Multi-device collaborative operation control system for jewelry 3D printing

Abstract

The invention relates to the technical field of 3D printing, and discloses a jewelry 3D printing multi-equipment collaborative operation control system which comprises an analysis parameter adjustment module, a model manufacturing module, a scanning inspection module, a data sharing module, a mold construction module, a finished product casting module, a finished product quality inspection module and a recovery module. The analysis parameter adjustment module acquires and sorts triangular patches for a complex structure of a jewelry STL model, and adaptively generates a layered parameter table; the model manufacturing module prints a jewelry resin model based on the layered parameter table; the scanning inspection module inspects the model, a defect identification network is adopted to generate a problem label of an unqualified model, an adjustment scheme is autonomously selected, and the stability and the automation degree of the printing technological process are improved; the finished product quality inspection module is combined with an image recognition technology and a traditional measurement technology to monitor the quality of jewelry finished products, inferior products are smelted through the recovery module, precious metal raw materials are obtained again, the quality of the jewelry finished products is ensured, and the cost of the lost wax casting process is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of 3D printing, and in particular to a multi-device collaborative operation control system for jewelry 3D printing. Background Art

[0002] 3D printing technology is a technology that uses a gradual accumulation of materials to manufacture physical parts. It is a "bottom-up" manufacturing method that can solve the forming of many complex structural parts that were difficult to manufacture in the past and greatly reduce the processing steps. As 3D printing technology matures, it is being applied to various industries, such as jewelry production.

[0003] At present, there are two main technical solutions for 3D printing in the jewelry industry. One is laser sintering technology, which is to directly print precious metal entities and is less used. The other is to use 3D printed resin molds and cast finished products using the lost wax casting method. This solution can achieve rapid mass production and is the mainstream application solution.

[0004] The existing Chinese patent application with publication number CN118144276A proposes a 3D printing system, in which the server receives resin tank model data, automatically switches the consumable type and process parameters, and generates printing data. The device control board is used to send the resin tank model data provided by the control board to the server, and control the printing module to perform printing actions according to the received printing data, thereby improving the intelligence and convenience of the 3D printing system.

[0005] However, jewelry often has a relatively complex structure. The existing solution of generating printing data based on resin materials will cause the loss of the jewelry's precise structure or the stepped surface, which seriously affects the appearance of the jewelry. It is necessary to adaptively generate printing and light-curing data for the complex structure of a single piece of jewelry. In addition, the jewelry 3D printing process involves multiple steps of model printing and lost wax casting, and the existing technology lacks an effective monitoring mechanism to automatically detect defects and make adjustments. Summary of the invention

[0006] The purpose of the present invention is to provide a multi-device collaborative operation control system for jewelry 3D printing, so as to provide an adaptive layered printing and light curing method for complex jewelry structures, and at the same time, through model inspection and finished product quality inspection feedback results, the adjustment of the printing plan and the recycling of precious metal raw materials can be achieved.

[0007] The technical solution to achieve the purpose of the present invention is:

[0008] A multi-device collaborative operation control system for jewelry 3D printing, including an analysis and parameter adjustment module, a model manufacturing module, a scanning and inspection module, a data sharing module, a mold construction module, a finished product casting module, a finished product quality inspection module and a recycling module:

[0009] The parameter analysis module exports the jewelry STL model, obtains and sorts the triangles, and adaptively generates a hierarchical parameter table;

[0010] The model manufacturing module prints and photocures the single slice layer in sequence according to the preset printing parameters and the layered parameter table to obtain the jewelry resin model;

[0011] The scanning inspection module analyzes the jewelry resin model, selects qualified jewelry resin models, assigns problem labels to unqualified jewelry resin models based on the defect recognition network, generates adjustment instructions and transmits them to the model manufacturing module;

[0012] The data sharing module has a built-in tag correspondence table to store and share the uploaded data of other modules within the system;

[0013] The mold building module manufactures jewelry molds using qualified jewelry resin models;

[0014] The finished product casting module pours precious metal raw materials into the jewelry mold. After the precious metal raw materials solidify, they are taken out and mechanically processed to obtain the finished jewelry.

