3D printer intelligent control system based on cloud service

Abstract

The present invention discloses a 3D printer intelligent control system based on cloud service, which relates to the field of intelligent 3D printing technology. The system includes: collecting target printing tasks to obtain multi-target constraint conditions; performing multi-target balanced optimization in the control storage unit of the cloud platform to obtain target control parameters; transmitting the target control parameters to the cloud platform to perform printing control; performing printing parameter mapping analysis to determine the associated control parameter set; performing parameter adjustment index analysis to obtain the parameter adjustment index set; performing parameter optimization on the associated control parameter set to generate a target adjustment parameter set; and performing printing parameter adjustment on the target 3D printer. The present invention solves the technical problems of the lack of intelligence, low printing accuracy and efficiency of 3D printing in the prior art, and achieves the technical effects of realizing intelligent printing, improving the accuracy and quality of 3D printing, and optimizing printing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent 3D printing, and in particular to an intelligent control system for a 3D printer based on cloud services. Background Art

[0002] As an emerging manufacturing method, 3D printing technology has been widely used in many fields such as medicine, aviation, construction, and automobiles since its implementation in the 1990s. It processes objects layer by layer and point by point through digital models, making it possible to manufacture parts with complex shapes. However, current 3D printing technology still faces many challenges.

[0003] The existing technology has technical problems such as lack of intelligence, low printing accuracy and efficiency in 3D printing. Summary of the invention

[0004] The present application provides a cloud service-based 3D printer intelligent control system, which is used to solve the technical problems of lack of intelligence, low printing accuracy and efficiency in the prior art 3D printing.

[0005] In view of the above problems, the present application provides a 3D printer intelligent control system based on cloud services.

[0006] The present application provides a 3D printer intelligent control system based on cloud services, the system comprising:

[0007] A multi-objective constraint condition acquisition module, the multi-objective constraint condition acquisition module is used to collect target printing tasks, perform multi-objective extraction on the target printing tasks, and obtain multi-objective constraint conditions, wherein the multi-objective constraint conditions include size constraints, self-supporting structure constraints, and manufacturing defect constraints; a target control parameter acquisition module, the target control parameter acquisition module performs multi-objective balanced optimization in the control storage unit of the cloud platform based on the size constraints, self-supporting structure constraints, and the manufacturing defect constraints to obtain target control parameters; a printing control module, the printing control module is used to transmit the target control parameters to the control unit of the cloud platform, use the control unit to perform printing control on the target 3D printer, and use the quality inspection module to perform real-time surface defect recognition on the target printed product to generate A surface defect type set and a surface defect feature value set; an associated control parameter set determination module, the associated control parameter set determination module is used to perform printing parameter mapping analysis according to the surface defect type set, and determine the associated control parameter set; an adjustment parameter index set acquisition module, the adjustment parameter index set acquisition module is used to traverse the surface defect feature value set and the preset standard feature value set to perform adjustment parameter index analysis and obtain the adjustment parameter index set; a target adjustment parameter set generation module, the target adjustment parameter set generation module performs parameter optimization on the associated control parameter set based on the adjustment parameter index set, and generates a target adjustment parameter set; a printing adjustment parameter module, the printing adjustment parameter module is used to send the target adjustment parameter set to the control unit to perform printing adjustment on the target 3D printer.

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

[0009] A multi-objective constraint condition acquisition module is used to collect target printing tasks, perform multi-objective extraction on the target printing tasks, and obtain multi-objective constraint conditions; a target control parameter acquisition module performs multi-objective balanced optimization in the control storage unit of the cloud platform based on the size constraints, self-supporting structure constraints and manufacturing defect constraints to obtain target control parameters; a printing control module is used to transmit the target control parameters to the control unit of the cloud platform, use the control unit to perform printing control on the target 3D printer, and use the quality inspection module to perform real-time surface defect identification on the target printed product to generate a surface defect type set and a surface defect feature value set; an associated control parameter set determination module is used to perform printing parameter mapping analysis according to the surface defect type set to determine the associated control parameter set; a parameter adjustment index set acquisition module is used to traverse the surface defect feature value set and the preset standard feature value set to perform parameter adjustment index analysis to obtain the parameter adjustment index set; a target adjustment parameter set generation module performs parameter optimization on the associated control parameter set based on the parameter adjustment index set to generate a target adjustment parameter set; a printing parameter adjustment module is used to send the target adjustment parameter set to the control unit to perform printing parameter adjustment on the target 3D printer. The technical effect of realizing intelligent printing, improving the accuracy and quality of 3D printing, and optimizing printing efficiency has been achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 A schematic diagram of the structure of a 3D printer intelligent control system based on cloud services is provided for an embodiment of the present application.

