Arrangement for forming a three-dimensional structure and method relating thereto

Abstract

Embodiments relate to a method for forming a three-dimensional structure. The method includes determining one or more locations for positioning a sensing device based on structure coordinate information related to the three-dimensional structure to be formed. The method also includes forming a portion of the three-dimensional structure based on the structure coordinate information. The method further includes positioning the sensing device at a location of the one or more locations.

Description

[0001] Cross-references to related applications

[0002] This application claims priority to European patent application 20155109.0, filed with the European Patent Office on 3 February 2020, the entire contents of which are incorporated herein for all purposes. Technical Field

[0003] The embodiments disclosed herein relate to arrangements for forming three-dimensional structures and related methods. Background Technology

[0004] The fabrication of 3D structures can be monitored using either indirect or direct measurements of printing errors. Indirect measurements measure printing parameters and environmental variables, while direct measurements monitor defects in the printed structure itself. For example, in fused deposition modeling (FDM) printing, various sensors can be used for indirect measurements because many measurable variables can represent the printing state. For instance, acoustic emission (AE) technology can be used to identify abnormal printing states and detect filament breakage. In some applications, hidden semi-Markov models and k-means clustering can be used to differentiate printing states. In some applications, nozzle clogging monitoring can be implemented. In some applications, the material deposition state can be determined by measuring the current of the filament feed motor. In some applications, a two-dimensional laser triangulation system can be used to scan the dimensions of the extrusion track. In some applications, ultrasonic excitation can be used to detect bonding failures during printing. In some applications, fiber Bragg grating sensors and thermocouples can be embedded inside the sample to monitor residual strain and temperature profiles generated during printing. Thermocouples can be used to evaluate temperature conditions in the printing process. In some applications, the temperature field of a fused filament fabricated sample can be reconstructed by applying compression sensing measurements using four temperature readings in one time step. In some applications, sensor arrays with multiple sensors can monitor the printing process. Different process states (normal operation, abnormal operation, and build failure) can be classified by analyzing the sensor data using nonparametric Bayesian models.

[0005] For direct measurement, augmented reality-based techniques can be used to compare images of the printed model with CAD models. In some applications, single and dual-camera systems can be used to detect nozzle clogging, project incompleteness, and filament loss. In some applications, two cameras can be used to reconstruct 3D images of the printed part to identify differences between the point cloud of the ideal model and the point cloud of the printed part. In some applications, a 2D camera can be used to image the printed part. The center of the printed part can be detected and compared to the center of the ideal geometry. In some applications, boundary methods can be applied to detect geometric deviations in the external geometry of simple entities, comparing the ideal profile and the printed part profile layer by layer. In some applications, a USB microscope camera can be used to measure and control slippage.

[0006] Online inspection methods can be used in other printing technologies. For example, computed tomography (CT) scans can be performed on parts manufactured using metal-based powder bed fusion, and two-dimensional images of each layer can be analyzed using multifractal analysis. In some applications, fringe projection can be used to measure the surface morphology of AM-manufactured parts. Furthermore, online signature analysis can be performed on layers of ceramic sensors. In some applications, features of extruded tracks can be compared with template features to detect defects. In some applications, to image selective laser melting errors, statistical methods can be used to capture individual pixel intensity variations in consecutive images from video frames to detect pixels with abnormal intensity distributions over time. Additionally, singularities in printed parts or labels printed on printed objects can be identified to capture the appearance of additively manufactured parts. Summary of the Invention

[0007] Various embodiments relate to providing methods and arrangements for monitoring the quality of three-dimensional structures during the manufacturing process. These methods and arrangements provide improved processes for classifying and locating defects and their sources during printing.

[0008] Various embodiments relate to a method for forming a three-dimensional structure. The method includes: determining one or more locations for locating a sensing device based on structural coordinate information associated with the three-dimensional structure to be formed. The method further includes: forming a portion of the three-dimensional structure based on the structural coordinate information. The method also includes: positioning the sensing device at one of the one or more locations.

[0009] Various embodiments relate to an arrangement for forming a three-dimensional structure. The arrangement includes: a forming device for forming a three-dimensional structure based on structural coordinate information associated with the three-dimensional structure. The arrangement further includes a movable sensing device. The arrangement also includes a processor configured to determine one or more locations for positioning the sensing device based on the structural coordinate information associated with the three-dimensional structure to be formed; and to control the movable sensing device to be positioned at the one or more locations.

[0010] Various embodiments relate to an arrangement for forming a three-dimensional structure. The arrangement includes: forming apparatus for forming a three-dimensional structure comprising multiple layers. The arrangement also includes a movable sensing device. The arrangement further includes: a processor configured to determine at least one location for positioning the sensing device for a set of layers. The processor is further configured to: control the apparatus to form the set of layers based on structural coordinate information associated with the three-dimensional structure. The processor is further configured to control the positioning of the movable sensing device to at least one location associated with the set of layers after the formation of the set of layers and before the formation of another set of layers. The processor is further configured to determine a process state based on acquired structural data of the formed set of layers, wherein the acquired structural data is acquired by the sensing device at the at least one location associated with the formed set of layers. Attached Figure Description

[0011] The foregoing and other features of this disclosure will become more apparent when read in conjunction with the accompanying drawings and the appended claims. It should be understood that the drawings only illustrate a few embodiments according to this disclosure and should therefore not be considered as limiting its scope. Further specific details of this disclosure will be described using the drawings, thereby making its advantages more readily apparent, wherein:

[0012] Figure 1A A flowchart of a method 100 for forming a three-dimensional structure is shown;

[0013] Figure 1B A flowchart of a method 160 for forming a three-dimensional structure is shown;

[0014] Figure 2A and Figure 2B Examples of different layers of a three-dimensional structure are shown respectively;

[0015] Figure 3 A flowchart of at least a portion of a method 300 for forming a three-dimensional structure is shown;

[0016] Figure 4A and Figure 4B An example of a chain that is mislocated is shown;

[0017] Figure 5A and Figure 5B Images showing training data for a convolutional neural network (CNN) are displayed;

[0018] Figure 6 An illustration shows edge detection using a convolutional neural network (CNN);

[0019] Figure 7A The image shows the edge points found during the edge clustering process, which includes both correct and incorrect edges 721.

[0020] Figure 7B The results of using edge clustering to determine the main edge are shown;

[0021] Figure 7C The further clustering process used to identify the upper and lower edges of clusters is shown;

[0022] Figure 8 The distance l measured with reference to the actual axis is shown. real The distance l measured with reference to the ideal axis ideal The difference;

[0023] Figures 9A to 9D Different printable 3D structures are shown;

[0024] Figures 10A to 10F Images of a 50ml breast stent are shown using gradient-based edge detection for visual monitoring.

[0025] Figures 11A to 11F Images of chains in a cell culture network are shown for visual monitoring.

[0026] Figures 12A to 12F Images of a 50ml breast stent are shown using a convolutional neural network (CNN) for visual monitoring.

[0027] Figures 13A to 13C The first-level assessment is shown in histograms of the chain diameters (minimum, maximum, and average diameters) measured using a 50ml breast implant stent;

[0028] Figures 14A to 14C The second-level assessment is shown, such as the layer-by-layer assessment of a 50ml breast implant stent;

[0029] Figures 15A to 15B The third-level assessment is shown, in which chain-by-chain assessment can be performed on the chains of a 50ml stent;

[0030] Figure 16 An illustration of the arrangement 150 for forming a three-dimensional structure is shown. Detailed Implementation

[0031] In the following detailed description, reference is made to the accompanying drawings, which illustrate by way of illustration specific embodiments in which the claimed subject matter can be practiced. These embodiments are described in sufficient detail to enable those skilled in the art to practice the subject matter. It should be understood that the various embodiments, while different, are not necessarily mutually exclusive. The terms “embodiment,” “example embodiment,” “exemplary embodiment,” and “this embodiment” do not necessarily refer to a single embodiment, although they may be referred to as such, and various example embodiments can be readily combined and / or interchanged without departing from the scope or spirit of the example embodiments. For example, a particular feature, structure, or characteristic described herein in conjunction with one embodiment may be implemented in other embodiments without departing from the spirit and scope of the claimed subject matter. References to “an embodiment” or “an embodiment” in this specification mean that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one implementation covered by this specification. Therefore, the use of the phrase “an embodiment” or “in one embodiment” does not necessarily refer to the same embodiment. Furthermore, it should be understood that the position or arrangement of the various elements within each disclosed embodiment may be modified without departing from the spirit and scope of the claimed subject matter. Therefore, the following detailed description should not be construed as limiting, and the scope of the subject matter is defined only by the appended claims as properly interpreted and by the full range of their equivalents. In the drawings, the same numerals indicate the same or similar elements or functions in multiple views, and the elements depicted therein are not necessarily drawn to scale, but may be enlarged or reduced to facilitate understanding of these elements within the context of this specification.

[0032] As used herein, the terms “above,” “to,” “between,” and “above” can indicate the relative position of a layer with respect to other layers. A layer located “above” or “above” another layer may be in direct contact with that other layer, or one or more intervening layers may exist. A layer located “between” multiple layers may be in direct contact with those layers, or one or more intervening layers may exist.

[0033] The terms “a,” “an,” and “the” can include both singular and plural references. Furthermore, as used in this disclosure and the appended claims, the word “and / or” can indicate and cover any and all possible combinations of one or more of the associated listed items. As used in this disclosure, the phrase “A and / or B” refers to (A), (B), or (A and B). As used in this disclosure, the phrase “A, B, and / or C” refers to (A), (B), (C); (A and B); (A and C); (B and C) or (A, B, and C). As used in this disclosure, when used in the phrase “A, B, or C,” the term “or” means that (A) does not exclude (B) and (C), (B) does not exclude (A) and (C), and (C) does not exclude (A) and (B).

[0034] Figure 1A A flowchart of a method 100 for forming a three-dimensional structure is shown.

[0035] Method 100 includes: determining one or more locations 110 for positioning a sensing device based on structural coordinate information associated with the three-dimensional structure to be formed. Method 100 further includes: forming a portion of the three-dimensional structure 120 based on the structural coordinate information. Method 100 further includes: positioning the sensing device 130 at one of the one or more locations.

