A fast metal printing system and method based on machine vision feedback

Abstract

The present application discloses a rapid metal printing system and method based on machine vision feedback, and the present application belongs to the field of 3D printing technology. The system includes: a data acquisition module, which is used to determine the surface height difference, temperature gradient and density fluctuation of each preset area in the first layer of the rapid metal printing object; a defect probability prediction module, which is used to obtain the defect probability of each preset area in the first layer of the rapid metal printing object; an adjustment scheme determination module, which is used to take the preset area corresponding to the defect probability exceeding the preset defect probability threshold as the target preset area, and determine the defect type and adjustment scheme of the target preset area; a printing adjustment module, which is used to print and adjust the first layer of the rapid metal printing object. The accurate defect prediction, real-time adjustment and automatic control of this scheme can not only improve the quality and efficiency of metal printing, but also greatly reduce resource waste, reduce costs, and enhance the flexibility and intelligence level of the production process.

Description

Technical Field

[0001] The present application belongs to the field of 3D printing technology, and specifically relates to a rapid metal printing system and method based on machine vision feedback. Background Art

[0002] With the rapid development of the manufacturing industry, metal 3D printing technology has attracted much attention because of its ability to directly convert digital models into metal parts. This technology not only significantly shortens the production cycle and improves manufacturing efficiency, but also reduces costs by saving materials and reducing manual intervention. More importantly, metal 3D printing technology can produce more precise and complex metal parts, improve product quality, and meet the personalized needs of customers. However, while achieving these advantages, existing metal 3D printing technology also faces a series of challenges.

[0003] Traditional metal 3D printing systems use metal powder or wire as raw materials and build parts through steps such as powder spreading, melting / sintering, and layer-by-layer stacking. First, a layer of metal powder is spread, and then a high-energy heat source such as a laser or electron beam is used to selectively melt or sinter specific areas of the powder according to the 3D model data. After completing a layer, the platform is lowered and new powder is spread, and this process is repeated until the part is completed. Subsequent processing is required after printing, such as support removal, grinding and polishing, and heat treatment to improve part quality and performance.

[0004] However, existing metal 3D printing systems often lack real-time feedback mechanisms and are unable to monitor and correct abnormalities during the printing process in real time, which may affect the quality and accuracy of the final product. In addition, without machine vision feedback, operators need to manually monitor and adjust the printing process, which not only increases labor costs, but may also cause errors and defects due to human factors. Therefore, there is an urgent need for a fast metal printing system based on machine vision feedback that can perform accurate defect prediction, real-time adjustment and automatic control, which can not only improve the quality and efficiency of metal printing, but also significantly reduce resource waste, reduce costs, and enhance the flexibility and intelligence level of the production process. Summary of the invention

[0005] The embodiments of the present application provide a rapid metal printing system and method based on machine vision feedback, which solves the problem that existing metal 3D printing systems often lack a real-time feedback mechanism and cannot monitor and correct abnormal conditions during the printing process in real time, which may affect the quality and accuracy of the final product. In addition, in the absence of machine vision feedback, the operator needs to manually monitor and adjust the printing process, which not only increases labor costs, but may also cause errors and defects due to human factors.

[0006] In a first aspect, an embodiment of the present application provides a rapid metal printing system based on machine vision feedback, the system comprising:

[0007] A data acquisition module, used to obtain surface height data, temperature data and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determine the surface height difference according to the surface height data and the preset printing standard, determine the temperature gradient according to the temperature data and the preset printing standard, and determine the density fluctuation according to the powder deposition data and the preset printing standard;

[0008] A defect probability prediction module, used to obtain the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and a preset defect prediction formula;

[0009] An adjustment scheme determination module is used for, if there is a defect probability exceeding a preset defect probability threshold, taking a preset area corresponding to the defect probability exceeding the preset defect probability threshold as a target preset area, obtaining printing parameters, temperature and environmental data of the first layer of the rapid metal printing object, and obtaining machine vision feedback data, powder data and local geometric data of the target preset area, inputting the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data into a preset printing adjustment model, and obtaining a defect type and an adjustment scheme of the target preset area; wherein the number of the target preset area is at least one;

[0010] The printing adjustment module is used to control the printing device to make printing adjustments to the first layer of the rapid metal printing object according to the adjustment scheme of each target preset area.

[0011] Furthermore, the preset defect prediction formula is:

[0012]

[0013] in, is the defect probability; The preset surface height difference weight is the surface height difference; is the preset temperature gradient weight; is the temperature gradient; is the preset density fluctuation weight; is the density fluctuation; The coordinates of the center point of each preset area.

[0014] Furthermore, the system further includes a printing continuous adjustment module, and the printing continuous adjustment module is used to:

[0015] The surface height data, temperature data and powder deposition data of each preset area in each printing layer of the rapid metal printing object transmitted by the machine vision feedback system are respectively obtained according to the sequence of the printing layers, and the defect probability of each preset area in each printing layer is determined based on the above data. In the case of a defect probability exceeding a preset defect probability threshold, the preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as the target preset area, and the defect type and adjustment plan of the target preset area are determined. Based on the adjustment plan, the printing device is controlled to perform printing adjustments on each printing layer in the sequence until the printing adjustment of all printing layers is completed.

[0016] Furthermore, the system further comprises a laser power determination module, wherein the laser power determination module is used to:

[0017] Powder particle size data and powder mass data are obtained, and a preset initial laser power is adjusted according to the powder particle size data, the powder mass data and a preset laser power calculation formula to obtain an adjusted laser power, and printing parameters are updated according to the adjusted laser power.

[0018] Furthermore, the preset laser power calculation formula is:

[0019]

[0020] in, To adjust the laser power; is the preset particle size adjustment factor; is the powder particle size data; is the preset quality adjustment factor; is the powder quality data; is the preset laser power adjustment factor; is the preset initial laser power.

[0021] Furthermore, the training process of the preset print adjustment model is:

[0022] Acquire historical adjustment records, determine historical defect types, historical adjustment solutions, historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data according to the historical adjustment records, and create a first data set according to the historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data;

[0023] labeling the defect type label of the first data set according to the historical defect type, and labeling the adjustment solution label of the first data set according to the historical adjustment solution;

[0024] A printing adjustment model is constructed, and the printing adjustment model is trained according to the first data set, the defect type label, and the adjustment solution label until the printing adjustment model reaches a preset training standard.

[0025] Furthermore, the system further comprises a fault detection module, wherein the fault detection module is used to:

[0026] Acquire the operating parameters of the printing device in real time, input the operating parameters into a preset fault diagnosis and repair model, and determine whether there is a fault risk;

[0027] If there is a risk of failure, the fault type is output through a preset fault diagnosis and repair model, and the fault type is transmitted to the control center;

[0028] Accordingly, the system further comprises a maintenance detection module, and the maintenance detection module is used to:

[0029] If there is no risk of failure, the health score is output through the preset fault diagnosis and repair model;

[0030] If the health score is lower than a preset health score threshold, a maintenance plan is output through a preset fault diagnosis and repair model, and the maintenance plan is transmitted to a control center.

[0031] In a second aspect, an embodiment of the present application provides a rapid metal printing method based on machine vision feedback, the method comprising:

[0032] Obtaining surface height data, temperature data, and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determining a surface height difference according to the surface height data and a preset printing standard, determining a temperature gradient according to the temperature data and the preset printing standard, and determining a density fluctuation according to the powder deposition data and the preset printing standard;

[0033] Obtaining the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and a preset defect prediction formula;

[0034] If there is a defect probability exceeding a preset defect probability threshold, a preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as a target preset area, and printing parameters, temperature and environmental data of the first layer of the rapid metal printing object are obtained, and machine vision feedback data, powder data and local geometric data of the target preset area are obtained, and the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data are input into a preset printing adjustment model to obtain the defect type and adjustment plan of the target preset area; wherein the number of the target preset area is at least one;

[0035] The printing device is controlled to make printing adjustments to the first layer of the rapid metal printing object according to the adjustment scheme of each target preset area.

[0036] In a third aspect, an embodiment of the present application provides an electronic device, which includes a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the method described in the second aspect.

[0037] In a fourth aspect, an embodiment of the present application provides a readable storage medium, on which a program or instruction is stored, and when the program or instruction is executed by a processor, the steps of the method described in the second aspect are implemented.

