Intelligent process control method and system for 3D printing of metal parts

Abstract

The invention discloses an intelligent process control method and system for 3D printing of metal parts, and relates to the technical field of intelligent control. The method comprises the steps that after a CAD model is imported into a 3D printing robot, equipment parameters are intelligently configured, process simulation is carried out based on the equipment parameters, and key process identifiers are established; when the 3D printing robot is started, the collaborative sensing unit is activated to execute multi-mode sensing recognition, and a sensing database is established; establishing a calibration response data set and a control deviation, configuring 3D printing demand data according to the CAD model, and establishing a deviation following target; synchronizing the control deviation and the deviation following target to a collaborative optimization channel to execute control optimization, and establishing a control optimization result; and intelligent process control management of the 3D printing robot is carried out according to a control optimization result. The technical problem that in the prior art, 3D printing is difficult to adjust in time according to real-time working conditions, and consequently the printing quality is unstable is solved, and the technical effect of improving the printing quality is achieved.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and more specifically to an intelligent process control method and system for 3D printed metal parts. Background Technology

[0002] Metal 3D printing technology, also known as additive manufacturing, overcomes the limitations of traditional subtractive manufacturing by directly constructing three-dimensional solid parts by layering metal materials. However, traditional 3D printing process control methods have some shortcomings, especially during the printing process. The inability to adjust in a timely manner according to actual working conditions often leads to unstable print quality. Traditional metal 3D printing processes typically rely on preset process parameters and offline model control, lacking real-time status perception and dynamic adjustment capabilities during the printing process. This results in quality problems such as unstable molten pools, warping deformation, and porosity defects under complex structures or changing working conditions, seriously affecting the quality and consistency of the printed parts. Especially when facing multi-material printing, complex geometries, or high-speed printing requirements, existing control methods struggle to respond promptly to actual deviations and cannot achieve intelligent optimization control of key process steps. Summary of the Invention

[0003] This application provides an intelligent process control method and system for 3D printing metal parts, which solves the technical problem in the prior art that 3D printing is difficult to adjust in a timely manner according to real-time working conditions, resulting in unstable printing quality.

[0004] The first aspect of this application provides an intelligent process control method for 3D printing metal parts, the method comprising:

[0005] After importing the CAD model into the 3D printing robot, automatic geometric optimization is performed, and equipment parameters are intelligently configured. Based on the equipment parameters, process simulation is conducted, and key process identifiers are established. After starting the 3D printing robot, a collaborative sensing unit is synchronously activated to perform multimodal perception and recognition, and a perception database is established. The collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. A calibration response dataset is established using the process simulation results, and a control deviation is established using the calibration response dataset and the perception database. 3D printing demand data is configured using the CAD model, and a deviation following target is established based on the 3D printing demand data, the perception database, and the key process identifiers. The control deviation and the deviation following target are synchronized to the collaborative optimization channel to perform control optimization and establish control optimization results. Intelligent process control management of the 3D printing robot is performed based on the control optimization results.

[0006] A second aspect of this application provides an intelligent process control system for 3D printing metal parts, the system comprising:

[0007] The system comprises the following modules: Parameter Configuration Module: After importing the CAD model into the 3D printing robot, it performs automatic geometric optimization and intelligently configures the equipment parameters; Identification Module: Based on the equipment parameters, it performs process simulation and establishes key process identifiers; Perception and Recognition Module: After starting the 3D printing robot, it synchronously activates the collaborative perception unit to perform multimodal perception and recognition, and establishes a perception database. The collaborative perception unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer; Deviation Establishment Module: It uses the process simulation results to establish a calibration response dataset, and uses the calibration response dataset and the perception database to establish a control deviation; Target Establishment Module: It uses the CAD model to configure 3D printing demand data, and establishes a deviation following target based on the 3D printing demand data, the perception database, and the key process identifiers; Optimization Module: It synchronizes the control deviation and the deviation following target to the collaborative optimization channel to perform control optimization and establish control optimization results; Control Module: It performs intelligent process control management of the 3D printing robot based on the control optimization results.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By importing the CAD model into the 3D printing robot and performing automatic geometric optimization, the equipment parameters are intelligently configured, and process simulation can be performed based on these parameters to establish key process identifiers. Then, when the 3D printing robot is started, the collaborative sensing unit is synchronously activated to perform multimodal perception and recognition, establishing a perception database. This collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. By utilizing the process simulation results, a calibration response dataset can be established, and this dataset, along with the perception database, can be further used to establish control deviations. Subsequently, 3D printing requirement data is configured based on the CAD model, and based on this data, the perception database, and key process identifiers, a deviation following target is established. Finally, the control deviation and deviation following target are synchronized to the collaborative optimization channel to perform control optimization, ultimately obtaining the control optimization result. Based on the control optimization result, intelligent process control management of the 3D printing robot is implemented. This solves the technical problem in existing technologies where 3D printing is difficult to adjust in a timely manner according to real-time working conditions, leading to unstable printing quality, and achieves the technical effect of improving printing quality. Attached Figure Description

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0011] Figure 1A schematic flowchart of an intelligent process control method for 3D printing metal parts provided in an embodiment of this application;

[0012] Figure 2 This is a schematic diagram of an intelligent process control system for 3D printing metal parts provided in an embodiment of this application.