[0036] Structural coordinate information may include, or may be, coordinate information related to the layout of the 3D structure to be formed. Structural coordinate information may include coordinate information related to the position, orientation, size, shape, and / or form of the physical structure and / or the layout of the 3D structure to be formed. Structural coordinate information may be coordinate system-based information that uses numerical scales or axes to uniquely define and / or describe the position, orientation, size, shape, and / or form of the features of the 3D structure to be formed. Optionally, structural coordinate information may be based on a three-dimensional Cartesian coordinate system with three mutually perpendicular axes (e.g., x-axis, y-axis, and z-axis) and three mutually orthogonal planes. Structural coordinate information may include information related to the internal and / or external structure of the 3D structure to be printed relative to the three-dimensional Cartesian coordinate system. Alternatively, structural coordinate information may be based on any other coordinate system, such as a spherical coordinate system, polar coordinate system, elliptical coordinate system, or cylindrical coordinate system, or any coordinate system whose coordinates can be transformed between said coordinate systems and the Cartesian coordinate system.

[0037] Structural coordinate information may include (or may be) computer-aided design (CAD) information related to the architecture, drawing, form, or layout of the three-dimensional structure. Additionally, alternatively, or optionally, structural coordinate information may include computer-aided manufacturing (CAM) information for controlling forming equipment (e.g., a 3D printing device or arrangement) used to form the three-dimensional structure. Additionally, alternatively, or optionally, structural coordinate information may include (or may be) toolpath instructions for controlling the forming equipment used to form the three-dimensional structure. For example, structural coordinate information may include (or may be) numeric code instructions, such as G-code instructions. Numeric code toolpath instructions and / or G-code instructions may be based on and / or may include coordinate information related to the structure and / or architecture of the three-dimensional structure to be formed. Such instructions, when executed by a processor, can control the movement and / or path of the equipment used to form the three-dimensional structure (e.g., a dispensing device).

[0038] Structural coordinate information may include information related to the internal and / or external structure of the three-dimensional structure. For example, the three-dimensional structure to be formed may include, but is not limited to, a three-dimensional scaffold structure and / or a mesh structure. Alternatively, additionally, or optionally, the three-dimensional structure may include infill lines (also referred to herein as chains), or may be any structure comprising a lattice network of chains or lines. In addition to infill lines, chains may also include outfill lines forming the outer boundary of the structure.

[0039] The three-dimensional structure to be formed (such as a scaffold structure and / or a mesh structure) may comprise multiple layers or sets of layers. During (or while) the three-dimensional formation (or printing), the three-dimensional structure may be formed by first forming a first layer, and then continuously forming multiple layers on top of each other, such that the successive layers are stacked perpendicularly to each other in the printing direction (which may be referred to herein as the z-direction). Optionally, such a printed layer may be understood or referred to as a lateral layer, horizontal layer, planar layer, or flat layer. By these terms, the layer is understood to be a layer located in or within the xy-plane. Such a printed layer may have dimensions extending along the x-axis (x-direction) and along the y-axis (y-direction), which are larger than its dimension in the z-direction (e.g., at least 10 times, or at least 50 times, or at least 100 times). It is understood that the terms lateral layer and horizontal layer as used herein may refer to a direction perpendicular to the printing direction (which may be considered as the z-direction or the vertical direction). Once a lateral (or horizontal) layer has been printed, the printer can then continue printing subsequent lateral (or horizontal) layers on top of (e.g., above, on, or above) the previous layer, resulting in the lateral layers being stacked perpendicularly to each other in the printing direction (e.g., vertically stacked) relative to the printing direction (e.g., in the case of 3D printing). Optionally, a layer set (or layer) may include a first sublayer (or a first group of sublayers) of chains oriented in a first direction and a second sublayer of chains oriented in a second direction different from the first direction. Optionally, the layer set is not limited to including only the first group of sublayers or the second group of sublayers, but may include any possible number of sublayers. Each group of sublayers may include (or may indicate) one or more sublayers. Optionally, intersecting (or, for example, crisscrossing) lines or chains may form multiple repeating cells of the layer set.

[0040] The external structure of a three-dimensional structure can be the outer surface region of the three-dimensional structure. The outer surface region can refer to (or can be) the outermost surface, outermost layer, and / or outermost contour of the three-dimensional structure. The outermost surface and outermost contour can be formed by one or more layers or chains. The outer surface region can refer to (or can be) the outermost set of layers of the three-dimensional structure (e.g., a single outermost layer, or multiple outermost layers, or boundary layers). The outer surface region can refer to the outward-facing (or, for example, outward-facing) surface of the three-dimensional structure.

[0041] The internal structure of a three-dimensional structure can be any part of the three-dimensional structure not located at its outermost boundary. Additionally, alternatively, or optionally, a cell whose all sides are surrounded by other cells within the layer set can be considered part of the internal structure of the three-dimensional structure. Conversely, a boundary cell not surrounded by another cell on at least one side can be considered part of the external structure of the three-dimensional structure. The internal structure of a three-dimensional structure can be any part of the three-dimensional structure with the characteristic that its distance from the outermost surface is more than twice the desired chain width or thickness of the printed chain. For example, the desired chain width or thickness can be the ideal thickness of the chain in the three-dimensional structure to be printed. Optionally, the average thickness (or width or diameter) of the chain can be, but is not limited to, between 0.001 mm and 30 mm (or, for example, between 0.001 mm and 1 mm, or, for example, between 0.01 mm and 0.5 mm). Optionally, the spacing between adjacent lines can be, but is not limited to, between 0.001 mm and 30 cm (or, for example, between 0.001 mm and 50 mm, or, for example, between 0.01 mm and 50 mm). Similarly, the spacing between each sublayer oriented in the second direction and its adjacent (or consecutive) sublayer oriented in the (same) second direction is between 1% and 100% of the average chain thickness.

[0042] Method 100 includes: determining one or more locations (or, for example, multiple locations) for locating a sensing device based on structural coordinate information associated with the three-dimensional structure to be formed. The processor can be configured to determine (or calculate) one or more locations based on or according to the structural coordinate information. For example, one or more locations can be determined based on the structural coordinate information by calculating or determining the intersection regions or points between chains (or segments) of the three-dimensional structure. Figures 2A to 2B (As shown).

[0043] The portion forming the 120 three-dimensional structure may include: printing the portion of the three-dimensional structure using a three-dimensional printing device or by arranging the three-dimensional structure. The three-dimensional structure may be formed by a three-dimensional (3D) printing process. The three-dimensional printing process may include at least one of stereolithography (SLA), digital light processing (DLP), fused deposition modeling (FDM), selective laser sintering (SLS), selective laser melting (SLM), electron beam melting (EBM), layered object fabrication (LOM), bond jetting (BJ), and material jetting (MJ). The processor may execute digital code instructions based on structural coordinate information to control the movement and / or path of at least a portion of the forming device used to form the three-dimensional structure. In some examples, the device portion whose movement is controlled by structural coordinate information may be a dispensing portion for dispensing the printing material (e.g., chains of the three-dimensional structure) used to form the three-dimensional structure, for example, in fused deposition modeling. In other examples, the controlled portion of the device may be a laser, where the structural coordinate information controls the selection of a location where at least one of sintering, hardening, melting, bonding, laminating, and / or curing of the material used to form the three-dimensional structure will be performed. It should be understood that the device portion whose movement is controlled by digital code instructions can be a part of a device responsible for forming a three-dimensional structure at a selected location defined by the structural coordinate information of the digital code instructions (e.g., a chain responsible for forming a three-dimensional structure). Alternatively or optionally, the three-dimensional structure can be formed by a printing process based on more than three dimensions of movement, such as a five-dimensional (5D) printing process or a six-dimensional (6D) printing process.

[0044] Since method 100 involves an on-site monitoring process, it is understood that the portion formed herein (formed in 120) may be smaller than the entire three-dimensional structure to be formed. For example, this portion of the three-dimensional structure may be a set of layers (e.g., one, or more than one set of layers) of the three-dimensional structure. The formed portion may be or may include the internal structure of the three-dimensional structure. For example, the formed portion may include one or more set of layers, which may include multiple intersecting lines or chains forming a lattice of unit cells within the three-dimensional structure. Figure 1B The process for defect detection across multiple layers is further described.

[0045] Method 100 may include determining a set of locations for locating sensing devices associated with the formation of a portion of the three-dimensional structure before or even during the formation process. For example, this set of locations may be determined before or after the formation of a portion of the three-dimensional structure. Method 100 may include sequentially positioning the sensing devices at locations from a set of determined locations associated with the already formed portion before forming another (or, for example, subsequent) portion of the three-dimensional structure. This set of locations may include multiple locations. The number of locations may be random, sequential, array-based, and / or may be determined by the user. This set of locations may refer to (or may be, or may include) one or more (e.g., multiple) locations.

[0046] At each of the set of locations (e.g., at each location), method 100 may include acquiring data of at least a portion of the formed portion via a sensing device. The sensing device may be any sensing device that acquires the data, including structural information of the formed portion. For example, the sensing device may be at least one of a set of devices including: imaging devices, cameras, thermal imagers, microscopes, laser scanners, and 3D scanning devices. Optionally, the acquired data may be two-dimensional or three-dimensional images of the formed portion acquired at each of the set of locations.

[0047] The acquired data may include visual data, which may include or may be structural information related to and / or relating to one or more structural features of the forming portion of the three-dimensional structure. The structural information acquired and / or determined based on the acquired data may include information related to the location, orientation, size, shape, and / or form of physical structural features and / or the layout of the forming portion of the three-dimensional structure. For example, structural features may be lines, line segments, chains, chain segments, holes, and / or walls of the forming portion of the three-dimensional structure.

[0048] Alternatively or additionally, method 100 may further include determining a process state based on acquired data relating to at least a portion of the formed portion of the three-dimensional structure. The process state may be determined after acquiring data relating to the formed portion (e.g., a set of layers) and (optionally) before forming another portion of the three-dimensional structure (e.g., another set of layers). This other portion may be, but is not limited to, a directly subsequent portion to be formed. For example, the forming apparatus may continue the printing process until the process state is determined, depending on the time required to determine the process state, meaning that one or more additional portions of the three-dimensional structure may have already been printed.

