Process compensation method and system for additive manufacturing recycled powder property evolution

Abstract

The application relates to a process compensation method and system for evolution of additive manufacturing recycled powder characteristics, including component shaping of powders with different recycling times and new powder proportions, and characteristic detection of recycled powders and shaped components, to establish a decay relationship between the recycling times and the powder characteristic parameters; according to the decay relationship, a flowability decay coefficient, a heat absorption decay rate and a satellite powder proportion increase coefficient are calculated to construct a compensation strategy for a scanning task; in the process of executing the additive manufacturing scanning task, the compensation strategy is triggered every preset layer, compensation process parameters are generated based on the current recycling times, and an optimal new powder proportion is predicted by using a machine learning decision model; and the compensation parameters and the new powder proportion are input into an additive manufacturing equipment for execution. The application can accurately quantify the powder characteristic decay law under the condition of powder recycling and establish a dynamic matching compensation mechanism with the process parameters, so as to effectively guarantee the shaping quality, reduce defects and improve the utilization rate of recycled powders.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of additive manufacturing technology, and more particularly to a process compensation method and system for evolution of recycled powder characteristics in additive manufacturing. BACKGROUND

[0002] In recent years, with the rapid increase in the demand for the integration and lightweight manufacturing of high-performance complex components of new-generation aerospace equipment, the powder bed fusion (PBF) technology is increasingly widely used in the integration and lightweight manufacturing of high-end aerospace equipment due to its ability to realize high-performance forming and manufacturing of highly complex components. However, due to the process characteristics of dynamic powder laying and local rapid fusion of powder layers, the characteristics of the powder raw material and its process matching are particularly important for the improvement of the process effect and product quality of the powder bed fusion technology, especially the powder characteristics such as particle flowability, particle morphology, and surface properties, which can significantly affect the heat and mass transfer processes such as powder bed laying, molten pool metallurgical solidification and cooling.

[0003] Currently, in actual production, metal powder usually needs to be recycled multiple times. However, as the number of recycling increases, the surface oxide film of the powder gradually thickens, the proportion of satellite powder and hollow powder increases, resulting in a decrease in powder flowability (manifested as an increase in Hall flow rate), a decrease in heat absorption rate, and insufficient density. These degradation effects not only affect the quality of the powder bed, but also exacerbate the risk of insufficient or excessive laser molten pool energy input, ultimately leading to a decrease in the quality of the formed components. SUMMARY

[0004] The present application provides a process compensation method and system for evolution of recycled powder characteristics in additive manufacturing, which can accurately quantify the degradation law of powder characteristics with the number of recycling under recycling conditions, and then establish a dynamic matching compensation mechanism between the degradation law and the process parameters, to ensure the forming quality of the formed components and improve the utilization rate of recycled powder.

[0005] In a first aspect, the application provides a process compensation method for evolution of additive manufacturing recycled powder characteristics, the method comprising: forming a component by using powders with different recycling times and new powder proportions, and detecting the characteristics of the recycled powders and the component after forming, establishing a decay relationship between the recycling times and the powder characteristic parameters; calculating a decay coefficient of the powder according to the decay relationship between the recycling times and the powder characteristic parameters, and constructing a compensation strategy for a scanning task according to the decay coefficient, wherein the decay coefficient comprises a flowability attenuation coefficient, a heat absorption attenuation rate, and a satellite powder proportion increase coefficient; during execution of an additive manufacturing scanning task, triggering the compensation strategy every preset scanning layer to generate a compensation process parameter according to the current recycling time, and simultaneously predicting a new powder proportion corresponding to an optimal forming quality index of the current recycling time by using a machine learning decision model; and inputting the compensation process parameter and the new powder proportion into an additive manufacturing device for execution.

[0006] In an optional implementation of the first aspect, the detecting the characteristics of the recycled powders and the component after forming comprises: detecting flowability parameters of the powders by using a Hall flowmeter test and a dynamic angle of repose test; testing the oxide film thickness and oxygen content parameters of the powders by using an X-ray photoelectron spectroscopy; and measuring the surface roughness of a cladding layer of the component by using a non-contact three-dimensional profilometer, and measuring the density of the component by using a drainage method.

[0007] In an optional implementation of the first aspect, the calculating the decay coefficient of the powder according to the decay relationship comprises: calculating a flowability attenuation coefficient of the powder according to the decay relationship: wherein is the recycling time; calculating a heat absorption attenuation rate of the powder according to the decay relationship: wherein, is the energy beam absorption rate of the powder recycled in the Nth recycling time, is the energy beam absorption rate of the new powder; and calculating a satellite powder proportion increase coefficient of the powder according to the decay relationship: wherein is the satellite powder proportion of the powder recycled in the Nth recycling time, is the satellite powder proportion of the new powder.

[0008] In an optional implementation of the first aspect, the constructing a compensation strategy for a scanning task according to the decay coefficient comprises: constructing a powder layer thickness increase strategy according to the flowability attenuation coefficient, wherein the adjustment amplitude formula of the powder layer thickness increase strategy is: wherein is the increased powder layer thickness, is the flowability attenuation coefficient, is the reference powder layer thickness; and constructing an energy beam power promotion strategy according to the heat absorption attenuation rate, wherein the adjustment amplitude formula of the energy beam power promotion strategy is: wherein, is the energy beam power, is the heat absorption decay rate, is the reference energy beam power; a powder laying speed reduction strategy is constructed according to the satellite powder proportion increase coefficient, and a regulation range formula of the powder laying speed reduction strategy is: wherein, is the reduced powder laying speed, is the satellite powder proportion increase coefficient, is the reference powder laying speed.

[0009] In an optional solution of the first aspect, after the compensation strategy of the scanning task is constructed according to the decay coefficient, the method comprises: in the powder layer thickness increase strategy, setting the adjustment threshold range to ±10 μm; in the energy beam power increase strategy, setting the adjustment threshold range to ±30 W; and in the powder laying speed reduction strategy, setting the adjustment threshold range to ±50 mm / s.

