unit 250 included in the defect prediction device 200. The defect determination unit 250 determines whether a defect occurs in the narrow part based on the stacking conditions and feature quantity information of the input narrow part. In addition, if a defect occurs, the size of the defect can also be predicted.

[0110] Here, the defect size refers to the size of the cross-sectional area or length in the cross section orthogonal to the long side direction of the weld bead. Specific defect sizes include, for example, the diameter when the shape of the defect is approximated to a perfect circle (equivalent circle diameter), the cross-sectional area of ​​the approximate circle, the area of ​​the observed defect, and the major axis length and minor axis length of the ellipse when the shape of the defect is approximated to an ellipse.

[0111] The defect prediction by the defect determination unit 250 can also be implemented by pre-processing the relationship between the stacking conditions and the information of the narrow part and the presence or absence of defects or the size of the defects to generate a prediction model, and using the prediction model. In this case, the defect determination unit 250 includes a data acquisition unit 55 and a learning unit 57, and the learning unit 57 repeatedly learns various learning data obtained from the data acquisition unit 55 to generate a prediction model 59. As a unit for the learning unit 67 to generate the prediction model 59, known means such as decision trees, linear regression, random forests, support vector machines, Gaussian process regression, and neural networks can be cited. If the generated prediction model 59 inputs the stacking conditions and feature quantity information of the narrow part mentioned above, it outputs the information of the presence or absence of defects or the size of the defects.

[0112] In addition to machine learning, the defect determination unit 250 may simply compare the penetration depth corresponding to the lamination condition and the depth of the narrow portion and determine that a defect has occurred when the difference is greater than a given threshold. In this way, various methods can be used as a prediction method for defect prediction.

[0113] Next, a verification result of forming a plurality of rows of weld beads, forming a narrow portion in a part of the weld beads, and predicting defects in the narrow portion will be described.

[0114] Fig.13This is a model diagram showing the contents of a stacking plan of a modeled object based on multiple weld beads. The modeled object is formed by adjacent multiple weld beads parallel to each other along one direction (x direction) on the xy plane. In a part of the multiple weld beads, gaps 61 and 63 of widths Wa and Wb are provided as defects in the total length Wt.

[0115] Fig.14 It means along Fig.13 A graph showing the height distribution of the model in the section of line XIV-XIV. The profile Prf shown here represents the height distribution of the model in the section of line XIV-XIV. Fig.13 The height information of the dot matrix extracted by the model shown in FIG.

[0116] Fig.15 It means from Fig.13 The image is the result of visualizing the information of the dot matrix extracted from the model diagram. In this image, it is clearly seen that Fig.13 The gaps 61 and 63 shown in the figure can be understood from the pixel values ​​that they affect the height of the surrounding model.

[0117] Fig.16 Yes Fig.15 The image shown is an image of the result of image processing to emphasize the stenosis. Fig.15 The image shown is subjected to image processing such as binarization and differentiation, and the narrow part with a steep slope is particularly emphasized. Fig.16 It can be seen from the image that the pixel value of the gap 61 is larger and the range is wider than that of the gap 63.

[0118] Then, from Fig.16 The feature quantity of the narrow part is extracted from the image, and the defect is predicted using the learned prediction model. As a result, it is possible to predict that the defect size of the gap 61 is 1.73mm, and the defect size of the gap 63 is 0.73mm. Regarding the actual measured size of the defect (gap), the gap 61 is 1.77mm, and the gap 63 is 0.55mm. Therefore, according to this method, it can be verified that the presence or absence of defects and the defect size can be predicted with sufficient accuracy.

[0119] As described above, even if the object has a complicated three-dimensional shape and is difficult to grasp, by modeling its shape as information of a dot matrix, it is possible to obtain image information that can be easily grasped for each layer of the stacked model. In addition, by performing appropriate image processing on the image information, it is possible to visually easily grasp information such as the contour of the concave and convex, the excess or insufficient stacking height, etc. As a result, it is easy to evaluate the flatness of the stacked surface, and it is easy to determine the defect position and state of the defect such as the missing meat part and the excess welding part in the stacked model. Therefore, it is possible to feedback to the stacking plan so that no defects are generated, and it is easy to correct it to an appropriate stacking plan.