[0015] The finished product quality inspection module inspects finished jewelry and determines whether it is good or bad.

[0016] The recycling module smelts inferior products and regains precious metal raw materials.

[0017] Further, the analysis and parameter adjustment module includes a parsing unit, an adaptive layering unit, a track determination unit, a light adjustment unit and a first communication unit;

[0018] The parsing unit imports the jewelry STL model in the spatial rectangular coordinate system to obtain triangular facets and sort them. triangles to construct an adjacency undirected graph ;

[0019] Adaptive layering unit adaptively adjusts the layer thickness to divide the jewelry STL model into slice layers;

[0020] The orbit determination unit uses NURBS curve fitting to obtain the profile of a single slice layer;

[0021] The dimming unit adaptively adjusts the light-solidification parameters of a single slice layer;

[0022] The first communication unit stores the Layer thickness , No. Slice Layer of Order NURBS curve equation , No. Exposure duration and Light intensity distribution To build a hierarchical parameter table and upload jewelry STL models and adjacency matrices and hierarchical parameter tables to the data sharing module.

[0023] Furthermore, the parsing unit constructs an adjacency undirected graph include:

[0024] Sorting triangles, constructing a point set , build a A zero-matrix of dimension , defined as the adjacency matrix ;

[0025] Lock triangles , get the triangle patch Triangles with two identical vertex coordinates , define the triangle patch With triangles Adjacent and with an edge relationship , set the adjacency matrix No. Line Column Elements and Line Column Elements is 1;

[0026] Traverse the triangle patch to obtain all edge relationships and build an edge set , with the adjacency matrix An adjacency undirected graph is established in the form of .

[0027] Furthermore, adaptive hierarchical unit adaptive division The slice layers include:

[0028] Get the minimum printing thickness of the Model Maker module and maximum printing thickness , get the total height of the jewelry STL model ,set up Plane is the starting tangent plane ,along The starting tangent plane is translated in the positive direction of the axis Get the actual Slice Layer ;

[0029] judge Second remaining model height Is it less than the maximum printing thickness? ;

[0030] like Second remaining model height Greater than or equal to the maximum printing thickness ,along Maximum printing thickness of axis positive direction translation Get the initial Slice Layer , get the initial Slice Layer A collection of intersecting triangles ;

[0031] Traversing a collection of triangles For a single triangle in the image, calculate the normal vector and layering direction of each triangle. , get the angle Minimum angle And calculate the theoretical Layer thickness ;

[0032] Selection Theory Layer thickness and minimum print thickness The larger one is the actual Layer thickness , get the actual Slice Layer ;

[0033] like Second remaining model height Less than the maximum printing thickness And greater than or equal to the minimum printing thickness , then set the actual Layer thickness for Second remaining model height and obtain the actual Slice Layer ,definition ;

[0034] like Second remaining model height Less than the minimum printing thickness , then cancel the actual Slice Layer , get Second remaining model height , set the actual Layer thickness Minimum printing thickness , the rest are the actual Slice Layer , and define .

[0035] Furthermore, the orbit determination unit uses NURBS curve fitting to obtain the profile of a single slice layer including:

[0036] Get and actual Slice Layer A collection of intersecting triangles And calculate The intersection coordinates of the intersection points are used to select the starting triangle based on the adjacency matrix Constructing a collection of triangles The triangle face chain, select the traversal direction, sort Intersection Points ;

[0037] choose NURBS curves of order, based on basis functions The recursive formula determines The node interval of the intersection point is increased by adding the front supplementary node and the rear supplementary node to construct the actual Slice Layer The node vector ;

[0038] Determine the actual Slice Layer The node vector The value of the middle node is obtained Slice Layer of Order NURBS curve equation , and as the layer contour.