[0010] In an optional solution of the first aspect, when the compensation strategy is triggered, the method comprises: dividing a forming area of the additive manufacturing component into multiple sub-regions according to a scanning path of the scanning task; obtaining the powder recycling frequency of each sub-region and independently calculating the flowability decay coefficient, the heat absorption decay rate and the satellite powder proportion increase coefficient of the powder of each sub-region; and independently executing the corresponding powder layer thickness increase strategy and / or the energy beam power increase strategy and / or the powder laying speed reduction strategy for each sub-region.

[0011] In an optional solution of the first aspect, when the compensation strategy of the scanning task is constructed, the method further comprises: collecting sample parameters of a first preset number of recycling frequencies, new powder proportions, powder characteristic parameters, corresponding component surface roughnesses and densities and constructing a data set; using a machine learning method to iteratively train the data set, constructing a machine learning decision model, and the process of iterative training comprises: taking the recycling frequency and the powder characteristic parameter as input, taking the forming quality index and the new powder proportion as output and setting a constraint condition.

[0012] In an optional solution of the first aspect, the data set is constructed by: generating a second preset number of new sample parameters for adjacent two sample parameters; using linear interpolation to calculate the new powder proportion, the corresponding component surface roughness and the density of the new sample parameter, while adding small amplitude Gaussian noise to simulate experimental measurement error, constructing enhanced sample parameters; adding the constructed enhanced sample parameters to the data set to form an enhanced data set, and dividing the data set into a training set and a test set according to a preset proportion by using a random index method.

[0013] In an optional implementation of the first aspect, the method further includes: using an input data standardization method to eliminate dimensional differences of the enhanced data set; setting a kernel function of the machine learning decision model as a radial basis function; and setting a constraint condition to maximize a proportion of the recycled powder while ensuring that the surface roughness of the build-up layer is not greater than a preset value.

[0014] In a second aspect, the application provides an additive manufacturing recycled powder property evolution process compensation method system using the method of any one of the first aspect, including: a decay building module for building components using powders with different recycling times and new powder proportions, and detecting properties of the recycled powders and components after building, to establish a decay relationship between the recycling times and the powder property parameters; a strategy building module for calculating a decay coefficient of the powder according to the decay relationship between the recycling times and the powder property parameters, and building a compensation strategy for a scanning task according to the decay coefficient, wherein the decay coefficient includes a flowability decay coefficient, a heat absorption decay rate, and a satellite powder proportion increase coefficient; a process compensation module for triggering the compensation strategy every preset number of scanning layers during execution of the additive manufacturing scanning task, to generate a compensation process parameter according to the current recycling time; a machine learning module for building a machine learning decision model and predicting an optimal recycled powder proportion corresponding to the current recycling time using the machine learning decision model; and an execution control module for inputting the compensation process parameter and the optimal recycled powder proportion into an additive manufacturing device for execution.

[0015] It should be understood that the general description above and the following detailed description are only exemplary and do not limit the application. BRIEF DESCRIPTION OF DRAWINGS

[0016] The accompanying drawings, which are incorporated herein and form part of the specification, illustrate one or more embodiments of the application and, together with the description, explain the principles of the application and the practices for making and using the same.

[0017] Figure 1 is a flowchart of an exemplary additive manufacturing recycled powder property evolution process compensation method according to some embodiments of the application.

[0018] Figure 2 is a structural schematic diagram of an exemplary three-dimensional profilometer according to some embodiments of the application.

[0019] Figure 3 is a flowchart of an exemplary method for calculating a decay coefficient of a powder according to some embodiments of the application.

[0020] Figure 4 is a flowchart of an exemplary SVM model building method according to some embodiments of the application.

[0021] Figure 5 is an exemplary optimization map of an SVM model according to some embodiments of the present application.

[0022] Figure 6 is an exemplary surface cladding quality map of uncompensated recycled powder at different cycle numbers according to some embodiments of the present application, wherein (a) to (j) are cycle numbers 1 to 10, respectively.

[0023] Figure 7 is an exemplary surface cladding quality map of 40% compensated recycled powder at different cycle numbers according to some embodiments of the present application, wherein (a) to (j) are cycle numbers 1 to 10, respectively.

[0024] Figure 8 is an exemplary surface cladding quality map of 50% compensated recycled powder at different cycle numbers according to some embodiments of the present application, wherein (a) to (j) are cycle numbers 1 to 10, respectively.

[0025] Figure 9 is an exemplary surface cladding quality map of 60% compensated recycled powder at different cycle numbers according to some embodiments of the present application, wherein (a) to (j) are cycle numbers 1 to 10, respectively.

[0026] Figure 10 is an exemplary SEM image of powder particle morphology of uncompensated recycled powder to 60% compensated recycled powder at different cycle numbers according to some embodiments of the present application.

[0027] Figure 11 is an exemplary morphology map of a component sample measured by a three-dimensional optical profiler according to some embodiments of the present application; wherein (a) is the boundary between the substrate and the cladding layer, (b) is the surface profile in the X direction, and (c) is the surface profile in the Y direction.

[0028] Figure 12 is an exemplary flowchart of a partitioned compensation method according to some embodiments of the present application.

[0029] Figure 13 is an exemplary module connection diagram of a process compensation system for additive manufacturing recycled powder property evolution according to some embodiments of the present application.

[0030] Figure 14 is an exemplary connection diagram of an electronic device according to some embodiments of the present application. DETAILED DESCRIPTION

[0031] Example implementations will now be described more fully with reference to the accompanying drawings. Example implementations can, however, be implemented in many different forms and should not be considered limited to the examples set forth herein; rather, these implementations are provided as a full and enabling disclosure of example implementations by being fully descriptive, and will fully convey the essence of the example implementations to those skilled in the art. Numbered features described can be combined in any suitable manner in one or more implementations. In the following description, numerous specific details are provided, in order to provide a thorough understanding of the example implementations.

