Heat-aware toolpath reordering for 3D printing of physical parts

Abstract

A computing system (100) can include an access engine (108) and a tool path reordering engine (110). The access engine (108) can be configured to access an original layer tool path of a slice of a 3D CAD object and a thermal criticality metric (240, 340, 540) of the original layer tool path. The thermal criticality metric can indicate a thermal impact for different points on a plurality of tool path segments of the original layer tool path for 3D printing of a physical part using the original layer tool path. The tool path reordering engine (110) can be configured to reorder the plurality of tool path segments into a modified layer tool path (250, 350, 550), and the modified layer tool path (250, 350, 550) can have a thermal criticality metric that is less thermally impactful to the physical part than the thermal criticality metric (240, 340, 540) of the original layer tool path.

Description

Background Technology

[0001] Computer systems can be used to create, use, and manage data for products and other items. For example, Computer-Aided Technology (CAx) systems can be used to assist in the design, analysis, simulation, or manufacturing of products. Examples of CAx systems include Computer-Aided Design (CAD) systems, Computer-Aided Engineering (CAE) systems, visualization and computer-aided manufacturing (CAD) systems, Product Data Management (PDM) systems, and Product Lifecycle Management (PLM) systems. These CAx systems may include components (e.g., CAx applications) that facilitate the design and simulation testing of product structure and manufacturing. Attached Figure Description

[0002] Some examples are described in the following detailed description and with reference to the accompanying drawings.

[0003] Figure 1 An example of a computational system is shown that supports the reordering of thermal sensing toolpaths for 3D printing of physical parts.

[0004] Figure 2 An example of heat sensing tool path reordering is shown, from the original layer tool path to the modified layer tool path.

[0005] Figure 3 An example of heat-sensing toolpath reordering performed by the toolpath reordering engine using a path optimization algorithm is shown.

[0006] Figure 4 An example training of a machine learning model is shown to support path reordering for heat-sensing tools used in 3D printing of physical parts.

[0007] Figure 5 This demonstrates an example application of a machine learning model supporting the reordering of heat-sensing toolpaths for 3D printing of physical parts.

[0008] Figure 6 An example of the logic that the system can implement to support the reordering of heat-sensing toolpaths for 3D printing of physical parts is shown.

[0009] Figure 7An example of a computational system is shown that supports the reordering of thermal sensing toolpaths for 3D printing of physical parts. Detailed Implementation

[0010] Additive manufacturing (sometimes referred to as 3D or 3D printing) can be performed via a 3D printer capable of constructing objects layer by layer. Examples of additive manufacturing include multi-axis 3D printing and laser bed fusion processes. In multi-axis 3D printing, the 3D printer can adjust (e.g., tilt) the axis along which the 3D construction is performed via material deposition. In laser bed fusion, a laser can be used as a power source to sinter / melt powdered material (e.g., metal powder) laid on a powder bed or build platform. 3D printing can involve the incremental, continuous formation of material using 3D printing tools (e.g., a material deposition head or energy beam for incrementally building 3D parts in an ordered manner). As used herein, toolpath can refer to any path, route, or route used by a 3D printer to construct any part of a 3D part via additive manufacturing, whether as a path for the continuous deposition of material for material deposition 3D printing techniques, or as a path guiding a laser (or other energy emission) for energy application via LPBF-type 3D printing techniques, and so on.

[0011] One challenge facing modern 3D printing systems is managing the heat generated during the 3D printing process. For example, multi-axis 3D printing technologies may require heating the 3D printing material sufficiently into a malleable form (e.g., metal beads), and this heat can be amplified when using metals or other substrates that can accumulate, retain, and emit heat. Applying energy via an LBPF laser to sinter metal powder also utilizes heat and injects it into the 3D printing environment as part of the process. Excessive heat can adversely affect 3D part construction, for example, due to part warping caused by heat hotspots, inaccurate part construction, part failure, insufficient material quality, and print job interruptions due to damage to the recoating system. Many current toolpath generation algorithms for 3D printing are optimized for printing speed without considering heat generation, and may therefore suffer from increased part deformation, reduced print yield, or decreased printing efficiency. Simple solutions that pause the 3D printing process during part construction may attempt to address heat-related part problems, but at the cost of increased 3D part construction time (also known as build time) and reduced efficiency.

[0012] The disclosure herein provides systems, methods, apparatus, and logic for heat-aware reordering of toolpaths for 3D printing of physical parts. As described in more detail herein, various heat-aware toolpath reordering features can support the reordering of 3D printing toolpaths to reduce the impact of heat-based issues in 3D parts. As described herein, toolpaths can be reordered and optimized to account for thermal criticality metrics and reduce the thermal impact on the 3D part construction. Reordering can be performed using path optimization algorithms or machine learning models, point-by-point or segment-by-segment, leveraging input thermal analysis and thermal criticality metrics. Therefore, the heat-aware toolpath reordering features described herein can reduce the thermal impact of conventionally generated toolpaths while also reducing build time to optimize 3D part construction.

[0013] These and other path reordering features and technological benefits of heat sensing tools are described in more detail in this paper.

[0014] Figure 1 An example of a computational system 100 that supports thermal sensing toolpath reordering for 3D printing of physical parts is illustrated. The computational system 100 may take the form of a single or multiple computing devices (e.g., application servers, compute nodes, desktop or laptop computers, smartphones or other mobile devices, tablets, embedded controllers, etc.). In some embodiments, the computational system 100 implements CAx tools, applications, or programs to assist users in designing, analyzing, simulating, or 3D manufacturing products, including thermal sensing toolpath reordering.

[0015] This serves as an example implementation of any combination of the path reordering features used in the heat sensing tool described herein. Figure 1 The computing system 100 shown includes an access engine 108 and a tool path reordering engine 110. The computing system 100 can implement engines 108 and 110 (including their components) in various ways, such as as hardware and programming. Programming for engines 108 and 110 can take the form of processor-executable instructions stored on a non-transient machine-readable storage medium, and the hardware for engines 108 and 110 can include processors that execute these instructions. The processor can take the form of a single-processor or multi-processor system, and in some examples, the computing system 100 uses the same computing system features or hardware components (e.g., a common processor or a common storage medium) to implement multiple engines.

