Method and system for predicting heating in additive manufacturing

Abstract

A method and system for predicting overheating in layers of build material in an additive manufacturing system are provided. The method comprises accessing a representation of a three-dimensional object as a plurality of cross-sectional layers for printing in an additive manufacturing system, accessing scan path data for a region of a selected one of the plurality of cross-sectional layers, sub-dividing the region into a plurality of cells and for each cell, generating a vector of features based on the representation of the object and scan path data and evaluating a trained predictive model based the vector of features, to obtain an output comprising a predicted melting time for the cell.

Description

METHOD AND SYSTEM FOR PREDICTING HEATING IN ADDITIVE MANUFACTURINGTECHNICAE FIELD

[0001] The present disclosure relates to a method and system for predicting melting times of build material during manufacture of an object in an additive manufacturing system.BACKGROUND

[0002] Additive manufacturing is a technology for constructing three-dimensional objects from digital models. In contrast to traditional subtractive manufacturing methods such as milling or machining, where material is removed to create a final product, additive manufacturing adds material on a layer-by-layer basis. The build material may include plastics, metals, ceramics, or many other types of materials.

[0003] Powder bed fusion (PBF) technology is an additive manufacturing technology that facilitates the creation of high-quality parts with complex geometries as well as optimization for specific properties such as weight and strength. PBF systems comprise a build platform on to which powdered build material is deposited in layers. Once a layer of build material has been deposited, a high-energy source such as a laser beam is controlled to selectively heat the powdered material, causing the material to heat up and solidify to form a cross-sectional layer of an object.

[0004] Common PBF methods include selective laser sintering (SLS) and selective laser melting (SLM). In SLS, a laser methodically sinters particles of a polymer-based powder and builds parts layer by layer. SLM systems may operate on metal alloy powders. SLM is used for the production of metal components in the aerospace, automotive, and healthcare industries.

[0005] In PBF systems, the high-energy source is controlled via software, which directs the source to heat the build material along scan paths. A scan path is a predetermined path the heat source follows to create a layer of the object. The scan path taken by the heat source and the traversal time and power of the heat source all play a significant role in determining the material properties of the build and avoiding defects in manufacturing.

[0006] Creating optimal scan paths is a complex task. Traditional scan path algorithms use sets of globally optimized printing parameters that are derived from tests on large numbers of part specimens. These parameters are selected to globally optimize material properties such as density and tensile strength as well as build times for the entire build. Such globally optimized parameters may lead to local part-specific print problems. Parameters obtained from trial-and-error experiments cannot guarantee the local building quality of geometries with complex topology and intricate internal structures. Build parameters that are optimal for printing simple test parts may be sub-optimal or even problematic for printing specific geometry features in actual parts.

[0007] A major cause of local part-specific problems is localized overheating. Overheating may occur where, for example, the energy source is repeatedly directed over the same area, or where the energy source is directed to one point for too long. Local overheating has also been observed to occur in overhanging regions. Overhanging regions, which are unsupported by the underlying layers, may be prone to overheating as there is no build material beneath the overhanging layers to conduct heat away, causing heat to accumulate.

[0008] Finite Element Analysis (FEA) simulation may be used to analyze and optimize various aspects of PBF processes. FEA may be used to simulate the thermal behavior of the PBF process including prediction of temperature gradients. This enables an engineer to identify potential areas of overheating in a build. Despite this, the application of FEA remains limited. It is not efficient to perform FEA at path scale: an individual build file may be terabytes of data owing to the small layer thickness and hatching distances involved in the build. Consequently, detecting local problems for complex part geometry with conventional FEA would take many months, which makes it impractical.SUMMARY

[0009] It is an object of the disclosure to enable quick identification of regions of overheating when manufacturing an object in an additive manufacturing system.

[0010] The foregoing and other objects are achieved by the features of the independent claims. Further implementation forms are apparent from the dependent claims, the description, and the figures.