[0032] Powder Bed Fusion (PBF) is an additive manufacturing technology driven by energy beams such as laser beams or electron beams, which melts metal or polymer powder materials layer by layer selectively to manufacture three-dimensional solid parts with complex geometrical structures. PBF technology is particularly suitable for preparing metal components with high density, high precision and complex structure, and is widely used in fields such as aerospace, biomedicine, energy and power, which have extremely high performance requirements.

[0033] At present, according to the different energy sources used, powder bed fusion technology is mainly divided into several types such as Selective Laser Melting (SLM), Laser Powder Bed Fusion (L-PBF), Electron Beam Melting (EBM), Selective Laser Sintering (SLS), etc. The process flow of powder bed fusion technology usually includes powder laying, energy beam scanning melting, interlayer lifting and post-processing, etc. The specific process is as follows: a powder laying mechanism is used to lay a layer of powder material with a thickness of tens to hundreds of microns on the forming platform to form a uniform powder bed, then according to the slicing data of the three-dimensional component model, a laser beam (Laser) or an electron beam (Electron Beam) is used to selectively scan, melt and solidify the current powder layer according to the pre-set scanning path, forming a solid cross section, and then the forming platform is lowered by one layer thickness unit after scanning a layer thick, the next layer of powder is laid, and the above steps are repeated to realize layer-by-layer construction, until the entire three-dimensional component is completed. After the component is completed, it is taken out, and after the post-processing steps such as heat treatment, support removal and surface treatment, the final product is obtained. The performance and quality of powder bed fusion technology are affected by many parameters such as energy beam power, scanning speed, scanning spacing, layer thickness, scanning strategy and powder characteristics.

[0034] The powder characteristics affect the energy transfer mechanism of the particle group (powder bed) under the action of the laser heat source, and then change the quality of the cladding forming. The thermal physical properties of the PBF powder layer, such as energy absorption rate, thermal conductivity, etc., will change with the changes of the characteristics of the powder layer, such as the bulk density, specific surface area and particle surface properties. At present, the numerical modeling technology is used to model and analyze the energy absorption and heat transfer behavior in the PBF powder layer, which provides a key theoretical basis for revealing the interaction mechanism of high-energy beam heat source and powder and makes an excellent demonstration. However, the discussion of powder characteristics related to most studies is mainly limited to powder particle size distribution, so there is still some limitation in their views. In addition, the highly dynamic moving molten pool formed after the powder is heated and melted, and the defect formation tendency will change with the powder characteristics, and then different processing window ranges are generated. The convective heat and mass transfer in the molten pool affect the temperature field at the solidification interface and affect the solidification conditions (temperature gradient and solidification speed), thereby changing the final solidification organization characteristics. However, due to the difficulty in evaluating the thermal properties of the powder bed and obtaining the information of the cladding process, the differences in energy transfer and phase transition behavior of the powder bed under the interaction of laser heat for different characteristic powders still lack deep understanding.

[0035] A uniform, dense and continuous powder bed is an important prerequisite for the quality assurance of PBF forming. Therefore, the present application clarifies the control factors of powder flow and dynamic powder laying process stability related to powder characteristics, to establish the correlation mechanism between particle flowability, particle morphology, surface properties and powder recycling times and new powder ratio, and to provide an important reference for process parameter compensation.

[0036] Figure 1 An exemplary process compensation method 10 for additive manufacturing recycled powder characteristic evolution of some embodiments of the present application is shown, referring to Figure 1 The process compensation method 10 provided has the following steps 11 to 14.

[0037] 11: Forming components with powders of different recycling times and new powder ratios, and detecting the characteristics of the recycled powders and components after forming, to establish the decay relationship between recycling times and powder characteristic parameters.

[0038] Among them, the present application selects metal powder as experimental material, then mixes according to the preset new powder and recycled powder ratio (i.e. new powder ratio, for example 40%, 50%, 60%), respectively under the condition of different recycling times (N=1-10) to carry out additive manufacturing component forming; then in turn, the characteristics of the recycled powders and the formed components after forming are detected; and then based on the experimental detection results, the decay law between the recycling times and the powder characteristic parameters is established, including the flowability, energy absorption rate, satellite powder ratio, oxide film thickness and oxygen content of the powder, and the surface roughness and density of the cladding layer of the component, and the relationship between the changes with the recycling times.

[0039] In particular, in step 11, the property detection includes detecting the flowability parameter of the powder, the oxide film thickness, the oxygen content parameter, and the surface roughness and density of the component cladding layer.

[0040] In particular, in detecting the flowability parameter of the powder, the Hall flowmeter test and the dynamic repose angle test are adopted. The specific implementation process of the Hall flowmeter is as follows.

[0041] (1) Place the Hall flowmeter on a stable test bench and keep it horizontal, and then check whether the instrument funnel, nozzle, and powder receiving cylinder are clean and free of residual powder, and confirm that the nozzle aperture meets the standard (usually φ2.5mm±0.05mm).

[0042] (2) Take the powder sample to be tested (usually 50g); the powder sample to be tested should be dried at 105°C for 1 hour and then cooled to room temperature, and then use a standard sieve to remove large impurities to ensure uniform powder particle size.

[0043] (3) Adjust the weighing balance to zero, prepare to record the time of the powder sample to be tested, and confirm that the stopwatch can be used normally.

[0044] (4) Place the powder receiving cylinder under the nozzle and ensure that it is centered, then quickly pour the weighed powder sample to be tested into the funnel, start the stopwatch at the same time, and stop the timing and record the time used (unit: seconds) after the powder flows freely until it is completely discharged.

[0045] (5) Calculate the Hall flow rate where 50 is the sample mass (g), is the flow-out time (s).

[0046] The specific implementation process of the dynamic repose angle test is as follows.

[0047] (1) Use a dynamic repose angle measuring instrument, take 40g of the powder sample to be tested, and ensure that the powder sample to be tested is dry, free of lumps and impurities; if the powder sample to be tested is easily affected by static electricity, it needs to be operated in a dry environment.

[0048] (2) For the rotating dynamic repose angle measuring instrument, set the cylinder rotation speed (for example, refer to 25rpm); for the vibrating device, set the vibration frequency and amplitude (for example, the vibration frequency is 50Hz, and the amplitude is 2mm).