[0016] In operation, access engine 108 can access the original layer toolpaths of slices of a 3D CAD object. As used herein, CAD objects (including 3D CAD objects) can include any type of CAx object data related to part design, simulation, analysis, or manufacturing. CAD objects can therefore include 3D object designs, models, model slices, toolpaths, etc. 3D CAD objects can represent physical parts, slices can represent physical layers for 3D printing of physical parts, and the original layer toolpaths can control the 3D printing of physical layers and include the 3D printing order of multiple toolpath segments of the original layer toolpaths. The original toolpath can refer to any 3D printing toolpath accessed before applying or being used as input to the heat-sensing toolpath reordering feature described herein. In operation, access engine 108 can also access thermal criticality metrics of the original layer toolpaths. Thermal criticality metrics can indicate the thermal impact at different points on multiple toolpath segments of the original layer toolpaths for 3D printing of physical parts using the original layer toolpaths.

[0017] In operation, the toolpath reordering engine 110 can reorder multiple toolpath segments of the original layer toolpath into a modified layer toolpath. The modified layer toolpath can have a different 3D printing order than the original layer toolpath and a thermal criticality metric with a smaller thermal impact on the physical part compared to the original layer toolpath's thermal criticality metric. The toolpath reordering engine 110 can also provide modified layer toolpaths to support the 3D printing of the physical part.

[0018] The following describes these and other path reordering features for heat sensing tools in more detail.

[0019] Figure 2 An example of heat sensing tool path reordering is shown, from the original layer tool path to the modified layer tool path. Figure 2 The example in this document is illustrated via a computing system that implements access engine 108 and tool path reordering engine 110. However, various other implementations are contemplated herein.

[0020] Access engine 108 can access any CAx data related to the generation, reordering, analysis, or processing of 3D printing toolpaths. The heat-sensing toolpath reordering feature described herein generates heat-sensing toolpaths. As used herein, a heat-sensing toolpath can refer to any toolpath that takes into account the thermal effects in the 3D printing of a physical part, including modified toolpaths reordered from conventionally generated non-heat-sensing or any other input or original toolpath as described herein. In some implementations, heat-sensing toolpath reordering is performed on a per-layer basis. In such an example, access engine 108 can access any number of slices of the 3D CAD object to support heat-sensing toolpath reordering.

[0021] Access engine 108 can access layer toolpaths for various slices of 3D CAD object 210, and slices can be generated via slice plane 220, which intersects CAD object 210 along any build axis supported for 3D printing of the physical part represented by 3D CAD object 210. Figure 2 In this implementation, access engine 108 accesses the original layer toolpath 230, which may represent the toolpath for the physical fabrication of a given slice of 3D CAD object 210. In some implementations, access engine 108 itself may perform an intersection operation on 3D CAD object 210 to obtain part slices and generate the original layer toolpaths for the part slices. In other examples, access engine 108 may obtain the original layer toolpaths for slices of 3D CAD object 210 generated by other path logic (e.g., conventional layer toolpath generation tools) in other ways.

[0022] The original layer toolpath can refer to any initial, predetermined, or non-thermally-aware toolpath generated for slicing a 3D CAD object. Therefore, the original layer toolpath can take the form of any conventionally generated toolpath that does not consider the heat in its path (referred to herein as a non-thermally-aware toolpath). Examples of conventionally generated toolpaths include toolpaths optimized for 3D printing speed, such as continuous line scan material deposition paths, conventional laser trailing, or path vectors generated by conventional 3D printing systems. Conventionally generated toolpaths can also include topology-dependent local tool parameters (e.g., laser power, write speed) suitable for topology-dependent requirements (e.g., overhang strategies, etc.).

[0023] As described herein, a toolpath (including layer toolpaths) can include toolpath segments. A toolpath segment can be any sub-segment of a toolpath and can be in the form of a hatch vector (and thus have a start and an end point). At this point, a combination of toolpath segments of a given toolpath can form an ordered path through which a 3D printer travels to print a portion of a 3D part (e.g., the physical layers of a 3D part).

[0024] Access engine 108 can support the analysis of any type of toolpath, including original and modified layer toolpaths as described herein. Specifically, access engine 108 can access the thermal criticality metric 240 of the original layer toolpath 230. The thermal criticality metric can include any analysis, measure, data, statistics, or other evaluation indicating the thermal impact of a given toolpath on the 3D printing of a physical part. In some embodiments, access engine 108 can calculate the thermal criticality metric of the layer toolpath itself. In other embodiments, access engine 108 can access the calculated thermal criticality metric based on separate analysis logic configured to process the layer toolpath and generate the thermal criticality metric.

[0025] Various forms of thermal criticality measurement are envisioned herein, including any thermal analysis, process, implementation method or feature described in International Patent Application No. PCT / EP2019 / 085918, filed November 18, 2019, by inventors Daniel Reznik, Frank Heinrichsdorff, Darya Kastsian and Katharina Eissing, which is incorporated herein by reference in its entirety.

[0026] For example, the accessed thermal criticality metric can specify the thermal effect for a specific toolpath point-by-point, and can depict a point as any division, depiction, or other sub-part of a specific toolpath or a toolpath segment (e.g., a vector) that includes that specific toolpath. Taking a layer toolpath in the form of a ray vector generated for a slice during laser bed powder melting (LBPF) 3D printing as an example, the thermal criticality metric can specify the point-by-point thermal criticality of multiple points on each ray vector. The thermal criticality metric can utilize an integral that calculates the mass below and near a given point (e.g., an objectively defined surrounding area) and the amount of energy applied to the given point, as well as the 3D history of a predetermined number of points before and / or after the given point in the ordered ray vector.

[0027] In some examples, the thermal criticality metric generated by or otherwise accessed by access engine 108 can indicate a gradient exceeding the critical temperature for 3D printing of the physical part at a specific point, multiple points in a neighborhood of a given objective measurement, a segment of the toolpath, another sub-section of a layer toolpath, or the entire layer toolpath. This critical temperature can be determined experimentally for a given build environment and can vary depending on the 3D part material (e.g., deposited beads, metal powder, or other construction materials), 3D printing process parameters, build environment temperature, 3D printer characteristics, etc. The critical temperature can refer to the threshold temperature at which the 3D part will be burned, oxidized, deformed, or otherwise adversely affected to alter the quality or design of the 3D printed part. For example, the critical temperature can take the form of a threshold temperature at the melt pool when producing dimensions exceeding the critical size of LBPF type or other 3D printing technologies. Access engine 108 or other entities can configure critical temperature values ​​when determining the thermal criticality metric of the toolpath.