[0049] Determining the process state may include: determining parameter values ​​of structural features of the formed portion of the three-dimensional structure based on the acquired data. The parameter values ​​of the structural features may be at least one of the following: length, width or diameter, height, roughness, color, uniformity, thickness, and tilt angle of the structural features of the formed portion of the three-dimensional structure. Optionally, the parameter values ​​of the structural features may be determined based on edge detection of chain segments (e.g., by detecting the edges of chain segments). For example, the parameter value may be the width of a chain segment between two detected edges. For example, the edges of chain segments may be detected by implementing an artificial neural network process or a gradient-based detection process. Optionally, the parameter values ​​of the structural features may be determined based on the difference between the structural feature and a given baseline or comparison structure (e.g., by subtracting image values) and / or by determining the mismatch between the structural feature and a given baseline or comparison structure.

[0050] Optionally, determining the process state may include: determining (e.g., calculating, or generating) the ideal axis of a segment of a portion of the three-dimensional structure based on structural coordinate information. Determining the process state may also include determining (e.g., measuring, or calculating) the actual axis of the segment of the formed portion based on acquired data of the formed portion. The process state may be determined based on a comparison between the ideal axis and the actual axis. If the difference between the ideal axis and the actual axis exceeds a threshold, a defective process state may be identified.

[0051] Optionally, determining the process state may include comparing the ideal parameter values ​​of a segment of a portion of the three-dimensional structure with the determined parameter values ​​of the segments forming the portion. The determined parameter values ​​of the segments may be determined based on acquired data of at least a portion of the forming portion of the three-dimensional structure. The ideal parameter values ​​of the segments may be determined based on at least one of structural coordinate information and input values. Input values ​​may be one or more user input values ​​and / or one or more values ​​from a dataset or database (e.g., data from multiple printouts or a collection of previous processes). If the difference between the ideal parameter value and the determined parameter value exceeds a threshold, a defective process state can be determined.

[0052] Additionally, optionally, or alternatively, determining the process condition may include determining multiple parameter values ​​associated with multiple segments of the forming part. Statistical parameters (e.g., at least one of standard deviation, variance, median, mode, range, correlation, frequency, maximum, minimum, quartiles, mean, and error) may be determined for the multiple parameter values. If the difference between the statistical parameter and the comparison parameter exceeds a threshold, a defective process condition can be determined.

[0053] Method 100 may further include: adapting process parameters to form a three-dimensional structure based on a determined process state. The process parameters may include at least one process parameter from the group consisting of: forming temperature, ambient temperature, cooling process, interlayer gap, nozzle cleanliness, flow rate of printing material, printing speed, tool path, and ambient humidity and / or laser power. The ambient temperature may be room temperature and / or chamber temperature or the temperature of the enclosed space where the three-dimensional structure is formed. The forming temperature (e.g., printing temperature) may be the temperature of the container holding the material (e.g., polymer) to be dispensed for forming the three-dimensional structure. The laser power may be the power and / or energy of a laser used to sinter, harden, melt, bond, laminate, and / or cure the material forming the three-dimensional structure.

[0054] Figure 1B A flowchart of a method 160 for forming a three-dimensional structure is shown. Method 160 may include components that have already been combined Figure 1A One or more of the features described. Figure 1B The process of the defect detection system is shown, and how on-site monitoring of the process used to form three-dimensional structures can be achieved.

[0055] Such as combination Figure 1A As described, method 160 may include: determining a corresponding set of positions for positioning sensing devices for a set of layers (or, for example, for each layer set) of multiple layers of a three-dimensional structure to be printed. The position sets may be determined based on structural coordinate information of the three-dimensional structure to be printed, and may be selectively determined before the start of the three-dimensional structure forming process (e.g., the printing process). Alternatively, the processor may be configured to determine the position sets during the printing process, as long as the position sets of the corresponding portions are determined before or at the time the corresponding portions of the three-dimensional structure begin to form or are fully formed. Optionally or alternatively, the processor may even be configured to determine the position sets of the corresponding portions after the corresponding portions have been formed.

[0056] Method 160 may further include generating adaptive digital code instructions for forming a three-dimensional structure. For example, the adaptive digital code instructions may include instructions for controlling the movement of the forming device and the movement of the movable sensing device. The adaptive digital code instructions may include structural coordinate information related to the three-dimensional structure to be formed and information related to positioning the sensing device at one or more locations. For example, information related to the positioning of the sensing device may be information for controlling the positioning or movement of the movable sensing device to one or more locations. Furthermore, the adaptive digital code instructions may be adapted to the timing or movement sequence of the forming device to take into account the movement of the movable sensing device to one or more locations.

[0057] By executing adapted digital code instructions via a processor, method 160 may include: forming a portion of a 120-dimensional structure based on structural coordinate information (as described in conjunction with method 100). Forming a portion of the 120-dimensional structure may include, or may mean, forming (e.g., printing) a set of multiple layers (e.g., a first layer set, or, for example, any first layer set).

[0058] After forming the (first) layer set and before forming another (or, for example, a subsequent, second, or any second) layer set, method 160 may include: determining the process state (e.g., manufacturing state) for forming the three-dimensional structure 140. Determining the process state 140 may include: sequentially positioning sensing devices at locations within a set of locations determined for the formed layer set. Determining the process state 140 may further include: acquiring data 130 (e.g., image acquisition) associated with the formed layer set at each location within the set of locations determined for the formed layer set. Determining the process state 140 may further include processing the acquired data 141 (e.g., by processing each acquired image).

[0059] Determining the 140 process state may include: based on a comparison between ideal parameter values ​​and acquired parameter values, determining the 142 defective process state (e.g., determining whether a significant printing defect has occurred). If the processor determines a defective process state, the processor may execute instructions to stop the 180 forming process or adapt the 170 printing parameters. If no defective process state is determined, the processor executes instructions to continue the forming process and may print the next layer set.

[0060] Method 160 may include: sequentially and alternately determining process states (which may include positioning a sensing device at a location within a set of locations of the formed layer sets) between forming one layer set (or one or more layer sets) of multiple layer sets, and after forming the layer set and before forming another layer set. Optionally, the subsequent alternating process may be performed until the printing process is complete and a three-dimensional structure is formed, or until a significant defect is identified that causes the processor to execute a stop printing instruction. Optionally, even if no defect is identified, the alternating process may be performed until a specific portion (e.g., a predefined portion) of the three-dimensional structure has been printed. Such a predefined portion may be a specified or specific number of layers within the total number of layers in the structure, or it may be a specified or specific location or portion within the three-dimensional structure. Such a predefined portion may be a portion within the three-dimensional structure with a complexity higher or lower than other portions of the three-dimensional structure, and may be selected according to the needs of the manufacturing process.

[0061] Sensing devices (such as commercial digital microscopes) can be used to capture images (pictures) of regions of interest layer by layer. These images can be processed to detect defects. Figure 1BThe layer-by-layer characteristics of detection method 160 are illustrated. After each printed layer, the microscope can be automatically positioned above a designated region of interest via the printer's axes. Depending on the field of view and the size of the monitored feature, one or more images can be captured and processed. For automation of camera positioning and image capture, the image capture time and optimal camera position are required. This information can be extracted from digital codes (e.g., G-codes) that include all printer movements used to print the 3D scaffold structure. The digital codes can be modified, adapted, or used to determine potential camera positions and modified via commands for camera positioning. Optionally, the sensing device can be positioned in a separate or different coordinate system from the printing device (e.g., the printhead). Adapted digital codes can be used to control two independent coordinate systems: the sensing device coordinate system and the printing device coordinate system. For example, the adapted digital codes can be sent to multiple different processors or controllers, each controlling a different coordinate system.

[0062] Subsequent image processing can be used to measure the chain diameter. If the measurement is within a predetermined tolerance, printing can continue. In the case of significant deviation, printing can stop at 180°. When the deviation is within a predetermined tolerance (e.g., the deviation is insufficient to stop printing but can be prevented in subsequent layers), the printing parameters can be adjusted at 170°. After adjusting at 170°, the printing process can continue. For example, methods 100 and 160 can be used to detect the chain diameter, and additionally, alternatively, or optionally, to detect breakage or geometric deviations.

[0063] Figure 2A and Figure 2B They respectively showed about Figures 1A to 1B Examples of different layer sets described by the method.

[0064] Figures 2A to 2B A diagram of a layer set of a unit cell is shown. The layer set may include: a first sublayer (or a first set of sublayers) comprising chains 228 oriented in a first direction; and a second sublayer (or a second set of sublayers) comprising chains 229 oriented in a second direction different from the first direction.

[0065] Optionally, the chain of each sublayer can be part of a continuous sublayer chain extending continuously from the starting point S of the sublayer to the ending point E of the sublayer. For example, the chain 228 of the first sublayer can be part of a continuous sublayer chain that is continuously tortuous from the starting point S to the ending point E of the first sublayer. For example, the chain of the second sublayer can be part of a continuous sublayer chain that is continuously tortuous from the starting point S to the ending point E of the second sublayer. It is understood that when the terms “line” or “chain” are used herein, they can include not only straight lines but also curved or free-form chains. For example, intersecting chains forming a unit cell can be straight lines, or these chains can be sinusoidal or curved lines, where the unit cell can have a “free-form” shape. Alternatively or optionally, these layers can be printed as paths with more than one starting point and one ending point. For example, gaps may exist between each layer, or a layer may include different regions, or the layer may include different structures formed simultaneously.

[0066] Optionally, some of the chains 228, 229 within each respective sublayer may be parallel to each other (e.g., the acute angle between chains within a sublayer, or the acute angle between chains best suited for a sinusoidal chain, is within + / - 5°). Alternatively, the chains may be curved or may extend in random directions without being parallel to each other.

[0067] Optionally, the multiple chains 228 of the first sublayer and the multiple chains 229 of the second sublayer may intersect at intersections or cross regions to form a two-dimensional lattice arrangement of two-dimensional unit cells of that layer. The two-dimensional unit cells of each layer in the stacked layer arrangement can form a three-dimensional lattice structure. Each unit cell of the layer set may include or may be formed by cross chains 228, 229 from adjacent sublayers defining the cell aperture. For example, two adjacent (parallel) chains 228 of the first sublayer may intersect with two adjacent (parallel) chains 229 of the second (adjacent) sublayer. The unit cell region surrounded by the intersecting chains may be a rhombic unit cell, a polygonal unit cell, a triangular unit cell, a diamond-shaped unit cell, a free-form unit cell, a square unit cell, a parallelogram unit cell, and / or a hexagonal unit cell.