[0049] (3) Set to collect the pile angle of the powder sample to be tested once per second; the collection of the pile angle of the powder sample to be tested can also be carried out according to the required sampling frequency of the instrument.

[0050] (4) Pour the powder sample to be tested evenly into the measuring container or cylinder and gently shake the container to make the surface of the powder sample flat and avoid the initial tilt affecting the measurement; during the measurement process, ensure that the container is stable to prevent it from sliding or tipping over during the test.

[0051] (5) Start the running program of the dynamic angle of repose measuring instrument to make the powder sample to be tested gradually form a natural flow and accumulation state. The dynamic angle of repose measuring instrument records the change of the tilt angle formed by the accumulation of the powder sample to be tested over time through a sensor or high-speed camera system, thereby collecting and recording the accumulation angle of the powder sample to be tested during the dynamic angle of repose stabilization period.

[0052] (6) Take the average value of the angle of repose of the powder sample under steady flow as the dynamic angle of repose; wherein, each powder sample is repeated at least 3 times, the average value is taken and the standard deviation is calculated.

[0053] Specifically, when detecting the oxide film thickness and oxygen content parameters of the powder, X-ray photoelectron spectroscopy (XPS) is used to test the oxide film thickness and oxygen content parameters of the powder; the process of X-ray photoelectron spectroscopy testing of oxide film thickness and oxygen content is as follows.

[0054] (1) Select a representative powder sample to be tested and spread it gently on the conductive tape to ensure that the powder is evenly distributed and to minimize agglomeration; in order to avoid the powder sample to be tested from adsorbing moisture and organic pollution, the powder sample to be tested is pretreated in a vacuum drying oven before testing (for example, vacuum drying at 60°C for 0.5 to 3 hours).

[0055] (2) Test in ultra-high vacuum (~ The experiment was conducted in an environment of mbar, using monochromatic Al Kα rays (1486.6 eV) as the excitation source, with the spot diameter controlled at around 400 μm and the energy resolution set at 0.1 eV to ensure the resolution of fine spectral peaks.

[0056] (3) First, a full spectrum (Survey Scan) is performed to obtain the types and relative contents of the main elements on the surface of the powder sample to be tested. Then, high-resolution spectral (High-resolution Scan) tests are performed on the characteristic energy level regions such as O 1s, Ti 2p, and Ni 2p, and the oxygen in different valence states (such as lattice oxygen, hydroxyl oxygen, and adsorbed oxygen) are distinguished by peak fitting.

[0057] (4) The ratio of the peak intensity of the substrate metal to its oxide in XPS is combined with the corresponding electron mean free path (IMFP) and geometric correction factor to quantitatively estimate the thickness of the oxide film using an improved film thickness calculation model (such as Tougaard or Beer-Lambert attenuation model).

[0058] (5) The average oxygen content of the surface oxide film of the powder sample to be measured is obtained by quantitatively analyzing the oxygen element by integrating the O 1s spectrum peak and combining the sensitivity factor (Sensitivity Factor) and the relative correction factor.

[0059] (6) To improve accuracy, the C 1s peak (284.8 eV) is used as the binding energy calibration standard, and the Shirley method is used to subtract the background to exclude the interference of the secondary electron background. Thus, the thickness range of the oxide film and the accurate value of the oxygen content are finally obtained.

[0060] Specifically, when detecting the surface roughness and density of the component cladding layer, a non-contact three-dimensional profiler is used to measure the surface roughness of the component cladding layer, and a drainage method is used to measure the density of the component. The style of the three-dimensional profiler is shown in Figure 2 , Figure 2 An exemplary structure schematic diagram of a three-dimensional profiler is shown.

[0061] The non-contact three-dimensional profiler measures the surface roughness of the component cladding layer as follows.

[0062] (1) Select a non-contact three-dimensional profiler, such as a three-dimensional optical profiler.

[0063] (2) Calibrate the non-contact three-dimensional profiler horizontally and calibrate the height according to the instrument instructions, and use a standard roughness block or a standard flat plate for measurement verification to ensure that the accuracy of the non-contact three-dimensional profiler meets the experimental requirements.

[0064] (3) Use a fiber-free cloth and alcohol to clean the surface of the component sample to remove surface oil, dust and particulate impurities.

[0065] (4) Place the component sample on the sample stage and use a clamp or vacuum suction to fix it to ensure that it does not move during measurement.

[0066] (5) Set the maximum measurement field of view to 2.5x2mm, the Z-direction measurement range to 10nm~10mm, and to ensure measurement accuracy, select different areas of each component sample to measure its surface roughness 3 times.

[0067] (6) Start the scanning program of the non-contact three-dimensional profilometer. The non-contact three-dimensional profilometer acquires the three-dimensional height information of the component sample surface along the specified path. Confirm that there are no obvious missing or abnormal points during the scanning process, and repeat the scan if necessary.

[0068] (7) Select the appropriate filter length according to the standard (e.g., ISO 4287 / 4288) to separate roughness and ripple.

[0069] (8) Calculate the surface roughness parameters of the component sample. Based on the three-dimensional height information of the component sample surface, calculate the average roughness of the entire component sample surface, calculate the root mean square value of the height of the entire component sample surface, and calculate the height difference between the highest and lowest points of the entire component sample surface. Then, perform three measurements and calculations on different areas of each component sample and calculate the average value and standard deviation of the three measurement results. Record the final surface roughness parameters of the component sample in this way.

[0070] The density of each component sample was determined by measuring it five times using the water displacement method and taking the average value as the density of the component sample.

[0071] In the actual implementation process, this application also uses scanning electron microscopy (SEM) to observe the surface morphology of powder particles and to count the proportion of satellite powder and hollow powder in the powder sample to be tested.

[0072] 12: Calculate the decay coefficient of the powder based on the decay relationship and construct a compensation strategy for the scanning task based on the decay coefficient, where the decay coefficient includes the flowability decay coefficient, the thermal absorption decay rate and the satellite powder ratio amplification coefficient.