[0120] In addition, since defect information is extracted by image processing, defect information can be obtained with high accuracy and simplicity, compared to the case where defects are directly determined by applying the model shape of the stenosis based on the measured value of the shape or the information of the dot matrix obtained from the model. Furthermore, by using a prediction model that obtains the characteristic value of the stenosis and the relationship between the stacking conditions and the defect, the occurrence of defects and the size of the defects can be predicted in the planning stage before modeling, and the stacking plan can be easily re-evaluated. That is, the occurrence of defects can be simulated, and the quality of the stacking plan can be analytically judged without actually performing modeling. In addition, by obtaining a prediction model through machine learning, defects can be reliably predicted even if the details of the mechanism of defect occurrence cannot be grasped.

[0121] The present invention is not limited to the above-mentioned embodiments, and the combination of the structures of the embodiments and the changes and applications made by those skilled in the art based on the description in the specification and known technologies are also the intended contents of the present invention and are included in the scope of protection required.

[0122] As described above, the following matters are disclosed in this specification.

[0123] (1) An image information generating device for generating image information representing the shape of a shaped object in which layers composed of a plurality of weld beads are stacked, comprising:

[0124] A coordinate information acquisition unit that uses a model that simulates the shape of the weld bead to approximate the design shape of each layer of the shaped object, and acquires coordinate information of a plurality of points of the approximated model;

[0125] a contour extraction unit that obtains a contour representing a surface shape of the designated layer based on the coordinate information; and

[0126] The image information conversion unit converts the concavo-convex shape of the surface of the designated layer into image information using the contour.

[0127] According to the image information generating device, the design shape of the model is approximated by a model imitating the shape of the weld bead, and the contour representing the surface shape of the model is obtained. The concave and convex shape of the surface of the model is converted into image information using the contour. Thus, even for a model having a complex three-dimensional shape, the surface shape of each layer can be confirmed by the image information, so it is easy to visually grasp the contour of the concave and convex, the excess or deficiency of the stacking height, etc.

[0128] (2) The image information generating device according to (1), wherein the image information includes difference information between a stacking height of the designated layer and a planned height of the layer.

[0129] According to this image information generating device, it is easier to identify a portion where a weld is missing in the layered modeling, a portion where an excess or deficiency of weld beads occurs, and the like.

[0130] (3) The image information generating device according to (1), wherein the image information includes information indicating a distribution of inclinations formed by the concavities and convexities of the surface shape.

[0131] According to this image information generating device, it is easy to evaluate the flatness of the surface shape of the shaped object.

[0132] (4) An image processing device comprising:

[0133] a narrow portion extracting unit that extracts a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the image information generating device according to any one of (1) to (3); and

[0134] The feature quantity calculation unit calculates a feature quantity including any one of the position and the size of the extracted stenosis portion.

[0135] According to this image processing device, since the stenosis is extracted by image processing and the feature amount is calculated, it is possible to perform processing more accurately and simply compared with a method of determining the feature amount by fitting a model shape of the stenosis, for example.

[0136] (5) A defect prediction device comprising:

[0137] The image information generating device according to any one of (1) to (3);

[0138] a narrow portion extracting unit for extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the image information generating device;

[0139] a feature quantity calculation unit that calculates a feature quantity including any one of the position and size of the extracted stenosis;

[0140] a stacking condition acquisition unit that acquires a stacking condition of the weld bead that forms the narrow portion; and

[0141] The defect determination unit predicts the presence or absence of a defect generated in the narrow portion of the object or the size of the defect based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

[0142] According to the defect prediction device, the narrow part is extracted by image processing to calculate the feature value, so the feature value can be obtained easily and with high accuracy. In addition, by predicting defects based on the feature value and stacking conditions, defects can be predicted in the design stage before modeling, and the stacking plan can be re-evaluated.