[0039] Furthermore, we determine the actual Slice Layer The node vector The values ​​of the midpoints include:

[0040] Assign the values ​​of the preceding supplementary node and the following supplementary node to 0 and 1 respectively;

[0041] Get Intersection Points The total length of the polyline segments ;

[0042] Computational Control Intersection of segment curves Constructed Segment polyline length ;

[0043] No. The node interval of the segment curve is , then the node The value of Total length of polyline segments and total length of polyline segments The ratio of nodes The value of Total length of polyline segments and total length of polyline segments ratio.

[0044] Furthermore, the dimming unit adaptively adjusts the light solid parameters of a single slice layer including:

[0045] Set the maximum print thickness The corresponding exposure time is ;

[0046] According to the actual Layer thickness Maximum printing thickness The ratio of Exposure duration ;

[0047] Further obtain the actual Layer thickness , calculate the Curing depth ;

[0048] Based on Curing depth Calculate the Light intensity distribution .

[0049] Further, the model manufacturing module includes a printing unit, a light curing unit, an adjustment unit and a second communication unit;

[0050] Print unit gets the actual Layer thickness and Slice Layer of Order NURBS curve equation And they are used as the additive thickness and boundary contour of the nozzle, and 3D printing is performed with preset printing parameters;

[0051] Photocuring unit obtains Exposure duration and Light intensity distribution , set the UV laser light curing parameters for light curing;

[0052] The adjustment unit locks the abnormal slice layer based on the adjustment instruction, and adjusts the preset printing parameters and / or exposure time of the abnormal slice layer;

[0053] The second communication unit is used to receive the adjustment instruction.

[0054] Further, the scanning and inspection module includes a scanning unit, a detection and analysis unit, a text generation unit and a third communication unit;

[0055] The scanning unit obtains the resin STL model by scanning the jewelry resin model;

[0056] The detection and analysis unit determines whether there is an abnormality in the resin STL model. If there is no abnormality, it is determined to be a qualified jewelry resin model. If there is an abnormality, a problem label is generated based on the defect recognition network;

[0057] The text generation unit searches the label correspondence table to obtain the adjustment plan corresponding to the problem label, based on the abnormal sequence Get the abnormal slice layer and generate adjustment instructions;

[0058] The third communication unit calls the jewelry STL model and the label correspondence table, and feeds back the adjustment instruction to the model manufacturing module.

[0059] Furthermore, the detection and analysis unit determines whether there is an abnormality in the resin STL model including:

[0060] Obtain the resin STL model in the same way as the analysis and parameter adjustment module triangular patches and sort them;

[0061] Traverse the single triangle face of the resin STL model and the jewelry STL model according to the sorting order;

[0062] Get the resin STL model Triangles The coordinates of the three vertices , get the jewelry STL model Triangles The coordinates of the three vertices , calculate and judge the Deviation of triangles Is it greater than the deviation threshold? If so, then in the abnormal sequence Record the triangle patch number in ;

[0063] After the traversal is completed, the abnormal sequence is counted The number of numbers in , if the ratio of the number of numbers to the number of triangles If the ratio is less than or equal to the ratio threshold, the resin STL model is judged to be a qualified jewelry resin model, otherwise it is judged to be abnormal.

[0064] Furthermore, the detection and analysis unit assigns problem labels through the defect recognition network, including:

[0065] Convert the jewelry STL model and the resin STL model into a first image and a second image;

[0066] Input the first image and the second image into the downsampling subnetwork, reduce the dimension by pooling, further extract deep features by multiple convolutions, and output the first feature and the second feature;

[0067] Input the first feature and the second feature into the residual coding subnetwork, generate the first reconstructed feature and the second reconstructed feature through the attention mechanism, and generate the first code and the second code through residual structure superposition;

[0068] The first code and the second code are input into the linear sub-network, and the aggregate code is generated by combining the linear fully connected layer and the ReLU activation function;

[0069] The aggregated code is input into the upsampling sub-network, and the expanded dimension features are generated through multiple deconvolutions;

[0070] The expanded dimension features are input into the multi-layer perceptual sub-network, and the softmax function is used to nonlinearly activate the multiple linear modulations of the expanded dimension features to generate a probability vector, and the question label is selected to be assigned to the resin STL model.