[0073] Specifically, in step 12, the decay coefficient of the powder is calculated based on the decay relationship, including the following steps 12A-12C, with reference to... Figure 3 As shown, Figure 3 A schematic flowchart illustrating an exemplary method for calculating the decay coefficient of powder according to some embodiments of this application is shown.

[0074] 12A: Based on the decay relationship in step 11, define and calculate the powder flowability decay coefficient:

[0075] ,

[0076] in The coefficient represents the number of recycling cycles. This coefficient reflects the decrease in powder flowability as the number of recycling cycles increases. The larger the value of the flowability decay coefficient, the more obvious the flowability decline.

[0077] 12B: Based on the decay relationship in step 11, define and calculate the thermal absorption decay rate of the powder:

[0078] ,

[0079] wherein, is the energy beam absorption rate of the Nth recycling times, is the energy beam absorption rate of the new powder; this coefficient represents the decrease of effective laser energy absorption due to the thickening of the oxide film and the change of composition.

[0080] 12C: On the basis of the decay relationship in step 11, define and calculate the satellite powder proportion increase coefficient of the powder:

[0081] ,

[0082] wherein is the satellite powder proportion of the powder recycled in the Nth recycling times, is the satellite powder proportion of the new powder; this coefficient represents the increase of the proportion of small particles adhering to the surface of the recycled powder particles, reflecting the degree of deterioration of the powder morphology.

[0083] Thus, the flowability decay coefficient , the heat absorption decay rate and the satellite powder proportion increase coefficient are used to construct an adaptive compensation strategy for the process parameters of the scanning task. For example, when the flowability decay coefficient increases, it indicates that the flowability parameter decreases and the spreading uniformity of the powder on the powder bed decreases. In this way, the decrease in flowability is offset by constructing a powder layer thickness increase strategy, and the specific adjustment amplitude formula is: wherein is the increased powder layer thickness, i.e. the powder layer thickness increase amount; is the flowability decay coefficient, is the reference powder layer thickness; the compensated powder layer thickness is updated as: .

[0084] When the heat absorption decay rate increases, it indicates that the thickness of the surface oxide film increases, resulting in a decrease in the heat absorption rate of the effective energy beam. In order to ensure that the molten pool has sufficient input energy, the effective energy input deficiency is compensated by constructing an energy beam power improvement strategy, and the adjustment amplitude formula is: wherein, is the improved energy beam power, i.e. the energy beam power improvement amount; is the heat absorption decay rate, is the reference energy beam power; the compensated energy beam power is updated as: .

[0085] When the satellite powder proportion increase coefficient The increase indicates that the number of satellite powder adhering to the surface of the recovered powder particles increases, resulting in insufficient powder bed bulk density or the generation of voids, so as to improve the powder bed density or reduce the voids by reducing the powder laying speed during construction, and the adjustment range formula is: , wherein, is the reduced powder laying speed, i.e., the amount of reduction in the powder laying speed; is the satellite powder proportion increase coefficient, is the reference powder laying speed; and the updated powder laying speed after compensation is: .

[0086] In some examples of the present application, after constructing the compensation strategy for the scanning task according to the decay coefficient, in order to avoid excessive deviation of the compensated process parameters, which may lead to instability of the additive manufacturing process, the present application sets a corresponding compensation threshold range in each of the above-mentioned compensation strategies.

[0087] Specifically, in the powder layer thickness increase strategy, the adjustment threshold range is set to ±10μm; in the energy beam power increase strategy, the adjustment threshold range is set to ±30W; in the powder laying speed reduction strategy, the adjustment threshold range is set to ±50mm / s; when the calculated , or exceeds the set threshold range, the maximum or minimum value in the range is taken as the actual executed compensation value.

[0088] Thus, based on the above-mentioned step content, a powder decay relationship model is constructed.

[0089] 13: During the execution of the additive manufacturing scanning task, the compensation strategy is triggered every preset scanning layer number (for example, the preset scanning layer number can be 1-20 layers, which can be adjusted according to the device characteristics and the powder decay speed), so as to generate a compensation process parameter according to the current recycling number, and simultaneously predict the new powder proportion corresponding to the optimal forming quality index of the current recycling number by using the machine learning decision model.

[0090] Wherein, when the compensation strategy is triggered, the powder recycling number corresponding to the current scanning layer number is input into the constructed powder decay relationship model and combined with the powder characteristic parameters corresponding to the recycling number, to calculate the corresponding flowability decay coefficient , heat absorption decay rate and satellite powder proportion increase coefficient , and then according to the flowability decay coefficient , heat absorption decay rate and satellite powder proportion increase coefficient , the compensation strategy is called to calculate the corresponding process parameter compensation amount and the process parameter compensation amount is threshold constrained to automatically clip the compensation value exceeding the set threshold range to the maximum or minimum threshold, and finally the compensated process parameter is input to the additive manufacturing equipment; at the same time, the corresponding recycling number of the current scanning layer and the powder characteristic parameter corresponding to the recycling number are input to the machine learning decision model, and the machine learning decision model is used to predict the optimal forming quality index and the corresponding new powder ratio under the recycling number; thereby realizing the dual adaptive regulation and control of the process parameter and the powder supply.

[0091] Wherein, the machine learning decision model can adopt one of support vector machine (SVM) model, logistic regression model, decision tree model, random forest model, etc., in this application, the machine learning decision model refers to the SVM model, for example, as shown in Figure 4 and Figure 5 , Figure 4 shows an exemplary SVM model construction method of some embodiments of the application, Figure 5 shows an exemplary SVM model optimization atlas of some embodiments of the application. The construction process of the SVM model consists of the following steps 13A to 13D.

[0092] 13A: Collecting a first preset number of recycling numbers, new powder ratios, powder characteristic parameters, corresponding component surface roughness and density sample parameters to construct a data set; wherein the first preset number of sample parameters can be set by the debugging personnel according to the actual demand, or the experimental data generated in step 11 can be directly used as the sample parameters.