[0012] Explanation of the accompanying drawings: multi-objective constraint condition acquisition module 10, target control parameter acquisition module 20, printing control module 30, associated control parameter set determination module 40, parameter adjustment index set acquisition module 50, target adjustment parameter set generation module 60, printing parameter adjustment module 70. DETAILED DESCRIPTION

[0013] The present application provides a cloud-based 3D printer intelligent control system to solve the technical problems of the lack of intelligence, low printing accuracy and efficiency in the prior art 3D printing. The technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of them. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0014] Embodiment 1, as Figure 1 As shown, the present application provides a 3D printer intelligent control system based on cloud services, the system comprising:

[0015] A multi-objective constraint condition acquisition module 10 is used to collect target printing tasks, perform multi-objective extraction on the target printing tasks, and obtain multi-objective constraint conditions, wherein the multi-objective constraint conditions include size constraints, self-supporting structure constraints, and manufacturing defect constraints.

[0016] Specifically, the main function of the multi-objective constraint condition acquisition module 10 is to collect target printing tasks, which contain various detailed requirements and information about the objects to be printed, and perform multi-objective extraction on these tasks. In this process, focus on and identify multiple key constraints. Dimensional constraints refer to strict regulations on the length, width, height, thickness and other dimensions of the printed object to ensure that the size of the printed object meets specific design requirements. Self-supporting structure constraints involve the ability of the printed object to maintain its own structure stable and intact without the need for additional supporting materials. This is particularly important for printed objects of complex shapes, affecting the feasibility and efficiency of printing. Production defect constraints are restrictions and requirements on quality problems such as holes, cracks, and deformations to ensure the quality and performance of printed products. Ensure that the size of the printed object is highly consistent with the design requirements, reduce dimensional errors, and improve printing accuracy.

[0017] The target control parameter acquisition module 20 performs multi-objective balanced optimization in the control storage unit of the cloud platform based on the size constraint, the self-supporting structure constraint and the manufacturing defect constraint to obtain the target control parameters.

[0018] Specifically, the target control parameter acquisition module 20 performs complex calculations and analyses in the control storage unit of the cloud platform based on the information such as the size constraints, self-supporting structure constraints and manufacturing defect constraints extracted by the multi-objective constraint condition acquisition module. Multi-objective balanced optimization is a process of comprehensively considering multiple interrelated and potentially conflicting goals and finding the optimal solution. In this module, a large number of control parameter combinations are evaluated and compared to weigh multiple aspects such as dimensional accuracy, stability of the self-supporting structure, and minimization of manufacturing defects. The target control parameters finally obtained are the result of careful calculation and optimization, which can meet the multi-objective constraints to the greatest extent, provide ideal parameter settings for subsequent printing control, and significantly improve the quality of 3D printed products.

[0019] The printing control module 30 is used to transmit the target control parameters to the control unit of the cloud platform, use the control unit to control the printing of the target 3D printer, and use the quality inspection module to perform real-time surface defect identification on the target printed product to generate a surface defect type set and a surface defect feature value set.

[0020] Specifically, the primary task of the printing control module 30 is to accurately transmit the target control parameters obtained above to the control unit of the cloud platform. This transmission process ensures that the control unit can receive the latest and most optimized parameter settings. The control unit of the cloud platform is used to implement precise printing control of the target 3D printer, which means that the control unit will adjust and monitor the key parameters such as the nozzle movement speed, material extrusion amount, heating temperature, etc. of the 3D printer in real time according to the received target control parameters to ensure that the printing process proceeds as expected. During the printing process, the quality inspection module plays an important role. It will perform real-time surface defect recognition on the target product being printed. Through detection, a set of surface defect types is generated to determine their types and possible causes. Common surface defects include but are not limited to: rough surface, caused by excessively large printing layer thickness, too fast printing speed or uneven material extrusion; cracks or step lines, caused by large tolerance of consumables, unstable extrusion temperature, incorrect assembly of the nozzle heating block or problems with the machine motion components; wire drawing, caused by the failure to retract the filaments formed by the nozzle during movement in time; spots or depressions, caused by excessively high printing temperature, damp consumables or impurities in the nozzle. At the same time, a surface defect feature set is generated to provide a specific quantitative description for each defect type, such as the length, width, depth of the crack, and the specific value of the roughness. The defect information detected in real time can be quickly fed back to the system, allowing the system to adjust the printing parameters in time, make real-time corrections and optimizations, and improve the stability of printing quality.

[0021] The associated control parameter set determination module 40 is used to perform printing parameter mapping analysis according to the surface defect type set to determine the associated control parameter set.

[0022] Specifically, after receiving the surface defect type set, the associated control parameter set determination module 40 will conduct an in-depth printing parameter mapping analysis. This analysis aims to find the intrinsic correlation between the surface defect type and the printing control parameters. For example, the surface defect type set includes "surface roughness". After analysis, the module determines that the control parameter set related to the defect includes the printing layer thickness setting, printing speed or material extrusion speed. Through this precise mapping analysis, it is ensured that the determined associated control parameter set can effectively target the surface defects that have occurred, and provide a clear direction for subsequent parameter adjustment and optimization. By analyzing the surface defect type and accurately determining the control parameters related to it, it is helpful to quickly find the root cause of the defect and improve the stability of printing quality.

[0023] The parameter index set acquisition module 50 is used to traverse the surface defect feature value set and the preset standard feature value set to perform parameter index analysis to obtain the parameter index set.