[0038] In an embodiment of the present application, a data acquisition module is used to obtain surface height data, temperature data and powder deposition data of each preset area in the first layer of a rapid metal printing object transmitted by a machine vision feedback system, determine the surface height difference according to the surface height data and the preset printing standard, determine the temperature gradient according to the temperature data and the preset printing standard, and determine the density fluctuation according to the powder deposition data and the preset printing standard; a defect probability prediction module is used to obtain the defect probability of each preset area in the first layer of a rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and the preset defect prediction formula; an adjustment scheme determination module is used to determine the defect probability of each preset area in the first layer of a rapid metal printing object if there is a defect probability threshold value exceeding the preset defect probability threshold value. The defect probability of the value is determined, and the preset area corresponding to the defect probability exceeding the preset defect probability threshold is taken as the target preset area, then the printing parameters, temperature and environmental data of the first layer of the rapid metal printing object are obtained, and the machine vision feedback data, powder data and local geometric data of the target preset area are obtained, and the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data are input into the preset printing adjustment model to obtain the defect type and adjustment plan of the target preset area; wherein the number of the target preset areas is at least one; the printing adjustment module is used to control the printing equipment to adjust the printing of the first layer of the rapid metal printing object according to the adjustment plan of each target preset area. Through the above-mentioned rapid metal printing system based on machine vision feedback, accurate defect prediction, real-time adjustment and automatic control can not only improve the quality and efficiency of metal printing, but also greatly reduce resource waste, reduce costs, and enhance the flexibility and intelligence level of the production process. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is a structural schematic diagram of a rapid metal printing system based on machine vision feedback provided in Example 1 of the present application;

[0040] Figure 2It is a structural schematic diagram of a rapid metal printing system based on machine vision feedback provided in Example 2 of the present application;

[0041] Figure 3 It is a flowchart of a rapid metal printing method based on machine vision feedback provided in Example 3 of the present application;

[0042] Figure 4 It is a schematic diagram of the structure of the electronic device provided in Example 4 of the present application. DETAILED DESCRIPTION

[0043] In order to make the purpose, technical scheme and advantages of the present application clearer, the specific embodiments of the present application are further described in detail below in conjunction with the accompanying drawings. It is understood that the specific embodiments described herein are only used to explain the present application, rather than to limit the present application. It should also be noted that, for the convenience of description, only part of the present application is shown in the accompanying drawings, but not all of the content. Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow chart describes each operation (or step) as a sequential process, many of the operations therein can be implemented in parallel, concurrently or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but it can also have additional steps not included in the accompanying drawings. The process can correspond to a method, a function, a procedure, a subroutine, a subprogram, etc.

[0044] The following will be combined with the drawings in the embodiments of the present application to clearly describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments in the present application belong to the scope of protection of this application.

[0045] The terms "first", "second", etc. in the specification and claims of the present application are used to distinguish similar objects, and are not used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described here, and the objects distinguished by "first", "second", etc. are generally of one type, and the number of objects is not limited. For example, the first object can be one or more. In addition, "and / or" in the specification and claims represents at least one of the connected objects, and the character " / " generally indicates that the objects associated with each other are in an "or" relationship.

[0046] In combination with the accompanying drawings, an RSMC chip, a chip multi-stage startup method and a Beidou communication navigation device provided in the embodiments of the present application are described in detail through specific embodiments and their application scenarios.

[0047] Embodiment 1

[0048] Figure 1 Schematic diagram of the structure of the rapid metal printing system based on machine vision feedback provided in the first embodiment of the present application. Figure 1 As shown, specifically including the following:

[0049] The data acquisition module 101 is used to obtain the surface height data, temperature data and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determine the surface height difference according to the surface height data and the preset printing standard, determine the temperature gradient according to the temperature data and the preset printing standard, and determine the density fluctuation according to the powder deposition data and the preset printing standard;

[0050] A defect probability prediction module 102, for obtaining the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and a preset defect prediction formula;

[0051] The adjustment scheme determination module 103 is used for, if there is a defect probability exceeding a preset defect probability threshold, taking a preset area corresponding to the defect probability exceeding the preset defect probability threshold as a target preset area, obtaining the printing parameters, temperature and environmental data of the first layer of the rapid metal printing object, and obtaining the machine vision feedback data, powder data and local geometric data of the target preset area, inputting the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data into a preset printing adjustment model, and obtaining the defect type and adjustment scheme of the target preset area; wherein the number of the target preset area is at least one;

[0052] The printing adjustment module 104 is used to control the printing device to perform printing adjustment on the first layer of the rapid metal printing object according to the adjustment scheme of each target preset area.

[0053] In this embodiment, this solution can be applied to various types of metal materials, but it needs to be adjusted according to the characteristics of different materials. Each metal material will have different thermal behavior, physical properties and processing characteristics during the printing process, and these factors may affect the performance of data such as surface height difference, temperature gradient and powder deposition during the printing process. Specifically, aluminum alloy: aluminum alloy has higher thermal conductivity and lower melting point, which may make it more prone to heat accumulation, warping or surface quality problems during the printing process. Therefore, surface height difference, temperature gradient, powder deposition, etc. need to be adjusted according to the characteristics of aluminum alloy. For example, aluminum alloy printing may require stricter temperature control and lower printing speeds to avoid excessive heat accumulation.

[0054] Stainless steel: Stainless steel has a high melting point and low thermal conductivity, and is relatively prone to large temperature gradients and thermal stresses. Therefore, for the printing of stainless steel materials, the prediction of temperature gradients and powder deposition data may be more critical. Special attention should be paid to the cooling rate of the material during the printing process to avoid cracks or warping caused by uneven thermal expansion.

[0055] Titanium alloy: Titanium alloy has low thermal conductivity and high reflectivity, which easily leads to large temperature gradient and unstable molten pool. Therefore, the control of temperature and powder deposition is particularly important. The surface quality and molten pool stability of such materials directly affect the printing accuracy and the final material properties.

[0056] Nickel-based alloys: Nickel-based alloys have a high melting point and poor thermal conductivity. They are usually used in high-temperature and high-load applications, such as aircraft engines. Therefore, during their printing process, the control of temperature and surface quality is particularly critical, and the printing process needs to be adjusted according to the printing parameters and material properties.

[0057] Machine vision feedback system can refer to a system that uses cameras, sensors and image processing technology to monitor and analyze the printing process in real time. In metal printing, the machine vision system is used to capture image data during the printing process, such as the printing surface, shape, quality, temperature, etc., and feed this data back to the system for real-time analysis.

[0058] Rapid metal printing objects can be objects made using metal 3D printing technology. Usually, metal powder or metal wire is deposited layer by layer on the print bed through additive manufacturing methods, and the material is heated and fused at each layer to finally complete the object.

[0059] Each preset area can refer to the system dividing the printing area into multiple areas on each layer of the printed object. The division of these areas helps to monitor and control the printing quality of each part more carefully. Each area will be individually detected for its surface height, temperature and powder deposition. It is usually divided according to the size of the print bed, the shape of the printed object and the area that needs to be monitored.

[0060] Surface height data can be obtained by measuring the height of each area on the surface of the printed object through a machine vision system or a laser scanner or other device to evaluate the flatness and quality of the printed surface.

[0061] Temperature data can be the temperature information of different areas in the printing process collected by temperature sensors. Especially in the metal printing process, temperature is critical to the printing quality.

[0062] The powder deposition data may be measurement data of the actual amount and distribution of powder deposited layer by layer on the substrate by the printing device in each printing area.

[0063] Preset printing standards can refer to a series of standard parameters set during the design phase based on material properties, printing process, quality requirements, etc. These standards define the ideal printing process, such as temperature range, powder deposition uniformity, surface height tolerance, etc.

[0064] Surface height difference refers to the difference between the maximum and minimum surface heights in a specific area. This is an important indicator for evaluating the quality of printed surfaces. Large surface height differences may indicate uneven printing, defects, or design problems.

[0065] Temperature gradient refers to the temperature variation at different points in a specific area. It reflects the uneven temperature distribution during the printing process. Large temperature gradients may cause problems such as thermal stress and deformation.

[0066] Density fluctuations can be differences in the density distribution of materials on the surface or in different areas of a metal printed object. In metal 3D printing, factors such as powder deposition, laser scanning, heating and cooling may lead to uneven density in local areas, which may affect the mechanical properties, thermal conductivity and surface quality of the printed object.