[0011] According to a first aspect, a computer-implemented method for predicting overheating in layers of build material in an additive manufacturing system is provided. The method comprises accessing a representation of a three-dimensional object as a plurality of cross-sectional layers for printing in an additive manufacturing system and accessing scan path data for a region of a selected one of the plurality of cross-sectional layers, sub-dividing the region into a plurality of cells. Further, the method comprises, for each cell: generating a vector of features based on the representation of the object and scan path data; and evaluating a trained predictive model based the vector of features, to obtain an output comprising a predicted melting time for the cell.

[0012] According to a second aspect, a computer-implemented method for training a predictive model for predicting a melting time of build material in a deposited layer of build material in an additive manufacturing system is provided. The method comprises: accessing a set of model parameters for the predictive model; accessing object data for an object and print data for printing the object in the additive manufacturing system comprising a representation of the object as a plurality of cross-sectional layers and scan path data for each of the plurality of cross-sectional layers; accessing training data for the object comprising a simulation of the melting time in each of the plurality of cross- sectional layers; sub-dividing the object into a plurality of cells, based on the object data; generating a vector of features for each of the plurality of cells, based on the print data; evaluating the predictive model to obtain a predicted melting time for each cell; and modifying the set of model parameters based on the training data and predicted melting time in each cell.

[0013] According to a third aspect, a data processing system comprising a processor and a memory is provided. The memory stores instructions that, when executed by the processor cause the processor to execute the method according to the first aspect.

[0014] According to a fourth aspect, a non-transitory computer readable storage medium is provided. The non-transitory computer readable storage medium comprises program code that, when executed by a processor, provides instructions to perform the method according to the first aspect.

[0015] In a first implementation form of the method according to the first aspect, the melting time comprises a maximum period of time the build material for the cell exceeds a melting point of the build material.

[0016] In a second implementation form, the method comprises comparing the predicted melting time to a threshold value; and identifying the region as an overheating region, when the melting time exceeds the threshold value.

[0017] In a third implementation form, the method comprises modifying the scan path in response to identifying the region as an overheating region.

[0018] In a fourth implementation form, modifying the scan path comprises modifying an energy source power, scan speed, delay time, or re-generating portions of the scan path.

[0019] In a fifth implementation form, the trained predictive model comprise a neural network model.

[0020] In a sixth implementation form, the vector of features comprises an energy density value for the cell.

[0021] In a seventh implementation form, the vector of features comprises a value determined on the basis of energy source timing data for neighboring cells in the cross- sectional layer.

[0022] In an eighth implementation form, the selected cross-sectional layer is selected on the basis of an output of a thermal analysis test applied to each of the plurality of cross-sectional layers.

[0023] These and other aspects of the disclosure are apparent from the embodiments described below.BRIEF DESCRIPTION OF THE DRAWINGS

[0024] For a more complete understanding of the present disclosure, and the advantages thereof, reference is now made to the following descriptions taken in conjunction with the accompanying drawings, in which:

[0025] Figure 1 is a schematic diagram of a computing system, according to an example;

[0026] Figure 2 is a schematic perspective illustration showing scan paths on an object, according to an example;

[0027] Figure 3 A is a schematic perspective illustration showing a part for manufacture, according to an example;

[0028] Figure 3B is a cross-sectional slice of the part shown in Figure 3 A, according to an example;

[0029] Figure 3C is a scan strategy for the cross-sectional slice shown in Figure 3B, according to an example;

[0030] Figure 3D shows overheating on the cross-sectional slice, shown in Figure 3B, according to an example;

[0031] Figure 4 is a schematic diagram showing a Machine Learning Unit (MLU), according to an example;

[0032] Figure 5 is a flow diagram showing a method for training a predictive model for predicting melting times, according to an example;

[0033] Figure 6 shows examples of training objects for training a predictive model, according to an example;

[0034] Figure 7 is a flow diagram showing a method for predicting a melting time of build material, according to an example;

[0035] Figure 8 shows a simplified schematic diagram of a computing system, according to an example.DETAILED DESCRIPTION

[0036] Example embodiments are described below in sufficient detail to enable those of ordinary skill in the art to embody and implement the systems and processes herein described. It is important to understand that embodiments can be provided in many alternate forms and should not be construed as limited to the examples set forth herein.