[0093] 13B: According to the sample parameters, interpolation and noise adding method is used to construct enhanced sample parameters; wherein in the process of constructing enhanced sample parameters, a preset number (exemplary num_augments=3) of new sample parameters are generated for adjacent two samples, and then the new powder ratio, the corresponding component surface roughness and the density of the new sample parameters are calculated using linear interpolation, while small Gaussian noise (exemplary noise_level=0.03) is added to the interpolation result to simulate experimental measurement error, thereby the enhanced sample parameters can increase the data density while maintaining the original distribution characteristics.

[0094] 13C: The constructed enhanced sample parameters are added to the data set to form an enhanced data set, and the enhanced data set is divided into training set and test set by random index method (fixed random seed rng(42)) according to the preset proportion (exemplary 80%:20% proportion); wherein the training set is used for model fitting, and the test set is used for evaluating the model generalization ability.

[0095] 13D: Adopting the input data standardization method (Standardize=true), eliminating the dimensional difference of the enhanced data set and setting the kernel function of the machine learning decision model as the Gaussian radial basis function (RBF) to capture the nonlinear relationship between the recycling times, the proportion of new powder and the forming quality index, and then generating and training the SVM model, while ensuring that the surface roughness of the component cladding layer is not more than the preset value (the surface roughness is less than or equal to 10 μm), at the same time, the proportion of recycled powder is maximized (greater than or equal to 60%), which is set as the constraint condition of the SVM model, and the mean square error (MSE) is used as the performance evaluation index, the predicted value and the true value of the test set are compared, so as to evaluate the prediction performance of the SVM model.

[0096] 14: Input the compensation process parameters and the proportion of new powder into the additive manufacturing equipment for execution.

[0097] By adopting the above technical scheme, by detecting and analyzing the evolution of powder characteristics of different recycling times and establishing the decay relationship of flowability, heat absorption rate and satellite powder proportion, the decay trend of the powder can be identified, so as to adaptively compensate the process parameters in the additive manufacturing process, so that the recycled powder can still maintain high forming quality in multiple recycling use, and realize efficient use of powder resources; and using the machine learning decision model to predict the optimal proportion of new powder under the current recycling times, the proportion of recycled powder use can be maximized while ensuring the forming quality of the component, reducing the consumption of new powder, reducing the production cost, and realizing the optimization of economic benefit and material utilization.

[0098] Therefore, in some examples of the present application, the change rule of powder characteristics with the number of cycles is obtained according to experimental data, and specific results are shown in Tables 1 to 4 below:

[0099] Table 1: Decay rule of key performance parameters of recycled powder (uncompensated recycled powder)

[0100]

[0101] Table 2: Decay rule of key performance parameters of recycled powder (recycled powder after compensation 40%)

[0102]

[0103] Table 3: Decay rule of key performance parameters of recycled powder (recycled powder after compensation 50%)

[0104]

[0105] Table 4: Decay rule of key performance parameters of recycled powder (recycled powder after compensation 60%)

[0106]

[0107] Therefore, the surface roughness measurement data of this application are referenced Figures 6 to 9 As shown, Figure 6 This invention illustrates an exemplary schematic diagram of the surface cladding quality of uncompensated recycled powder with different cycle numbers according to some embodiments of this application, wherein... Figure 6 (a) to Figure 6 (j) represents the number of iterations from 1 to 10; Figure 7 This application illustrates an exemplary schematic diagram of 40% surface cladding quality of compensated recycled powder with different cycle numbers, according to some embodiments of this application. Figure 7 (a) to Figure 7 (j) represents the number of iterations from 1 to 10; Figure 8 This illustration shows an exemplary schematic diagram of 50% surface cladding quality of compensated recycled powder with different cycle numbers according to some embodiments of this application, wherein... Figure 8 (a) to Figure 8 (j) represents the number of iterations from 1 to 10; Figure 9 This invention illustrates an exemplary schematic diagram of 60% surface cladding quality of compensated recycled powder with different cycle numbers according to some embodiments of this application, wherein... Figure 9 (a) to Figure 9 The (j) values ​​represent the number of iterations from 1 to 10.

[0108] Therefore, the density measurement data of this application are shown in Table 5 below:

[0109] Table 5: Packing density at different cycle numbers

[0110]

[0111] In this application, scanning electron microscopy was used to photograph and characterize the powder morphology, and the specific morphology of the powder particles is as follows: Figure 10 As shown, Figure 10 The diagram shows an exemplary electron microscope image of the particle morphology of uncompensated recovered powder to 60% compensated recovered powder at different cycle counts according to some embodiments of this application.

[0112] In this application, a three-dimensional optical profilometer is used to measure and characterize the morphology of the component, and the specific morphology of the component is as follows: Figure 11 As shown, Figure 11 The illustration shows a schematic diagram of the morphology of a component sample measured by an exemplary three-dimensional optical profilometer according to some embodiments of this application, wherein, Figure 11 (a) represents the boundary between the substrate and the cladding layer. Figure 11 (b) is the surface profile in the X direction. Figure 11 (c) is the surface profile in the Y direction.

[0113] Reference Figure 12 as shown, Figure 12 A flowchart of an exemplary zoning compensation method is shown to illustrate some embodiments of the present application. In some embodiments of the present application, the method (specifically step 13) further comprises the following steps 13E to 13G when the compensation strategy is triggered.

[0114] 13E: Sub-regionally divide the forming area of the additive manufacturing equipment according to the scanning path of the scanning task to form a plurality of zones.

[0115] Specifically, in step 13E, the scanning area of the additive manufacturing equipment is divided according to the path of the scanning task to form a plurality of sub-regions (zones), each zone corresponding to a certain number of scanning tracks or scanning grid units; wherein the division of the sub-regions can be determined according to the scanning path, the component geometry, and the energy beam working area of the additive manufacturing equipment, to ensure that the powder layering and energy beam cladding process of each zone are independently controllable; and the current scanning layer number and the cumulative number of powder recycling of each zone are recorded, and the number of powder recycling can be obtained from the powder conveying system data of the additive manufacturing equipment or the historical scanning records.