[0028] In some examples, thermal criticality measures can be taken as gradient values ​​relative to the critical temperature. For example, Figure 2 The thermal criticality metric 240 can (e.g., point-by-point) indicate the degree / amount by which the 3D part temperature at each of the multiple toolpath segments of the original layer toolpath 230 exceeds (or falls below) the critical temperature for 3D printing of the physical part. In calculating the thermal criticality metric, the access engine 108 (or other analysis logic) can utilize a combination of finite element analysis (FEA), machine learning (ML), or other processing techniques to calculate the build temperature at each point in a given toolpath and determine the degree of difference between the calculated build temperature and the critical temperature for 3D printing. The access engine 108 can do this, for example, in any manner described in International Patent Application No. PCT / EP2019 / 085918. The gradient (e.g., the amount / amount of the temperature difference) can be objectively measured as the difference between the calculated build temperature at each point in a given toolpath and one or more corresponding critical temperatures.

[0029] Performing FEA analysis along the entire toolpath of a 3D-printed part can be impractical or infeasible. In some cases, the entire 3D part toolpath can be tens or hundreds of miles long, and FEA calculations can take days, weeks, or even years, depending on the capabilities and availability of the available computing resources. Access Engine 108 can leverage machine learning models to improve the speed and efficiency of thermal criticality metric calculations, for example, by training the machine learning model using detailed FEA input data from representative (e.g., small or selected) portions of the layer toolpath. These representative portions can be provided for training against consistent 3D printing parameters and can include representative FEA data from ray vectors or other toolpath segments of varying lengths and locations sampled from the entire 3D part toolpath. In some embodiments, hypothetical representative volume elements with reordered vectors can also be used as training data. Through such training, the machine learning model utilized by Access Engine 108 can support the calculation and determination of thermal criticality metrics with increased speed and efficiency.

[0030] Therefore, the accessed thermal criticality metric can identify points in the toolpath (and the corresponding parts of the manufactured 3D part) that may exceed the critical temperature and thus cause part deformation and / or other problems caused by localized overheating.

[0031] To address potential overheating of 3D parts for a given layer toolpath, the toolpath reordering engine 110 can generate modified layer toolpaths with reduced thermal impact compared to the given layer toolpath (which can therefore result in improved part quality). Figure 2 In the example shown, the toolpath reordering engine 110 can reorder the 3D printing order of toolpath segments in the original layer toolpath 230 to obtain a modified layer toolpath 250. Compared to the original layer toolpath 230, the modified layer toolpath can have a reduced thermal impact, and the reduced thermal impact can be determined by the toolpath reordering engine 110 through a comparison between the thermal criticality metric of the modified layer toolpath 250 and the thermal criticality metric 240 of the original layer toolpath 230.

[0032] To construct the modified layer toolpath 250, the toolpath reordering engine 110 can apply any number of optimization techniques to reorder toolpath segments to satisfy any number of predetermined objectives, such as those defined by a configured objective function. Example objectives that the toolpath reordering engine 110 may consider include minimizing toolpath build time (which can be measured as a function of the length (also called distance) of a given toolpath as a layer toolpath, including the effective printing distance of a toolpath segment and the travel distance and delay / pause of the 3D printer between consecutive toolpath segments), a threshold reduction of the thermal criticality metric, or an upper limit on the thermal criticality metric. The constraints and objectives applied by the toolpath reordering engine 110 in determining the modified layer toolpath 250 can be interdependent, for example, to optimize the 3D printing order of the toolpath segments of the original layer toolpath 230 such that the build time of the modified layer toolpath 250 is minimized under the constraint that the thermal criticality metric of the modified layer toolpath 250 is reduced to at least below the threshold thermal criticality metric.

[0033] In some implementations, the toolpath reordering engine 110 can reorder multiple toolpath segments of the original layer toolpath 230 into a modified layer toolpath 250 by iteratively optimizing candidate layer toolpaths using a path optimization algorithm. This path optimization algorithm is applied to optimize thermal criticality metrics and the overall build time of the candidate layer toolpaths to determine the modified layer toolpath 250. The modified layer toolpath 250 may have a smaller thermal impact on the physical part than the original layer toolpath 230, and the toolpath reordering engine 110 can objectively compare the thermal impact between layer toolpaths in various ways.

[0034] As an illustrative example, the toolpath reordering engine 110 can numerically compare the total number of points in the original layer toolpath 230 and the modified layer toolpath 250 that exceed the critical temperature for 3D printing of the physical part represented by the 3D CAD object 210. As another example, the toolpath reordering engine 110 can compare the maximum point-by-point gradient exceeding the critical temperature in the original layer toolpath 230 and the modified layer toolpath 250, and accordingly determine the reduced heat metric of the modified layer toolpath 250.

[0035] As another example, the tool path reordering engine 110 can measure a thermal criticality metric (e.g., a vector-based thermal criticality metric), which may be referred to herein as a segment thermal criticality metric. The tool path reordering engine 110 can determine the segment thermal criticality metric by, for example, averaging point-by-point criticalities (e.g., per-shadow vectors) on the tool path segment, defining an intervention threshold (e.g., where high criticality is measured by point-by-point construction temperatures exceeding a critical temperature at least a criticality threshold (e.g., exceeding 10°C)) for a given tool path segment, defining the criticality of the tool path segment by averaging adjacent sample points, etc. The tool path reordering engine 110 can use this per-tool path segment thermal criticality metric as yet another metric to compare the thermal impact between the original layer tool path 230 and the modified layer tool path 250, or to optimize the thermal criticality metric of the modified layer tool path 250 via segment reordering.