[0037] Accordingly, while embodiments can be modified in various ways and take on various alternative forms, specific embodiments thereof are shown in the drawings and described in detail below as examples. There is no intent to limit to the particular forms disclosed. On the contrary, all modifications, equivalents, and alternatives falling within the scope of the appended claims should be included. Elements of the example embodiments are consistently denoted by the same reference numerals throughout the drawings and detailed description where appropriate.

[0038] The terminology used herein to describe embodiments is not intended to limit the scope. The articles “a,” “an,” and “the” are singular in that they have a single referent, however the use of the singular form in the present document should not preclude the presence of more than one referent. In other words, elements referred to in the singular can number one or more, unless the context clearly indicates otherwise. It will be further understood that the terms “comprises,” “comprising,” “includes,” and / or “including,” when used herein, specify the presence of stated features, items, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, items, steps, operations, elements, components, and / or groups thereof.

[0039] Unless otherwise defined, all terms (including technical and scientific terms) used herein are to be interpreted as is customary in the art. It will be further understood that terms in common usage should also be interpreted as is customary in the relevant art and not in an idealized or overly formal sense unless expressly so defined herein.

[0040] The methods and systems described herein utilize machine learning to enable rapid identification of regions of local overheating in additive manufacturing. In addition the methods described herein provide intelligent scan path optimization where needed, to prevent problems in specific layers and region.

[0041] According to examples, each layer of a build is subdivided into a grid structure comprising a plurality of grid cells, where the size of the cells is determined by the hatching distance of the energy source. A machine learning simulation is applied to a unit, referred to as a Machine Learning Unit (MLU), which comprises the cell where the energy source is applied and the neighboring cells.

[0042] The methods described herein enable efficient simulation of thermal behavior for scan paths to identify regions of overheating. The simulation may be performed as soon as a scan path is output from a scan path generation algorithm. Simulation may be performed in parallel with scan path generation for the subsequent layer to minimize the overhead from the simulation on the print job generation time. The Machine Learning Unit (MLU)-based simulation described herein may be deployed on any material and multi-laser scan paths with any combination of scan path strategies.

[0043] Scan path optimization strategies are used in regions where the simulation predicts issues. In particular, in identified regions of overheating, scan strategies or buildparameters may be modified. This may include, for example, varying laser power, scan speed or delay times, locally re-ordering scan vectors, or even regenerating scan paths.

[0044] Figure 1 is a simplified schematic diagram of a computing system 100, according to an example. The computing system 100 may be used to implement scan path generation and optimization for additive manufacturing. In examples described herein, the computing system 100 may be an application server, a compute node, desktop, laptop, smart phone or other portable device, an embedded device, or similar. In other cases, the computing system 100 may represent a plurality of computing systems in a distributed or networked configuration.

[0045] The computing system 100 may communicate with other systems via wired and wireless interfaces (not shown in Figure 1). Examples of devices and systems with which the computing system 100 may communicate via a wired connection or wirelessly include a keyboard, a mouse, monitors, touch screen interfaces, trackballs, cameras, microphones, scanners, printers, speakers, pointers, touch pads, drawing tablets, joysticks, keypads, motion sensing devices that captures motion gestures, and the like. Some devices function as input devices facilitating user input into the computing system 100, whereas other devices function as output devices.

[0046] The computing system 100 may be configured to implement computer aided design (CAD) and computer aided manufacturing (CAM) applications for manufacture of objects in an additive manufacturing system. It will be appreciated that CAD and CAM applications may be used as a graphical user interface that facilitate interaction with a computer-generated model of an object.

[0047] According to examples, the computing system 100 comprises a storage device 110. The storage device 110 is configured to store object data for an object to be manufactured in an additive manufacturing system. The storage device 110 is communicatively coupled to a build processor 120. The build processor 120 is configured to access object data for the object from the storage device 110 and generate a print data to print the object in the additive manufacturing system. The build processor 120 may take the form of a dedicated hardware device, or software implemented by a general- purpose processor. A processor may take the form of single core or multicore processor system. The print data comprises a subdivision of the object into a plurality of cross-sectional layers and scan path data specifying scan paths for controlling the energy source of the additive manufacturing system during printing.