[0013] Explanation of reference numerals in the attached diagram: Parameter configuration module 11, Identification module 12, Perception and recognition module 13, Deviation establishment module 14, Target establishment module 15, Optimization module 16, Control module 17. Detailed Implementation

[0014] This application provides an intelligent process control method and system for 3D printing metal parts, which solves the technical problem in the prior art that 3D printing is difficult to adjust in a timely manner according to real-time working conditions, resulting in unstable printing quality.

[0015] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0016] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0017] Example 1, as Figure 1 As shown, this application provides an intelligent process control method for 3D printing metal parts, wherein the method includes:

[0018] After importing the CAD model into the 3D printing robot, automatic geometric optimization is performed, and the equipment parameters are intelligently configured.

[0019] In this embodiment, after importing the CAD model into the 3D printing robot, the CAD model is first automatically identified for geometric features, including identifying key geometric information such as the overall structural dimensions of the component, surface curvature changes, overhang areas, boundary complexity, cavity structures, and support requirement areas. Then, based on these geometric features, the structural printability optimization module is invoked to perform automatic geometric optimization operations. These optimization operations include, but are not limited to, adjusting the support structure layout, optimizing the construction angle, correcting excessive overhang areas, and smoothing minimal curvature boundaries, resulting in an optimized printable geometry. After obtaining the optimized geometry, its key influencing factor set is extracted, wherein… The set of influencing factors includes multiple quantitative indicators characterizing printing difficulty, such as construction size parameters (e.g., maximum outer contour size, number of layers), boundary complexity factors (e.g., edge line density, frequency of curvature abrupt changes), support structure density indicators, and vertical length ratio. Furthermore, a pre-set 3D printing robot performance mapping rule library is invoked to input key printing influencing factors into the performance mapping model for equipment capability adaptation analysis, obtaining recommended configuration values ​​for each key equipment parameter. The equipment parameters include multiple adjustable parameters such as printing path strategy, layer thickness setting, scanning speed, laser power, preheating temperature, powder feeding rate, and nozzle movement trajectory.

[0020] Furthermore, after importing the CAD model into the 3D printing robot, automatic geometric optimization is performed, and the equipment parameters are intelligently configured, including:

[0021] Automatic geometric feature recognition is performed on the CAD model, and structural printability optimization is performed to establish an optimized printable geometry. Based on the optimized printable geometry, a set of key printing influencing factors is extracted, including build size, boundary complexity, support structure density, and vertical length. The preset performance mapping rules of the 3D printing robot are invoked, and the equipment parameters are configured based on the set of key printing influencing factors.

[0022] Specifically, geometric analysis algorithms are used to extract key structural features from the CAD model, identifying parameters such as the outer contour dimensions of components, internal cavity structures, overhang angles, boundary curvature variation areas, thin-walled areas, and surface concavity and convexity features. Based on the identification results and combined with printability evaluation criteria, structural printability optimization operations are performed, including adjusting the model construction direction to reduce the need for support structures, reconstructing the geometry of non-printable areas, eliminating overhang features, optimizing heat conduction paths, and improving structural continuity, thereby establishing an optimized printable geometry.

[0023] Based on the optimized printing geometry, a set of key printing influencing factors is extracted. This set of factors is used to characterize the potential impact of the geometry on the printing process. Specifically, it includes: build size (extracting the maximum size values ​​of the model in the X, Y, and Z directions), boundary complexity (quantitatively evaluated based on indicators such as the number of segments of the boundary contour, the rate of curvature change, and the number of edges), support structure density (evaluating the density level of the required support structure based on the area ratio, angle distribution, and spatial location of the model's overhanging region), and vertical length (the maximum height of the continuous vertical structure of the model in the build direction).

[0024] The performance mapping rule is a parameter mapping model established based on the correlation between historical printing samples, equipment operating conditions, and printing quality results. By calling the preset 3D printing robot performance mapping rule and inputting a set of key printing influencing factors into the rule, the system automatically completes the equipment parameter configuration. These parameters include, but are not limited to, laser power, scanning speed, layer thickness, preheating temperature, powder feeding rate, cooling interval, and trajectory mode.

[0025] Furthermore, the system invokes preset performance mapping rules for 3D printing robots and configures equipment parameters based on the set of key printing influencing factors, including:

[0026] A ternary relation database of structure, process, and parameters is established through historical printing records; sample matching of the current component feature vector with historical printing records is performed using structural similarity matching rules to establish sample matching results; the corresponding ternary relation database is called based on the sample matching results to establish the first parameter configuration result; the performance mapping rule of the preset 3D printing robot is triggered by the set of key printing influencing factors to establish the second parameter configuration result; the first parameter configuration result and the second parameter configuration result are fused from multiple sources to complete the equipment parameter configuration.