[0068] The layer set may include at least one first sub-layer and at least one second sub-layer. A position among multiple locations can be determined based on structural coordinate information based on the intersection point 231 between the first and second sub-layers (e.g., the intersection point between chain 228 of the first sub-layer and chain 229 of the second sub-layer). For example, position 232 among multiple positions 232 may be between two intersection points 231 of the first and second sub-layers.

[0069] The printing process (e.g., FDM process) can follow a layer-by-layer approach. The structure of the support S can be represented by a layer set L:

[0070] S={L j} j＝1,…,j

[0071] Where j is the total number of layers, Each movement of the printer can be defined by digital code instructions (e.g., G-code) with a start and end point, where the print head can move along a straight line. Circular motion can be approximated by a short, continuous line. Therefore, each layer L j It can be defined as a line set l:

[0072]

[0073] The number of buses in the layer is G, where Each line It can be used as its starting point and the end point express:

[0074]

[0075] The geometric description of the support allows for the definition of appropriate camera positions (or locations) or regions of interest within structural layers. A region of interest can be the area where the connection points between the current and previous layers occur. Assuming a symmetrical nozzle, the centers of these connection points can be determined by two consecutive layers L projecting onto a plane (x, y). j and L j-1 The intersection point is represented by the intersection point, where j>1. The intersection point can be calculated as:

[0076]

[0077] in

[0078] There are different methods to calculate these points and determine the corresponding intersection lines. If it is assumed that the number of intersection points is significantly less than the square of the number of lines, the Bentley-Ottmann algorithm can be appropriately applied. Before calculating the intersection points, the lines can be filtered to eliminate lines in the boundary regions and reduce computation time. This operation can filter lines based on their length and can be adapted to the aperture of the printed portion. For example, lines shorter than the minimum aperture of the printed structure can be eliminated. Using the calculated intersection points, the camera position can be calculated based on the field of view of the camera used. If the field of view is large enough to capture the entire layer, the center of all intersection points can be selected as the camera point. If the field of view is too small to monitor the entire layer, the camera position above the smaller region where the intersection point is located can also be calculated. In this case, the layer can be sequentially scanned using overlapping or boundary scan regions to evaluate the entire layer. Alternatively, the layer can be partially monitored. Partial imaging of the layer can be performed based on randomly selected intersection points. The chain portion between two intersection points can be used as the region of interest. This technique is suitable because deviations in chain diameter often occur on the bridge between intersection points. Optionally, the camera position can be calculated between two consecutive intersection points (e.g., at the center). To calculate these positions, each line in layer j... Lines from the previous layer The intersections can be determined based on their online... Sort the points according to their positions. The sorted points can be defined as:

[0079]

[0080] Here, r is the index of the intersection point, g is the index of the line, and j is the index of the layer. After sorting, the center of the line segment can be determined. The range of possible camera locations can also be limited, for example, by ignoring locations near boundaries. Then, some of these possible locations can be randomly selected. The number of camera locations depends on quality control requirements and can be specified by the user. Alternatively, the camera location can be calculated at the intersection itself, rather than between two intersections. Alternatively, instead of positioning the camera based on intersections, it can be placed based on other factors (such as the layer's external coordinates) and / or on toolpath codes.

[0081] exist Figure 2A The diagram illustrates a set of layers with two consecutive sub-layers, these sub-layers having multiple intersection points and a large number of possible camera positions. For example, from all possible camera positions, five points can be randomly selected (in...). Figure 2A (Identified by X in Chinese). Figure 2A The results of structural separation and location calculations are shown in the diagram. Five monitoring locations can be identified, each with two intersection points at their distances from the boundary. Figure 2B The diagram shows a set of layers with two sub-layers, whose intersection ratio is... Figure 2AThe number is small. For example, among a small number of possible camera locations, a camera position can be randomly selected (in...). Figure 2B (Identified by X in the image). Alternatively, since the distance from the camera position to the outer boundary is fixed, only one monitoring position is found for this layer. The x-axis and y-axis show the coordinates of the printer's movement. This example shows that the number of positions can be random, sequential, array-based, and / or user-determined.

[0082] Figure 3 A flowchart of at least a portion of a method 300 for forming a three-dimensional structure is shown. Figure 3 The combination is shown Figures 1A to 2B This is part of the image processing 140 described in the method. Image processing 140 can be combined with... Figures 4A to 15B Further description. Additionally, method 300 may include components already combined with... Figures 1A to 2B One or more of the features described.

[0083] Method 300 may include performing image processing procedure 140 after forming a set of layers and before forming another set of layers. Image processing procedure 140 (e.g., it may be executed by a processor performing an image processing algorithm) may include or contain... Figure 3 Several processes 301 to 312 are described (e.g., continuous processes).

[0084] Image processing procedure 140 may include processing structural coordinate information 301 (e.g., G-code or numeric code) to determine an ideal axis 302 based on the structural coordinate information. For example, image processing procedure 140 may include determining the ideal axis of 302 or extracting the chain. Optionally, the ideal axis can be obtained by structural extraction 301 using numeric code (e.g., G-code), wherein the extracted axis can be determined from its starting point P. s And the finish line P e express.

[0085] Image processing 140 may include fitting the determined ideal axis 302 to the acquired data after determining the ideal axis 302. When the acquired data is an image, since the image of the chain can be captured at a specific location and scale, the extracted ideal axis can be fitted to the captured image. For example, fitting can be performed by rotating the image 303 so that the extracted ideal axis can be fitted to the image. This can be done because the camera position and pixel size are known.

[0086] After scaling and positioning the ideal axis as in 303, image processing 140 may include evaluating 302 whether the axis is on a chain to avoid detecting erroneous edges. If the printed chain is missing, misplaced, or has significant curling, the ideal axis may be assessed as not being on a chain. In this case, an edge not belonging to the monitored chain may be detected.

[0087] Figure 4A and Figure 4B An example of a mislocated chain is shown.

[0088] exist Figure 4A In the middle, because the chain rotates relative to the ideal axis, the chain is not located on the ideal axis 415.

[0089] exist Figure 4B In this case, the chain is not on the ideal axis because the chain is translated relative to the ideal axis. For example, edges 416 and 417 found next to the ideal axis are not the true edges of the chain and can be disregarded when measuring chain dimensions.

[0090] Method 300 may further include converting the image 304 to grayscale to identify these cases. Characteristics of the grayscale values ​​along the ideal axis can be analyzed at the start of image processing. These characteristics may include: the difference in average grayscale values ​​across consecutive axis segments and the complexity of the grayscale value function along the axis. Both represent the color uniformity of the image along the axis. If the chain is not on the ideal axis, it may be impossible to measure the chain size based on the ideal axis. In this application, a displaced chain can be defined as a printing defect. A displaced chain means that the ideal axis is not within any part of the chain that is expected to be located at that position.

[0091] Method 300 may include: detecting the edges of the chain 305 after converting the image to grayscale if the ideal axis of the chain segment lies on the actual printing axis (e.g., the ideal axis lies between the edges of the actual printing chain). For example, if the ideal axis is found to lie between the edges of the actual printing chain, edge detection 305 may include rotating and / or translating the image of the actual chain to obtain the desired alignment. For example, in Figure 4A and 4B In this case, the ideal axis is not located between the edges of the actual printing chain, therefore, edge detection is not performed further. Figure 8 The illustration shows a scenario where the ideal axis is located between the edges of the actual printed chain. However, the actual axis is not yet aligned in the desired direction. The required alignment can depend on the imaging sensing and image processing techniques used. Since the angle of the ideal axis can be known via G-code, the image can be rotated to align and / or level the ideal axis (e.g., in a direction parallel to the axis coordinate system, such as the x-direction). Alternatively, the required alignment can be the alignment of the chain edges with the axis coordinate system (e.g., parallel and / or horizontal alignment). The term "alignment" here refers to a situation where it is horizontal or parallel to a desired direction (such as a defined axis). It is understood that the term "alignment" used in conjunction with the required alignment can refer to any predefined or desired angle based on the required manufacturing tolerances.

[0092] After rotating the image, the direction of the ideal axis can be defined as the x-direction, and the y-direction can be orthogonal to the ideal axis. Edge detection process 305 can be performed after rotating the image and converting it to grayscale. Edge detection process 305 may include implementing a gradient detection process (V1*) or an artificial neural network process (V2*), such as a convolutional network (CNN) process.

[0093] Gradient-based edge detectors (V1*) are sensitive to high-frequency features such as edges. Features can be detected as high-amplitude points by taking the first derivative of the image. For this reason, a convolution operation can be performed on the image using a filter kernel H. A simple gradient filter can be a Prewitt kernel, whose kernel for filtering in the x and y directions can be described as:

[0094]

[0095] Noise in an image can also be high-frequency features, and therefore can be detected as edges. To avoid detecting false edges, the image can be preprocessed before the edge detection operation. For example, a filter can be used to blur the image to attenuate noise. An example of such an operator could be a median filter, which can be an edge-preserving filter that attenuates noise while preserving edges. A local nonlinear contour-preserving filter can be placed before edge detection to improve the results of gradient-based Prewitt filter operations. The preservation filter can eliminate noise that is detected as an edge. Since the direction of the edge is known, a filter that detects edges in only one direction is sufficient. The edge to be detected may be relatively long compared to the image size. Therefore, the kernel of the Prewitt operator can be scaled up in the horizontal direction.

[0096]

[0097] For example, the kernel can be expanded to nine elements in a row, resulting in a 3×9 kernel. The filter size can be adapted depending on the edge length. The filtered image can then be scanned along the positive and negative y-directions within a selected range. A constant value for the ideal axis can serve as a starting point. If an edge with a certain thickness is found, it can be noticed or detected, and the y-value can be saved. This process can be performed for each x-value of the ideal axis.

[0098] Alternatively, convolutional neural networks (CNNs) (V2*) can be used to process multi-array input data, as well as for image classification. Compared to fully connected multi-layer networks, input data preprocessing can be integrated into the network, and feature extraction can be performed by the network itself. CNNs may be able to consider the correlation of local groups in an image. They can detect features regardless of their location because weights are shared across different regions of the image.