[0048] Figure 2 is a schematic perspective illustration 200 showing scan paths in an additive manufacturing system. Figure 2 shows an energy source 210 which is part of an additive manufacturing system. The energy source 210 may be a laser or other form of high energy source such as an electron beam. The energy source 210 is controlled to direct energy towards a layer 220 comprising deposited build material. The layer 220 may be deposited on a build platform of the additive manufacturing system.

[0049] The energy source 210 is controlled by the additive manufacturing system to traverse scan paths 230 illustrated by the arrows in the layer 220 so that energy is applied to the layer along the scan paths 230. The scan paths 230 are generated by a build processor such as the build processor 120 shown in Figure 1. In Figure 2, the scan paths 230 are parallel vectors which span the length of the layer 220. In other examples, scan paths may follow a meandering, zig-zag, spiral, chessboard, contour, island, or raster pattern. The distance, d, between adjacent scan paths is referred to herein as the hatch distance.

[0050] Figure 3 A is a schematic perspective illustration 300 showing a part 310 for manufacture in an additive manufacturing system, according to an example. Figure 3 A further shows a cross-sectional slice 320 of part 310. A view of the same cross-sectional slice 320 is also illustrated in Figure 3B. The part 310 may be represented as object data in a computer-aided design system. The object data for part 310 may be stored in the storage 110 shown and accessed by the build processor 120. The object data may be in stereolithography file format (STL) or additive manufacturing file format (AMF).

[0051] To illustrate the problem of overheating, Figure 3C shows a scan strategy for the cross-sectional layer 320 of part 310 shown in Figure 3. In Figure 3 A, the scan paths form a chessboard pattern across the cross-sectional layer 320. Figure 3D illustrates the thermal effects of the scan strategy from Figure 3D when a laser with a uniform powerdistribution is applied. In the case of uniform power, the region 330 overheats due to the fact that the region 330 overhangs with respect to the build platform during manufacture of the part 300.

[0052] Figure 4 is a schematic diagram showing a Machine Learning Unit (MLU) 400, according to an example. The MLU 400 comprises a grid representation comprising a plurality of cells 410. In the example shown in Figure 4, the MLU comprises a 3x3x3 grid of cells. In other examples, the MLU 400 may comprise a 4x4x4 or 5x5x5 grid. The MLU 400 is defined for each point of each scan path in a sub-region of a layer. The center cell 420 represents the point to which the energy source is being applied. Each MLU is associated with parameters. The parameters include the energy source settings: the power, scan speed, and time at the center cell 420 of the grid, as well as the density of material in each cell.

[0053] Figure 4B shows a top view of the MLU 410. The data values tt in each grid cell represent the energy source timing at the j’th cell of the i ’th nearest neighbor to the center cell 420. A vector of features, v = (Vi)l ii V+1may be defined with respect to each MLU as follows: vt=Pi / pNfor i = l.. N - 1, where comprises the summation of the mass at the i’th layer.

[0054] Similar features may be specified where the MLU is defined over a larger grid. In some examples, the feature vector comprises an additional feature specifying the energy density at the center cell 420:

[0055] In this equation, P comprises the energy source power, £ is ratio of the power absorbed by the material, V is the scan speed, dhatchis the hatch distance, and Liayeris the layer thickness. In some cases, the feature vector may also comprise a material identifier.

[0056] Figure 5 shows a flow diagram of a method 500 for training a predictive model for predicting a melting time of build material in a deposited layer of build material in an additive manufacturing system. Herein, the term melting time refers to a maximum period of time the build material for the cell exceeds a melting point of the build material. The method 500 may be used in conjunction with the other methods and systems described herein, and in particular, the MLU 400 and associated feature vectors previously described. According to examples described herein, the predictive model may comprise a neural network model.