[0116] For example, the X-Y plane of the forming area is divided into several squares according to 10mm x 10mm, and each square is a zone; or a linear zone can also be formed according to each scanning track of the scanning path.

[0117] 13F: Obtain the number of powder recycling of each zone and independently calculate the flowability decay coefficient, heat absorption decay rate, and satellite powder proportion increase coefficient of the powder of each zone.

[0118] Specifically, in step 13F, for the powder of each zone, the flowability decay coefficient , heat absorption decay rate , and satellite powder proportion increase coefficient are independently calculated according to the cumulative number of recycling, and then the corresponding powder layer thickness increase , energy beam power increase , and powder laying speed reduction are calculated according to the flowability decay coefficient , heat absorption decay rate , and satellite powder proportion increase coefficient of each zone, while the powder layer thickness increase , energy beam power increase , and powder laying speed reduction are threshold-constrained to ensure that the compensation values do not exceed the set maximum / minimum range,

[0119] 13G: Execute the corresponding powder layer thickness increase strategy and / or energy beam power increase strategy and / or powder layer speed reduction strategy independently for each partition.

[0120] Specifically, in step 13G, the calculated increase in the powder layer thickness is... Energy beam power increase And the amount of powder spreading speed reduction The baseline process parameters for each partition are added to generate the compensated process parameters for each partition. The compensated process parameters for each partition are then input into the control system of the additive manufacturing equipment, and adjustments are made accordingly when the scanning task reaches that partition.

[0121] In actual implementation, after each preset number of scanning layers is completed, steps 13E to 13G are triggered again to re-count and update the number of powder recovery times for each zone. , , and compensation amount , , ; Repeat the process until the scanning task is completely finished.

[0122] By employing the above technical solutions to locally compensate for powder in different zones, it is possible to regulate powder areas with poor flowability, ensuring uniform powder distribution and preventing localized sparse or accumulated powder layers; it is possible to regulate powder areas with low heat absorption rates, improving powder melting sufficiency and preventing localized over-melting or under-melting; and it is possible to regulate powder areas with increased satellite powder quantity, improving the quality of the formed surface.

[0123] In some embodiments of this application, reference is made to Figure 13 As shown, Figure 13 The diagram illustrates a modular connection of an exemplary additive manufacturing process compensation system 20 for the evolution of characteristics of recycled powder, according to some embodiments of this application. The process compensation system 20 is configured to have a decay building module 201, a strategy building module 202, a process compensation module 203, a machine learning module 204, and an execution control module 205.

[0124] The decay construction module 201 is configured to form components from powders with different recycling times and new powder ratios, and to perform characteristic testing on the recycled powder and components after forming, thereby establishing a decay relationship between the recycling times and powder characteristic parameters.

[0125] The strategy construction module 202 is configured to calculate the decay coefficient of the powder based on the decay relationship and construct a compensation strategy for the scanning task based on the decay coefficient, wherein the decay coefficient includes the flowability decay coefficient, the thermal absorption decay rate and the satellite powder ratio amplification coefficient.

[0126] The process compensation module 203 is configured to trigger a compensation strategy every preset number of scanning layers during execution of the additive manufacturing scanning task, to generate a compensation process parameter according to the current recycling number.

[0127] The machine learning module 204 is configured to build a machine learning decision model and predict an optimal recycled powder ratio corresponding to the current recycling number by using the machine learning decision model.

[0128] The execution control module 205 is configured to input the compensation process parameter and the optimal recycled powder ratio to the additive manufacturing device for execution.

[0129] In some embodiments, with reference to Figure 14 as shown, Figure 14 A connection diagram of an electronic device for implementing the embodiments of the present application is shown. The electronic device 30 includes a memory 301 and a processor 302, and the memory 301 stores a computer program capable of running on the processor 302. The processor 302 implements the method in the above embodiments when executing the computer program. The number of the memory 301 and the processor 302 can be one or more.

[0130] The electronic device 30 further includes a communication interface 303 for communicating with external devices and performing data transmission.

[0131] If the memory 301, the processor 302 and the communication interface 303 are independently implemented, the memory 301, the processor 302 and the communication interface 303 can be connected to each other through a bus and complete communication therebetween.

[0132] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 14 only one thick line is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0133] Optionally, in specific implementation, if the memory 301, the processor 302 and the communication interface 303 are integrated on a chip, the memory 301, the processor 302 and the communication interface 303 can complete communication therebetween through an internal interface.

[0134] The embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the program is executed by the processor 302 to implement the method provided in the embodiment of the present application.

[0135] The embodiment of the present application further provides a chip, which comprises the processor 302, is used for calling and running the instruction stored in the memory 301 from the memory 301, so that the communication device installed with the chip executes the method provided in the embodiment of the present application.

[0136] The embodiment of the present application further provides a chip, which comprises an input interface, an output interface, a processor 302 and a memory 301, and the input interface, the output interface, the processor 302 and the memory 301 are connected through internal connection paths, the processor 302 is used for executing the code in the memory 301, and when the code is executed, the processor 302 is used for executing the method provided in the embodiment of the present application.

[0137] It should be understood that the processor 302 described above can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. It should be noted that the processor 302 can be a processor supporting an advanced RISC machine (ARM) architecture.

[0138] Further, the aforementioned memory 301 can include a read-only memory and a random access memory, and can also include a non-volatile random access memory. The memory 301 can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories. Among them, the non-volatile memory can include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory can include a random access memory (RAM) used as an external cache. By way of example, and not limitation, many forms of RAM can be used. For example, a static random access memory (SRAM), a dynamic random access memory (DRAM), a synchronous dynamic random access memory (SDRAM), a double data rate synchronous dynamic random access memory (DDR SDRAM), an enhanced synchronous dynamic random access memory (ESDRAM), a synchlink dynamic random access memory (SLDRAM), and a direct rambus random access memory (DRRAM) can be used.

[0139] Other embodiments of the application will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. It is intended that the specification and examples be considered as exemplary only, with the true scope of the application being indicated by the following claims.