[0067] Machine vision systems can be used to collect the height, temperature, and powder deposition of the printed object surface. For example, surface topography data can be obtained using laser scanners, structured light sensors, or other height measurement tools. These devices can scan the surface of the object and generate a height map (also called a depth map or point cloud) relative to the print bed, thereby calculating the surface height of each area. Use infrared temperature sensors or thermal imaging cameras to obtain temperature data. Based on the imaging effect of the thermal imaging camera, the temperature distribution of each area can be obtained, and then the temperature change during the printing process can be evaluated. The acquisition of powder deposition data relies on visual inspection and image processing technology. By using a high-speed camera or a dedicated powder detection sensor, the powder deposition quality of each layer is recorded. The data is processed by image analysis software to determine the uniformity of powder distribution, deposition thickness, etc. The surface height data is then subtracted from the standard surface height of the preset printing standard to determine the surface height difference. For example, if the surface height of a certain area is too low or too high, the surface height difference is the deviation value. This can identify which areas may have uneven surfaces or too high / too low conditions. The temperature data is subtracted from the standard temperature of the preset printing standard to determine the temperature gradient. A model between deposition data and density can be established in advance:

[0068]

[0069] in, The actual metal density printed for the preset area; fIt is a mapping function of powder deposition data to density, which can be constructed through regression analysis, machine learning methods or physical models; For example, if the effect of powder deposition data on density is logarithmic (for example, the rate of increase of density slows down as the powder deposition data increases), a logarithmic function can be used to describe it:

[0070]

[0071] The preset printing standard can include the ideal density for each area. The density fluctuation can be obtained by subtracting the calculated density from the ideal density.

[0072] The preset defect prediction formula can be a mathematical model for defect prediction based on multiple factors (such as surface height difference, temperature gradient, density fluctuation, etc.). It is usually a formula or algorithm obtained through historical data, physical model or machine learning method to predict possible defects in the printing process.

[0073] Defect probability indicates the possibility of defects in certain areas during the printing process. Defects can include surface unevenness, pores, cracks, holes, and shedding, which may lead to quality problems of the printed object, such as insufficient strength and material fatigue.

[0074] The surface height difference, temperature gradient, and density fluctuation of each preset area can be substituted into the preset defect prediction formula to calculate the defect probability of each preset area in the first layer of the rapid metal printing object.

[0075] The preset defect probability threshold can be a criterion for determining when printing adjustments are required. When the defect probability of a certain area exceeds this threshold, it is considered that the printing quality of the area is at risk and further optimization or adjustment is required.

[0076] The target preset areas are those preset areas where the defect probability exceeds the preset threshold.

[0077] Printing parameters can include all parameters used in the printing process, such as laser power, scanning speed, layer thickness, powder deposition rate, etc.

[0078] Temperature and data data can include temperature changes in various areas during the printing process, usually measured by thermal imaging equipment or infrared sensors. The key is the change in temperature gradient. Excessive or uneven temperature may cause defects such as thermal cracks. And external environmental factors such as temperature, humidity, air pressure, ventilation, etc., which can affect the process and results of metal printing.

[0079] Machine vision feedback data can be the surface morphology, shape, texture and other data captured by scanning the printed surface through a machine vision system or camera. These feedback data are used to evaluate surface defects such as pores, cracks or other physical defects.

[0080] Powder data can refer to data such as powder size distribution, particle shape, powder layer thickness, etc. The quality and deposition of powder will directly affect the density, surface quality and structural integrity of the printed part.

[0081] Local geometric data can refer to the geometric shape data of each preset area in the metal printing process, including the thickness, curvature, edge shape, etc. of the printed layer. These data help to determine whether there are geometric deviations, such as warping, offset, etc.

[0082] The preset printing adjustment model can be a model established based on historical data and expert experience using machine learning or other data analysis methods. The model is used to predict possible defect types and provide corresponding adjustment solutions based on input printing parameters, temperature, environmental data, machine vision feedback data, powder data, and local geometry data.

[0083] Defect type can refer to the specific defect type that may occur during printing in a specific area or situation. For example: Surface defects: such as pores, cracks, surface unevenness, etc. Uneven density: If there is a low density in a local area, it may be due to uneven powder deposition or insufficient laser power. Thermal damage: Cracks, deformation, etc. caused by excessive temperature or uneven cooling.

[0084] The adjustment plan can be an optimization or correction plan based on the defect type. For example: Adjust the laser power: If the defect is caused by a molten pool that is too large or too small, the size of the molten pool can be controlled by adjusting the laser power. Adjust the printing speed: If the temperature is too high or the temperature gradient is too large, the heat conduction rate can be changed by adjusting the printing speed to slow down the thermal gradient. Adjust the powder deposition amount: If the density problem is caused by uneven powder deposition, the powder supply amount or spraying method can be adjusted.

[0085] If the defect probability of a certain area exceeds the preset threshold, the area is marked as a target preset area. The collected printing parameters, temperature and environment data, machine vision feedback data, powder data and local geometry data of the target preset area are input into the trained printing adjustment model. Based on the prediction of the model, the defect type and corresponding adjustment plan of the target preset area are obtained.

[0086] The printing device can be a 3D printer or additive manufacturing device for printing metal materials. It uses a laser, electron beam or other heating source to melt metal powder and deposit it layer by layer in a layer-by-layer printing manner to finally form a three-dimensional object. According to the adjustment plan obtained from the model, the parameters of the printing device can be adjusted accordingly. The printing device usually includes a control system that receives external input (such as an adjustment plan) and automatically adjusts various parameters during the printing process according to the plan. Specific adjustment operations may include laser power adjustment: by controlling the power regulator of the laser, the output power of the laser is increased or decreased, thereby affecting the depth and width of the printed molten pool. Scanning path and speed adjustment: according to the preset adjustment plan, the scanning path and scanning speed of the print head are changed. The scanning path may need to be adjusted according to the shape of the defective area. Temperature and cooling control: by adjusting the heating device (such as bed heater, laser power, etc.) to control the local temperature during the printing process to ensure appropriate thermal gradients. Control of powder injection amount: adjusting the delivery amount of the powder injection device to ensure uniform powder deposition in each area and avoid excessive or insufficient powder accumulation. The modified printing parameters are then transmitted to the printing device through the control system. Most modern metal printing equipment has an automatic control system that can automatically adjust the printing process according to the input parameters. Start the printing equipment and start printing according to the adjustment plan. The equipment will make real-time adjustments based on the new printing parameters to ensure that defects in the target preset area are repaired.

[0087] The core principle of this solution involves real-time monitoring and defect prediction based on data such as surface height, temperature, and powder deposition, and optimizing printing quality by adjusting printing parameters. This solution is not only applicable to laser melting (such as laser powder bed melting), but can also be extended to other metal printing technologies such as electron beam melting and direct energy deposition.

[0088] In the embodiment of the present application, the response time of real-time feedback is the time interval from the acquisition of data (such as surface height, temperature, powder deposition, etc.) by the machine vision system to the processing and generation of adjustment plans by the computer model. This time needs to be short enough to make timely adjustments during the printing process to avoid corrections after defects occur, thereby improving the printing quality. Therefore, the response time requirement is generally required to be between 100 milliseconds and 1 second to ensure that data collection and feedback adjustments during fast metal printing can be completed within a single printing cycle, especially in the case of high-precision printing and fast process requirements.

[0089] The maximum allowable delay refers to the maximum time delay from the system obtaining feedback data to executing printing adjustments during the printing process. Too long a delay will cause the system to be unable to respond to process changes in a timely manner, resulting in printing defects. The maximum allowable delay should be considered comprehensively based on printing speed, material characteristics and system performance. A reasonable maximum delay time can balance system response and processing capabilities, and avoid excessive errors caused by frequent adjustments.

[0090] Therefore, the maximum allowable delay is usually no more than 2 seconds, ensuring that each adjustment of the system during the printing process can be completed in a short time, avoiding uncorrectable defects caused by excessive delays during the printing process.

[0091] The following is an example of the implementation process of this program:

[0092] For example, there is a demand for quality optimization of aluminum alloy 3D printed automotive parts.

[0093] Printing goal: Printing aluminum alloy automobile engine housing requires the printed parts to have high mechanical strength and surface smoothness, while reducing internal pores and cracks. The goal is to optimize the printing results and ensure high-quality printing by adjusting the temperature gradient, powder deposition and surface height difference during the printing process.

[0094] Print parameter settings:

[0095] Laser power: 250W

[0096] Scanning speed: 850mm / s

[0097] Layer thickness: 50µm

[0098] Scan Path: Interleaved Mode

[0099] Printing environment temperature: 25°C

[0100] System input and processing:

[0101] Surface height difference: Through the machine vision feedback system, it is detected that there are large differences in the surface of certain areas. Specifically, the surface height difference between area A and area B exceeds the preset standard, showing obvious depressions.