[0027] The ternary relational database is based on data collected from historical printing tasks, forming a ternary mapping relationship between structural features (including build size, boundary complexity, support density, vertical length, etc.), process conditions (such as layer thickness, scanning path strategy, cooling control method, etc.) and printing parameters (such as laser power, scanning speed, powder feed, preheating temperature, etc.). It supports the retrieval of optimal or high-success-rate combinations of printing parameters based on structural features.

[0028] For the set of key printing influencing factors extracted from the current CAD model, a feature vector of the current component is constructed based on standardization and vectorization encoding methods. This feature vector is then subjected to structural similarity matching analysis with the component feature vectors in historical printing records stored in a ternary relation database. Matching is performed using Euclidean distance, cosine similarity, or approximate nearest neighbor methods based on locality-sensitive hashing to obtain sample matching results. Based on these results, the corresponding historical process parameter configuration records in the ternary relation database are retrieved, and a set of parameter configurations with similar structure and excellent printing effects to the current component are extracted as the first parameter configuration result. Simultaneously, based on the current set of key printing influencing factors, a preset 3D printing robot performance mapping rule is triggered. This rule, based on machine learning models (such as decision trees, random forests, or neural networks) or expert experience rules, takes the set of key influencing factors as input and outputs a set of recommended equipment parameters to establish the second parameter configuration result. Finally, the first and second parameter configuration results are fused from multiple sources, including parameter weighted averaging and credibility assessment weighting, to form the final equipment parameter configuration result.

[0029] After performing process simulation based on the equipment parameters, key process identifiers are established.

[0030] In this embodiment of the application, based on the configured equipment parameters, process simulation is performed in the simulation platform. Specifically, based on the CAD model and its optimized printing geometry, combined with the configured equipment parameters (such as laser power, scanning speed, layer thickness, path strategy, etc.), a proposed printing path is generated, and dynamic simulation is performed on the temperature field, stress field, and melt pool evolution behavior during the printing process. By analyzing the time series data during the simulation process, key process parameter fluctuation points, stable intervals, and abnormal thermal behavior regions are identified.

[0031] Based on simulation data, key process identifiers are extracted and established to reflect typical process segments and parameter characteristics that significantly affect printing quality. These include: thermal accumulation critical point identifiers (recording regions and layers where the thermal field continuously rises and exceeds the critical stable temperature threshold); stress concentration zone identifiers (identifying areas where residual stress changes drastically due to geometric complexity or scanning sequence); path change node identifiers (recording the time and location of process actions such as scanning trajectory switching and cross-region jumps); construction speed fluctuation zone identifiers (identifying areas where printing rate changes due to laser power or movement speed adjustments); and molten pool dynamic instability risk identifiers (marking stages in the simulation where molten pool size deviation or irregular diffusion is predicted).

[0032] After the 3D printing robot is started, the collaborative sensing unit is activated simultaneously to perform multimodal perception and recognition and establish a perception database. The collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer.

[0033] After the 3D printing robot is started, the collaborative sensing unit is activated simultaneously to perform multimodal perception and recognition. The collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. The three work together to achieve high spatiotemporal resolution monitoring of the printing process.

[0034] The molten pool infrared camera is used to acquire real-time information on the temperature field changes of the molten pool. The collected data includes indicators such as the center temperature of the molten pool, the edge gradient, and the thermal diffusion rate. It generates a series of temperature heat maps to identify heat accumulation trends, ablation risks, and the consistency of interlayer temperature control.

[0035] High-speed cameras are used to perform high-speed imaging of the printing area, capturing the laser scanning path, changes in the molten pool morphology, the metal powder spreading process, and the surface features after forming. By using image frame sequences, abnormal behaviors such as metal spatter, forming defects (such as pores, cracks, wire hanging, etc.) and trajectory deviations can be identified.

[0036] Laser interferometers are used to monitor microscopic height changes and displacement fluctuations of the formed layer with high precision, obtain data on interlayer thickness consistency, warping deformation data, and microscale changes such as local expansion, and assist in judging the interlayer deposition quality and deformation trend.

[0037] By fusing thermal, visual, and structural information collected by the aforementioned multiple sensing devices, a sensing database is established that includes time series, spatial coordinates, image labels, and parameter states. The sensing database, indexed by the printing path, stores thermal field images, topographic images, and interferometric measurements for each printing unit area at each time point.

[0038] A calibration response dataset is established using process simulation results, and a control deviation is established using the calibration response dataset and the sensing database.

[0039] Based on the completed process simulation, simulation data such as molten pool thermal field evolution, residual stress distribution, scanning path thermal response, and structural strain changes are extracted. This data is then archived and structured according to time series, spatial location, and process stage to construct a calibration response dataset. The calibration response dataset serves as a standard reference set for key response indicators during ideal metal 3D printing, including but not limited to: molten pool temperature distribution templates under different scanning strategies; a baseline range of stress and strain fields within the constructed region; a mapping between interlayer thermal diffusion time constant and path sequence; and response curves of laser power and melt depth and width.