[0024] Specifically, the parameter index set acquisition module 50 acquires a surface defect feature value set, which includes various detailed quantitative descriptions of the surface defects of the printed product. At the same time, a standard feature value set is preset in the system, representing an ideal, defect-free or acceptable feature value range. Then a detailed traversal and analysis process is started, and each feature value in the surface defect feature value set is taken out one by one. For each feature value taken out, it will find the corresponding standard feature value in the preset standard feature value set, and compare and analyze this pair of surface defect feature values ​​and the preset standard feature value. Exemplarily, if the surface defect feature value is "surface roughness is 2 microns", and the preset standard feature value is "surface roughness is 1 micron", a value representing the degree of roughness deviation is calculated, such as a difference of 1 micron, or a ratio of 2 times. After performing such analysis and calculation on each pair of feature values, the module will organize all the analysis results obtained into a set, which is the parameter index set. The parameter index set provides a specific and quantitative basis for subsequent parameter adjustment and optimization, helping the system to adjust the 3D printing process more accurately to reduce surface defects and improve printing quality.

[0025] The target adjustment parameter set generation module 60 performs parameter optimization on the associated control parameter set based on the parameter adjustment index set to generate a target adjustment parameter set.

[0026] Specifically, the target adjustment parameter set generation module 60 undertakes the important task of generating precise adjustment parameters for the 3D printing system. For the parameter index set, the ratio of each parameter index to the sum of the indexes of the parameter index set is calculated respectively. Through such calculation, multiple parameter adjustment factors are generated. Each parameter adjustment factor corresponds to a specific parameter adjustment index, which reflects the relative importance or weight of the parameter adjustment index in the entire parameter adjustment requirement. A special parameter optimization network layer is used to conduct an in-depth analysis of these parameter adjustment factors and associated control parameter sets. During the analysis process, the network layer will continuously try different parameter combinations and evaluate the effect of each combination on meeting the parameter adjustment requirement based on the parameter adjustment factor. Through a large number of calculations and comparisons, the parameter combination that can optimally balance the various parameter adjustment requirements is finally determined to generate a target adjustment parameter set. It helps to make the 3D printing process more stable and reduce printing interruptions or quality fluctuations caused by improper parameter adjustment.

[0027] The printing parameter adjustment module 70 is used to send the target adjustment parameter set to the control unit to perform printing parameter adjustment on the target 3D printer.

[0028] Specifically, the printing parameter adjustment module 70 is the final execution link of the entire 3D printing adaptive optimization control system. The main responsibility of this module is to accurately send the target adjustment parameter set generated by the target adjustment parameter set generation module 60 to the control unit. After receiving these parameters, the control unit will perform real-time printing parameter adjustment on the target 3D printer based on them. This means that the key parameters of the 3D printer, such as the nozzle movement speed, material extrusion speed, printing temperature, filling density, etc., will be changed accordingly according to the settings in the target adjustment parameter set. For example, if the target adjustment parameter set indicates that the nozzle movement speed needs to be reduced to improve printing accuracy, the printing parameter adjustment module 70 will pass this parameter change to the control unit, and the control unit will then adjust the nozzle movement speed. Through this timely and accurate parameter transmission and adjustment, it is ensured that the 3D printer can apply the optimized parameters in the subsequent printing process, thereby improving the printing quality, reducing defects, and improving the overall printing effect.

[0029] In a possible implementation, the target control parameter acquisition module 20 further includes:

[0030] Using the size constraint, the self-supporting structure constraint and the manufacturing defect constraint as indexes, frequent item mining is performed in the cloud server to generate a topology structure optimization control library; the topology structure optimization control library is stored in the control storage unit, and the size constraint, the self-supporting structure constraint and the manufacturing defect constraint are used as retrieval conditions to perform retrieval in the topology structure optimization control library to generate a size retrieval control parameter set, a self-supporting structure retrieval control parameter set and a manufacturing defect retrieval control parameter set; the size retrieval control parameter set, the self-supporting structure retrieval control parameter set and the manufacturing defect retrieval control parameter set are summarized to generate a balanced optimization space; multi-objective balanced optimization is performed in the balanced optimization space to generate the target control parameters.

[0031] Specifically, when the system receives the three key conditions of size constraint, self-supporting structure constraint and manufacturing defect constraint, it will use them as indexes or keywords to mine the huge data resources in the cloud server. The cloud server stores a large amount of relevant data of past 3D printing projects, including various printing parameters, structural design, material usage and final printing results. Frequent item mining is a data mining technology that aims to discover frequently occurring patterns or combinations in data. In this scenario, the system will look for frequently occurring parameter combinations and design patterns based on three aspects: size constraint, self-supporting structure constraint and manufacturing defect constraint. For example, for size constraint, the system will find out the combination of parameters such as printing speed, material type, nozzle temperature, etc. that often appear in printing projects that meet specific size requirements. Through such separate mining, the system will organize and summarize the collected related frequent items to form a control library specifically for 3D printing topology optimization.