[0143] (6) According to the defect prediction device described in (5), the defect judgment unit predicts the presence or absence of the defect or the size of the defect from a prediction model based on the feature value of the narrow part calculated by the feature value calculation unit and the stacking condition obtained by the stacking condition acquisition unit, and the prediction model is a model obtained by machine learning the relationship between the feature value and the stacking condition of the narrow part and the presence or absence of the defect or the size of the defect.

[0144] According to this defect prediction device, by using a prediction model that has been subjected to machine learning, it is possible to predict defects even if the details of the mechanism by which the defects occur cannot be understood.

[0145] (7) An image information generating method for generating image information representing the shape of a shaped object in which layers composed of a plurality of weld beads are stacked,

[0146] Using a model that simulates the shape of the weld bead, the design shape of each layer of the object is approximated, and coordinate information of a plurality of points of the approximated model is obtained.

[0147] deriving a contour representing the surface shape of the designated layer based on the coordinate information,

[0148] The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

[0149] (8) An image processing method,

[0150] Based on the image information generated by the image information generation method described in (7), a concave narrow portion formed by the weld bead is extracted by image processing,

[0151] A feature quantity including either the position or the size of the extracted stenosis is calculated.

[0152] (9) A defect prediction method,

[0153] Based on the image information generated by the image information generation method described in (7), a concave narrow portion formed by the weld bead is extracted by image processing,

[0154] calculating a feature quantity including any one of the position and size of the extracted stenosis,

[0155] obtaining the lamination conditions of the weld bead forming the narrow portion,

[0156] The presence or absence of a defect generated in the narrow portion of the object or the size of the defect is predicted based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

[0157] (10) A program for executing the step of generating image information representing the shape of a molded object in which layers composed of a plurality of weld beads are stacked,

[0158] Instruct the computer to perform the following steps:

[0159] Using a model that mimics the shape of the weld bead to approximate the design shape of each layer of the object, and obtaining coordinate information of a plurality of points of the approximated model;

[0160] deriving a contour representing a surface shape of the designated layer based on the coordinate information; and

[0161] The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

[0162] (11) A program for causing a computer to execute the following steps:

[0163] Based on the image information generated by the program according to (10), a concave narrow portion formed by the weld bead is subjected to image processing to extract the concave narrow portion; and

[0164] A feature quantity including either the position or the size of the extracted stenosis is calculated.

[0165] (12) A program for causing a computer to execute the following steps:

[0166] Based on the image information generated by the program according to (10), extracting the concave narrow portion formed by the weld bead by performing image processing;

[0167] calculating a feature quantity including any one of the position and size of the extracted stenosis;

[0168] obtaining a lamination condition of the weld bead forming the narrow portion; and

[0169] The presence or absence of a defect generated in the narrow portion of the object or the size of the defect is predicted based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

[0170] In addition, this application is based on the Japanese patent application (Japanese Patent Application No. 2022-115682) filed on July 20, 2022, and the contents are incorporated herein by reference.

[0171] -Explanation of symbols-

[0172] 11. Modeling Department

[0173] 13 Control Devices

[0174] 15. Robot

[0175] 17 Robot control unit

[0176] 19 Filling material supply unit

[0177] 19a Scroll

[0178] 19b Extraction mechanism

[0179] 21 Heat source control department

[0180] 23 Welding torch

[0181] 25 Base

[0182] 31 Coordinate information acquisition unit

[0183] 33. Contour extraction unit

[0184] 35 Image information conversion unit

[0185] 37 Stenosis extraction

[0186] 39 Feature quantity calculation unit

[0187] 51 images

[0188] 53 recess

[0189] 55 Data Acquisition Department

[0190] 57 Learning Department

[0191] 59 Prediction Model

[0192] 61, 63 gap

[0193] 65 Stenosis

[0194] 100 Layered Modeling Device

[0195] 200 Defect Prediction Device

[0196] 210 Cascade condition acquisition unit

[0197] 220 Image information generating device

[0198] 230 Image processing device

[0199] 240 Display unit

[0200] 250 Defect Assessment Department

[0201] B Weld pass

[0202] BM, BM1, BM2 unit models

[0203] BMa, BMb models

[0204] FL Footprint

[0205] H Average height

[0206] M Filling material

[0207] Point P

[0208] Prf Profile

[0209] Pb1, Pb2 model lower end

[0210] Pt spacing

[0211] Pt1, Pt2 Top

[0212] U Bottom Spacing

[0213] W Weld bead spacing

[0214] Wa, Wb Width

[0215] Wk Modeling

[0216] Wt full length.