[0057] At block 510, the method 500 comprises accessing a set of model parameters for the predictive model. According to examples, the set of model parameters may comprise weights for a neural network. At block 520, the method comprises accessing object data for an object and print data for printing the object in the additive manufacturing system. The print data comprises a representation of the object as a plurality of cross-sectional layers and scan path data for each of the plurality of cross- sectional layers. The scan path data specifies a scan path for a heat source in the additive manufacturing system to selectively heat a layer of deposited build material to cause the build material to melt and coalesce to form a cross-sectional layer.

[0058] At block 530, the method comprises accessing training data for the object where the training data comprising a simulation of the melting time in each of the plurality of cross-sectional layers. According to examples described herein, the simulation may be a Finite Element Analysis (FEA) simulation in each of the layers. The training data may be obtained by evaluating the simulation in each of the plurality of cross- sectional layers.

[0059] At block 540, the method comprises sub-dividing the object into a plurality of cells, based on the object data. In examples, sub-dividing the object into a plurality of cells, comprises sub-dividing the object into a plurality of MLUs similar to the MLU 400 previously described.

[0060] At block 550, a vector of features is generated for each cell, based on the print data. In examples, the feature vector may have vectors v = (Vi)lii V+1for each MLU, as previously described herein. At block 560, the method 500 comprises evaluating the predictive model to obtain a predicted melting time for each cell. Evaluating the predictive model may comprise evaluating a neural network model and observing an output comprising a predicted melting time. At block 570, the method comprises modifying the set of model parameters based on the training data and predicted melting time in each cell. According to examples, modifying the set of model parameters may comprise iteratively modifying model parameters via gradient descent to minimize a loss function of a neural network model.

[0061] Figure 6 shows examples 600 of objects that may be used to train the predictive model. The object 610 is an example of a rectangular block. Objects 620, 630, 640 comprise blocks with sloping sides having varying inclinations. In examples, the method 500 may be repeated for rectangular blocks similar to object 610, with rotations between 0 to 90 degrees, and objects similar to objects 620, 630, 640 with sloping sides of varying angles between 15 and 30 degrees. In addition, the method 500 may be repeated for objects with different materials and a range of energy source powers and speeds. According to examples, the energy source power may be varied from 0.5 to 1.5 x Pref, a power reference value. According to examples, the speed may be varied from 0.5 to 1.5 x Vre^, a speed reference value.

[0062] Figure 7 shows a flow diagram of a method 700 for predicting a melting time of build material in a deposited layer of build material in an additive manufacturing system. The method 700 may be used in conjunction with the other methods and systems described herein, and in particular, the MLU 400 and associated feature vectors and the training method 500 previously described. According to examples described herein, the predictive model may comprise a neural network model.

[0063] At block 710, the method 700 comprises accessing a representation of three- dimensional object as a plurality of cross-sectional layers for printing in an additive manufacturing system. At block 720, the method comprises accessing scan path data for a region of a selected one of the plurality of cross-sectional layers. The region may comprise the entire selected cross-sectional layer or a portion of the layer. The scan pathdata specifies a scan path for an energy source in the additive manufacturing system to selectively heat a layer of deposited build material, to cause the build material to melt and coalesce to form the selected cross-sectional layer.

[0064] At block 730, the method 700 comprises sub-dividing the region into a plurality of cells. According to examples, the region may be sub-divided according to the method previously described, by specifying a grid, with dimensions equal to the hatch distance of the energy source.

[0065] At block 740, the method 700 comprises, for each cell, generating a vector of features based on the representation of the object and scan path data. The vector of features may be the vector v previously defined. At block 750, the method comprises evaluating a trained predictive model to obtain an output comprising a predicted melting time for the cell. The trained predictive model may be the predictive model previously described in relation to the method 500 and trained based on objects and data as described in relation to Figure 6.

[0066] According to examples, based on the evaluation of the predictive model, the region may be determined to be an overheating region. In such a case, the scan path may be optimized, for example, by changing the power of the energy source, the scan speed, the delay time, or re-generating a portion of the scan path in the region.

[0067] In one example, the method 700 is repeated for each cross-sectional layer. However, in another example, a rapid approximate thermal analysis check may be performed for each layer before execution of the method 700, thus enabling the user to quickly identify those layers which may be problematic, from those which are not.