[0102] Temperature gradient: According to the feedback from the temperature sensor, it was found that the temperature gradient in some areas was large and exceeded the preset range. The temperature fluctuations in areas A and C were large, which may cause thermal cracks.

[0103] Density fluctuation: Analysis of powder deposition data showed that insufficient powder deposition in some areas resulted in lower density than expected. The density in area B was only 90%, lower than the standard 95%.

[0104] Defect probability calculation: Based on the above data, the defect probability of areas A, B and C is calculated using the preset defect prediction formula. The defect probability of area B is 30%, which exceeds the set defect probability threshold (20%).

[0105] Adjustment plan:

[0106] For areas with a higher probability of defects, the system recommends the following adjustments:

[0107] Laser power: Increased to 280W to increase the molten pool depth and improve the fusion of powder.

[0108] Scanning speed: Reduced to 750 mm / s to reduce thermal stress, control temperature gradient and avoid cracks.

[0109] Cooling rate: Adjust the temperature of the print bed to reduce the local cooling rate and avoid excessive thermal stress and thermal cracks.

[0110] Print result comparison:

[0111] Before adjustment (original parameters):

[0112] Surface roughness: Ra=15µm

[0113] Density: 92% (lower than the standard 95%)

[0114] There are pores and local cracks.

[0115] Defective area: 25% (surface unevenness, cracks).

[0116] After adjustment:

[0117] Surface roughness: Ra=5µm (significantly improves surface quality)

[0118] Density: 96% (conforms to the standard)

[0119] No obvious pores or cracks.

[0120] Defect area: 5% (slight surface unevenness, cracks and pores have been basically eliminated).

[0121] Through the system's defect prediction and printing adjustments, cracks, pores and surface roughness were successfully reduced, and the printing quality was improved. The adjusted printed parts have higher density and surface smoothness, meeting the quality standards of automotive parts.

[0122] In the embodiment of the present application, the data acquisition module is used to obtain the surface height data, temperature data and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determine the surface height difference according to the surface height data and the preset printing standard, determine the temperature gradient according to the temperature data and the preset printing standard, and determine the density fluctuation according to the powder deposition data and the preset printing standard; the defect probability prediction module is used to obtain the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and the preset defect prediction formula; the adjustment scheme determination module is used to determine the defect probability of each preset area in the first layer of the rapid metal printing object if there is a defect probability threshold value exceeding the preset defect probability threshold value. The defect probability of the metal printing object is determined by the preset area corresponding to the defect probability exceeding the preset defect probability threshold as the target preset area, and the printing parameters, temperature and environmental data of the first layer of the metal printing object are obtained, as well as the machine vision feedback data, powder data and local geometry data of the target preset area. The printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometry data are input into the preset printing adjustment model to obtain the defect type and adjustment plan of the target preset area; wherein the number of the target preset areas is at least one; the printing adjustment module is used to control the printing equipment to make printing adjustments to the first layer of the metal printing object according to the adjustment plan of each target preset area. Through the above-mentioned rapid metal printing system based on machine vision feedback, accurate defect prediction, real-time adjustment and automatic control can not only improve the quality and efficiency of metal printing, but also greatly reduce resource waste, reduce costs, and enhance the flexibility and intelligence level of the production process.

[0123] On the basis of the above technical solution, an optional, preset defect prediction formula is:

[0124]

[0125] in, is the defect probability; The preset surface height difference weight is the surface height difference; is the preset temperature gradient weight; is the temperature gradient; is the preset density fluctuation weight; is the density fluctuation; The coordinates of the center point of each preset area.

[0126] In this scheme, the weight parameter , , The source and determination of can be obtained through the following two main methods:

[0127] 1. Experimental calibration:

[0128] Experimental design: Design a series of experiments based on actual needs, control variables to change the values ​​of factors such as Δh(x, y), ΔT(x, y), Δρ(x, y) one by one, and measure their specific impact on D(x, y).

[0129] Impact analysis: Statistical analysis of experimental data is performed to quantify the contribution of each factor to the results, thereby determining preliminary weights.

[0130] Verify and adjust: Apply the initial weights to the actual scenario, compare the model predictions with the actual values ​​for verification, and further fine-tune the weights.

[0131] For example, the research object is the factors affecting the surface quality deviation (D(x, y)) during laser printing, including:

[0132] Δh(x, y): surface height difference

[0133] ΔT(x, y): temperature gradient

[0134] Δρ(x, y): density fluctuation

[0135] step:

[0136] In the control experiment, only the value of Δh(x, y) is changed (such as ±0.1, ±0.2 mm), other parameters are kept constant, and the corresponding D(x, y) is measured.

[0137] Repeat the experiment, varying the values ​​of ΔT(x, y) and Δρ(x, y) individually, and record D(x, y)D(x, y)D(x, y).

[0138] Impact Analysis:

[0139] Perform statistical analysis on experimental data, for example:

[0140] Calculate the contribution ratio of each factor change to D(x, y).

[0141] If Δh(x, y) contributes 50% to the bias, ΔT(x, y) is 30%, and Δρ(x, y)\Delta \rho(x, y)Δρ(x, y) is 20%, then the preliminary weights are set to w1=0.5, w2=0.3, and w3=0.2.

[0142] Verify and adjust:

[0143] Apply these weights to calculate D(x, y) in actual production scenarios.

[0144] Compare the model prediction value and the actual measurement value. If the error is large, adjust the weight, such as =0.4, =0.35, =0.25.

[0145] 2. Optimization based on historical data:

[0146] Data collection: Collect a large amount of historical data, including the changes in Δh(x, y), ΔT(x, y), Δρ(x, y) and the corresponding D(x, y) output.

[0147] Regression analysis: Use linear regression, multivariate regression, or machine learning models (such as gradient descent optimization) to calculate weights so that the error between the model output D(x, y) and the historical true value is minimized.

[0148] Optimization process: introduce the objective function (such as mean square error) and constraints, and use the optimization algorithm (such as least squares method or genetic algorithm) to further optimize the weights.

[0149] For example, data collected:

[0150] Assume that the historical data includes:

[0151] Δh(x, y): value range [0.05, 0.5 mm]

[0152] ΔT(x, y): Value range [±5°C]

[0153] Δρ(x, y): value range [±0.1 g / cm³]

[0154] Corresponding D(x, y): Obtained through the statistics of surface deviation results in historical records.

[0155] Regression Analysis:

[0156] Using multiple linear regression:

[0157] The formula is .

[0158] The historical data is fed into the model and the preliminary weights are calculated using linear regression, e.g. =0.45, =0.35, =0.2.

[0159] Optimization process:

[0160] Define the objective function:

[0161]

[0162] Among them, MSE is the mean square error, which is used to measure the deviation between the model prediction value and the true value. The smaller the value, the closer the model's prediction result is to the actual situation; N is the number of samples; is the output value predicted by the model, that is, D(x, y) calculated according to the formula; is the actual output value, i.e. D(x, y) recorded in the experiment or historical data; is the square of the error of a single data point, which is used to emphasize the impact of data points with larger errors on the total error.

[0163] On the basis of the above technical solution, optionally, the system further includes a printing continuous adjustment module, and the printing continuous adjustment module is used to:

[0164] The surface height data, temperature data and powder deposition data of each preset area in each printing layer of the rapid metal printing object transmitted by the machine vision feedback system are respectively obtained according to the sequence of the printing layers, and the defect probability of each preset area in each printing layer is determined based on the above data. In the case of a defect probability exceeding a preset defect probability threshold, the preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as the target preset area, and the defect type and adjustment plan of the target preset area are determined. Based on the adjustment plan, the printing device is controlled to perform printing adjustments on each printing layer in the sequence until the printing adjustment of all printing layers is completed.

[0165] In this scheme, the surface height data, temperature data and powder deposition data of the preset areas in each printing layer of the rapid metal printing object transmitted by the machine vision feedback system can be obtained layer by layer according to the order of the number of printing layers, and the defect probability of each preset area in each printing layer can be calculated based on these data. When the defect probability exceeds the preset threshold, the preset areas where the defect probability exceeds the threshold are identified as the target preset areas, and the defect types and corresponding adjustment plans are determined based on the specific data of these areas. Subsequently, based on the preset adjustment plan, the printing device is controlled to adjust the printing of each layer to ensure that the printing process meets the quality standards. This process is carried out step by step until the adjustment of all printing layers is completed to ensure that the quality of the entire printing process is optimized. For example, after the adjustment of the first layer is completed, the second layer is adjusted, and after the adjustment of the second layer is completed, the third layer is adjusted, and so on, until the adjustment of the printing of all layers is completed.