Claims

1. A process compensation method for additive manufacturing recycled powder property evolution, characterized in that, The method comprises: component forming on powders with different recycling times and new powder proportions, and property detection on the formed components and recycled powders, to establish a decay relationship between the recycling times and the powder property parameters; calculating decay coefficients of the powders according to the decay relationship between the recycling times and the powder property parameters, and constructing a compensation strategy for the scanning task according to the decay coefficients, wherein the decay coefficients include a flowability decay coefficient, a heat absorption decay rate, and a satellite powder proportion increase coefficient; during the execution of the additive manufacturing scanning task, triggering the compensation strategy every preset scanning layer to generate compensation process parameters according to the current recycling time, and simultaneously predicting the new powder proportion corresponding to the optimal forming quality index of the current recycling time by using a machine learning decision model; and inputting the compensation process parameters and the new powder proportion into an additive manufacturing device for execution; wherein the calculation of the decay coefficients of the powders according to the decay relationship comprises: calculating the flowability decay coefficient of the powders according to the decay relationship: wherein is the number of recoveries; calculating the heat absorption decay rate of the powders according to the decay relationship: wherein, is the energy beam absorptivity for the Nth recycling pass, is the energy beam absorptivity for fresh powder; calculating the satellite powder proportion increase coefficient of the powders according to the decay relationship: wherein the proportion of satellite powder recovered for the Nth recovery is the proportion of satellite powder in the new powder; the construction of the compensation strategy for the scanning task according to the decay coefficients comprises: According to the flowability attenuation coefficient, a powder layer thickness increasing strategy is constructed, and a regulation range formula of the powder layer thickness increasing strategy is: wherein is the increased powder layer thickness, is the flowability attenuation coefficient, is the reference powder layer thickness; According to the heat absorption attenuation rate, an energy beam power promotion strategy is constructed, and a regulation amplitude formula of the energy beam power promotion strategy is: wherein, is the promoted energy beam power, is the heat absorption attenuation rate, is a reference energy beam power; According to the satellite powder proportion amplification coefficient, a powder laying speed reduction strategy is constructed, and a regulation range formula of the powder laying speed reduction strategy is: wherein, is the reduced powder laying speed, is the satellite powder proportion amplification coefficient, is the reference powder laying speed.

2. The method of claim 1, wherein, the property detection on the formed components and recycled powders comprises: detecting the flowability parameters of the powders by using a Hall flowmeter test and a dynamic angle of repose test; detecting the oxide film thickness and oxygen content parameters of the powders by using an X-ray photoelectron spectroscopy test; measuring the surface roughness of the cladding layer of the component by using a non-contact three-dimensional profilometer and measuring the density of the component by using a drainage method.

3. The method of claim 1, wherein, after the construction of the compensation strategy for the scanning task according to the decay coefficients, the method comprises: in the powder layer thickness increasing strategy, setting the adjustment threshold range to ±10 μm; in the energy beam power increasing strategy, setting the adjustment threshold range to ±30 W; in the powder laying speed decreasing strategy, setting the adjustment threshold range to ±50 mm / s.

4. The method of claim 1, wherein, when the compensation strategy is triggered, the method further comprises: dividing the forming area of the additive manufacturing component into sub-areas according to the scanning path of the scanning task to form multiple partitions; obtaining the powder recycling times of each partition and independently calculating the flowability decay coefficient, the heat absorption decay rate, and the satellite powder proportion increase coefficient of the powder of each partition; independently executing the corresponding powder layer thickness increasing strategy and / or energy beam power increasing strategy and / or powder laying speed decreasing strategy for each partition.

5. The method of claim 1, wherein, when the compensation strategy for the scanning task is constructed, the method further comprises: collecting sample parameters of a first preset number of recycling times, new powder proportions, powder property parameters, corresponding component surface roughness, and density, and constructing a data set; using a machine learning method to iteratively train the data set to construct a machine learning decision model, and the process of iterative training comprises: taking the recycling times and the powder property parameters as inputs, taking the forming quality index and the new powder proportion as outputs, and setting constraint conditions.

6. The method of claim 5, wherein, the construction of the data set comprises: generating a second preset number of new sample parameters from adjacent two sample parameters; A new powder ratio, a corresponding component surface roughness and a density of a new sample parameter are calculated using linear interpolation, and a small amount of Gaussian noise is added to simulate experimental measurement error to construct an enhanced sample parameter; The constructed enhanced sample parameter is added to a data set to form an enhanced data set, and the data set is divided into a training set and a test set according to a predetermined proportion by using a random shuffling index method.

7. The method according to claim 5 or 6, characterized in that, The machine learning decision model is constructed, including: An input data standardization method is used to eliminate the dimensional differences of the enhanced data set; The kernel function of the machine learning decision model is set as a radial basis function; The surface roughness of the component cladding layer is ensured to be less than a predetermined value, and the maximum recovery powder ratio is set as a constraint condition.

8. A process compensation system for the evolution of the properties of a recycled powder for additive manufacturing using the method according to any one of claims 1 to 7, characterized in that, It includes: A decay construction module for component forming of powders with different recycling times and new powder ratios, and for characteristic detection of the powders and components after forming, to establish a decay relationship between the recycling times and the powder characteristic parameters; A strategy construction module for calculating a decay coefficient of the powder according to the decay relationship between the recycling times and the powder characteristic parameters, and for constructing a compensation strategy for the scanning task according to the decay coefficient, wherein the decay coefficient includes a flowability decay coefficient, a heat absorption decay rate and a satellite powder proportion increase coefficient; A process compensation module for triggering the compensation strategy every predetermined number of scanning layers during the execution of the additive manufacturing scanning task to generate a compensation process parameter according to the current recycling time; A machine learning module for constructing a machine learning decision model and predicting a new powder ratio corresponding to an optimal forming quality indicator of the current recycling time using the machine learning decision model; An execution control module for inputting the compensation process parameter and the new powder ratio to an additive manufacturing device for execution.

Images