[0040] During the printing process, the perception database acquired by the collaborative perception unit is compared and analyzed with the calibration response dataset mentioned above. This includes: aligning the actual perception data at the current moment according to time, space, and process stage; using the corresponding simulation standard data as a benchmark, calculating the deviation of key indicators, such as temperature deviation, displacement deviation, thickness deviation, trajectory deviation angle, etc.; classifying and mapping the above deviation data, and expressing it in a structured manner according to the printing unit to form the control deviation dataset of the printing process.

[0041] The CAD model is used to configure 3D printing demand data, and a deviation from the target is established based on the 3D printing demand data, the perception database, and the key process identifier.

[0042] After importing the CAD model, design requirements such as the three-dimensional spatial arrangement, wall thickness distribution, and structural complexity of the components are extracted based on their geometric features and material properties. These requirements are then combined with printing strategies (such as scan path planning, layering strategies, and support structure design) to generate 3D printing requirement data. This 3D printing requirement data includes: layer path planning diagrams; component accuracy requirements (such as tolerance grades and surface roughness); thermal stress control requirements (such as maximum allowable residual stress and warpage limits); forming efficiency targets (such as time constraints and scan rate limits); and printing physical parameter constraints such as power density and energy input limitations.

[0043] Combining the aforementioned 3D printing demand data, the perception database from the actual production process, and key process markers generated during simulation (such as high-thermal-sensitive areas, high-stress-concentration areas, and boundary shrinkage-sensitive sections), a deviation-following target is established. Specifically, this includes: comparing each key indicator in the printing demand data with the data acquired in real time from the perception database to identify key points with deviations or risk evolution trends; combining key process markers to screen out areas and indicators that are sensitive to or require key control of printing quality; and integrating the above content into a fusion model to form a set of deviation-following targets to guide the target definition of the control optimization channel.

[0044] Deviation from the target includes: target deviation in a specified dimension (e.g., maintaining temperature within 5°C); control accuracy target within a specified spatial range (e.g., maintaining trajectory deviation <50μm at the edge of the support area); dynamic adjustment requirements for a specified structural region (e.g., implementing an intermittent power control strategy in the overhanging region); and response constraints of the target layer segment (e.g., limiting the heat accumulation factor from the Nth to the Mth layers to not exceed a threshold).

[0045] The control deviation and the deviation from the target are synchronized to the collaborative optimization channel to perform control optimization and establish the control optimization result.

[0046] The control deviation (including multi-dimensional data such as temperature, stress, deformation, and trajectory error) and the deviation from the target (including expected response index, control target value, target area range, etc.) are used together as optimization input parameters and input into the collaborative optimization channel for control optimization. The optimal control strategy is output, which is the control optimization result.

[0047] Furthermore, the control deviation and the deviation from the target are synchronized to the collaborative optimization channel to perform control optimization, and the control optimization result is established, including:

[0048] The deviation from the target is analyzed to obtain a deviation dataset, which includes the deviation type, degree of deviation, and location of the deviation area. A state space is constructed based on the deviation dataset. The adjustable process behavior set of the current node is obtained. After constructing an action space based on the adjustable process behavior set, the control deviation is mapped to the action space. A reward function is constructed based on the control deviation. Control optimization is performed based on the reward function, the action space, and the state space to establish the control optimization result.

[0049] By analyzing deviations from the target, a deviation dataset is extracted, containing deviation types (e.g., temperature deviation, trajectory displacement deviation, melt pool size fluctuations), deviation degrees (i.e., differences or proportions from set thresholds), and deviation location areas (e.g., build layer number, scan path segment). Based on this deviation dataset, a current process state space is constructed to comprehensively describe the dynamic state characteristics of the system. Simultaneously, combined with the executable controllable behaviors of the current printing node (e.g., laser power adjustment, scan speed variation, interlayer cooling time adjustment), a process action space is constructed, and the detected control deviations are mapped into this action space, forming the initial constraint boundary for control behavior selection. Based on this, a reward function is constructed to evaluate the effectiveness of process adjustments. This reward function comprehensively considers multiple objective functions, including deviation suppression efficiency, print quality improvement, resource consumption changes, and the rationality of heat input. Subsequently, under the joint constraints of the state space, action space, and reward function, a collaborative optimization calculation process is executed. This process can employ optimization methods such as reinforcement learning, adaptive dynamic programming, genetic algorithms, or policy gradient search to iteratively explore and evaluate the value of different control path combinations. Finally, a set of optimal control parameters and corresponding policies under the current operating conditions are output, constituting the control optimization result.