[0071] Furthermore, the data sharing module includes a storage unit and a fourth communication unit. The storage unit pre-stores a label correspondence table and stores the uploaded data of the remaining units. The fourth communication unit receives the uploaded data of the remaining units and transmits the data called by the remaining units.

[0072] Furthermore, the mold building module fixes the qualified jewelry resin model in the medium, and the qualified jewelry resin model must be fully immersed in the medium, and the heating temperature and heating time are set, and the jewelry mold is heated to obtain.

[0073] Furthermore, the finished product quality inspection module includes an image acquisition unit, an image analysis unit, a parameter measurement unit and an evaluation unit;

[0074] The image acquisition unit captures the surface image of the finished jewelry product;

[0075] The image analysis unit obtains the surface image of the finished jewelry product and determines whether there are defects on the surface of the finished jewelry product through image recognition technology. If there are no defects, the surface qualified label is transmitted to the evaluation unit;

[0076] The parameter measurement unit measures the size and quality of the finished jewelry and calculates the size error and quality error. If the sum of the size error and the quality error is less than or equal to the preset error threshold, the parameter compliance label is transmitted to the assessment unit;

[0077] After the assessment unit obtains the surface qualified label and parameter compliance label of the same finished jewelry product, it will be judged as a superior product, and the rest will be judged as inferior products.

[0078] Compared with the prior art, the present invention has the following significant advantages:

[0079] 1. Aiming at the complex structure of jewelry STL model, obtain and sort the triangular facets, establish the triangular facet adjacency matrix, and use adaptive stratification to adjust the thickness of a single slice layer to avoid the step effect in printing;

[0080] 2. For a single slice layer, NURBS curve fitting is used to generate accurate layer contours. At the same time, the light curing parameters are adjusted in real time based on the layer thickness of the single slice layer to achieve fine printing of jewelry resin models;

[0081] 3. Analyze the jewelry resin model and screen the qualified jewelry resin model, assign problem labels to the unqualified jewelry resin models based on the defect recognition network, select the corresponding adjustment plan based on the problem label, and improve the stability of the process of printing jewelry resin models;

[0082] 4. Combine image recognition technology and traditional measurement technology to determine the quality of finished jewelry products, ensure the quality of finished jewelry products, and recycle and smelt inferior products to regain precious metal raw materials, thereby reducing the cost of jewelry lost wax casting process. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 A schematic diagram of a multi-device collaborative operation control system for jewelry 3D printing;

[0084] Figure 2 This is a flow chart of adaptively dividing slice layers in the present invention;

[0085] Figure 3 A flow chart for obtaining a layer profile of a single slice layer in the present invention;

[0086] Figure 4 This is a diagram of the defect identification network model in the present invention. DETAILED DESCRIPTION

[0087] The present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0088] like Figure 1 As shown, the embodiment provided by the present invention is: a multi-device collaborative operation control system for jewelry 3D printing, including an analysis and parameter adjustment module, a model manufacturing module, a scanning and inspection module, a data sharing module, a mold construction module, a finished product casting module, a finished product quality inspection module and a recycling module:

[0089] The analysis and parameter adjustment module exports the jewelry STL model, obtains and sorts the triangular faces, and adaptively generates a hierarchical parameter table based on the triangular faces;

[0090] The model manufacturing module prints and photocures the single slice layer in sequence according to the preset printing parameters and the layered parameter table to obtain the jewelry resin model;

[0091] The scanning inspection module analyzes the jewelry resin model, selects qualified jewelry resin models, assigns problem labels to unqualified jewelry resin models based on the defect recognition network, generates adjustment instructions and transmits them to the model manufacturing module;

[0092] The data sharing module has a built-in label correspondence table to receive and store the uploaded data from other modules, and respond to the data calls from other modules at any time to realize the sharing of uploaded data among various units in the system;

[0093] The mold building module manufactures jewelry molds using qualified jewelry resin models;

[0094] The finished product casting module pours precious metal raw materials into the jewelry mold. After the precious metal raw materials solidify, they are taken out and mechanically processed to obtain the finished jewelry.