[0099] Figure 5A and Figure 5B Images of CNN training data are shown, where the CNN can be trained in a supervised manner using portions of training images acquired during one or more previous printing processes of a previous 3D structure. The training images show portions of chains or edges. Therefore, training images can be labeled as "chains" or "edges." After training the network, the trained network can be used to classify edges in unknown images acquired from one or more locations determined based on the currently monitored 3D structure.

[0100] Figure 5A An example of training data that can be labeled or identified as edges is shown.

[0101] Figure 5B An example of a training image that can be labeled as a chain is shown.

[0102] Figure 6 An illustration of edge detection using CNN 631 is shown.

[0103] The image obtained based on the formed portion can be cropped along the y-direction to produce cropped images 633. For example, an entire image taken at one of the defined locations during printing can be divided (or cut) into smaller images 633. These smaller cropped images 633 can be fed into a trained CNN 631, which can classify the images based on whether they contain edges.

[0104] Similar to gradient methods, the ideal axis can be used as the starting point for searching edges. At a constant (same) x-coordinate, the y-coordinate of the ideal axis (y0) can be used as the starting point to crop the entire image into smaller images along the positive and negative y-directions. Figure 6 Multiple cropped images 633 corresponding to the same x-coordinate are shown, where each cropped image may correspond to a different y-coordinate. Regions within the cropped images can be analyzed by a CNN 631. As an example, starting with cropped image I(x,y0) at the ideal axis, the CNN can determine that cropped image I(x,y0) has no edges. In cropped images 633 along the positive y-direction, cropped image I(x,y0)... k The first cropped image (or first region) containing the edges can be identified. Therefore, the first region (or cropped image I(x,y)) containing the edges can be saved. k The center point location of )). Similarly, in the cropped image 633 in the negative y direction, the cropped image I(x,y) n The first cropped image (or first region) containing the edges can be identified. Therefore, the first region (or cropped image I(x,y)) containing the edges can be saved. kThe center point location of the edge detection is determined. This process can be repeated for each x-coordinate of the ideal axis. In both gradient-based and CNN-based edge detection, the result of edge detection is finding all edge points located above and below the ideal axis.

[0105] Method 300 (e.g.) Figures 7A to 7C (As shown) may also include identifying 306 (or finding) one or more principal edges, and performing an edge clustering process 307 after identifying (or finding) the 305 edge, because the edge or edge point found in 305 contains correct and / or incorrect edges.

[0106] Figure 7A The diagram shows edge points found during the edge clustering process, including correct and incorrect edges 721. Incorrect edges 721 may have been found due to reflection, noise, and adjacent chains. Therefore, further analysis of the found edges is required.

[0107] Figure 7B The results of edge clustering are shown to determine 306 principal edges. To determine the principal edges of the upper and lower boundaries, the angles between adjacent edge points can be calculated. Points can be clustered in order of their x-values ​​as long as the angle between two adjacent points is below a threshold. If the angle exceeds the threshold, a new cluster can be created. Assuming the longest continuous edge belongs to the true boundary of the chain, the principal edge can be defined as the maximal cluster. A principal edge can be the entire boundary of the chain or a part of the boundary. The results of this clustering are shown in... Figure 7B As shown in the diagram. The longest consecutive edge point cluster can form the main edge 722, such as the main upper edge and the main lower edge. Points 1 and 2 are the starting points for another (or subsequent) clustering process.

[0108] Figure 7C A further (or final) clustering process 307 is illustrated for identifying the upper edge 723 and the lower edge 723 of the cluster. In this further clustering process 307, the remaining edge points can be collected using the outer points of the main edge as starting points. This clustering is based on a nearest neighbor method. If the x and y differences between neighboring parts are below a set threshold, the nearest neighbor can be added to the cluster. In this way, outliers can be removed during the clustering process by using the main edge as a reference. In this manner, the actual edges 723 of the target chain (e.g., the actual upper edge and the actual lower edge) can be identified.

[0109] Method 300 may also include approximating the 308 chain boundaries after performing edge clustering processes 306 and 307. The 308 edge function can be approximated using the cluster edge points by minimizing the squared error. To detect erroneous approximations, the characteristics of the edge function can be analyzed.

[0110] If the two boundaries of the chain are found, the chain diameter can be measured. To achieve a correct measurement process, the measurement axis of the chain (e.g., the actual axis) can be determined. The actual axis of the chain can be calculated using an approximate boundary function. An ideal axis of the chain can lead to incorrect measurements because the chain may shift due to translation and rotation during manufacturing.

[0111] Figure 8 This illustrates how measurements taken with an ideal axis 415 as a reference can lead to erroneous measurements due to the rotation angle α of the print chain relative to the ideal axis (e.g., the angle between the ideal axis 415 and the actual axis 834 can be α, optionally where 0° < α < 180°). The distance between the upper and lower edges can be measured orthogonally to the reference axis. Figure 8 This shows the distance l measured with the actual axis as a reference. real and the distance I measured with an ideal axis as a reference ideal There are differences between them, so errors may occur.

[0112] Method 300 may also include rotating the image 309 to perform the measurement after approximating the chain boundary 308. For example, method 300 may be rotated so that the actual axis matches the desired orientation.

[0113] Method 300 may also include measuring the dimensions of chain 311. For example, the diameter of chain 311 can be measured by calculating the distance between the upper edge 723 and the lower edge 723 in a direction perpendicular to the actual axis.

[0114] Figures 9A to 9D The diagram illustrates various printable 3D structures. Different structures, including breast implants and cell culture nets, can be printed and monitored during the manufacturing process. Structures can be manufactured with a variety of parameters; for example, they can be printed with different nozzle diameters and orifices, as well as different numbers of layers.

[0115] Figure 9A A breast implant scaffold (50 ml volume) is shown. At the bottom, the average pore size can be 6 mm. The pore size decreases with increasing height, so that the average pore size reaches 3 mm at the top. To avoid forming a barrier that hinders tissue growth, these layers are printed offset relative to the previous layer. This technique allows for the formation of sloping channels and a unique overall shape. The outer boundary evolves from a circular structure, while the internal structure contains only orthogonally intersecting straight chains. This structure can be printed using a nozzle diameter of 350 μm.

[0116] Figure 9B A cell culture mesh with 350 μm chain width, 700 μm pore size, and 36° chain orientation offset per layer is shown. This structure can be printed using a nozzle with a diameter of 350 μm.

[0117] Figure 9CA cell culture mesh with a chain width of 150 μm, a pore size of 400 μm, and a chain orientation offset of 36° is shown. This structure can be printed using a nozzle with a diameter of 150 μm.

[0118] Figure 9D A cell culture mesh with a chain width of 150 μm, a pore size of 400 μm, and a chain orientation offset of 90° is shown. This structure can be printed using a nozzle with a diameter of 150 μm.

[0119] To obtain a processable image, LED lights can be used to illuminate the structure from below. Figures 9A to 9D The LED lights can be placed beneath the substrate of the printed structure. To automate camera positioning, the digital code of each structure can be modified as described above to include information associated with one or more locations. The camera can move to the calculated location and pause there until an image is captured and saved by image capture software. For example, 40x magnification can be used to measure the chain diameter of a 50ml breast implant. For example, four chains per layer can be monitored during printing. When printing the mesh, depending on the mesh aperture, 10 chains per layer can be monitored using magnifications of 140 and 200. In other words, the number of locations can be determined by the user and can vary depending on the stringency of manufacturing requirements.

[0120] Figures 10A to 10F This demonstrates the use of gradient-based edge detection for visual monitoring of a 50ml breast stent. Figure 9A (Description) of the image.

[0121] Figure 10A The overlapping vertical chains in the third layer of the three-dimensional structure are shown.

[0122] Figure 10B The horizontal chain in the fifteenth layer of the three-dimensional structure is shown.

[0123] Figure 10C The overlapping horizontal chains in the forty-third layer of the three-dimensional structure are shown.

[0124] Figure 10D The diagram shows the missing horizontal chains next to the boundaries in the fifty-three layers of the three-dimensional structure.

[0125] Figure 10E The horizontal chain in the seventy-fifth layer of the three-dimensional structure is shown.

[0126] Figure 10F The vertical chains in the eightieth layer of the three-dimensional structure are shown.

[0127] Gradient-based methods can be used to detect chain boundaries. An example is shown... Figure 10A The images begin at different heights from the third layer. (The third layer of the image...) Figure 10AIn the image, a paper structure can be seen beneath the substrate, which becomes out of focus as the scaffold grows. Furthermore, the illumination changes as the structure grows. This is because the scaffold can scatter light through its structure. Due to the implant's architecture, overlapping chains may occur during printing, meaning that a portion of the chain in the current layer lies above the chain in the penultimate layer. These layers are not physically connected but are separated by gaps between them, yet they appear to overlap due to the two-dimensional view. This effect occurs... Figure 10A and Figure 10C In the middle. The boundary of the chain (line 1023) can be detected correctly regardless. In Figure 10D In this image, chains that should be observed are missing (e.g., chains expected to be observed based on structural coordinate information). Some dragged chains may be visible in the background of this image. The axis evaluation process correctly indicated the incorrect location of the axes, meaning the chains are not printed on their ideal axes. This displacement can be recorded as a defect. In this image, the ideal axis is represented by line 1024. Figure 10E and Figure 10F The images were taken of the higher layers of the support structure. In these layers, the pore size is significantly smaller than that in the lower layers.

[0128] For example, in the processed image, the minimum (min) or minimum diameter or width and the maximum (max) or maximum diameter or width of the chain, along with their locations, can be visualized and measured. Optionally, the average diameter or width (average), in μm, can be determined by measuring multiple diameters along the monitored chain segment.

[0129] Figures 11A to 11F Visual monitoring is shown Figures 9B to 9D Images depicting chains within a cell culture network. Gradient-based methods can be used to detect boundaries. This is compared to the observation scaffold (in...). Figures 10A to 10F Compared to the previous method, a higher magnification can be set because the structure being measured is smaller.

[0130] Figure 11A and Figure 11B The images show chains with a diameter of 350 μm within a 700 μm mesh, with the chain orientation changing by 36° in each layer. Chains in consecutive layers can intersect each other at 36° angles. These images were taken at 140x magnification.