[0050] Furthermore, based on the reward function, the action space, and the state space, control optimization is performed to establish the control optimization result, including:

[0051] An initial following strategy is established based on the calibration response dataset and the 3D printing demand data. The initial following strategy is a process behavior reference path strategy. A deviation response strategy is constructed based on the control deviation. The initial following strategy and the deviation response strategy are input into the strategy fusion sub-channel. A random exploration factor is introduced to perform strategy updates in the action space. The strategy update results are evaluated through a reward function to establish a strategy evaluation result and complete one round of control optimization.

[0052] First, by combining a pre-set calibration response dataset with 3D printing demand data, an initial following strategy is established for the execution of the printing path under normal operating conditions. This initial following strategy represents a set of process behavior reference paths, reflecting the optimal combination of various control dimensions (such as laser energy, scanning speed, and path trajectory) under ideal operating conditions. Then, based on the aforementioned control deviation information, a deviation response strategy is constructed to simulate the process behavior adjustment scheme to be taken under deviation conditions. Next, the initial following strategy and the deviation response strategy are input into the strategy fusion subchannel. A random exploration factor is introduced into the strategy fusion subchannel to expand the search range of the action space and achieve diversified exploration of strategy combination paths, avoiding getting trapped in local optima. By introducing this randomness during the strategy update process, the adaptability of the optimization process to complex deviation dynamics can be effectively improved. After each round of strategy update, the effect of the updated strategy execution result is evaluated using the previously constructed multi-dimensional reward function, forming a strategy evaluation result. This strategy evaluation result reflects the comprehensive performance indicators of the current strategy in terms of printing deviation correction, forming accuracy improvement, and process stability enhancement. Based on the above process, one round of control optimization can be completed.

[0053] Furthermore, the strategy evaluation results include:

[0054] Before executing the next round of control optimization actions, the reward level is determined based on the policy evaluation results, and a reward level determination result is established. Based on the reward level determination result, the search range and search granularity of the next round action space are configured. After constraining the policy fusion sub-channel based on the configuration result, the policy update of the action space is executed.

[0055] After the optimization process is complete, a reward function is invoked to score the output strategy update results. The score includes multiple dimensions such as deviation correction effect score, energy consumption impact factor, process stability index, and component quality consistency level. Based on the score, a reward level positioning analysis is performed, mapping the score value to multiple preset level ranges, such as high reward level, medium reward level, and low reward level. The reward level positioning results are used to guide the selection of the search mode in the next round of optimization.

[0056] Based on this, the search range and granularity of the next round of action space are dynamically configured according to the reward level positioning results: if the strategy evaluation is in the high-level range, it indicates that the current strategy is close to optimal. At this time, the control system will automatically narrow the action space search range and refine the search granularity to achieve fine adjustment and improve local control accuracy; if the strategy is in the low-reward-level range, it indicates that the current strategy has significant optimization space. At this time, the system will widen the search range, increase the search step size, and turn to the global optimization direction to avoid getting trapped in the local minimum region. The above search configuration results will be input into the strategy fusion sub-channel as a constraint on the strategy update mechanism, guiding the strategy update operation in the next round of action space.

[0057] Furthermore, based on the reward level positioning result, the search range and granularity of the next round's action space are configured, including:

[0058] If the reward level positioning result meets the first level threshold, a focus instruction is generated, and the search range and search granularity of the next round's action space are configured according to the focus instruction; if the reward level positioning result meets the second level threshold, a global exploration instruction is generated, and the search range and search granularity of the next round's action space are configured according to the global exploration instruction.

[0059] When the reward level positioning result meets the first level threshold, it indicates that the current control strategy is in a high-performance range, meaning that the printing deviation is small, the process behavior is stable, and the printing quality is high. At this time, the system generates a focusing instruction. The focusing instruction is used to control the action space to shrink to the area near the current strategy and refine the search granularity to a tiny step size, so that the strategy update process is locally fine-tuned around the optimal solution, thereby improving control accuracy and printing stability.

[0060] Conversely, when the reward level positioning result meets the second-level threshold, it indicates that the current control strategy is at a low performance level, such as having significant deviations, insufficient response, abnormal formation process, or poor structural consistency. At this time, the system generates a global exploration instruction. The global exploration instruction will drive the action space to expand to a larger range and adjust the search granularity to a larger step size, so as to realize global jump exploration of the strategy update path, jump out of the local inefficiency zone, and quickly seek the globally optimal control strategy.

[0061] Furthermore, the control optimization based on the reward function, the action space, and the state space, and the establishment of the control optimization result, also includes:

[0062] The perception database is analyzed to establish a thermal time-series dataset, which is then synchronized to the collaborative optimization channel as supplementary data. Temperature clustering and gradient analysis are performed based on the supplementary data to identify heat accumulation regions and establish a heat accumulation region identification map. After configuring intermittent cooling constraints using the heat accumulation region identification map, optimization constraints in the action space are executed.

[0063] The sensor database is analyzed, and a thermal time series dataset is extracted and established. The thermal time series dataset records the temperature distribution information of each key area during the printing process over time, and is synchronously input into the collaborative optimization channel as additional data.