[0036] The toolpath reordering engine 110 can provide modified layer toolpaths to support 3D printing of physical parts represented by 3D CAD objects. For example, the toolpath reordering engine 110 can transmit modified layer toolpaths 250 as control data to the 3D printer, causing deposition tools, lasers, or other energy sources, or other 3D printing instruments to traverse the modified layer toolpaths 250 to physically fabricate physical layers of slices of the 3D part represented by the 3D CAD object 210. In some embodiments, the toolpath reordering engine 110 is implemented locally as part of the 3D printer itself, so heat-sensing toolpath reordering can occur on the same physical machine or computing environment as the 3D printing of the physical part. In other embodiments, the toolpath reordering engine 110 can be implemented remotely to the 3D printer (e.g., via a remote CAD system or in a cloud computing environment), and the modified layer toolpaths 250 can be transmitted to the 3D printer across a communications network.

[0037] Therefore, original toolpaths (e.g., conventionally generated toolpaths) can be reordered into modified layer toolpaths via the various heat-sensing toolpath reordering features described herein. Physical parts constructed using modified layer toolpaths (compared to the original layer toolpaths) can have reduced thermal effects and improved part quality. Thus, the heat-sensing toolpath reordering features described herein can improve the quality of 3D part constructions using 3D printing, and can do so while reducing, minimizing, or otherwise optimizing build time, to control, balance, and consider build efficiency and part quality. (See below for further details.) Figures 3 to 5 This describes some examples of reordering techniques that can be applied to the Tool Path Reordering Engine 110.

[0038] Figure 3 An example of heat-sensing toolpath reordering performed by the toolpath reordering engine 110 using a path optimization algorithm is shown. Figure 3 In the example shown, the tool path reordering engine 110 implements a path optimization algorithm 310, which can be any optimization algorithm that the tool path reordering engine 110 can apply to reorder tool path segments of the input tool path. For example, the tool path reordering engine 110 can reorder the original layer tool path 330 and the original layer tool path 330's thermal criticality metric 340 (each of which can be respectively compared with...) Figure 2 The original layer tool path 230 and thermal criticality metric 240 described herein are used as input. The tool path reordering engine 110 can apply the path optimization algorithm 310 to the original layer tool path 330 and thermal criticality metric 340, and determine a modified layer tool path 350 with reduced thermal impact compared to the original layer tool path 330.

[0039] The application of path optimization algorithm 310 can be iterative, generating and evaluating candidate layer toolpaths until a set of algorithmic constraints are satisfied. As described herein, as an example, algorithmic constraints can be specified in terms of minimizing the build time affected by the threshold hot criticality metric. For a given iteration in the application of path optimization algorithm 310, toolpath reordering engine 110 can access the output layer toolpaths determined from previous iterations of path optimization algorithm 310 as candidate layer toolpaths. For the initial iteration of path optimization algorithm 310, toolpath reordering engine 110 can identify the original layer toolpath 330 as the first candidate layer toolpath.

[0040] Each candidate layer toolpath can include a specific 3D printing order of multiple toolpath segments (e.g., the same toolpath segments of the original layer toolpath 330, but in a different order) and a corresponding thermal criticality metric. The thermal criticality metric of each candidate layer toolpath can effectively provide an evaluation of the candidate layer toolpath, and the toolpath reordering engine 110 can iterate over various candidate layer toolpaths using the path optimization algorithm 310 to optimize the layer toolpaths to satisfy algorithmic constraints and determine the optimal toolpath as the modified layer toolpath output. After converging to the optimal toolpath (e.g., evaluated via algorithmic constraints such as minimum build time and threshold thermal criticality metric), the toolpath reordering engine 110 can identify and generate the modified layer toolpath 350. In some cases where the number of toolpath segments in the original layer segment is relatively large (e.g., above a threshold number, such as 150, 2000, or any other configurable threshold), the toolpath reordering engine 110 can divide the layer into a set of regions / sublayers (e.g., assigning the toolpath segments of that layer to different regions / sublayers). In this case, the tool path reordering engine 110 can determine the modified tool path for each region / sublayer and generate the modified layer tool path 350 as a combination of the modified tool paths determined for each region / sublayer of a given layer.

[0041] In some examples, the tool path reordering engine 110 may apply an ant colony optimization algorithm as the path optimization algorithm 310, in which the pheromone parameter of the ant colony optimization algorithm is specified as the reciprocal of the segment thermal criticality metric calculated for the tool path segment of the candidate layer tool path. When applying the ant colony optimization algorithm (or any other path optimization algorithm 310), the tool path reordering engine 110 may evaluate tool path segments individually based on the segment thermal criticality metric, which can refer to any thermal criticality metric specified on a per-tool path segment basis. Example forms of segment tool path thermal criticality metrics include averaging point-by-point criticality (e.g., per-shadow vector) on the tool path segment, defining an intervention threshold by multiple high criticality points in a given tool path segment having a construction temperature exceeding a critical temperature critical gradient threshold (e.g., exceeding 10°C), defining the criticality of a tool path segment by averaging the criticality metrics of adjacent sample points, and so on.

[0042] Typically, the tool path reordering engine 110 can apply an optimization process that iterates across candidate tool paths via a random distribution and measures changes in pheromone parameter values ​​to evaluate the optimal path. Ant colony optimization can be designed to identify the “shortest” path in a weighted graph via pheromone distribution, and in this case, the “shortest” path criterion can be specified based on build time and / or segment hot criticality metric. Specifically, the tool path reordering engine 110 can measure pheromone trails by the reciprocal of a segment-by-segment (e.g., vector-by-vector) hot criticality metric. When testing different candidate layer tool paths, “ants” advance through possible paths (in this case, tool path segments) as represented via a weighted graph, and deposit pheromone and weighted edges accordingly in doing so.

[0043] Using these information pheromone parameter configurations, the tool path reordering engine 110 can further configure another heuristic as the probability that the k-th ant follows a given tool path segment or a set of tool path segments from point i to point j, as follows:

[0044]

[0045] In this example, the tool path reordering engine 110 can configure the parameter τ as a pheromone parameter, which in this example is the reciprocal of the segment criticality metric (and thus the thermal metric criticality with a reduced thermal effect / gradient compared to the critical temperature acts as a positive indicator, and shifts the ant colony path traversal toward the smaller segment thermal criticality metric).