[0131] The following four images visualize the measurement of chains in a smaller mesh structure. The required mesh aperture is 400 μm, and the required chain diameter is 150 μm. These images were taken at a magnification of 200.

[0132] Figure 11C and Figure 11DThe diagram shows chains with a diameter of 150 μm in a mesh with a pore size of 400 μm, and the chain orientation changes by 36° in each layer.

[0133] Figure 11E and Figure 11F The diagram shows chains with a diameter of 150 μm in a grid with a pore size of 400 μm, and the direction of the chains changes by 90° in each layer.

[0134] Figures 12A to 12F The results of image processing are shown, where a CNN is used to detect edges or boundaries. Figures 12A to 12F The image shown is a raw image taken during the manufacturing process of a 50ml stent, showing the chain and its connection. Figures 10A to 10F The corresponding chains in the gradient-based method described are the same. Similar results were obtained, but some discrepancies can be observed between the measurements using gradient-based edge detection and CNN-based edge detection. These discrepancies range up to 30 μm. Due to these discrepancies, the minimum, maximum, and average diameters measured by the CNN-based edge detection method and the gradient-based method may differ. These discrepancies may be due to the size of the cropped images fed into the CNN network. Since the location of edges on the classified images is not fully known, there may be a discrepancy as large as the size of the cropped images. This effect is observed by examining the prototype approximate edge function. Figure 12F This becomes particularly noticeable because, in most cases, edges are detected earlier (once they appear at the boundaries of the image), resulting in less deviation than when the image is cropped. Due to the size of the images fed into the network, the CNN-based technique shown here appears less accurate than gradient-based methods, which achieve a resolution of 5 μm at a 40x magnification. However, if higher resolution is needed, smaller portions of the image can be fed into the CNN if necessary. For this purpose, the CNN network can also be trained using smaller images. Furthermore, better resolution can be achieved by increasing the training data and / or by training the network with specific images (which include or contain edges that the network could not previously recognize) when necessary.

[0135] In CNN-based edge detection, edges can be found or observed along the entire ideal axis. This can be compared... Figures 10A to 10F as well as Figures 12A to 12F When observing the corresponding image, this was noted. In contrast, using gradient-based edge detection, no edges were found or observed in the first and last segments of the ideal axis. However, these edges can be detected by CNNs. For example, in Figure 10C In this context, gradient-based methods may fail to find edges in the final segment of the ideal axis, while these edges can be detected using CNNs (see...). Figure 12CWhen considering processing time, gradient-based techniques are faster than CNN-based methods. However, whether the larger time consumption of CNN-based methods is critical depends on the application. On the other hand, trained CNNs are able to classify edges in low-contrast regions better than gradient-based methods. Depending on the image size used in CNN-based methods, gradient-based methods may be more accurate. Supports with a 6mm large aperture and a 400μm small aperture can be monitored. However, different magnifications in digital microscopy can be used to compensate for differences in feature size.

[0136] When analyzing data, the average diameter of the chain and the average of all average diameters in a layer are meaningful measurement variables. Insufficient chain diameter limits the mechanical properties of the scaffold. On the other hand, chains that are too thick will stiffen the print and reduce the aperture, which will limit the tissue growth process in medical applications. If the average diameter is not within a given tolerance, the printing parameters must be adjusted. An insufficient average diameter can be compensated for by increasing the material flow rate or decreasing the printing speed. Conversely, the diameter can be reduced by decreasing the flow rate or increasing the speed. Clogged nozzles that require cleaning can also lead to insufficient diameter. Other important measurable variables can be the maximum (max) and minimum (minimum) diameter or width of the chain and its location, as well as the variation of the diameter along the chain. With this information, defects can be classified. A diameter that is within tolerance but varies along the chain can be inferred to be caused by a narrowing effect due to insufficient interlayer spacing or unfavorable temperature conditions. In this case, printing conditions can be improved by increasing the interlayer spacing, extending the cooling process, and / or adjusting the ambient temperature. In addition, the stability of the printing process can be evaluated. To check stability, the standard deviation of one or more or all average diameters, minimum diameters, or maximum diameters in a layer or the entire printed part can be determined. A low standard deviation indicates a consistent and stable printing process, which can be a good starting point for optimizing printing parameters. Conversely, a high standard deviation suggests an unstable process. In this case, it's necessary to ensure a reliable and consistent process before optimizing the chain diameter's dimensional accuracy. Furthermore, an unmeasurable diameter indicates geometric problems, such as a displaced chain. These errors can be caused by warping, nozzle clogging, chain dragging, or inaccurate movement of the printer shaft.

[0137] Some of these assessment methods can be relative to combinations Figure 9A The results extracted during the monitoring process of the breast implant stent are described and interpreted. Three different levels of analysis are illustrated. The first level typically verifies the entire printed part, considering all measured diameters. This allows for an assessment of the overall quality of the printed object. The second level is a layer-by-layer assessment, used to determine whether individual layers meet quality requirements. In the third level, the dimensions of individual chains can be analyzed. This assessment level can be used to classify defects and systematically optimize printing parameters.

[0138] Figures 13A to 13C The first-level evaluation is shown in histograms of all measured chain diameters (minimum, maximum, and average diameters) using a breast support. To assess the quality of the entire printed part or all printed chains at a specific time, the frequency of the measured diameters can be displayed using histograms. This is a suitable technique because histograms can show the entire range of measured diameters. Furthermore, the dispersion of measured diameters can be visualized. This evaluation must be performed not only after printing is complete but may also be performed during printing.

[0139] Figure 13A A histogram is shown, illustrating the relationship between the average chain diameter dispersion and frequency in an example of a breast support structure. Figure 13A As shown, the difference between the ideal diameter of 350 μm and the mean (in μm) of all measured average diameters does not exceed 10 μm. The standard deviation (σ = 9.5) is low.

[0140] Figure 13B A histogram is shown, which illustrates the relationship between the maximum diameter dispersion of the chain and frequency in the example of the breast support.

[0141] Figure 13C A histogram is shown, which illustrates the relationship between the minimum diameter dispersion of the chain and frequency in the example of the breast stent.

[0142] When it comes to the maximum and minimum diameters being measured, the standard deviations are relatively high (σ = 21.8 and σ = 19.8, respectively). Depending on the permissible tolerances, the frequency of chain diameters below the minimum permissible diameter or above the maximum permissible diameter can be determined.

[0143] Figures 14A to 14C The second-level evaluation is shown, such as the layer-by-layer evaluation of a 50ml stent. Figures 14A to 14C The measured diameter is shown layer by layer. Figure 14A The minimum diameter is shown. Figure 14B The maximum diameter is shown. Figure 14C The average diameter is shown. With these charts, it's easy to determine which layers have diameters outside the specified tolerance range. Depending on the quality requirements, if a certain number of diameters within a layer exceed the required tolerance, that layer is marked as a non-conforming layer. Figures 14A to 14C In the examples shown, different thresholds can be determined by the user. For example, the minimum chain diameter should not be less than 225 μm. The maximum diameter should not exceed 475 μm. The average diameter should be in the range of 310 μm to 390 μm. Chains with diameters exceeding the tolerances are considered defects, and a defective process condition can be identified. Whether a single layer or the entire structure meets the required quality depends on the desired quality. Chain diameters exceeding the permissible diameter range are determined to be non-compliant with the specified quality requirements.

[0144] For the purpose of defect classification and optimization, it is necessary to evaluate the diameter length along a single chain.

[0145] Figures 15A to 15B The third-level assessment is shown, in which the chains of a 50ml stent can be evaluated chain by chain.

[0146] Figure 15A The visual results of the measurement process are shown.

[0147] Figure 15B A graph showing all deviations from the ideal chain diameter (350 μm) along chain length L is displayed. For example, the tolerance range can be set to ±50 μm. Measurements less than or equal to ±50 μm from the ideal diameter can be determined as meeting specified quality requirements, while deviations greater than ±50 μm can be determined as not meeting them. The characteristics of the displayed graph will vary depending on the defect type, thus allowing for defect classification. If the cause of the classified defect is known, this classification can be used to systematically adjust printing parameters. Figure 15B In this process, the critical portion of the chain being inspected can be located. The most critical portion of the chain is its minimum diameter, which deviates from the ideal diameter by more than 50 μm. Because the minimum width of the chain is located outside its center, this defect may be classified as a narrowing effect, which could be due to an inappropriate cooling process or insufficient interlayer spacing.

[0148] Therefore, it can be understood that whether the printing process stops (e.g., stops printing 180) or does not stop (e.g., instead adjusts printing parameters 170) is based on the rigor required for the evaluation process. For example, the decision to stop printing 180 depends on the frequency and / or magnitude of the deviations between the determined parameter values ​​and the ideal parameter values. The thresholds for these deviations can be determined by the user.

[0149] Figure 16 An illustration shows an arrangement 150 for forming a three-dimensional structure. Arrangement 150 can be configured to perform or implement combinations. Figures 1A to 15B The method described.

[0150] Arrangement 150 includes a device 101 for forming a three-dimensional structure 104 based on structural coordinate information associated with the three-dimensional structure 104. Arrangement 100 also includes a movable sensing device 102. Arrangement 100 also includes a processor 103. The processor 103 is configured to: determine one or more positions for positioning the sensing device 102 based on the structural coordinate information associated with the three-dimensional structure to be formed, and control the positioning of the movable sensing device 102 to one or more positions.

[0151] The apparatus 101 for forming three-dimensional (3D) structures can be configured as a stereolithography (SLA) apparatus, a digital light processing (DLP) apparatus, a fused deposition modeling (FDM) apparatus, a selective laser sintering (SLS) apparatus, a selective laser melting (SLM) apparatus, an electron beam melting (EBM) apparatus, a layered object manufacturing (LOM) apparatus, a bond jetting (BJ) apparatus, and / or a material jetting (MJ) apparatus.

[0152] For example, sensing device 102 may be an imaging device, a camera or 3D scanning device, a thermal imager, a digital microscope, an acoustic emission sensor and / or a CT device.

[0153] Processor 103 can be any computer or machine capable of executing instructions from a computer-readable storage medium. Such a computer-readable storage medium may include instructions that, when executed by the computer (or processor) 103, cause the computer 103 to perform actions in conjunction with… Figures 1A to 15B The method described.