[0032] The carefully generated topology optimization control library is safely stored in the control storage unit. This control storage unit can be a specific storage area of ​​the local server or a storage space in the cloud. The size constraint, self-supporting structure constraint and manufacturing defect constraint are used as clear search conditions to carry out precise search operations in the stored topology optimization control library. When the size constraint is used as the search condition, the system searches the control library for all control parameter records related to meeting specific size requirements. These records include parameters such as specific printing resolution settings, material dosage control, and nozzle movement accuracy. After screening and sorting, a size retrieval control parameter set is generated. The parameters in this set are all effective control parameters for achieving specific size goals. For the retrieval of self-supporting structure constraints, the system will look for parameter information related to building a structure that can stand stably without additional support, such as specific geometric shape design parameters, internal filling density settings, and structural center of gravity control. Finally, a self-supporting structure retrieval control parameter set is formed. When using manufacturing defect constraints as search conditions, the system will focus on the parameters in the control library that are related to reducing various types of manufacturing defects, such as adjusting the printing temperature to avoid defects caused by overheating or overcooling of the material, controlling the printing speed to ensure uniform deposition of the material, etc., thereby generating a set of manufacturing defect retrieval control parameters.

[0033] The size retrieval control parameter set, the self-supporting structure retrieval control parameter set and the manufacturing defect retrieval control parameter set represent the optimal or effective parameter ranges in terms of size accuracy, self-supporting structure stability and manufacturing defect control, respectively. When they are summarized, the parameter ranges of these different aspects are actually combined together. After summarizing, a comprehensive set of all these different types of parameters is obtained, which constitutes a broad parameter space, namely the equilibrium optimization space. In this space, each control parameter is regarded as an independent particle, which is distributed in the space, and each particle represents a possible solution. By generating such an equilibrium optimization space, the system provides a comprehensive search range for the subsequent multi-objective equilibrium optimization process, which allows the system to explore various possible parameter combinations under the consideration of multiple constraints and objectives such as size, self-supporting structure and manufacturing defects, so as to find an optimal solution that can achieve the best balance among these aspects.

[0034] The mean shift algorithm is an iterative algorithm based on kernel density estimation, which is used to find the mode (density peak) of data points in the data space. When the mean shift algorithm is applied in the equilibrium optimization space, some initial points are first randomly selected in this space as candidate control parameter combinations. Then, for each candidate point, the weighted average position of other points within a certain range around it is calculated, where the weight is inversely proportional to the distance. This weighted average position is the mean shift vector. Next, the candidate point is moved a certain distance in the direction of the mean shift vector. After multiple iterations, the candidate points will gradually converge to the area with higher density, which is the possible optimal solution area. In the context of multi-objective equilibrium optimization, the evaluation of each point (control parameter combination) is not only based on a single objective, but also comprehensively considers multiple objectives such as dimensional accuracy, self-supporting structure stability and manufacturing defect control. After a series of iterative movements, the point or set of points that finally converges is the control parameter combination that achieves relative balance and optimization on multiple objectives, that is, the generated target control parameters.

[0035] In a possible implementation, the target control parameter acquisition module 20 further includes:

[0036] Extract the centroid particle in the balanced optimization space, take the centroid particle as the starting point, and construct a first search area according to a preset balanced search step, wherein the balanced optimization space includes multiple particles; perform centroid iteration on the centroid particle in combination with the centroid particle coordinates of the centroid particle in the first search area to determine the stage centroid particle, wherein the stage centroid particle includes the stage particle coordinates; perform centroid update iteration based on the stage centroid particle and the stage particle coordinates until a preset number of iterations is met to obtain a target centroid particle; and use the retrieval control parameter corresponding to the target centroid particle as the target control parameter.

[0037] Specifically, in the equilibrium optimization space, the centroid particle of the particle distribution is first calculated by the mean shift algorithm. The mean shift algorithm determines the center of gravity based on the position and density distribution of the particles. After obtaining the centroid particle, use it as the starting point. The preset equilibrium search step determines the size range of the search area. Assume that the position of each particle in space can be represented by coordinates (x, y, z), the coordinates of the centroid particle are (x_c, y_c, z_c), and the step size is d. Then the first search area is a spherical area with a radius of d centered on the coordinates of the centroid particle. In this area, the density of particles is relatively high, which means that it is more likely to find a better combination of control parameters. Through further search and analysis in this area, the efficiency and accuracy of finding the optimal solution can be improved.

[0038] Determine the coordinates of the centroid particle of the centroid particle, and based on this coordinate, calculate the offsets of other particles in the first search area relative to this centroid particle. By assigning certain weights (inversely proportional to the distance) to these offsets, they are weighted summed. Add the weighted summed offsets to the coordinates of the centroid particle to obtain a new position. This new position is the coordinate of the stage centroid particle after one centroid iteration. Repeat this process continuously, each time recalculating the offsets of the particles in the area based on the newly obtained coordinates of the stage centroid particle and performing weighted summation and accumulation, gradually determining a more accurate stage centroid particle until convergence or reaching the preset number of iterations.