[0068] The methods and systems described herein provide a number of advantages. In particular, the methods described leverage the power of machine learning techniques and extensive simulation data to detect and mitigate the local building problems. This may be performed in parallel with the scan path generation process. Sub-optimal or problematic paths may be rectified in the path generation process. The methods described do not require expensive monitoring hardware for feedback control and reduce the number of test prints iterations to get to a good build process and high part quality. This saves significant time, material, and money and quality parts are printed as fast as possible without introducing longer print times.

[0069] Figure 8 illustrates an example of a data processing system in which an embodiment of the present disclosure may be implemented. The data processing system 800 comprises a processor 810 connected to a local system bus 820. The local system bus connects the processor to a main memory 830 and graphics display adaptor 840, which may be connected to a display 850. The data processing system may communicate with other systems via a wireless user interface adapter connected to the local system bus 820, or via a wired network, for example, to a local area network. Additional memory 860 may also be connected via the local system bus 820.

[0070] A suitable adaptor, such as wireless user interface adapter 870, for other peripheral devices, such as a keyboard 880 and mouse 890, or other pointing device, allows the user to provide input to the data processing system. Other peripheral devices may include one or more I / O controllers such as USB controllers, Bluetooth controllers, and / or dedicated audio controllers (connected to speakers and / or microphones). It should also be appreciated that various peripherals may be connected to the USB controller (via various USB ports) including input devices (e.g., keyboard, mouse, touch screen, trackball, camera, microphone, scanners), output devices (e.g., printers, speakers), or any other type of device that is operative to provide inputs or receive outputs from the data processing system.

[0071] Further, it should be appreciated that many devices referred to as input devices or output devices may both provide inputs and receive outputs of communications with the data processing system. Further it should be appreciated that other peripheral hardware connected to the I / O controllers may include any type of device, machine, or component that is configured to communicate with a data processing system.

[0072] An operating system included in the data processing system enables an output from the system to be displayed to the user on the display and the user to interact with the system. Examples of operating systems that may be used in a data processing system may include Microsoft Windows™, Linux™, UNIX™, iOS™, and Android™ operating systems.

[0073] In addition, it should be appreciated that data processing system 800 may be implemented as in a networked environment, distributed system environment, virtual machines in a virtual machine architecture, and / or cloud environment. For example, theprocessor and associated components may correspond to a virtual machine executing in a virtual machine environment of one or more servers. Examples of virtual machine architectures include VMware ESCi, Microsoft Hyper-V, Xen, and KVM.

[0074] Those of ordinary skill in the art will appreciate that the hardware depicted for the data processing system 800 may vary for particular implementations. For example, the data processing system 800 in this example may correspond to a computer, workstation, and / or a server. However, it should be appreciated that alternative embodiments of a data processing system may be configured with corresponding or alternative components such as in the form of a mobile phone, tablet, controller board or any other system that is operative to process data and carry out functionality and features described herein associated with the operation of a data processing system, computer, processor, and / or a controller discussed herein. The depicted example is provided for the purpose of explanation only and is not meant to imply architectural limitations with respect to the present disclosure.

[0075] The data processing system 800 may be connected to the network (not a part of data processing system 800), which can be any public or private data processing system network or combination of networks, as known to those of skill in the art, including the Internet. The data processing system 800 can communicate over the network with one or more other data processing systems such as a server (also not part of the data processing system 800). However, an alternative data processing system may correspond to a plurality of data processing systems implemented as part of a distributed system in which processors associated with several data processing systems may be in communication by way of one or more network connections and may collectively perform tasks described as being performed by a single data processing system. Thus, it is to be understood that when referring to a data processing system, such a system may be implemented across several data processing systems organized in a distributed system in communication with each other via a network.

[0076] The data processing system 800 is adapted to carry out the methods in accordance with the embodiments described herein. For example, the keyboard 880 and mouse 890 may function as a user input device for receiving information from the user, the processor 810 may be adapted to carry out the steps of the method and the display 850adapted to display a particular view to the user. A computer product comprising instructions that, when run on a computer such as the data processing system 800, may be provided to cause the computer to execute the steps of the methods of the embodiments outlined above.