[0046] The tool path reordering engine 110 can configure the parameter η as a heuristic to be solved / optimized by the path optimization algorithm 310 (ant colony optimization algorithm in this example). In a specific example, the tool path reordering engine 110 can be set... The tool path reordering engine 110 can define such a travel time term as the non-3D printing time required to traverse from the end of a given tool path segment to the start of a subsequent tool path segment within a traversed path (in fact, measuring the non-3D printing traversal time of the 3D printer between the end and start of consecutive tool path segments). Therefore, the tool path reordering engine 110 can optimize the total exposure time of the layer tool path to T = ∑ i t i (where t) i As the travel time after the i-th tool path segment in a given tool path 3D printing sequence, it can depend on the distance between subsequent tool path segments. As described herein, the tool path reordering engine 110 can optimize the build time T by utilizing the thermal criticality metric of layer tool paths (or included tool path segments) to satisfy the algorithmic constraints on the threshold thermal criticality metric.

[0047] To further explain, the tool path reordering engine 110 can set exponents α and β as weights for the pheromone parameter values ​​and construct a temporal heuristic about the ant traversal probability. For the initial iteration, the tool path reordering engine 110 can use the thermal criticality metric 340 of the original layer tool path 330 to calculate the segment thermal criticality metric, and then calculate the pheromone parameter values ​​for each tool path segment (as the reciprocal of the calculated or accessed segment thermal criticality metric), which may include some or all possible transition pheromone parameter values ​​from one tool path segment to another from the tool path segments of the original layer tool path 330.

[0048] Based on the calculated pheromone parameter values ​​and the heuristic value η, the tool path reordering engine 110 can determine the probability of all permissible transitions between tool path segments. In doing so, the tool path reordering engine 110 can limit (e.g., exclude or otherwise constrain) the transitions between tool path segments using any physical, technical, or process-specific boundary conditions (e.g., part boundaries) applied to the 3D printing process. Thus, the tool path reordering engine 110 can iterate across a weighted graph of an ant colony optimization algorithm and accordingly determine the optimal 3D printing order of the tool path segments. Under constraints limiting the thermal criticality metrics of the tool path segments (e.g., all segment thermal criticality metrics are below a threshold, the sum of segment thermal criticality metrics is below a threshold, or any other configurable algorithmic constraints), the determined optimal 3D printing order minimizes the build time. The tool path reordering engine 110 can then recalculate the criticality of the determined candidate layer tool paths (and corresponding 3D printing orders), which allows the tool path reordering engine 110 to update the pheromone parameter values ​​until the thermal criticality metrics are reduced as much as possible, thereby limiting the build time.

[0049] In the manner described herein, the tool path reordering engine 110 can iteratively optimize candidate layer tool paths using a path optimization algorithm 310, and thus reorder the original layer tool paths 330 according to specified algorithmic constraints (e.g., build time and thermal criticality metrics). The output of the path optimization algorithm 310 may include a reordered set of tool path segments optimized for algorithmic constraints, which can thus form a modified layer tool path 350.

[0050] In some implementations (e.g., Figure 3As illustrated in the example, the tool path reordering engine 110 can apply path optimization algorithm 310 individually to each of the original layer tool paths (and corresponding thermal criticality metrics) received as input. In other embodiments, instead of applying optimization algorithms individually to each of the individual original layer tool paths for reordering, the tool path reordering engine 110 can alternatively train a machine learning model to control the reordering of tool path segments used for the received original layer tool paths. (See below for further details.) Figure 4 and Figure 5 This describes the training and application of an example machine learning model for path reordering in a heat sensing tool.

[0051] Figure 4 An example training of a machine learning model is shown to support heat-sensing toolpath reordering for 3D printing of physical parts. Figure 4 In this context, the toolpath reordering engine 110 may implement or otherwise access a machine learning model 410, which can be trained based on any combination of machine learning techniques, processes, or algorithms. The machine learning model 410 may implement or provide any number of machine learning techniques to analyze and interpret toolpaths used for 3D printing. For example, the machine learning model 410 may implement any number of supervised, semi-supervised, unsupervised, or reinforcement learning models to interpret the original layer toolpaths and / or corresponding thermal criticality metrics. The machine learning model 410 may include Markov chains, context trees, support vector machines, neural networks, Bayesian networks, or various other machine learning components. In some cases, the machine learning model 410 may apply any number of reinforcement learning techniques used for model training to support the thermal sensing toolpath reordering features described herein.

[0052] The toolpath reordering engine 110 can train a machine learning model 410 by configuring a parametric reordering algorithm 420 that considers multiple parameters in the 3D printing of physical parts. The parametric reordering algorithm 420 can take the form of any function, technique, thermal sensing criterion, or other logic that can be applied to reorder toolpath segments of the input layer toolpath (e.g., the original layer toolpath conventionally generated from slices of a 3D CAD object). In some implementations, the parametric reordering algorithm 420 may also take a thermal criticality metric of the input layer toolpath as input. Examples of parametric reordering algorithms include logic that specifies reordering criteria for toolpath segments, such as rule-based toolpath interleaving schemes, distance-based toolpath segment reordering schemes, random distributions, etc., where the rule-based toolpath interleaving scheme reorders the input layer toolpath to alternate between every three (3) toolpath segments in the original 3D printing order.

[0053] The toolpath reordering engine 110 can be configured to include any number of parameters in the parametric reordering algorithm 420, which can be used to control the reordering of toolpath segments in the input layer toolpath. Some examples of parameterizable properties of the parametric reordering algorithm 420 include toolpath segment lengths (e.g., shadow vector lengths), segment thermal criticality metrics, point-by-point thermal criticality metrics, regions of interest or hot spots in the layer or layer toolpath (e.g., identified via thermal criticality metrics of the input toolpath), 3D printing process parameters (e.g., deposition tool, LBPF parameters), 3D printing material properties, any geometric and boundary constraints limiting the physical printing or ordering of toolpath segments, restrictions on jump or interlacing directions, etc. Any aspect of the 3D printing process of a physically measurable part can be parameterized by the toolpath reordering engine 110 as part of the parametric reordering algorithm 420.