[0039] The stage centroid particle and its stage particle coordinates are obtained, and the position of the stage centroid particle is used as a new starting point, and the calculation is performed again according to the centroid iteration method described above. In each iteration, a new stage centroid particle and its corresponding stage particle coordinates are re-determined based on the particle distribution around the current stage centroid particle. This is continuously updated and iterated round by round, and each time it is closer to the optimal centroid position. After each iteration is completed, it is checked whether the preset number of iterations has been reached. If not, the next round of iterations will be continued; if the preset number of iterations has been reached, the stage centroid particle obtained at this time is determined as the target centroid particle.

[0040] After the target centroid particle is finally determined through a series of iterative operations, the search control parameters corresponding to the particle in the initial equilibrium optimization space are selected and identified as the target control parameters that can achieve multi-objective balance and optimization. In the 3D printing scenario, each particle in the initial equilibrium optimization space represents a specific set of control parameters such as printing speed, nozzle temperature, and material usage. After multiple centroid iterations, the target centroid particle is found, and the specific set of parameters such as printing speed, nozzle temperature, and material usage associated with this particle are extracted as the final target control parameters for actual printing operations, in the hope of achieving the best comprehensive effect in terms of dimensional accuracy, self-supporting structure stability, and manufacturing defect control.

[0041] In a possible implementation, the target control parameter acquisition module 20 further includes:

[0042] Construct a centroid iteration formula, wherein the centroid iteration formula is: ;in, is the stage center-of-mass particle, is the center-of-gravity particle, is the tth particle in the first search area, is the number of particles distributed in the first search area, is an integer greater than or equal to 1, is an empirical parameter; the centroid particle coordinates of the centroid particle are input into the center of mass iteration formula for calculation to generate the stage particle coordinates; and the stage centroid particle is determined according to the stage particle coordinates.

[0043] Specifically, we construct a centroid iteration formula, in which is the phase center particle, which is the current The center of gravity particle and other related parameters are calculated; represents the tth particle in the first search area; represents the number of particles distributed in the first search area, and is an integer greater than or equal to 1, It is an empirical parameter, which plays a regulating role in the formula.

[0044] The specific calculation process is: Substitute the coordinates of the centroid particle into the centroid iteration formula. First, for each particle in the first search area , calculate its difference with the current center of gravity particle The difference , and then multiply this difference by , and sum it for all particles. Finally, multiply the sum by , and accumulated to the current center of gravity particle The coordinates of the stage particles are obtained.

[0045] According to the calculated stage particle coordinates, the specific position of the stage centroid particle can be determined. This stage centroid particle will serve as the basis for the next iteration, or as the final target centroid particle when the preset number of iterations is met. By repeating this process, the centroid is updated and iterated until the preset number of iterations is met, and finally the target centroid particle is obtained.

[0046] In a possible implementation, the printing control module 30 further includes:

[0047] Call multiple cameras in the quality inspection module to capture images of the target printed product from multiple preset angles to obtain multiple sub-angle surface images; identify overlapping areas of the multiple sub-angle surface images to obtain multiple identification overlapping area sets, wherein the multiple identification overlapping area sets have multiple overlapping image mapping identification sets; identify surface defects of the multiple sub-angle surface images to generate an initial surface defect type set and an initial surface defect feature value set; calculate the defect feature value mean based on the multiple overlapping image mapping identification sets and the multiple identification overlapping area sets to determine the surface defect type set and the surface defect feature value set.

[0048] Specifically, in this step, multiple cameras in the quality inspection module will be activated to capture images of the target printed product from multiple pre-set angles. Each camera will capture a portion of the surface image of the target printed product. Due to different angles, these images show different parts of the product surface. In this way, multiple sub-angle surface images can be obtained, which together constitute a comprehensive description of the surface of the target printed product. For example, suppose there are three cameras, shooting from the front, side and top angles, then three sub-angle surface images of the front, side and top will be obtained. These images can provide surface information of the product at different angles, which is helpful for the subsequent detection and analysis of product surface defects.

[0049] First, the system acquires multiple angled surface images, which are obtained by taking pictures of the target printed product from different angles by multiple cameras in the quality inspection module. Then, for each image, the system compares it with other images one by one. During the comparison process, image matching and feature recognition technology are used to determine whether there is an overlapping area between the two images. When an overlapping area is found, the system assigns a unique identifier to the overlapping area and adds it to the set of identifier overlapping areas. For example, if image A and image B overlap in a certain area, the system creates an identifier, such as overlapping area 1, and adds it to the set. At the same time, for each identifier overlapping area, the system records the associated overlapping image mapping identifier. This means that the system clearly indicates which two images overlap in this area. For example, for overlapping area 1, the system records the mapping identifiers of image A and image B. In this way, by comparing and analyzing all angled surface images, the system can obtain multiple identifier overlapping area sets, and each set has a corresponding overlapping image mapping identifier set, so as to clearly understand the overlap between each image.