[0154] Deployment 150 can be a vision-based monitoring system used to perform combined... Figures 1A to 15B The method described herein. A sensing device 102 (e.g., a commercial digital microscope) can be configured to acquire data (e.g., images) layer by layer. The acquired data can be processed to detect defects. After printing one (or each) layer, the sensing device 102 can be positioned at one or more locations along the axis of the printer. Depending on the field of view and the size of the feature being monitored, one or more images can be captured and processed.

[0155] For the automation of camera positioning and image capture, it is necessary to know the time of image capture and the optimal camera position. This information can be extracted from digital codes, which include all printer movements used to print the 3D scaffold structure. The digital codes can be separated to determine potential camera positions and modified with commands for camera positioning. Therefore, processor 103 can be configured to process structural coordinate information related to the layout and / or internal and / or external structure of the 3D structure to be formed. For example, structural coordinate information may include CAD-based information or digital code information. Based on the structural coordinate information, processor 103 can be configured to generate adaptive digital code instructions for forming the 3D structure 104. The adaptive digital code instructions may include structural coordinate information related to the 3D structure to be formed and information related to positioning the sensing device at one or more locations.

[0156] Device 101, movable sensing device 102 and processor 103 can be interconnected so that the movement of device 101 controlled by digital code instructions can be part of device 101 responsible for forming a three-dimensional structure at a selected location defined by the structural coordinate information of the digital code instructions (e.g., printhead, dispenser, nozzle, extruder or laser).

[0157] The processor 103 can also be configured to determine the process state based on acquired data associated with at least a portion of the part forming the three-dimensional structure. The acquired data may be acquired or generated by sensing devices 102 at determined locations in one or more locations.

[0158] It is understood that, in some embodiments, arrangement 150 includes forming apparatus 101 for forming a three-dimensional structure comprising multiple layers. Arrangement 150 includes a movable sensing device 102. Arrangement 150 includes a processor configured to:

[0159] (a) For a set of multiple layers, determine at least one location for locating the sensing device.

[0160] (b) Based on structural coordinate information related to the three-dimensional structure, the control device forms the layer set.

[0161] (c) After forming a set of layers and before forming another set of layers, control the positioning of the movable sensing device to at least one location associated with the set of layers, and

[0162] (d) Determine the process state based on the data obtained from the formed set of layers, wherein the data is acquired by a sensing device at at least one location associated with the formed set of layers.

[0163] The various embodiments described herein relate to vision-based systems for in-situ defect detection during the additive manufacturing process of porous scaffolds. They provide a complete process for defect detection systems based on measuring the diameter of the printed chain. Using a digital microscope with adjustable magnification as a sensor and vision-based data processing, small defects in the structure can be detected directly. Various concepts may include automated camera localization using digital codes and image processing of the captured images in which the diameter of the printed chain is measured. Image processing can be implemented using at least two different edge detection methods, such as gradient-based edge detection or CNN-based edge detection.

[0164] Various embodiments and examples can be integrated into existing printing environments, where monitored chains can be detected and measured. These embodiments can provide one or more options for measurement analysis. For example, based on quality control requirements, the quality of individual chains, layers, or the overall structure can be inspected.

[0165] Various embodiments can be implemented at different levels of automation. Manual quality control is also suitable. For example, a user can manually place the structure under a microscope and take an image of the upper layer. Computer-executable instructions can be configured to manually draw the axes of the chain to be measured by marking the start and end points of the axes. The chain can be measured via computer-executable instructions, and the output can be displayed and saved. In this case, no modification to the digital code is required. Optionally or alternatively, automation of the monitoring process can be performed through automatic camera positioning, image acquisition, and processing, but without closed-loop control. Here, the user can access the printing status at any time during and after the manufacturing process and intervene if necessary. A fully automated closed-loop system can also be implemented. In this case, printing can run as long as quality requirements are met, otherwise it will be stopped, or the printing parameters will be automatically adjusted to compensate.

[0166] Various embodiments can be used to monitor 3D structures using any 3D printing method. In particular, various embodiments can monitor the quality of FDM-printed chains in porous scaffolds during manufacturing. Measuring the chain diameter is useful for classifying and locating defects and their sources. Manufacturers may be able to analyze the quality of the printed object in situ. FDM-printed porous structures are becoming increasingly popular in tissue engineering applications, such as in cell culture networks and uniquely customized implants. For scaffold fabrication, PCL can be used as a material due to its good rheological and viscoelastic properties. PCL also exhibits good thermal stability, making it suitable for FDM printing processes. Especially for applications with a medical background, complex quality control processes are required. Although sensing devices have become essential components in manufacturing, FDM, as one of the most popular 3D printing technologies, lacks monitoring systems and open-loop control systems. The lack of sensing during AM manufacturing makes in-situ quality control very challenging. Post-printing quality control is also complex because the internal structures of the printed part are difficult or impossible to access.

[0167] Combination Figures 1A to 16 The described methods and arrangements may relate to a vision-based inspection process for in-situ monitoring of porous scaffolds in additive manufacturing. For example, the methods and arrangements may include assessing the quality of the printed scaffold by monitoring the geometry of its internal structure. A commercially available digital microscope with adjustable magnification can be used to capture a selected internal region of the printed object. The captured images can be analyzed using computer vision methods. At least one of two different edge detection techniques can be used to detect boundaries in the selected region of the image. The first edge detection technique may include or may be a gradient-based method. The second edge detection technique may be based on classification using a convolutional neural network (CNN). These methods and arrangements may include the use of automatic camera localization, which may be controlled based on and / or through the separation and modification of digital codes.

[0168] Structures to be formed, such as breast implants (e.g., biodegradable) and / or cell culture nets, can be monitored during the manufacturing process. These methods can be integrated into various stages of the quality control process. It can be part of a closed-loop system that automatically compensates for defects, or it can halt the manufacturing process upon the occurrence of a significant defect.

[0169] Sensing devices can be used in manufacturing processes to achieve better process reliability, repeatability, and automation. The ideal effects of monitoring the printing process are: detecting errors during printing; preventing further failures by adjusting printing parameters; automatically stopping the printing process when detected errors are severe; and detecting errors in manufactured products that are difficult to detect (such as defects in internal features). This effectively saves printing time, reduces material waste, and improves quality control and record-keeping in the manufacturing process.

[0170] The various embodiments described herein relate to in-situ monitoring of printed (e.g., FDM-printed) porous structures. Prior art primarily concerns solid printed objects because it monitors the dimensions of the external boundaries rather than the internal structure. Typically, prior art cannot achieve resolutions smaller than one millimeter. The quality and mechanical properties of porous structures are scientifically influenced by the appearance of the internal structure; therefore, the various embodiments described herein relate to monitoring systems, methods, and arrangements that can also monitor the internal structure. For a meaningful assessment of print quality, measurements in the sub-millimeter range can be performed.

[0171] In some embodiments, a digital code can be used to obtain the ideal structure of the printed object as a reference, instead of using CAD data. In such embodiments, no CAD program is used. A representation of the ideal appearance can be obtained using digital code because all machine movements and the diameter of the printing chain are known through the code.

[0172] The various embodiments described herein relate to the monitoring of chain diameter. A quality characteristic of the manufactured structure is the size of the printed chain. High-quality products are characterized by a consistent chain diameter with a specified value. Consistency in chain diameter also indicates a stable and reliable printing process. On the other hand, inconsistent chain diameters and deviations from the desired chain diameter indicate an unstable manufacturing process. Therefore, monitoring the chain diameter during printing not only enables reliable in-situ quality control but also optimizes the manufacturing process. For tissue engineering, chain diameter is an important variable because diameter deviations alter pore size, which may hinder normal tissue growth or change mechanical properties.

[0173] The present invention is further characterized by the following items.

[0174] Project 1: A method for forming a three-dimensional structure, the method comprising:

[0175] Based on the structural coordinate information related to the three-dimensional structure to be formed, one or more locations for locating the sensing device are determined.

[0176] Based on the structural coordinate information, a portion of the three-dimensional structure is formed; and

[0177] The sensing device is positioned at one of the one or more locations.

[0178] Project 2: According to the method described in Project 1, the structural coordinate information includes information related to the layout of the three-dimensional structure.

[0179] Project 3: According to the method described in Project 1 or 2, wherein the structural coordinate information includes information related to the internal structure of the three-dimensional structure.

[0180] Project 4: The method according to any one of Project 2 or 3, wherein the structural coordinate information includes tool path instructions for controlling the forming apparatus for forming the three-dimensional structure.

[0181] Project 5: The method according to any one of Projects 1 to 4 further includes: determining a process state based on acquired data related to at least a portion of the formed portion of the three-dimensional structure, wherein the acquired data is acquired by the sensing device at the determined location.

[0182] Project 6: According to the method described in Project 5, determining the process state includes: determining the parameter values ​​of the structural features of the formed part of the three-dimensional structure based on the acquired data.

[0183] Project 7: The method according to Project 6, wherein the parameter value is at least one of the structural features of the portion formed by the three-dimensional structure, namely, length, width or diameter, height, roughness, color, uniformity, thickness, and tilt angle.

[0184] Project 8: The method according to Project 6 or 7, wherein the structural feature includes segments of the internal structure of the three-dimensional structure.

[0185] Project 9: According to the method of Project 8, the parameter value of the feature is determined by detecting the edge of the chain segment.

[0186] Project 10: According to the method of Project 9, the edge of the chain segment is detected by implementing an artificial neural network process or a gradient-based detection process.

[0187] Item 11: The method according to any one of Items 5 to 9, wherein determining the process state includes:

[0188] Based on the structural coordinate information, the ideal axis of the chain segment of the part of the three-dimensional structure is determined;

[0189] Based on the acquired data of the formed portion, determine the actual axis of the chain segment of the formed portion; and

[0190] The process state is determined based on a comparison between the ideal shaft and the actual shaft.

[0191] Item 12: The method according to Item 11 includes: determining a defective process state if the difference between the ideal axis and the actual axis exceeds a threshold.

[0192] Item 13: The method according to any one of Items 5 to 12, wherein determining the process state includes:

[0193] Compare the ideal parameter values ​​of the chain segments of the portion of the three-dimensional structure with the determined parameter values ​​of the chain segments of the formed portion.