[0064] By performing temperature clustering and gradient change analysis on the thermal time-series dataset, regions of heat accumulation during the printing process were identified. Specifically, temperature clustering analysis grouped points with similar temperature change patterns into the same category to clearly distinguish between high-temperature concentrated areas and areas with uneven temperature distribution; gradient analysis was used to detect the rate and direction of temperature change, accurately locating key areas where heat accumulation might cause defects. Based on this, a heat accumulation region identification map was constructed to visually and structurally represent the spatial distribution and severity of heat accumulation regions. Subsequently, this heat accumulation region identification map was used to configure targeted intermittent cooling constraints as optimization constraints on the action space, limiting laser power and scanning speed in high-temperature areas, inserting cooling time intervals, and other control measures to prevent material deformation and defects caused by local overheating.

[0065] Intelligent process control management of 3D printing robots is carried out based on the control optimization results.

[0066] Based on the control optimization results, the system sends the optimal process parameters and control strategies to the 3D printing robot in real time, achieving intelligent process control management of the printing process. Specifically, this includes transmitting parameters such as laser power adjustment values, scanning speed changes, path correction instructions, and cooling interval settings from the control optimization results as input commands to the 3D printing robot's control unit. Based on the received commands, the printing robot dynamically adjusts the laser scanning trajectory, energy input, and powder spraying rate to ensure that the process parameters remain consistent with the optimization results during actual execution. Simultaneously, the system configures a closed-loop feedback mechanism between the control results and real-time sensing data, continuously monitoring printing quality and process response. If new deviations or anomalies are detected, a new control optimization process is triggered promptly, completing dynamic adaptive adjustment.

[0067] Furthermore, intelligent process control management of the 3D printing robot based on the control optimization results includes:

[0068] Configure a verification space for mapping the optimization results of the control system; verify the printing effect within the verification space and generate monitoring feedback; perform self-optimization compensation of the optimization results of the control system based on the monitoring feedback.

[0069] The verification space, which maps the configuration to the control optimization results, defines a corresponding physical or virtual printing area based on the process parameters and adjustment strategies output by the control optimization. This verification space is used to detect the effectiveness of the actual process execution. Within this verification space, real-time printing effect verification is implemented. Multimodal data collected by collaborative sensing units is used to monitor printing quality, including indicators such as melt pool temperature distribution, surface morphology, geometric dimensions, and structural integrity, generating accurate monitoring feedback information. Based on the monitoring feedback, the system analyzes the deviation between the actual printing effect and the control optimization results, executes a self-optimization compensation mechanism, dynamically corrects and adjusts the control optimization results, optimizes the control strategy to adapt to changes in actual working conditions, and achieves intelligent closed-loop control management, ensuring that the 3D printing robot continuously and stably completes the manufacturing of high-quality metal parts.

[0070] In summary, the embodiments of this application have at least the following technical effects:

[0071] By importing the CAD model into the 3D printing robot and performing automatic geometric optimization, the equipment parameters are intelligently configured, and process simulation can be performed based on these parameters to establish key process identifiers. Then, when the 3D printing robot is started, the collaborative sensing unit is synchronously activated to perform multimodal perception and recognition, establishing a perception database. This collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. By utilizing the process simulation results, a calibration response dataset can be established, and this dataset, along with the perception database, can be further used to establish control deviations. Subsequently, 3D printing requirement data is configured based on the CAD model, and based on this data, the perception database, and key process identifiers, a deviation following target is established. Finally, the control deviation and deviation following target are synchronized to the collaborative optimization channel to perform control optimization, ultimately obtaining the control optimization result. Based on the control optimization result, intelligent process control management of the 3D printing robot is implemented. This solves the technical problem in existing technologies where 3D printing is difficult to adjust in a timely manner according to real-time working conditions, leading to unstable printing quality, and achieves the technical effect of improving printing quality.

[0072] Example 2, based on the same inventive concept as the intelligent process control method for 3D printed metal parts in the foregoing examples, such as... Figure 2 As shown, this application provides an intelligent process control system for 3D printing metal parts, wherein the system includes:

[0073] Parameter Configuration Module 11: After importing the CAD model into the 3D printing robot, it performs automatic geometric optimization and intelligently configures the equipment parameters; Identification Module 12: Based on the equipment parameters, it performs process simulation and establishes key process identifiers; Perception and Recognition Module 13: After starting the 3D printing robot, it synchronously activates the collaborative perception unit to perform multimodal perception and recognition and establishes a perception database. The collaborative perception unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer; Deviation Establishment Module 14: It uses the process simulation results to establish a calibration response dataset and uses the calibration response dataset and the perception database to establish a control deviation; Target Establishment Module 15: It uses the CAD model to configure 3D printing demand data and establishes a deviation following target based on the 3D printing demand data, the perception database, and the key process identifiers; Optimization Module 16: It synchronizes the control deviation and the deviation following target to the collaborative optimization channel to perform control optimization and establish control optimization results; Control Module 17: It performs intelligent process control management of the 3D printing robot based on the control optimization results.