[0166] Based on the above technical solution, the optional training process of the preset print adjustment model is:

[0167] Acquire historical adjustment records, determine historical defect types, historical adjustment solutions, historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data according to the historical adjustment records, and create a first data set according to the historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data;

[0168] labeling the defect type label of the first data set according to the historical defect type, and labeling the adjustment solution label of the first data set according to the historical adjustment solution;

[0169] A printing adjustment model is constructed, and the printing adjustment model is trained according to the first data set, the defect type label, and the adjustment solution label until the printing adjustment model reaches a preset training standard.

[0170] In this solution, the historical adjustment record may refer to a record of all adjustments made in the past printing process. These adjustments may include adjustments to parameters such as laser power, printing speed, scanning strategy, layer thickness, etc., in order to optimize the printing quality or repair defects.

[0171] The historical defect type may refer to known defect types in the historical printing process, such as holes, warping, surface unevenness, etc.

[0172] Historical adjustment solutions may refer to adjustment solutions adopted to solve specific defects, such as adjusting laser power, changing printing speed, etc.

[0173] Historical printing parameters may refer to various printing settings used during historical printing, such as laser power, printing speed, layer thickness, scanning strategy, etc.

[0174] Historical temperature and environmental data can refer to environmental conditions related to the printing process, such as the temperature and humidity of the working environment, which may affect the printing quality.

[0175] Historical machine vision feedback data may refer to visual data about printed objects collected through a machine vision system, including defect detection results, surface quality assessments, etc.

[0176] Historical powder data may refer to historical data related to powder quality, such as powder particle size, purity, shape factor, etc.

[0177] The historical local geometric data may refer to historical data related to local geometric features of a printed object, such as the geometric shape of each layer, surface roughness, and the like.

[0178] The first data set can be a collection created based on various data in the historical printing process (printing parameters, temperature and environment data, machine vision feedback data, powder data, etc.). Each data record usually contains multiple features (such as temperature, printing parameters, etc.) and target labels (such as defect type and adjustment plan).

[0179] The defect type label can refer to the defect type corresponding to each data record, such as holes, surface unevenness, warpage, etc. It is a classification label used to guide the model to understand the defect type corresponding to each situation.

[0180] The Adjustment Plan label can refer to the adjustment measures taken to fix the defect, such as increasing laser power, reducing print speed, etc. It is also a classification label that indicates the adjustment measures that should be taken when a specific defect type is encountered.

[0181] The preset training standards can be the goals or performance standards that need to be achieved during the model training process. They usually include evaluation indicators such as the model's accuracy, recall, and F1 value. Meeting these standards means that the model is mature enough to be used in the actual printing adjustment process.

[0182] This can be done by collecting historical adjustment records, including historical defect types, historical adjustment plans, historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometry data. These historical data provide the basis for model training. On this basis, the first data set is constructed by analyzing historical printing parameters, temperature and environmental data, machine vision feedback data, powder data, and local geometry data. Next, the defect type label and adjustment plan label are annotated for each record in the data set, and these labels are used to supervise the training of the model. Then, the printing adjustment model is trained using these annotated data. The goal of the model is to learn how to predict the types of defects that may occur in the current printing process based on historical data, and recommend appropriate adjustment plans for each defect type. The training process will continue until the model reaches the preset standard in terms of prediction accuracy.

[0183] When building and training a preset print adjustment model, the quality and quantity of data are crucial to the training effect of the model. However, historical data may be insufficient or unevenly distributed, which will affect the accuracy and generalization ability of the model. To solve this problem, methods such as data enhancement and transfer learning can be used to improve the performance of the model when data is insufficient or unevenly distributed.

[0184] The solution to insufficient data: data augmentation

[0185] Data augmentation is a technique that generates new data samples by processing existing data. To address the problem of insufficient data in the print adjustment model, the following data augmentation methods can be used:

[0186] Noise injection: By adding tiny noise to historical data (such as surface height, temperature, powder data, etc.), simulating different environmental conditions or printing states, new data samples are generated. Noise injection can help the model better cope with the inevitable changes and errors in the printing process.

[0187] Simulation data generation: Use printing simulation software or physical models to generate data under different working conditions, such as different printing parameters, temperature gradients, powder deposition conditions, etc. Through simulation data, the lack of historical data can be made up, so that the model can learn more printing scenarios and defect modes.

[0188] Data transformation: By performing transformations such as rotation, scaling, and translation on existing data, especially enhancing images and local geometric data, for example, changing the local geometric shape of the printing area or the rotation angle of the machine vision feedback image to increase the diversity of training data.

[0189] Synthetic data: In the absence of certain types of data, synthetic data can be created by combining existing data. For example, printing parameters, environmental data, and machine vision feedback from different regions can be combined to synthesize new training data.

[0190] The solution to uneven data distribution: transfer learning

[0191] Transfer learning can help the model acquire knowledge from other related tasks, thereby improving the training effect, especially when data is insufficient. The specific technical solution is as follows:

[0192] Domain adaptation: When there is insufficient historical data for the target task, a large amount of data from similar tasks (such as other types of metal printing, printing of different materials, etc.) can be used to transfer the knowledge of the source domain (source data) to the target domain (target printing task) through domain adaptation technology. By adjusting the distribution difference between the source data and the target data, the model can better adapt to the target task.

[0193] Pre-trained model: You can use a model pre-trained from other similar tasks, such as a deep learning model trained on a large-scale printing dataset, as the initial model. Then fine-tune it for the specific data of the target task. For example, use a convolutional neural network model previously trained on other printing tasks to process machine vision feedback data and fine-tune it to adapt to the new printing task.

[0194] Knowledge distillation: Through knowledge distillation technology, knowledge can be extracted from a complex, data-rich large model and passed to the target model. In the case of insufficient data, a "teacher model" trained on a large dataset is used to guide the training of the "student model", so that the student model can perform better on a limited dataset.

[0195] After applying data augmentation and transfer learning, the training process still needs to be fine-tuned to further optimize model performance. The specific steps are as follows:

[0196] 1. Fine-tuning: In transfer learning, fine-tuning refers to further training for the target task (print adjustment model) based on the pre-trained model. In this process, you can choose to freeze some layers and only fine-tune some parameters of the model so that the model can better adapt to the new data.

[0197] 2. Adaptive learning rate: By adjusting the learning rate, the model can quickly adapt to different data sets, especially when using data enhancement and transfer learning. The appropriate learning rate can prevent overfitting and speed up the training process.

[0198] During the training process, the model's effect must be constantly verified to ensure its accuracy and stability in various scenarios. The data set can be divided into multiple subsets through cross-validation methods, and the model can be trained and validated on different subsets to ensure the generalization ability of the model. Especially in the case of uneven data distribution, cross-validation can help evaluate the robustness of the model. Or use a variety of evaluation indicators (such as accuracy, recall, F1 score, etc.) to comprehensively evaluate the model. For different printing tasks and materials, select appropriate indicators to evaluate model performance.

[0199] On the basis of the above technical solution, optionally, the system further includes a fault detection module, and the fault detection module is used to:

[0200] Acquire the operating parameters of the printing device in real time, input the operating parameters into a preset fault diagnosis and repair model, and determine whether there is a fault risk;

[0201] If there is a risk of failure, the fault type is output through a preset fault diagnosis and repair model, and the fault type is transmitted to the control center;

[0202] Accordingly, the system further comprises a maintenance detection module, and the maintenance detection module is used to:

[0203] If there is no risk of failure, the health score is output through the preset fault diagnosis and repair model;

[0204] If the health score is lower than a preset health score threshold, a maintenance plan is output through a preset fault diagnosis and repair model, and the maintenance plan is transmitted to a control center.

[0205] In this solution, operating parameters may refer to various data generated in real time by the printing device during operation, including but not limited to the device's operating temperature, vibration level, current, voltage, laser power, printing speed, laser focus position, etc. These parameters reflect the device's operating status and can be used to determine whether the device is in normal operation.