[0095] The finished product quality inspection module inspects finished jewelry and determines whether it is good or bad.

[0096] The recycling module smelts inferior products and regains precious metal raw materials.

[0097] Further, the analysis and parameter adjustment module includes a parsing unit, an adaptive layering unit, a track determination unit, a light adjustment unit and a first communication unit;

[0098] The parsing unit imports the jewelry STL model in the spatial rectangular coordinate system to obtain triangular facets and sort them. triangles to construct an adjacency undirected graph ;

[0099] Adaptive layering unit adaptively adjusts the layer thickness to divide the jewelry STL model into slice layers;

[0100] The orbit determination unit uses NURBS curve fitting to obtain the profile of a single slice layer;

[0101] The light-adjusting unit adaptively adjusts the light-curing parameters based on the layer thickness of a single slice layer;

[0102] The first communication unit builds a hierarchical parameter table and uploads the jewelry STL model and adjacency matrix and hierarchical parameter tables to the data sharing module.

[0103] Furthermore, the parsing unit constructs an adjacency undirected graph include:

[0104] Sorting triangles, constructing a point set ,in, Indicates triangular patches;

[0105] Lock triangles , determine the triangle patch Whether to match the triangle patch Adjacent, if yes, then define a triangular patch With triangles There is an edge relationship between ;

[0106] Traverse the triangles through recursive search, obtain all edge relationships and build edge sets , with the adjacency matrix An adjacency undirected graph is established in the form of .

[0107] Furthermore, the parsing unit determines the triangle patch Whether to match the triangle patch Adjacent triangles and triangular dough In the space rectangular coordinate system, the vertex coordinates are determined. If the triangle With triangles If the coordinates of two vertices coincide, then the triangle With triangles Adjacent.

[0108] Furthermore, the adjacency matrix is a dimensional matrix, adjacency matrix No. Line Column Elements Indicates the triangle patch With triangles Is there an edge relationship? , if it exists, then ,otherwise, , since a triangle cannot be adjacent to itself, the adjacency matrix is a symmetric matrix with all diagonal elements being 0.

[0109] like Figure 2 As shown, further, the adaptive hierarchical unit is adaptively divided The slice layers include:

[0110] Get the minimum printing thickness of the Model Maker module and maximum printing thickness , get the total height of the jewelry STL model ,set up Plane is the starting tangent plane ,along The starting tangent plane is translated in the positive direction of the axis Get the actual Slice Layer ;

[0111] judge Second remaining model height Is it less than the maximum printing thickness? ,in, For the actual Slice Layer The actual Layer thickness;

[0112] like Second remaining model height Greater than or equal to the maximum printing thickness ,Right now ,along Maximum printing thickness of axis positive direction translation Get the initial Slice Layer , get the same as the initial Slice Layer A collection of intersecting triangles ;

[0113] Traversing a collection of triangles For a single triangle in the image, calculate the normal vector and layering direction of each triangle. , get the angle Minimum angle ;

[0114] Theory of Computation Layer thickness , further judging the theory Layer thickness Is it less than the minimum printing thickness? ;

[0115] If the theory Layer thickness Less than the minimum printing thickness , then the actual Layer thickness Otherwise, the actual Layer thickness , get the actual Slice Layer ;

[0116] like Second remaining model height Less than the maximum printing thickness And greater than or equal to the minimum printing thickness ,Right now , then the actual Layer thickness , get the actual Slice Layer , and define this time ;

[0117] like Second remaining model height Less than the minimum printing thickness , then get Second remaining model height , easy to know Second remaining model height satisfy , adjust the actual Layer thickness , then the actual Layer thickness , respectively, to obtain the actual Slice Layer and the actual Slice Layer , and define this time .