[0054] To train the machine learning model 410 based on the parameterized reordering algorithm 420, the tool path reordering engine 110 can provide experimental parameter values ​​of multiple parameters of the parameterized reordering algorithm 420 as training input. The tool path reordering engine 110 can randomly determine the experimental parameters based on the actual previous 3D printing configuration, as a predetermined set of parameter values ​​configured through user input, or in various other ways. As training input, the tool path reordering engine 110 can also receive an initial layer tool path having a corresponding thermal criticality metric calculated for that initial layer tool path. Figure 4 In the example, the tool path reordering engine 110 provides an initial layer tool path 431, a thermal criticality metric 432 for the initial layer tool path 431, and experimental parameter values ​​433 as inputs for training the machine learning model 410.

[0055] Through training, the tool path reordering engine 110 (e.g., via machine learning model 410) can iterate on candidate layer tool paths using corresponding hot criticality metrics to optimize the parameterized reordering algorithm 420 based on the hot criticality metrics and the overall construction time of the candidate layer tool paths. Optimization can occur by adjusting the parameter values ​​of the parameterized reordering algorithm 420 and evaluating the corresponding generated candidate layer tool paths using the hot criticality metrics of the candidate layer tool paths in the current iteration of training. In a given iteration, the machine learning model 410 can generate candidate layer tool paths 441, and compute or otherwise access the hot criticality metrics 442 for candidate layer tool paths 441. The tool path reordering engine 110 can implement any hot criticality metric computation or determination capability as described herein.

[0056] Candidate layer tool paths 441 can be generated by machine learning model 410 using a set of candidate parameter values ​​443 that are determined or otherwise adjusted by machine learning model 410 for a given iteration (e.g., via reinforcement learning techniques). By evaluating the candidate layer tool paths 441 via a corresponding hot criticality metric 442, machine learning model 410 can continue to iterate, learn, process, and determine adjusted parameter values ​​to converge toward any number of configuration goals or constraints, for example, measured and evaluated via the hot criticality metric. Collecting input data and input features, as well as the reordering of tool path segments generated by each iteration, allows machine learning model 410 to find appropriate reordering strategies (e.g., parameter values ​​represented as parameterized reordering algorithm 420) for the reordering of tool path segments.

[0057] In some implementations, the toolpath reordering engine 110 can be configured with a machine learning model 410 to optimize the parameter values ​​of a parameterized reordering algorithm 420 based on a target metric, which can be specified as an objective function evaluated by a threshold of the overall build time and / or hot criticality metric of candidate layer toolpaths. Any of the various build time and hot criticality metric constraints described herein can be equally applied to optimizing the parameter values ​​of the parameterized reordering algorithm 420 via the machine learning model 410. For example, the machine learning model 410 learns parameter values ​​for the parameterized reordering algorithm 420 that minimize the overall build time of the reordered toolpath segments, keeping the hot criticality metric of the reordered layer toolpaths below a threshold (whether measured as a total, maximum segment, or point-by-point hot criticality metric, or according to any other configurable evaluation metric as described herein).

[0058] The output of machine learning model training can be a set of optimized parameter values ​​for the parameterized reordering algorithm 420 (or, in some examples, multiple sets based on varying parameter values ​​from 3D printing via the input layer toolpath). Therefore, the toolpath reordering engine 110 can obtain a trained model (e.g., a trained machine learning model 410) from iterations and model training, which specifies learned parameter values ​​for multiple parameters of the parameterized reordering algorithm 420, and these learned parameter values ​​can be applied to the input layer toolpath. After training, the toolpath reordering engine 110 can use the trained machine learning model to reorder toolpath segments used for the input layer toolpath, as described below. Figure 5 Described.

[0059] Figure 5 This demonstrates an example application of a machine learning model supporting heat-sensing toolpath reordering for 3D printing of physical parts. Figure 5In the example, the tool path reordering engine 110 can use a machine learning model 410 trained with input data for, for example... Figure 4 The parameterized reordering algorithm 420 is described above.

[0060] When applying the trained machine learning model, the tool path reordering engine 110 can provide the original layer tool path 530 and a hot criticality metric 540 determined for the original layer tool path 530 as input to the machine learning model 410. The machine learning model 410 can determine an appropriate reordering strategy for the original layer tool path 530 (further considering the hot criticality metric 540), as learned through training from the input data. Therefore, the machine learning model 410 can set the learned parameter values ​​of the parameterized reordering algorithm 420, apply the parameterized reordering algorithm 420 to the original layer tool path 430, and obtain a modified layer tool path 550 as the output of the machine learning model 410.

[0061] By pre-training the machine learning model 410, the tool path reordering engine 110 does not need to solve an optimization problem for each individual input layer tool path. Instead, it relies on the trained machine learning model 410 to optimize the reordering of tool path segments in the input layer tool paths. Therefore, the ML training and application features described in this paper can improve the efficiency of heat-sensing tool path reordering.

[0062] While many heat-sensing toolpath reordering features have been described herein with illustrative examples presented in various figures, access engine 108 and toolpath reordering engine 110 can implement any combination of the heat-sensing toolpath reordering features described herein.

[0063] Figure 6 An example of logic 600 that the system can implement to support thermally sensed toolpath reordering for 3D printing of physical parts is shown. For example, computing system 100 can implement logic 600 as hardware, executable instructions stored on a machine-readable medium, or a combination of both. Computing system 100 can implement logic 600 via access engine 108 and toolpath reordering engine 110, through which computing system 100 can execute or implement logic 600 as a method to support thermally sensed toolpath reordering for 3D printing of physical parts. The following description of logic 600 is provided using access engine 108 and toolpath reordering engine 110 as examples. However, various other implementation options for the system are possible.

[0064] When implementing logic 600, access engine 108 can access the original layer toolpath (602) of a slice of a 3D CAD object. The 3D CAD object can represent a physical part, the slice can represent a physical layer for 3D printing of the physical part, and the original layer toolpath can control the 3D printing of the physical layer and includes the 3D printing order of multiple toolpath segments of the original layer toolpath. When implementing logic 600, access engine 108 can also access the thermal criticality metric (604) of the original layer toolpath. The thermal criticality metric can indicate the thermal impact at different points on multiple toolpath segments of the original layer toolpath for 3D printing of the physical part using the original layer toolpath.