[0206] The preset fault diagnosis and repair model can be a machine learning or statistical model built from historical fault data, equipment operation status, and maintenance records. Its function is to analyze the operating parameters collected in real time, identify whether there is a potential fault through the learned pattern, or evaluate the health status of the equipment. In the case of a fault, the model will output the specific fault type and provide maintenance personnel with a maintenance plan.

[0207] The fault type can be a specific type of fault that may occur in the device. For example, a printing device may have various faults such as insufficient laser power, overheating, clogged print heads, circuit failures, wear of mechanical parts, etc. The fault type is determined by analyzing the operating parameters and device status.

[0208] The control center can be a system that centrally manages and monitors the operating status of equipment. It can receive real-time data from various devices, analyze and make corresponding processing. When a fault is detected, the control center will receive a notification of the fault type and arrange subsequent processing measures according to the preset process.

[0209] The health score can be an evaluation value of the current health status of the equipment, which is usually calculated by combining multiple operating parameters through a preset model. The higher the score, the healthier the equipment is; the lower the score, the higher the risk of equipment failure, and maintenance or repair may be required.

[0210] The preset health score threshold may be a set standard value used to determine whether the device needs maintenance. If the health score is lower than the threshold, it means that the device has certain health risks and needs maintenance and repair.

[0211] The maintenance plan can be a maintenance measure provided by the model based on historical data and the current status of the equipment when the health score of the equipment is lower than the preset threshold. The maintenance plan includes specific maintenance steps, required parts, arrangements for repair personnel, etc., with the aim of preventing equipment failures and extending the service life of the equipment.

[0212] The various operating parameters of the printing equipment, including temperature, current, laser power, vibration and other data, can be obtained through sensors and real-time monitoring equipment. These data will be transmitted in real time to the preset fault diagnosis and repair model, which has been trained with historical data and can identify the equipment status and predict the risk of failure. After the model analyzes these operating parameters, it determines whether there is a potential risk of failure. If there is a risk of failure, the model will output the specific fault type (such as excessive temperature, unstable laser power, etc.), and send these fault types to the control center to notify the maintenance personnel to handle it. If there is no risk of failure, the model will generate a health score based on the current working status of the equipment and compare it with the preset health score threshold. If the health score is lower than the set threshold, it indicates that the equipment may have a hidden fault or require maintenance. The model will automatically generate a maintenance plan, including specific maintenance steps, required tools and parts, and then transmit the maintenance plan to the control center, which will arrange the corresponding maintenance operations according to the plan.

[0213] The model training steps are:

[0214] Historical operation data of the equipment can be collected, including sensor data such as temperature, vibration, current, power, and the equipment's fault records and maintenance history. Based on this data, health scores and fault types are defined as target variables, where the health score quantifies the overall health of the equipment, while the fault type labels different fault categories (such as mechanical faults, electrical faults, etc.). Next, through data preprocessing and feature engineering, key features such as temperature fluctuations and vibration frequencies are extracted. Then, a training data set is constructed to pair the operation data with health scores or fault type labels. Suitable machine learning models are selected, such as regression models (for health score prediction) and classification models (for fault diagnosis), and the models are trained with labeled training data. During the training process, cross-validation is used to evaluate the model performance, and the performance of the model is optimized by adjusting hyperparameters. After training, the model can monitor the health status of the equipment in real time and predict the health score or diagnose the fault type based on real-time data. Finally, the trained model is deployed to the equipment monitoring system, and the model is continuously optimized to ensure its accuracy and robustness, so as to achieve efficient equipment management and maintenance decisions.

[0215] The calculation process of the health score may be affected by sensor errors and external environmental interference. Therefore, in practical applications, the sensor data must be corrected and filtered to ensure the accuracy and reliability of the health score. The following is a description of the health score correction and sensor data filtering methods:

[0216] 1. Sensor data correction

[0217] Correction of sensor data is an important step to ensure accurate health scores. To eliminate the systematic errors of sensors, the following measures can be taken:

[0218] Sensor calibration: Calibrate the sensor regularly to ensure that the data output by the sensor under different conditions is consistent with the true value. For example, use a known standard value or reference object for calibration to correct the zero point and gain of the sensor. Calibration eliminates drift or deviation that may occur during long-term use and ensures data consistency.

[0219] Temperature compensation: Temperature changes will affect the output of the sensor, so temperature compensation is required. By installing a temperature sensor next to the sensor, the temperature changes are monitored in real time and the sensor data is compensated according to the ambient temperature.

[0220] Nonlinear correction: For sensors with nonlinear responses, regression models or other nonlinear correction methods can be used to correct the sensor output based on known data to ensure that its relationship with the actual physical quantity is linear or conforms to other known laws.

[0221] 2. Sensor data filtering

[0222] In order to further eliminate the impact of external environmental interference, noise and instantaneous fluctuations on health scores, it is necessary to effectively filter and denoise the sensor data. Specifically, it can include:

[0223] Low-pass filter: Use a low-pass filter (such as a weighted moving average or Kalman filter) to remove high-frequency noise, especially fluctuations caused by transient disturbances or unstable factors. This smooths the sensor data so that it reflects long-term trends rather than short-term fluctuations.

[0224] Mean filtering: Mean filtering is a common noise removal method that reduces errors caused by short-term transient interference by locally averaging the sensor readings. This helps remove the effects of power supply fluctuations, sensor jitter, etc. on the data.

[0225] Outlier detection: Perform outlier detection on sensor data to identify and remove data points due to faults, sensor failures, or other external abnormalities. For example, use Z-Score or IQR methods to detect outliers in the data and remove them before health score calculation.

[0226] Multi-sensor fusion: If multiple sensors can provide the same or similar measurement data, sensor data fusion techniques (such as Kalman filtering or weighted averaging) are used to combine the data from multiple sensors. This not only helps reduce the error of a single sensor, but also improves the accuracy and robustness of the data.

[0227] 3. Environmental interference correction

[0228] External environmental interference, such as temperature and humidity changes, electromagnetic interference, etc., may affect the accuracy of the sensor, so environmental factors need to be considered and compensated. Specifically, it may include:

[0229] Environmental data monitoring: By installing additional environmental sensors (such as temperature and humidity sensors, electromagnetic interference monitors, etc.), the changes in the external environment are monitored in real time, and these environmental data are used as part of the health score calculation for compensation. For example, when the temperature or humidity changes significantly, the weight of the health score can be adjusted.

[0230] Environmental model compensation: By establishing a mathematical model of environmental interference, the impact of environmental changes on sensor output is taken into account. For example, using climate change models, mechanical vibration models, etc., data correction is performed based on known external environmental influences to reduce interference factors.

[0231] 4. Dynamic adjustment of health score

[0232] When calculating the health score, the score calculation method can be dynamically adjusted to cope with the impact of sensor errors and external interference. For example, the weight coefficient of the health score can be continuously revised based on historical data or real-time feedback to ensure that the score result can more accurately reflect the actual health status of the device.

[0233] 5. Model verification and correction

[0234] After obtaining the health score, the actual device status or fault diagnosis results are used to verify the accuracy of the score. By continuously comparing the difference between the health score and the actual device health status, the scoring model is optimized and corrected to further improve accuracy.

[0235] In this solution, by real-time monitoring of equipment operating parameters, failure risks can be detected in time and equipment health status can be predicted, reducing equipment downtime and production interruptions. If a failure is detected or the health score is below the threshold, the system automatically generates a maintenance plan to ensure that the equipment is maintained in a timely manner. This improves equipment reliability, operation and maintenance efficiency, and reduces failure and maintenance costs.

[0236] Embodiment 2

[0237] Figure 2 Schematic diagram of the structure of the rapid metal printing system based on machine vision feedback provided in the second embodiment of the present application. Figure 2 As shown, specifically including the following:

[0238] The system further comprises a laser power determination module 105, wherein the laser power determination module 105 is used to:

[0239] Powder particle size data and powder mass data are obtained, and a preset initial laser power is adjusted according to the powder particle size data, the powder mass data and a preset laser power calculation formula to obtain an adjusted laser power, and printing parameters are updated according to the adjusted laser power.

[0240] In this embodiment, as the powder particle size increases, more laser power is required to melt the powder, because larger powder particles require higher energy to completely melt. Conversely, powders with smaller particle sizes are easy to melt, so less laser power is required. High-quality powders usually have better fluidity and sphericity, and can be more easily melted by the laser. Therefore, under the same printing conditions, high-quality powders may require slightly lower laser power, while poorer quality powders may require more laser power to ensure that they are completely melted. Laser power is the parameter that most directly affects the quality of powder melting and deposition. It is closely related to the powder particle size and quality. Controlling the size of the laser power can adjust the degree of powder melting and ensure printing quality. Therefore, it is necessary to update the laser power in the printing parameters according to the powder particle size data and the powder quality data.