[0077] The present disclosure is described with reference to flow charts and / or block diagrams of the method, devices, and systems according to examples of the present disclosure. Although the flow diagrams described above show a specific order of execution, the order of execution may differ from that which is depicted. Blocks described in relation to one flow chart may be combined with those of another flow chart. In some examples, some blocks of the flow diagrams may not be necessary and / or additional blocks may be added.

[0078] The present disclosure can be embodied in other specific apparatus and / or methods. The described embodiments are to be considered in all respects as illustrative and not restrictive. In particular, the scope of the disclosure is indicated by the appended claims rather than by the description and figures herein. All changes that come within the meaning and range of equivalency of the claims are to be embraced within their scope.

Claims

CLAIMS1. A computer-implemented method for predicting overheating in layers of build material in an additive manufacturing system, the method comprising: accessing a representation of a three-dimensional object as a plurality of cross- sectional layers for printing in the additive manufacturing system; accessing scan path data for a region of a selected cross-sectional layer of the plurality of cross-sectional layers, the scan path data specifying a scan path for an energy source in the additive manufacturing system to selectively heat a layer of deposited build material, to cause the build material to melt and coalesce to form the selected cross- sectional layer; sub-dividing the region into a plurality of cells; generating, for each cell of the plurality of cells, a vector of features based on the representation of the three-dimensional object and the scan path data; and evaluating, for each cell of the plurality of cells, a trained predictive model based the vector of features to obtain an output comprising a predicted melting time for the respective cell.

2. The method of claim 1, wherein the predicted melting time comprises a maximum period of time the build material for the respective cell exceeds a melting point of the build material.

3. The method of claim 1, further comprising: comparing the predicted melting time to a threshold value; and identifying the region as an overheating region when the predicted melting time exceeds the threshold value.

4. The method of claim 3, further comprising: modifying the scan path in response to identifying the region as the overheating region.

5. The method of claim 4, wherein the modifying of the scan path comprises modifying an energy source power, scan speed, delay time, or re-generating portions of the scan path.

6. The method of claim 1, wherein the trained predictive model comprises a neural network model.

7. The method of claim 1, wherein the vector of features comprises an energy density value for the respective cell.

8. The method of claim 7, wherein the vector of features comprises a value determined on a basis of energy source timing data for neighboring cells of the plurality of cells in the cross-sectional layer.

9. The method of claim 1, wherein the selected cross-sectional layer is selected on a basis of an output of a thermal analysis test applied to each cross-sectional layer of the plurality of cross-sectional layers.

10. A computer-implemented method for training a predictive model for predicting a melting time of build material in a deposited layer of the build material in an additive manufacturing system, the method comprising: accessing a set of model parameters for the predictive model; accessing object data for an object and print data for printing the object in the additive manufacturing system, wherein the print data comprises a representation of the object as a plurality of cross-sectional layers and scan path data for each cross-sectional layer of the plurality of cross-sectional layers, wherein the scan path data specifies a scan path for an energy source in the additive manufacturing system to selectively heat a layer of deposited build material to cause the build material to melt and coalesce to form a cross-sectional layer of the plurality of cross-sectional layers;accessing training data for the object, wherein the training data comprises a simulation of the melting time in each cross-sectional layer of the plurality of cross- sectional layers; sub-dividing the object into a plurality of cells; generating a vector of features for each cell of the plurality of cells, based on the print data; evaluating the predictive model to obtain a predicted melting time for each cell; and modifying the set of model parameters based on the training data and predicted melting time in each cell.

11. The method of claim 10, further comprising: repeating steps of claim 10 for a plurality of objects and energy source parameters.

12. The method of claim 10, wherein the predictive model is a neural network model, wherein the model parameters comprise weights, and wherein the simulation is a Finite Element Analysis (FEA) simulation.

13. A data processing system comprising: a processor and a memory that stores instructions that, when executed by the processor, cause the processor to execute the method according to any one of claims 1 to 12.

14. A non-transitory computer readable storage comprising program code that, when executed by a processor, provides instructions to perform the method according to any one of claims 1 to 12.

Images