[0065] When implementing logic 600, the toolpath reordering engine 110 can reorder multiple toolpath segments of the original layer toolpath into a modified layer toolpath (606), doing so in any of the ways described herein. The modified layer toolpath can have a different 3D printing order than the original layer toolpath and a thermal criticality metric with a smaller thermal impact on the physical part compared to the thermal criticality metric of the original layer toolpath. When implementing logic 600, the toolpath reordering engine 110 can also provide modified layer toolpaths to support 3D printing of physical parts represented by 3D CAD objects (608).

[0066] Figure 6 The illustrated logic 600 provides an illustrative example of how computing system 100 can reorder thermally sensing toolpaths that support 3D printing of physical parts. Additional or alternative steps in logic 600 are contemplated herein, including any of the features described herein for access engine 108, toolpath reordering engine 110, or any combination thereof.

[0067] Figure 7 An example of a computing system 700 supporting heat-sensing toolpath reordering for 3D printing of physical parts is shown. The computing system 700 may include a processor 710, which may take the form of a single processor or multiple processors. One or more processors 710 may include a central processing unit (CPU), a microprocessor, or any hardware device adapted to execute instructions stored on a machine-readable medium. The system 700 may include a machine-readable medium 720. The machine-readable medium 720 may take the form of any non-transient electronic, magnetic, optical, or other physical storage device storing executable instructions, such as... Figure 7The access instruction 722 and tool path reordering instruction 724 are shown. Therefore, the machine-readable medium 720 can be, for example, random access memory (RAM) (e.g., dynamic RAM (DRAM)), flash memory, spin torque memory, electrically erasable programmable read-only memory (EEPROM), storage drive, optical disk, etc.

[0068] The computing system 700 can execute instructions stored on the machine-readable medium 720 via the processor 710. Executing instructions (e.g., access instruction 722 and / or tool path reordering instruction 724) can cause the computing system 700 to perform any of the heat-sensing tool path reordering features described herein, including any feature based on access engine 108, tool path reordering engine 110, or a combination of both.

[0069] For example, execution of access instruction 722 by processor 710 can enable computing system 700 to access the original layer toolpath of a slice of a 3D CAD object and access the thermal criticality metric of the original layer toolpath. The 3D CAD object may represent a physical part, the slice may represent a physical layer for 3D printing of the physical part, and the original layer toolpath can control the 3D printing of the physical layer and includes the 3D printing order of multiple toolpath segments of the original layer toolpath. The thermal criticality metric of the original layer toolpath can indicate the thermal impact at different points on the multiple toolpath segments of the original layer toolpath for 3D printing of the physical part using the original layer toolpath.

[0070] Execution of the toolpath reordering instruction 724 by the processor 710 enables the computing system 700 to reorder multiple toolpath segments of the original layer toolpath into a modified layer toolpath, doing so in any of the ways described herein. The modified layer toolpath may have a different 3D printing order than the original layer toolpath and a thermal criticality metric with a smaller thermal impact on the physical part compared to the thermal criticality metric of the original layer toolpath. Execution of the toolpath reordering instruction 724 by the processor 710 may also enable the computing system 700 to provide the modified layer toolpath to support the 3D printing of the physical part.

[0071] Any additional or alternative thermal sensing toolpath reordering features as described herein may be implemented via access command 722, toolpath reordering command 724, or a combination of both.

[0072] The systems, methods, devices, and logic described above, including access engine 108 and tool path reordering engine 110, can be implemented in many different ways in many different combinations of hardware, logic, circuitry, and executable instructions stored on a machine-readable medium. For example, access engine 108, tool path reordering engine 110, or combinations thereof may include circuitry in a controller, microprocessor, or application-specific integrated circuit (ASIC), or may be implemented using discrete logic or components or combinations of other types of analog or digital circuitry combined on a single integrated circuit or distributed among multiple integrated circuits. A product (e.g., a computer program product) may include a storage medium and machine-readable instructions stored on the medium that, when executed in a terminal, computer system, or other device, cause the device to perform operations according to any of the above descriptions (including any features of access engine 108, tool path reordering engine 110, or combinations thereof).

[0073] The processing power of the systems, devices, and engines described herein (including access engine 108 and tool path reordering engine 110) can be distributed across multiple system components, such as across multiple processors and memories, optionally including multiple distributed processing systems or cloud / network elements. Parameters, databases, and other data structures can be stored and managed separately, can be combined into a single memory or database, can be logically and physically organized in many different ways, and can be implemented in many ways, including data structures such as linked lists, hash tables, or implicit storage mechanisms. Programs can be parts of a single program (e.g., subroutines), standalone programs, distributed across several memories and processors, or implemented in many different ways, such as in libraries (e.g., shared libraries).

[0074] While various examples have been described above, many more implementations are possible.

Claims

1. A method for heat-sensory reordering of toolpaths for 3D printing of physical parts, comprising: Through the calculation system: Access the original layer toolpaths (230, 330, 530) of a slice of a 3D CAD object (210), wherein the 3D CAD object (210) represents a physical part, wherein the slice represents a physical layer for 3D printing of the physical part, and wherein the original layer toolpaths (230, 330, 530) control the 3D printing of the physical layer, and include the 3D printing order of multiple toolpath segments of the original layer toolpaths; Access (604) the thermal criticality measure (240, 340, 540) of the original layer toolpath, wherein the thermal criticality measure (240, 340, 540) of the original layer toolpath indicates the thermal effect at different points on the plurality of toolpath segments of the original layer toolpath (230, 330, 530) for the 3D printing of the physical part using the 3D printing sequence of the original layer toolpath (230, 330, 530); The plurality of toolpath segments are reordered (606) into modified layer toolpaths (250, 350, 550), wherein the modified layer toolpaths (250, 350, 550) have a different 3D printing order than the original layer toolpaths (230, 330, 530), and have a thermal criticality metric that has a smaller thermal impact on the physical part compared to the thermal criticality metric (240, 340, 540) of the original layer toolpaths; and Provide (608) the modified layer tool path (250, 350, 550) to support the 3D printing of the physical part.

2. The method according to claim 1, wherein, The thermal criticality measure (240, 340, 540) of the original layer toolpath indicates the degree to which the temperature of the 3D part at each of the different points on the plurality of toolpath segments exceeds the critical temperature for 3D printing of the physical part.