[0241] Powder particle size data can be the particle size distribution of powder materials. It is usually obtained through particle size analysis instruments (such as laser particle size analyzers) or sieving methods to describe the size, distribution range and uniformity of particles in the powder.

[0242] Powder quality data can refer to the density, purity, composition, and particle shape of the powder material. Higher quality powders have better flowability, more consistent particle shape, and lower impurity content, and generally provide better printing results.

[0243] The preset laser power calculation formula may refer to a mathematical expression for calculating the laser power required during the printing process, which adjusts the laser power based on some known factors (such as the particle size and quality of the powder) to optimize the printing quality and effect.

[0244] The preset initial laser power may refer to the laser power value set by the printing device by default without considering the actual powder parameters. It is usually given by the equipment manufacturer according to specific materials and printing requirements and serves as the basic power for printing.

[0245] Adjusting the laser power can mean adjusting the laser power through a calculation formula based on the powder particle size data and the powder quality data to optimize the printing process. The purpose of this step is to ensure that the laser power matches the physical properties of the powder to ensure the best effect of the melting process.

[0246] The particle size distribution information of the powder can be obtained by a particle size analyzer (such as a laser particle size analyzer, sieving method, etc.). Commonly used particle size parameters include D50 (i.e., the particle size at which 50% of the particles are smaller than this value), or D10 and D90 (representing the particle size at which 10% and 90% of the particles are smaller than this value, respectively). For example, assuming that the D50 of a certain powder is 20 µm obtained through particle size analysis, the powder particle size data is 20µm. Powder quality includes many aspects, mainly including the purity, fluidity, particle shape, etc. of the powder. It can be quantified in the following ways: Purity: Measure the content of the target metal in the powder, for example, the purity of a certain powder is 95%. Shape factor: Determine the shape of the powder particles through morphological analysis, and sphericity is commonly used. Powders with higher sphericity have better quality. For example, assuming that the purity of the powder = 95% and the sphericity = 0.9 (better sphericity). Then quantify it through the following formula:

[0247]

[0248] in, It is the quantified powder quality data; is the preset purity weight; for purity; is the preset sphericity weight; is the sphericity.

[0249] After the powder mass data is quantified, the powder particle size data and the quantified powder mass data are substituted into the formula to calculate the adjusted laser power. Then, the adjusted laser power is used to replace the laser power included in the original printing parameters.

[0250] In this embodiment, by adjusting the laser power according to the powder particle size and powder quality data, the stability, printing quality and efficiency of the printing process can be significantly improved, printing defects and waste can be reduced, and material utilization can be improved.

[0251] On the basis of the above technical solution, an optional, preset laser power calculation formula is:

[0252]

[0253] in, To adjust the laser power; is the preset particle size adjustment factor; is the powder particle size data; is the preset quality adjustment factor; is the powder quality data; Adjustment factor for preset printing parameters; is the preset initial laser power.

[0254] In this solution, for example, if the purity of the powder is 95% (i.e. 0.95), the sphericity is 0.9, the preset purity weight is 0.7, and the preset sphericity weight is 0.3, the above formula is used to quantify the powder mass data:

[0255] =0.7·95+0.3·0.9=66.77

[0256] After the quantified powder mass data is obtained, the powder particle size data and the quantified powder mass data are substituted into the formula to calculate and adjust the laser power.

[0257] and Indicates the degree of influence of powder particle size and powder quality on laser power adjustment. They are determined through experimental calibration and data analysis. The following is the specific determination process:

[0258] Experimental design:

[0259] Particle size range: When designing an experiment, choose metal powders with different particle size ranges for printing. For example, you can choose powders with particle sizes ranging from 10μm to 100μm, or a finer range (for example, 10μm to 30μm) to observe the need for laser power adjustment.

[0260] Powder quality: Powder quality includes characteristics such as purity and particle shape. You can choose powders of different qualities, such as 90%, 95%, 99% purity, etc., for experimental testing, and record the laser power adjustment required for each powder.

[0261] Experimental steps:

[0262] Controlled variable experiment: Keep other printing parameters (such as printing speed, temperature, etc.) unchanged and only change the particle size and quality of the powder.

[0263] Measurement results: For each powder particle size and mass, the adjusted laser power is recorded.

[0264] Regression analysis: Based on experimental data, linear regression or multiple regression methods are used to calculate and Generally, powders with larger particle sizes require more laser energy, so Usually a positive value; when the powder quality is poor, the laser power may need to be further increased, so It’s also positive.

[0265] Dealing with particle size differences between different batches of powder:

[0266] For the difference in powder particle size between different batches, the above experiment can be carried out for each batch separately, and the corresponding and .

[0267] The difference in powder particle size will affect the melting efficiency of the powder. Larger particles may require higher laser power for sufficient melting. Will be adjusted according to the particle size range.

[0268] Experimental conditions:

[0269] Powder particle size: For example, different particle sizes within the range of 10μm-100μm are selected (the specific particle size range is selected according to the actual application).

[0270] Powder quality: For example, choose a powder with 90%-99% purity for the experiment.

[0271] Indicates the preset laser power adjustment factor, which adjusts the relationship between laser power and environment, equipment and powder. The method is as follows:

[0272] Experimental design:

[0273] By controlling other printing parameters (such as printing speed, temperature, etc.), the actual printing effects of different types of powders were tested, and the laser power before and after adjustment was recorded.

[0274] In the experiment, the preset laser power adjustment factor corresponding to different powders (according to particle size and mass) was calculated through multiple measurements. .

[0275] Experimental steps:

[0276] Powders with different characteristics (e.g. different particle size, mass) were selected for testing.

[0277] The initial laser power and the adjusted laser power were recorded under each experimental condition.

[0278] Based on the actual results, regression analysis is used to determine The value of can be adjusted to meet the actual printing requirements.

[0279] Optimization based on historical data:

[0280] Collect data from multiple prints and analyze laser power variations for different powder batches, ambient temperatures, and equipment conditions.

[0281] Use machine learning algorithms (such as regression analysis, gradient descent) to calculate the optimal value, so that the adjusted laser power is optimized under different printing conditions.

[0282] Experimental conditions and scope:

[0283] Different environments: Tests can be performed under different temperature and humidity conditions to determine Scope of application.

[0284] Different powders: Select powders with a particle size range of 10μm to 100μm for experiments to ensure that the formula is applicable to powders of different particle sizes and masses.

[0285] In order to ensure that the formula is applicable to different powders and printing environments, the following boundary conditions need to be clarified:

[0286] Particle size range: For example, metal powders from 10μm to 100μm. If the particle size is outside this range, recalibration of parameters may be required.

[0287] Powder quality range: For example, choose metal powders with a purity between 90% and 99%. If the powder quality is very poor (for example, purity below 90%), a different laser power adjustment strategy may be required.

[0288] Environmental conditions: Environmental factors such as temperature and humidity will affect the laser power requirements and need to be controlled in the experiment to ensure that the formula is applicable under specific environmental conditions.

[0289] Embodiment 3

[0290] Figure 3 : is a flow chart of a rapid metal printing method based on machine vision feedback provided in Example 3 of the present application. Figure 3 As shown, the specific steps include:

[0291] S301, obtaining surface height data, temperature data and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determining the surface height difference according to the surface height data and the preset printing standard, determining the temperature gradient according to the temperature data and the preset printing standard, and determining the density fluctuation according to the powder deposition data and the preset printing standard.

[0292] S302, obtaining the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and a preset defect prediction formula.

[0293] S303, if there is a defect probability exceeding a preset defect probability threshold, the preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as the target preset area, and the printing parameters, temperature and environmental data of the first layer of the rapid metal printing object are obtained, as well as the machine vision feedback data, powder data and local geometric data of the target preset area, and the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data are input into a preset printing adjustment model to obtain the defect type and adjustment plan of the target preset area; wherein the number of the target preset areas is at least one.

[0294] S304, controlling the printing device to perform printing adjustment on the first layer of the rapid metal printing object according to the adjustment scheme of each target preset area.

[0295] The embodiment of the present application provides a rapid metal printing method based on machine vision feedback, which corresponds to the system provided in the above embodiments and has corresponding execution processes and beneficial effects, which will not be repeated here.