Claims

1. An image information generating device for generating image information representing the shape of a molded object having a plurality of layers formed by stacking weld beads, comprising: A coordinate information acquisition unit that uses a model that simulates the shape of the weld bead to approximate the design shape of each layer of the shaped object, and acquires coordinate information of a plurality of points of the approximated model; a contour extraction unit that obtains a contour representing a surface shape of the designated layer based on the coordinate information; and The image information conversion unit converts the concavo-convex shape of the surface of the designated layer into image information using the contour.

2. The image information generating device according to claim 1, wherein: The image information includes difference information between a stacking height of the designated layer and a planned height of the layer.

3. The image information generating device according to claim 1, wherein: The image information includes information indicating the distribution of inclinations formed by the concavities and convexities of the surface shape.

4. An image processing device comprising: a narrow portion extracting unit for extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the image information generating device according to any one of claims 1 to 3; and The feature quantity calculation unit calculates a feature quantity including any one of the position and the size of the extracted stenosis portion.

5. A defect prediction device comprising: The image information generating device according to any one of claims 1 to 3; a narrow portion extracting unit for extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the image information generating device; a feature quantity calculation unit that calculates a feature quantity including any one of the position and size of the extracted stenosis; a stacking condition acquisition unit that acquires a stacking condition of the weld bead that forms the narrow portion; as well as The defect determination unit predicts the presence or absence of a defect generated in the narrow portion of the object or the size of the defect based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

6. The defect prediction device according to claim 5, wherein: The defect determination unit predicts the presence or absence of the defect or the size of the defect from a prediction model based on the feature value of the narrow portion calculated by the feature value calculation unit and the stacking condition acquired by the stacking condition acquisition unit. The prediction model is a model obtained by machine learning the relationship between the feature value and the stacking condition of the narrow portion and the presence or absence of the defect or the size of the defect.

7. A method for generating image information, the method comprising generating image information representing the shape of a molded object in which layers composed of a plurality of weld beads are stacked, approximating the design shape of each layer of the object using a model that simulates the shape of the weld bead, and obtaining coordinate information of a plurality of points of the approximated model; deriving a contour representing the surface shape of the designated layer based on the coordinate information, The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

8. An image processing method, Based on the image information generated by the image information generating method according to claim 7, a concave narrow portion formed by the weld bead is subjected to image processing and extracted. A feature quantity including either the position or the size of the extracted stenosis is calculated.

9. A defect prediction method, Based on the image information generated by the image information generating method according to claim 7, a concave narrow portion formed by the weld bead is subjected to image processing and extracted. calculating a feature quantity including any one of the position and size of the extracted stenosis, obtaining the lamination conditions of the weld bead forming the narrow portion, The presence or absence of a defect generated in the narrow portion of the object or the size of the defect is predicted based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

10. A program for executing the step of generating image information representing the shape of a molded object in which layers consisting of a plurality of weld beads are stacked, Instruct the computer to perform the following steps: Using a model that mimics the shape of the weld bead to approximate the design shape of each layer of the object, and obtaining coordinate information of a plurality of points of the approximated model; deriving a contour representing a surface shape of the designated layer based on the coordinate information; and The concavo-convex shape of the surface of the designated layer is converted into image information using the contour.

11. A program for causing a computer to execute the following steps: extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the program according to claim 10; and A feature quantity including either the position or the size of the extracted stenosis is calculated.

12. A program for causing a computer to execute the following steps: extracting a concave narrow portion formed by the weld bead by performing image processing based on the image information generated by the program according to claim 10; calculating a feature quantity including any one of the position and size of the extracted stenosis; obtaining a lamination condition of the weld bead forming the narrow portion; as well as The presence or absence of a defect generated in the narrow portion of the object or the size of the defect is predicted based on the characteristic amount of the narrow portion and the stacking condition of the narrow portion.

Images