[0074] Furthermore, the optimization module 16 is used to perform the following method:

[0075] The deviation from the target is analyzed to obtain a deviation dataset, which includes the deviation type, degree of deviation, and location of the deviation area. A state space is constructed based on the deviation dataset. The adjustable process behavior set of the current node is obtained. After constructing an action space based on the adjustable process behavior set, the control deviation is mapped to the action space. A reward function is constructed based on the control deviation. Control optimization is performed based on the reward function, the action space, and the state space to establish the control optimization result.

[0076] Furthermore, the optimization module 16 is used to perform the following method:

[0077] An initial following strategy is established based on the calibration response dataset and the 3D printing demand data. The initial following strategy is a process behavior reference path strategy. A deviation response strategy is constructed based on the control deviation. The initial following strategy and the deviation response strategy are input into the strategy fusion sub-channel. A random exploration factor is introduced to perform strategy updates in the action space. The strategy update results are evaluated through a reward function to establish a strategy evaluation result and complete one round of control optimization.

[0078] Furthermore, the optimization module 16 is used to perform the following method:

[0079] Before executing the next round of control optimization actions, the reward level is determined based on the policy evaluation results, and a reward level determination result is established. Based on the reward level determination result, the search range and search granularity of the next round action space are configured. After constraining the policy fusion sub-channel based on the configuration result, the policy update of the action space is executed.

[0080] Furthermore, the optimization module 16 is used to perform the following method:

[0081] If the reward level positioning result meets the first level threshold, a focus instruction is generated, and the search range and search granularity of the next round's action space are configured according to the focus instruction; if the reward level positioning result meets the second level threshold, a global exploration instruction is generated, and the search range and search granularity of the next round's action space are configured according to the global exploration instruction.

[0082] Furthermore, the optimization module 16 is used to perform the following method:

[0083] The perception database is analyzed to establish a thermal time-series dataset, which is then synchronized to the collaborative optimization channel as supplementary data. Temperature clustering and gradient analysis are performed based on the supplementary data to identify heat accumulation regions and establish a heat accumulation region identification map. After configuring intermittent cooling constraints using the heat accumulation region identification map, optimization constraints in the action space are executed.

[0084] Furthermore, the parameter configuration module 11 is used to perform the following method:

[0085] Automatic geometric feature recognition is performed on the CAD model, and structural printability optimization is performed to establish an optimized printable geometry. Based on the optimized printable geometry, a set of key printing influencing factors is extracted, including build size, boundary complexity, support structure density, and vertical length. The preset performance mapping rules of the 3D printing robot are invoked, and the equipment parameters are configured based on the set of key printing influencing factors.

[0086] Furthermore, the parameter configuration module 11 is used to perform the following method:

[0087] A ternary relation database of structure, process, and parameters is established through historical printing records; sample matching of the current component feature vector with historical printing records is performed using structural similarity matching rules to establish sample matching results; the corresponding ternary relation database is called based on the sample matching results to establish the first parameter configuration result; the performance mapping rule of the preset 3D printing robot is triggered by the set of key printing influencing factors to establish the second parameter configuration result; the first parameter configuration result and the second parameter configuration result are fused from multiple sources to complete the equipment parameter configuration.

[0088] Furthermore, the control module 17 is used to perform the following methods:

[0089] Configure a verification space for mapping the optimization results of the control system; verify the printing effect within the verification space and generate monitoring feedback; perform self-optimization compensation of the optimization results of the control system based on the monitoring feedback.

[0090] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0091] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0092] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An intelligent process control method for 3D printed metal parts, characterized in that, The method includes: After importing the CAD model into the 3D printing robot, automatic geometry optimization is performed, and the equipment parameters are intelligently configured. After performing process simulation based on the equipment parameters, key process identifiers are established. After the 3D printing robot is started, the collaborative sensing unit is activated simultaneously to perform multimodal perception and recognition and establish a perception database. The collaborative sensing unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. A calibration response dataset is established using process simulation results, and a control deviation is established using the calibration response dataset and the sensing database. The CAD model is used to configure 3D printing demand data, and a deviation from the target is established based on the 3D printing demand data, the perception database, and the key process identifier. The control deviation and the deviation from the target are synchronized to the collaborative optimization channel to perform control optimization and establish the control optimization result. Intelligent process control management of 3D printing robots is carried out based on the control optimization results.

2. The intelligent process control method for 3D printed metal parts as described in claim 1, characterized in that, The step of synchronizing the control deviation and the deviation from the target to the collaborative optimization channel for control optimization and establishing the control optimization result includes: The deviation from the target is analyzed to obtain a deviation dataset, which includes the deviation type, the degree of deviation, and the location of the deviation area. A state space is constructed based on the deviation dataset. Obtain the adjustable process behavior set of the current node, construct the action space based on the adjustable process behavior set, and then map the control deviation to the action space; A reward function is constructed based on the control deviation. Control optimization is performed based on the reward function, the action space, and the state space to establish the control optimization result.