[0296] Embodiment 4

[0297] like Figure 4 As shown, the embodiment of the present application also provides an electronic device 400, including a processor 401, a memory 402, and a program or instruction stored in the memory 402 and executable on the processor 401. When the program or instruction is executed by the processor 401, each process of the above-mentioned embodiment of the rapid metal printing system method based on machine vision feedback is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0298] It should be noted that the electronic devices in the embodiments of the present application include the mobile electronic devices and non-mobile electronic devices mentioned above.

[0299] Embodiment 5

[0300] An embodiment of the present application also provides a readable storage medium, on which a program or instruction is stored. When the program or instruction is executed by a processor, each process of the above-mentioned cable installation process based on the tension adaptive control system embodiment is implemented, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0301] The processor is the processor in the electronic device described in the above embodiment. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0302] It should be noted that, in this article, the term "comprises", "includes" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or system including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, method, article or system. In the absence of further restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the process, method, article or system including the element. In addition, it should be pointed out that the scope of the method and system in the embodiment of the present application is not limited to performing functions in the order shown or discussed, and may also include performing functions in a substantially simultaneous manner or in reverse order according to the functions involved, for example, the described method may be performed in an order different from that described, and various steps may also be added, omitted, or combined. In addition, the features described with reference to certain examples may be combined in other examples.

[0303] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a computer software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present application.

[0304] The embodiments of the present application are described above in conjunction with the accompanying drawings, but the present application is not limited to the above-mentioned specific implementation methods. The above-mentioned specific implementation methods are merely illustrative and not restrictive. Under the guidance of the present application, ordinary technicians in this field can also make many forms without departing from the purpose of the present application and the scope of protection of the claims, all of which are within the protection of the present application.

[0305] The above are only preferred embodiments of the present application and the technical principles used. The present application is not limited to the specific embodiments described herein, and various obvious changes, readjustments and substitutions that can be made by those skilled in the art will not deviate from the scope of protection of the present application. Therefore, although the present application is described in more detail through the above embodiments, the present application is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present application, and the scope of the present application is determined by the scope of the claims.

Claims

1. A rapid metal printing system based on machine vision feedback, characterized in that: The system comprises: A data acquisition module, used to obtain surface height data, temperature data and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determine the surface height difference according to the surface height data and the preset printing standard, determine the temperature gradient according to the temperature data and the preset printing standard, and determine the density fluctuation according to the powder deposition data and the preset printing standard; The defect probability prediction module is used to obtain the defect probability of each preset area in the first layer of the rapid metal printing object according to the surface height difference, the temperature gradient, the density fluctuation and a preset defect prediction formula; wherein the preset defect prediction formula is: ； in, is the defect probability; The preset surface height difference weight is the surface height difference; is the preset temperature gradient weight; is the temperature gradient; is the preset density fluctuation weight; is the density fluctuation; The coordinates of the center points of each preset area; An adjustment scheme determination module is used for, if there is a defect probability exceeding a preset defect probability threshold, taking a preset area corresponding to the defect probability exceeding the preset defect probability threshold as a target preset area, obtaining printing parameters, temperature and environmental data of the first layer of the rapid metal printing object, and obtaining machine vision feedback data, powder data and local geometric data of the target preset area, inputting the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data into a preset printing adjustment model, and obtaining a defect type and an adjustment scheme of the target preset area; wherein the number of the target preset area is at least one; A printing adjustment module, used to control the printing device to make printing adjustments to the first layer of the rapid metal printing object according to the adjustment scheme of each target preset area; The system further comprises a laser power determination module, wherein the laser power determination module is used to: Obtain powder particle size data and powder mass data, adjust the preset initial laser power according to the powder particle size data, the powder mass data and a preset laser power calculation formula, obtain the adjusted laser power, and update the printing parameters according to the adjusted laser power; wherein the preset laser power calculation formula is: ； in, To adjust the laser power; is the preset particle size adjustment factor; is the powder particle size data; is the preset quality adjustment factor; is the powder quality data; is the preset laser power adjustment factor; is the preset initial laser power.

2. The rapid metal printing system based on machine vision feedback according to claim 1, characterized in that: The system further includes a printing continuous adjustment module, wherein the printing continuous adjustment module is used to: The surface height data, temperature data and powder deposition data of each preset area in each printing layer of the rapid metal printing object transmitted by the machine vision feedback system are respectively obtained according to the sequence of the printing layers, and the defect probability of each preset area in each printing layer is determined based on the above data. In the case of a defect probability exceeding a preset defect probability threshold, the preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as the target preset area, and the defect type and adjustment plan of the target preset area are determined. Based on the adjustment plan, the printing device is controlled to perform printing adjustments on each printing layer in the sequence until the printing adjustment of all printing layers is completed.

3. The rapid metal printing system based on machine vision feedback according to claim 1, characterized in that: The training process of the preset print adjustment model is: Acquire historical adjustment records, determine historical defect types, historical adjustment solutions, historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data according to the historical adjustment records, and create a first data set according to the historical printing parameters, historical temperature and environmental data, historical machine vision feedback data, historical powder data, and historical local geometric data; labeling the defect type label of the first data set according to the historical defect type, and labeling the adjustment solution label of the first data set according to the historical adjustment solution; A printing adjustment model is constructed, and the printing adjustment model is trained according to the first data set, the defect type label, and the adjustment solution label until the printing adjustment model reaches a preset training standard.

4. The rapid metal printing system based on machine vision feedback according to claim 1, characterized in that: The system further comprises a fault detection module, wherein the fault detection module is configured to: Acquire the operating parameters of the printing device in real time, input the operating parameters into a preset fault diagnosis and repair model, and determine whether there is a fault risk; If there is a risk of failure, the fault type is output through a preset fault diagnosis and repair model, and the fault type is transmitted to the control center; Accordingly, the system further comprises a maintenance detection module, and the maintenance detection module is used to: If there is no risk of failure, the health score is output through the preset fault diagnosis and repair model; If the health score is lower than a preset health score threshold, a maintenance plan is output through a preset fault diagnosis and repair model, and the maintenance plan is transmitted to a control center.

5. A fast metal printing method based on machine vision feedback, characterized in that: The method comprises: Obtaining surface height data, temperature data, and powder deposition data of each preset area in the first layer of the rapid metal printing object transmitted by the machine vision feedback system, determining a surface height difference according to the surface height data and a preset printing standard, determining a temperature gradient according to the temperature data and the preset printing standard, and determining a density fluctuation according to the powder deposition data and the preset printing standard; According to the surface height difference, the temperature gradient, the density fluctuation and the preset defect prediction formula, the defect probability of each preset area in the first layer of the rapid metal printing object is obtained; wherein the preset defect prediction formula is: ； in, is the defect probability; The preset surface height difference weight is the surface height difference; is the preset temperature gradient weight; is the temperature gradient; is the preset density fluctuation weight; is the density fluctuation; The coordinates of the center points of each preset area; If there is a defect probability exceeding a preset defect probability threshold, a preset area corresponding to the defect probability exceeding the preset defect probability threshold is used as a target preset area, and printing parameters, temperature and environmental data of the first layer of the rapid metal printing object are obtained, and machine vision feedback data, powder data and local geometric data of the target preset area are obtained, and the printing parameters, temperature and environmental data, machine vision feedback data, powder data and local geometric data are input into a preset printing adjustment model to obtain the defect type and adjustment plan of the target preset area; wherein the number of the target preset area is at least one; Control the printing device to adjust the first layer of the rapid metal printing object according to the adjustment plan of each target preset area; The method further comprises: Obtain powder particle size data and powder mass data, adjust the preset initial laser power according to the powder particle size data, the powder mass data and a preset laser power calculation formula, obtain the adjusted laser power, and update the printing parameters according to the adjusted laser power; wherein the preset laser power calculation formula is: ； in, To adjust the laser power; is the preset particle size adjustment factor; is the powder particle size data; is the preset quality adjustment factor; is the powder quality data; is the preset laser power adjustment factor; is the preset initial laser power.

6. An electronic device, characterized in that: It comprises a processor, a memory, and a program or instruction stored in the memory and executable on the processor, wherein the program or instruction, when executed by the processor, implements the steps of the rapid metal printing method based on machine vision feedback as claimed in claim 5.

7. A readable storage medium, characterized in that: The readable storage medium stores a program or instruction, and when the program or instruction is executed by the processor, the steps of the rapid metal printing method based on machine vision feedback as claimed in claim 5 are implemented.

Images