3. The method according to claim 1 or 2, wherein, Reordering the plurality of tool path segments into the modified layer tool paths (250, 350, 550) includes iteratively optimizing candidate layer tool paths using a path optimization algorithm (310), which is applied to optimize the thermal criticality metric and overall build time of the candidate layer tool paths in order to determine the modified layer tool paths (250, 350, 550).

4. The method according to claim 3, wherein, Iteratively optimizing the candidate layer tool path includes, for a given iteration of the path optimization algorithm (310): Access the output layer tool path determined from the previous iteration of the path optimization algorithm (310) as a candidate layer tool path, wherein the candidate layer tool path includes multiple tool path segments; A segment hot criticality metric is used to access each of the multiple tool path segments of the candidate layer tool path. The ant colony optimization algorithm is applied as the path optimization algorithm (310), in which the pheromone parameter of the ant colony optimization algorithm is specified as the reciprocal of the segment heat criticality metric; and The output layer tool path (310) of the current iteration of the path optimization algorithm is determined based on the ant colony optimization algorithm.

5. The method according to claim 4, wherein, The original layer tool path is a candidate layer tool path used for the initial iteration of the path optimization algorithm (310).

6. The method of claim 1 or 2, further comprising training a machine learning model (410) to determine the modified layer toolpath, including by: Configure a parameterized reordering algorithm (420) that takes into account multiple parameters in the 3D printing of physical parts; The experimental parameter values ​​(433) of the plurality of parameters and the initial layer tool path (431) with the corresponding thermal criticality metric (432) are provided as inputs to the training. Through the training, the parameterized reordering algorithm (420) is iterated on the candidate layer tool path (441) using the corresponding thermal criticality metric (442) to optimize the parameterized reordering algorithm based on the overall construction time of the thermal criticality metric (442) and the candidate layer tool path (441); and The machine learning model (410) obtains learned parameter values ​​for the specified plurality of parameters from the iteration.

7. The method according to claim 6, wherein, Reordering the plurality of tool path segments into the modified layer tool path includes: The original layer toolpath and the thermal criticality metric of the original layer toolpath are provided as inputs to the machine learning model (410); and The modified layer tool path is obtained as the output of the machine learning model (410).

8. A system for heat-sensing reordering of toolpaths for 3D printing of physical parts, comprising: Access engine, which is configured as follows: Access the original layer toolpath of a slice of a 3D CAD object, wherein the 3D CAD object represents a physical part, wherein the slice represents a physical layer for 3D printing of the physical part, and wherein the original layer toolpath controls the 3D printing of the physical layer, and includes the 3D printing order of multiple toolpath segments of the original layer toolpath; and Access the thermal criticality metrics (240, 340, 540) of the original layer toolpath, wherein the thermal criticality metrics indicate the thermal impact at different points on the plurality of toolpath segments of the original layer toolpath for the 3D printing of the physical part using the original layer toolpath; and The tool path reordering engine is configured as follows: The plurality of toolpath segments are reordered into modified layer toolpaths (250, 350, 550), wherein the modified layer toolpaths (250, 350, 550) have a different 3D printing order than the original layer toolpaths and a thermal criticality metric that has a smaller thermal impact on the physical part compared to the thermal criticality metric (240, 340, 540) of the original layer toolpaths; and The modified layer toolpaths (250, 350, 550) are provided to support the 3D printing of the physical parts.

9. The system according to claim 8, wherein, The thermal criticality measure (240, 340, 540) of the original layer toolpath indicates the degree to which the temperature of the 3D part at each of the different points on the plurality of toolpath segments exceeds the critical temperature for 3D printing of the physical part.

10. The system according to claim 8 or 9, wherein, The tool path reordering engine (110) is configured to reorder the plurality of tool path segments into the modified layer tool paths (250, 350, 550) by iteratively optimizing candidate layer tool paths using a path optimization algorithm (310), which is applied to optimize the thermal criticality metric and overall build time of the candidate layer tool paths in order to determine the modified layer tool paths (250, 350, 550).

11. The system according to claim 10, wherein, The tool path reordering engine (110) is configured to iteratively optimize the candidate layer tool paths, including through, for a given iteration of the path optimization algorithm (310): Access the output layer tool path determined from the previous iteration of the path optimization algorithm (310) as a candidate layer tool path, wherein the candidate layer tool path includes multiple tool path segments; A segment hot criticality metric is used to access each of the multiple tool path segments of the candidate layer tool path. The ant colony optimization algorithm is applied as the path optimization algorithm (310), in which the pheromone parameter of the ant colony optimization algorithm is specified as the reciprocal of the segment heat criticality metric; and The output layer tool path (310) of the current iteration of the path optimization algorithm is determined based on the ant colony optimization algorithm.

12. The system according to claim 11, wherein, The tool path reordering engine (110) is configured to identify the original layer tool path as a candidate layer tool path for the initial iteration of the path optimization algorithm (310).

13. The system according to claim 8 or 9, wherein, The tool path reordering engine (110) is also configured to train a machine learning model to determine the modified layer tool path, including by: Configure a parameterized reordering algorithm (420) that takes into account multiple parameters in the 3D printing of physical parts; The experimental parameter values ​​(433) of the plurality of parameters and the initial layer tool path (431) with the corresponding thermal criticality metric (432) are provided as inputs to the training; Through the training, the parameterized reordering algorithm (420) is iterated on the candidate layer tool path (441) using the corresponding thermal criticality metric (442) to optimize the parameterized reordering algorithm based on the overall construction time of the thermal criticality metric (442) and the candidate layer tool path (441); and The machine learning model (410) obtains learned parameter values ​​for the specified plurality of parameters from the iteration.

14. The system according to claim 13, wherein, The toolpath reordering engine (110) is configured to reorder the plurality of toolpath segments into the modified layer toolpath in the following manner: The original layer tool path and the thermal criticality metric of the original layer tool path are provided as inputs to the machine learning model (410); as well as The modified layer tool path is obtained as the output of the machine learning model (410).

15. A non-transient machine-readable medium (720) comprising instructions (722, 724) that, when executed by a processor (710), cause a computing system (700) to perform the method according to any one of claims 1 to 7.

Images