[0050] The system uses various image processing and analysis techniques to conduct in-depth inspections of multiple sub-angle surface images to identify defects on the surface of the target printed product shown in the image. Each sub-angle surface image is analyzed pixel by pixel, and each area in the image is compared with the preset defect feature pattern to determine whether there is a defect. If the characteristics of a certain area match the feature pattern of a certain defect type, the area will be marked as having the corresponding type of defect. After identifying the defect type, the system will also record the feature values ​​of these defects. These feature values ​​can include information such as the size, shape, color, texture, etc. of the defect, which can describe the specific characteristics of the defect in more detail. Through defect identification and feature value recording of all sub-angle surface images, the system will eventually generate two sets: the initial surface defect type set and the initial surface defect feature value set. The initial surface defect type set contains all detected defect types, and the initial surface defect feature value set correspondingly contains the specific feature value information of each defect. This processing process can comprehensively and accurately detect defects on the surface of the target printed product and provide important data support for subsequent defect evaluation and processing.

[0051] The system obtains multiple overlapping image mapping identification sets and multiple identification overlapping area sets. These sets contain the previously identified image overlap information and the identified overlapping areas. Then, for each identified overlapping area, the system will search for the corresponding overlapping image mapping identification set to determine the images that overlap in the overlapping area. Next, the system will extract the defect feature values ​​of the area in these overlapping images. Since images at different angles may have different feature value detection results for the same defect, it is necessary to perform mean calculation to obtain more accurate defect feature values. Specifically, the system will sum the feature values ​​of the same defect in different overlapping images and divide them by the number of overlapping images to obtain the mean feature value of the defect in the overlapping area. By performing mean calculation on the defect feature values ​​of all identified overlapping areas, the system finally determines the surface defect type set and the surface defect feature value set. The surface defect type set contains the integrated and confirmed defect types, while the surface defect feature value set contains more accurate defect feature values ​​after mean calculation. For example, for a certain overlapping area, there are three images that overlap in the area and all detect the same defect. The characteristic values ​​of the defect in these three images are A1, A2, and A3 respectively, so the average characteristic value of the defect in this overlapping area is (A1+A2+A3) / 3. By performing such calculations on defects in all overlapping areas, the final surface defect type set and surface defect characteristic value set are obtained.

[0052] In a possible implementation, the printing control module 30 further includes:

[0053] Constructing a defect recognition network layer; inputting the multiple angle-divided surface defect feature sets into the defect recognition network layer to obtain the initial surface defect type set and the initial surface defect feature value set.

[0054] Specifically, building a defect recognition network layer is a key step. Collect a large amount of image data containing surface defects, annotate these data, and mark information such as the type and location of the defects. According to the task requirements and data characteristics, select a suitable neural network architecture, such as convolutional neural network (CNN), recurrent neural network (RNN) or other deep learning architectures. According to the complexity of the data and the limitations of computing resources, determine the number of network layers and the number of neurons in each layer. Use the prepared data to train the network, and adjust the weights and biases of the network so that the network can learn the characteristics and patterns of the defects. Select a suitable optimization algorithm, such as stochastic gradient descent (SGD), to accelerate the network training process and improve the training effect. During the training process, use the validation set to evaluate the network, and adjust the parameters and architecture of the network based on the evaluation results to improve the performance of the network.

[0055] The defect recognition network layer that has been built is in a state of preparing to receive data, and the multiple angle-based surface defect feature sets extracted previously are input into this defect recognition network layer. These feature sets contain various feature information about surface defects obtained from different angles. The defect recognition network layer will process and analyze these input feature sets. It will use the knowledge and patterns learned in the training phase to identify and judge these features. Through the calculation and judgment of the network layer, the initial surface defect type set and the initial surface defect feature value set are finally output. The initial surface defect type set contains various defect types identified by the network layer. The initial surface defect feature value set contains the feature values ​​corresponding to these defect types, which can describe the specific attributes of the defects, such as the size, shape, depth, etc. of the defects. By inputting the angle-based surface defect feature set into the defect recognition network layer, the automatic recognition and feature extraction of surface defects can be realized, which provides a basis for subsequent defect analysis and processing.

[0056] In a possible implementation, the target adjustment parameter set generating module 60 further includes:

[0057] Respectively calculate the ratio of each tuning parameter index in the tuning parameter index set to the sum of the indices of the tuning parameter index set to generate multiple tuning parameter factors, wherein each tuning parameter factor corresponds to a tuning parameter index; and use a parameter optimization network layer to analyze the multiple tuning parameter factors and the associated control parameter set to generate the target adjustment parameter set.

[0058] Specifically, the parameter adjustment index set is a set of index values ​​related to parameter adjustment, for example, it may contain values ​​representing different defect levels or performance indicators. Calculate the sum of the exponents of the parameter adjustment index set, add all the parameter adjustment indexes in the set, and get a sum value. For each parameter adjustment index in the parameter adjustment index set, perform the following calculation: divide the parameter adjustment index by the sum of the exponents to get a ratio, which is the corresponding parameter adjustment factor. In this way, the parameter adjustment index can be converted into a relative proportional parameter adjustment factor, providing a basis for subsequent parameter optimization and adjustment.