[0194] Specifically, based on data acquired from at least a portion of the formed portion of the three-dimensional structure, the determined parameter value of the chain segment is determined, and

[0195] The ideal parameter value of the chain segment is determined based on at least one of the structural coordinate information and the input value.

[0196] Project 14: The method according to any one of Projects 5 to 13, comprising: determining a defective process state if the difference between the ideal parameter value and the determined parameter value exceeds a threshold.

[0197] Item 15: The method according to any one of Items 5 to 14, wherein determining the process state includes:

[0198] Determine multiple parameter values ​​associated with multiple chain segments of the formed portion;

[0199] Statistical parameters for determining the values ​​of the plurality of parameters, and

[0200] If the difference between the statistical parameter and the comparison parameter exceeds a threshold, a defective process condition is determined.

[0201] Item 16: The method according to any one of Items 1 to 15 further includes: adapting process parameters for forming the three-dimensional structure based on the determined process state.

[0202] Item 17: The method according to any one of Items 1 to 16, wherein the portion of the three-dimensional structure comprises a set of layers of the three-dimensional structure.

[0203] Item 18: The method described in Item 17 also includes...

[0204] For each layer set among the plurality of layers, a corresponding location group for locating the sensing device is determined, forming a layer set within the plurality of layers; and

[0205] After the set of layers is formed and before the subsequent set of layers is formed, the sensing device is sequentially positioned at one of the locations in the set of locations, and at each location, the data of the set of layers is acquired.

[0206] Item 19: The method according to Item 17 or 18, wherein the layer set includes at least a first sub-layer and at least a second sub-layer;

[0207] The location among the plurality of locations is determined based on the intersection between the first sub-layer and the second sub-layer.

[0208] Item 20: The method according to Item 18 or 19, wherein the location of the plurality of locations is located between two intersections of the first sub-layer and the second sub-layer.

[0209] Item 21: The method according to any one of Items 1 to 20, wherein the one or more locations include a plurality of locations, wherein the plurality of locations are random, continuous, array-based, or user-determinable.

[0210] Item 22: The method according to any one of items 17 to 21, wherein the method comprises:

[0211] Repeatedly alternate between the following:

[0212] Forming a layer set within the plurality of layer sets; and

[0213] After the formation of the layer set and before the formation of another layer set, the sensing device is sequentially positioned at the position in the position group of the layer set.

[0214] Item 23: A computer-readable storage medium comprising instructions that, when executed by a computer, cause the computer to perform the method according to any one of items 1 to 22.

[0215] Item 24: An arrangement for forming a three-dimensional structure, the arrangement comprising:

[0216] A forming device for forming the three-dimensional structure based on structural coordinate information associated with the three-dimensional structure;

[0217] Mobile sensing devices; and

[0218] processor,

[0219] The processor is configured as follows:

[0220] Based on the structural coordinate information related to the three-dimensional structure to be formed, one or more locations for locating the sensing device are determined.

[0221] The control positions the movable sensing device to one or more locations.

[0222] Item 25: According to the arrangement described in Item 24, the processor is further configured to generate adaptive digital code instructions for forming the three-dimensional structure, wherein the adaptive digital code instructions include structural coordinate information related to the three-dimensional structure to be formed and information related to the positioning of the sensing device at the one or more locations.

[0223] Item 26: According to the arrangement described in Item 24, wherein the processor is further configured to determine the process state based on acquired data associated with at least a portion of the formed portion of the three-dimensional structure, wherein the acquired data is acquired by the sensing device at a location in one or more of the locations.

[0224] Item 27: An arrangement for forming a three-dimensional structure, the arrangement comprising:

[0225] Forming equipment for forming three-dimensional structures comprising multiple layers;

[0226] Mobile sensing devices;

[0227] The processor is configured as follows:

[0228] For the set of multiple layers, determine at least one location for locating the sensing device.

[0229] Based on the structural coordinate information associated with the three-dimensional structure, the device is controlled to form the layer set.

[0230] After the formation of the layer set and before the formation of another layer set, control is used to position the movable sensing device to at least one location associated with the layer set, and

[0231] Based on the acquired data from the formed set of layers, the process state is determined, wherein the acquired data is obtained by the sensing device at the at least one location associated with the formed set of layers.

[0232] The embodiments of the invention have been described in detail. It will be understood that the invention as defined by the appended claims is not limited to the specific details set forth in the foregoing description, and many obvious changes may be made to the invention without departing from the spirit or scope thereof.

Claims

1. A method for forming a three-dimensional structure, the method comprising: Based on structural coordinate information related to the layout of the three-dimensional structure to be formed, one or more locations for locating the sensing device are determined. Based on the structural coordinate information, a portion of the three-dimensional structure is formed; Position the sensing device at one of the one or more locations; as well as The process state is determined based on the acquired data relating to at least a portion of the formed portion of the three-dimensional structure, wherein the acquired data is obtained by the sensing device at a location in one or more of the determined locations.

2. The method according to claim 1, wherein, The structural coordinate information includes information related to the internal structure of the three-dimensional structure.

3. The method according to claim 1 or 2, wherein, The structural coordinate information includes tool path instructions for controlling the forming equipment used to form the three-dimensional structure.

4. The method according to any one of claims 1 or 2, wherein, Determining the process state includes: based on the acquired data, determining the parameter values ​​of the structural features of the formed portion of the three-dimensional structure.

5. The method according to claim 4, wherein, The parameter value is at least one of the following: length, width or diameter, height, roughness, color, uniformity, thickness, and tilt angle of the structural feature of the portion formed by the three-dimensional structure.

6. The method according to claim 4, wherein, The structural features include chain segments of the internal structure of the three-dimensional structure.

7. The method according to claim 6, wherein, The parameter value of the feature is determined by detecting the edges of the chain segment.

8. The method according to claim 7, wherein, The edges of the chain segment are detected by implementing an artificial neural network process or a gradient-based detection process.

9. The method according to any one of claims 1, 5 to 8, wherein, Determining the process state includes: Based on the structural coordinate information, the ideal axis of the chain segment of the part of the three-dimensional structure is determined; Based on the acquired data of the formed portion, determine the actual axis of the chain segment of the formed portion; and The process state is determined based on a comparison between the ideal shaft and the actual shaft.

10. The method of claim 9, comprising: If the difference between the ideal axis and the actual axis exceeds a threshold, a defective process condition is determined.

11. The method according to any one of claims 1, 5 to 8 and 10, wherein, Determining the process state includes: Compare the ideal parameter values ​​of the chain segments of the portion of the three-dimensional structure with the determined parameter values ​​of the chain segments of the formed portion. Specifically, based on data acquired from at least a portion of the formed portion of the three-dimensional structure, the determined parameter value of the chain segment is determined, and The ideal parameter value of the chain segment is determined based on at least one of the structural coordinate information and the input value.

12. The method of claim 11, comprising: If the difference between the ideal parameter value and the determined parameter value exceeds a threshold, then a defective process state is determined.

13. The method according to any one of claims 1, 5 to 8, 10 and 12, wherein, Determining the process state includes: Determine multiple parameter values ​​associated with multiple chain segments of the formed portion; Statistical parameters for determining the values ​​of the plurality of parameters, and If the difference between the statistical parameter and the comparison parameter exceeds a threshold, a defective process condition is determined.

14. The method according to any one of claims 1, 5 to 8, 10 and 12, further comprising: Based on the determined process state, process parameters are adapted to form the three-dimensional structure.

15. The method according to any one of claims 1, 5 to 8, 10 and 12, wherein, The portion of the three-dimensional structure includes a set of layers of the three-dimensional structure.

16. The method of claim 15, further comprising: For each layer set among the plurality of layers, a corresponding set of locations for locating the sensing device is determined. To form a layer set within the multiple layer sets; as well as After the layer set is formed and before the subsequent layer set is formed, the sensing device is sequentially positioned at the locations in the location group, and at each location, the data of the formed layer set is acquired.

17. The method according to claim 15, wherein, The layer set includes at least a first sub-layer and at least a second sub-layer; The location among the plurality of locations is determined based on the intersection between the first sub-layer and the second sub-layer.

18. The method according to claim 17, wherein, The location of the plurality of locations is located between two intersections of the first sub-layer and the second sub-layer.

19. The method according to any one of claims 1, 5 to 8, 10, 12, 16 to 18, wherein, The one or more locations include multiple locations, wherein the multiple locations are random, sequential, array-based, or user-determinable.

20. The method of claim 16, wherein, The method includes: Repeatedly alternate between the following: Forming a layer set within the plurality of layer sets; and After the layer set is formed and before another layer set is formed, the sensing device is sequentially positioned at the position in the position group of the layer set.

21. An arrangement for forming a three-dimensional structure, the arrangement comprising: A forming device for forming the three-dimensional structure based on structural coordinate information associated with the three-dimensional structure; Mobile sensing devices; as well as processor, The processor is configured as follows: Based on the structural coordinate information associated with the three-dimensional structure to be formed, one or more locations for locating the movable sensing device are determined. Controlling the positioning of the movable sensing device to the one or more locations; and The process state is determined based on data acquired in relation to at least a portion of the formed portion of the three-dimensional structure, wherein the acquired data is obtained by the movable sensing device at one or more of the locations.

22. The arrangement according to claim 21, wherein, The processor is also configured to generate adaptive digital code instructions for forming the three-dimensional structure, wherein the adaptive digital code instructions include structural coordinate information related to the three-dimensional structure to be formed and information related to the positioning of the movable sensing device at one or more locations.

23. An arrangement for forming a three-dimensional structure, the arrangement comprising: Forming equipment for forming three-dimensional structures comprising multiple layers; Mobile sensing devices; The processor is configured as follows: For the set of multiple layers, determine at least one location for locating the movable sensing device. Based on the structural coordinate information associated with the three-dimensional structure, the layer set is formed by the forming device. After the formation of the layer set and before the formation of another layer set, control is used to position the movable sensing device to at least one location associated with the layer set, and Based on the data acquired from the formed set of layers, the process state is determined, wherein the acquired data is obtained by the movable sensing device at the at least one location associated with the formed set of layers.

24. A computer-readable storage medium comprising instructions that, when executed by a processor of an arrangement according to any one of claims 21 to 23, cause the arrangement to perform the method according to any one of claims 1 to 20.

Images