3. The intelligent process control method for 3D printed metal parts as described in claim 2, characterized in that, The control optimization based on the reward function, the action space, and the state space, and the establishment of the control optimization result, includes: An initial following strategy is established based on the calibration response dataset and the 3D printing demand data. The initial following strategy is a process behavior reference path strategy. Construct a deviation response strategy based on the control deviation; The initial following strategy and the deviation response strategy are input into the strategy fusion sub-channel, and a random exploration factor is introduced to perform strategy updates in the action space. The strategy update results are evaluated through a reward function, and a strategy evaluation result is established to complete one round of control optimization.

4. The intelligent process control method for 3D printed metal parts as described in claim 3, characterized in that, The evaluation results of the established strategy include: Before executing the next round of control optimization actions, the reward level is determined based on the evaluation results of the strategy, and the reward level determination results are established. Based on the reward level positioning result, configure the search range and search granularity of the next round action space, and after constraining the policy fusion sub-channel based on the configuration result, execute the policy update of the action space.

5. The intelligent process control method for 3D printed metal parts as described in claim 4, characterized in that, The step of configuring the search range and granularity of the next round's action space based on the reward level positioning result includes: If the reward level positioning result meets the first level threshold, a focus instruction is generated, and the search range and search granularity of the next round of action space are configured according to the focus instruction; If the reward level positioning result meets the second level threshold, a global exploration instruction is generated, and the search range and search granularity of the next round of action space are configured according to the global exploration instruction.

6. The intelligent process control method for 3D printed metal parts as described in claim 2, characterized in that, The step of performing control optimization based on the reward function, the action space, and the state space, and establishing the control optimization result, further includes: The perception database is analyzed to establish a hot time series dataset, which is then synchronized to the collaborative optimization channel as supplementary data. Based on the additional data, temperature clustering and gradient analysis are performed to identify heat accumulation regions and establish a heat accumulation region identification map. After configuring intermittent cooling constraints using the heat accumulation region identification map, optimization constraints are performed in the action space.

7. The intelligent process control method for 3D printed metal parts as described in claim 1, characterized in that, After importing the CAD model into the 3D printing robot, automatic geometric optimization is performed, and the equipment parameters are intelligently configured, including: The CAD model is automatically identified for geometric features, and structural printability optimization is performed to create the optimized printable geometry. Based on the optimized printing geometry, a set of key printing influencing factors is extracted, which includes construction size, boundary complexity, support structure density, and vertical length. The preset performance mapping rules for 3D printing robots are invoked, and the equipment parameters are configured based on the set of key printing influencing factors.

8. The intelligent process control method for 3D printed metal parts as described in claim 7, characterized in that, The step of invoking the preset performance mapping rules for 3D printing robots and configuring equipment parameters based on the set of key printing influencing factors includes: A ternary relational database of structure, process, and parameters is established by using historical print records; The current component feature vector is matched with historical printing records using structural similarity matching rules to establish the sample matching results; Based on the sample matching results, the corresponding ternary relation database is invoked to establish the first parameter configuration result; By using a set of key influencing factors for printing, a performance mapping rule-triggered analysis of a preset 3D printing robot is conducted to establish the configuration results of the second parameter. The first parameter configuration result and the second parameter configuration result are fused from multiple sources to complete the device parameter configuration.

9. The intelligent process control method for 3D printed metal parts as described in claim 1, characterized in that, The intelligent process control management of the 3D printing robot based on the control optimization results includes: Verification space for configuring and controlling the mapping of optimization results; The printing effect is verified within the verification space, and monitoring feedback is generated. The control optimization results are self-optimized and compensated based on the monitoring feedback.

10. An intelligent process control system for 3D printing metal parts, characterized in that, The system is used to implement the intelligent process control method for 3D printing metal parts according to any one of claims 1-9, the system comprising: Parameter configuration module: After importing the CAD model into the 3D printing robot, it performs automatic geometric optimization and intelligently configures the equipment parameters; Identification module: After performing process simulation based on the equipment parameters, establish key process identifiers; Perception and Recognition Module: After the 3D printing robot is started, the collaborative perception unit is activated synchronously to perform multimodal perception and recognition and establish a perception database. The collaborative perception unit includes a molten pool infrared camera, a high-speed camera, and a laser interferometer. Deviation establishment module: Establishes a calibration response dataset using process simulation results, and establishes control deviation using the calibration response dataset and the sensing database; Target establishment module: Utilizes the CAD model to configure 3D printing demand data, and establishes deviation and following targets based on the 3D printing demand data, the perception database, and the key process identifiers; Optimization module: Synchronizes the control deviation and the deviation from the target to the collaborative optimization channel to perform control optimization and establishes the control optimization results; Control module: Performs intelligent process control management of the 3D printing robot based on the control optimization results.

Images