[0059] The parameter optimization network layer is a model with a specific structure and function. It can process multiple input tuning factors and associated control parameter sets, and generate a target tuning parameter set through a series of calculations and analyses. In specific operations, the parameter optimization network layer associates and matches each tuning factor with the parameters in the associated control parameter set. It considers the degree of influence of the tuning factor on each control parameter, as well as the interaction between different control parameters. The parameter optimization network layer is implemented using a genetic algorithm, and intelligent search and optimization are performed based on the input data to find the optimal parameter combination. During the analysis process, the network layer continuously adjusts and optimizes the values ​​of each control parameter to optimize the objective function. The objective function is usually defined based on specific application scenarios and requirements, such as minimizing the defect rate, maximizing printing accuracy, etc. Through multiple iterations and optimizations, the parameter optimization network layer will eventually generate a target adjustment parameter set, which can automatically find the optimal parameter settings based on specific requirements and input data, thereby improving the performance and quality of 3D printing.

[0060] In a possible implementation, the target adjustment parameter set generating module 60 further includes:

[0061] A feedback monitoring window is set; stability monitoring is performed on the target adjustment parameter set within the feedback monitoring window to generate a stability monitoring factor set; the stability monitoring factor set is identified according to preset stability reliable monitoring factors to determine an abnormal adjustment parameter set; and an early warning instruction is generated based on the abnormal adjustment parameter set.

[0062] Specifically, the setting of the feedback monitoring window is to effectively monitor and analyze the target adjustment parameter set. Specifically, it includes the following considerations: time range, determining the monitoring time period, such as monitoring once every certain period of time, or continuous monitoring within a specific time period; data volume, specifying the amount of data to be processed within the monitoring window, such as monitoring a certain number of samples or data points; monitoring frequency, setting the monitoring frequency, that is, how often monitoring is performed to ensure that parameter changes and abnormal conditions can be discovered in a timely manner; related indicators, clarifying the indicators or parameters that need to be paid attention to within the monitoring window, in addition to the target adjustment parameter set itself, it may also include other indicators related to system performance; trigger conditions, setting the conditions for triggering monitoring, such as starting the feedback monitoring window when certain events occur in the system or certain thresholds are reached. By reasonably setting the feedback monitoring window, the system's operating status information can be obtained in a timely manner, providing a basis for subsequent stability monitoring and early warning.

[0063] In the feedback monitoring window, the system will conduct detailed stability monitoring of the target adjustment parameter set. This process usually includes analysis of the parameter change trend, fluctuation, and deviation from the expected value. The system will use a series of algorithms and methods to evaluate the stability of the parameters, for example, calculating statistical indicators such as the mean, variance, and standard deviation of the parameters to reflect the concentration and dispersion of the parameters. In addition, the parameters' change rate, periodicity and other characteristics will also be analyzed. Based on the results of these monitoring and analysis, the system will generate a set of stability monitoring factors. Each factor in this set represents the stability characteristics of the target adjustment parameter set in a certain aspect, for example, including mean stability factor, variance stability factor, change rate stability factor, etc. These stability monitoring factors can help the system understand the stability of the target adjustment parameter set more comprehensively and provide a basis for subsequent judgment and decision-making.

[0064] The preset stability reliability monitoring factors are pre-set standards or thresholds for measuring stability. These factors are determined based on the normal operating range of the system, historical data, experience or theoretical models. Compare and analyze the previously generated stability monitoring factor set with these preset stability reliability monitoring factors. For each factor in the stability monitoring factor set, check whether it exceeds the preset normal range or threshold. If a stability monitoring factor does not meet the preset reliability standard, it means that the adjustment parameters related to it may be abnormal. By identifying and judging all stability monitoring factors one by one, determine the adjustment parameters that cause the monitoring factors to be abnormal, and combine these abnormal adjustment parameters together to form an abnormal adjustment parameter set.

[0065] After determining the abnormal adjustment parameter set, the system will generate corresponding early warning instructions according to the specific conditions of these abnormal parameters. First, the system will conduct a detailed analysis of the abnormal adjustment parameter set, including the type of abnormal parameters, the degree of numerical deviation, and the possible impact. According to the analysis results, the specific content of the early warning instruction is generated. The early warning instruction contains the following information: clearly indicate which parameters are abnormal, such as abnormal printing temperature parameters. Describe the specific deviation of the abnormal parameters, such as the printing temperature is 30 degrees Celsius lower than the normal range. Estimate the consequences of abnormal parameters, such as product quality degradation and equipment damage. Give corresponding recommended measures, such as immediately stopping production and checking the temperature control system. The early warning instruction can be generated in various forms, such as a warning message popped up on the system interface, a text message or email sent to the mobile phone of the relevant personnel, and an audible and visual alarm on the equipment. By generating clear and unambiguous early warning instructions, relevant personnel can be reminded of abnormal conditions in a timely manner and take effective measures to solve the problem to avoid more serious consequences.

[0066] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0067] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0068] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A 3D printer intelligent control system based on cloud services, characterized in that: The system comprises: A multi-objective constraint condition acquisition module, which is used to collect target printing tasks, perform multi-objective extraction on the target printing tasks, and obtain multi-objective constraint conditions, wherein the multi-objective constraint conditions include size constraints, self-supporting structure constraints, and manufacturing defect constraints; A target control parameter acquisition module, wherein the target control parameter acquisition module performs multi-target balanced optimization in a control storage unit of a cloud platform based on the size constraint, the self-supporting structure constraint, and the manufacturing defect constraint to obtain target control parameters; A printing control module, the printing control module is used to transmit the target control parameters to the control unit of the cloud platform, use the control unit to control the target 3D printer to print, and use the quality inspection module to perform real-time surface defect recognition on the target printed product to generate a surface defect type set and a surface defect feature value set; An associated control parameter set determination module, the associated control parameter set determination module is used to perform printing parameter mapping analysis according to the surface defect type set to determine an associated control parameter set; A parameter index set acquisition module, wherein the parameter index set acquisition module is used to traverse the surface defect feature value set and the preset standard feature value set to perform parameter index analysis to obtain a parameter index set; a target adjustment parameter set generation module, wherein the target adjustment parameter set generation module performs parameter optimization on the associated control parameter set based on the parameter adjustment index set to generate a target adjustment parameter set; A printing parameter adjustment module, the printing parameter adjustment module is used to send the target adjustment parameter set to the control unit to perform printing parameter adjustment on the target 3D printer; The execution steps of the target control parameter acquisition module include: Using the size constraint, the self-supporting structure constraint and the manufacturing defect constraint as indexes, respectively performing frequent item mining in the cloud server to generate a topology structure optimization control library; The topology structure optimization control library is stored in the control storage unit, and the size constraint, the self-supporting structure constraint and the manufacturing defect constraint are used as retrieval conditions respectively to search in the topology structure optimization control library to generate a size retrieval control parameter set, a self-supporting structure retrieval control parameter set and a manufacturing defect retrieval control parameter set; Summarizing the size retrieval control parameter set, the self-supporting structure retrieval control parameter set, and the manufacturing defect retrieval control parameter set to generate a balanced optimization space; Perform multi-objective equilibrium optimization in the equilibrium optimization space to generate the target control parameters; Extracting a centroid particle in the balanced optimization space, taking the centroid particle as a starting point, and constructing a first search area according to a preset balanced search step length, wherein the balanced optimization space includes a plurality of particles; In the first search area, combining the centroid particle coordinates of the centroid particle with the centroid particle, and determining a stage centroid particle, wherein the stage centroid particle includes the stage particle coordinates; Performing mass center update iteration based on the mass center particle of the stage and the particle coordinates of the stage until a preset number of iterations is met to obtain a target mass center particle; Using the retrieval control parameter corresponding to the target centroid particle as the target control parameter; Construct a centroid iteration formula, wherein the centroid iteration formula is: ； in, is the stage center-of-mass particle, is the center-of-gravity particle, is the tth particle in the first search area, is the number of particles distributed in the first search area, is an integer greater than or equal to 1, is an empirical parameter; Inputting the centroid particle coordinates of the centroid particle into the centroid iteration formula for calculation to generate the stage particle coordinates; The stage centroid particle is determined according to the stage particle coordinates.

2. The cloud-based 3D printer intelligent control system according to claim 1, characterized in that: The execution steps of the printing control module include: Calling multiple cameras in the quality inspection module to collect images of the target printed product from multiple preset angles to obtain multiple surface images at different angles; Respectively identifying overlapping regions of the plurality of angle-divided surface images to obtain a plurality of identification overlapping region sets, wherein the plurality of identification overlapping region sets have a plurality of overlapping image mapping identification sets; Performing surface defect recognition on the multiple angle-divided surface images to generate an initial surface defect type set and an initial surface defect feature value set; The defect feature value mean is calculated according to the multiple overlapping image mapping identification sets and the multiple identification overlapping area sets to determine the surface defect type set and the surface defect feature value set.

3. The cloud-based 3D printer intelligent control system according to claim 2, characterized in that: The execution steps of the printing control module include: Build a defect recognition network layer; The multiple angle-divided surface defect feature sets are input into the defect recognition network layer to obtain the initial surface defect type set and the initial surface defect feature value set.

4. The cloud-based 3D printer intelligent control system according to claim 3, characterized in that: The target adjustment parameter set generation module performs the following steps: Respectively calculating the ratio of each tuning parameter in the tuning parameter set to the sum of the indices of the tuning parameter set to generate a plurality of tuning parameters, wherein each tuning parameter factor corresponds to a tuning parameter; The plurality of parameter adjustment factors and the associated control parameter set are analyzed using a parameter optimization network layer to generate the target adjustment parameter set.

5. The cloud-based 3D printer intelligent control system according to claim 1, characterized in that: The target adjustment parameter set generation module performs the following steps: Set feedback monitoring window; Performing stability monitoring on the target adjustment parameter set within the feedback monitoring window to generate a stability monitoring factor set; Identifying the stability monitoring factor set according to preset stability reliability monitoring factors, and determining an abnormal adjustment parameter set; An early warning instruction is generated based on the abnormal adjustment parameter set.

Images