[0118] like Figure 3 As shown, further, the trajectory determination unit uses NURBS curve fitting to obtain the layer profile of a single slice layer, including:

[0119] Determine the actual Slice Layer , confirm and actual Slice Layer A collection of intersecting triangles And calculate the intersection coordinates, assuming that a total of This intersection The intersection controls the actual Slice Layer The shape of the layer contour;

[0120] Sorting intersection points, defined as ;

[0121] choose NURBS curve, based on The node interval is confirmed by the intersection points, and the front supplementary nodes and the rear supplementary nodes are added to construct the actual Slice Layer The node vector ;

[0122] Determine the actual Slice Layer The node vector The value of the middle node is obtained Slice Layer of Order NURBS curve equation , and as the layer contour.

[0123] Furthermore, sorting The intersection points include:

[0124] Select a collection of triangles The smallest triangle patch in the sorting order is assumed to be ;

[0125] Based on the adjacency matrix , find the set of triangles Middle and triangle patch Two adjacent triangles are assumed to be and , construct a triangle patch chain ;

[0126] Further based on the adjacency matrix , find the set of triangles Middle and triangle patch Two adjacent triangles, one of which is a known triangle , and another assumption is , expand the triangle patch chain ;

[0127] Until the triangle patch collection All triangles in the search are completed. At this point, the triangle chain must form a closed loop.

[0128] Select the traversal direction, taking the clockwise direction as an example, to move the triangular face As a starting point, the triangle The two intersection points of are sorted as follows: and , and then search to the triangle based on the triangle chain , triangle patch One of the two intersection points must be , and another sorting ;

[0129] Until the triangle face chain searches for the triangle face , triangle patch One of the two intersection points must be , and the other sort is .

[0130] Furthermore, constructing the actual Slice Layer The node vector include:

[0131] Set the actual Slice Layer NURBS curve degree , so the actual Slice Layer Each segment of a NURBS curve needs The intersection control, that is, Segment curve from intersection point Control, the actual Slice Layer The number of curve segments of the NURBS curve is ;

[0132] Due to the actual Slice Layer of Order NURBS curve equation , where the basis function The recursive formula is as follows:

[0133] ,

[0134] ,

[0135] like , then Intersection Points In the node vector The node interval is ,like , we can derive the first Intersection Points In the node vector The node interval is ;

[0136] To ensure the Segment curves are composed only of intersection points Controlled by, but not controlled by other intersections, Segment curve The node interval of the intersection must not intersect with the node interval union of the other intersections. and intersection The node interval union of , therefore, Intersection of segment curves In the node vector The node interval is ;

[0137] Get it in sequence The node interval of the segment curve is obtained Nodes To ensure that the first curve and the The curve segments are just connected and do not overlap. Pre-supplementary nodes and Post-supplementary nodes ;

[0138] because , so construct the node vector .

[0139] Furthermore, we determine the actual Slice Layer The node vector The values ​​of the midpoints include:

[0140] Assign the value of the preceding supplementary node to 0, that is, , assign the value of the post-supplement node to 1, that is ;

[0141] Get Intersection Points The total length of the polyline segments ,in, For the The intersection and The Euclidean distance between the intersection points can be obtained by The intersection and Calculate the coordinates of the intersection points;

[0142] Computational Control Intersection of segment curves Constructed Segment polyline length ;

[0143] No. Intersection of segment curves In the node vector The node interval is , then the node The value of Total length of polyline segments and total length of polyline segments The ratio of ,node The value of Total length of polyline segments and total length of polyline segments The ratio of .

[0144] Furthermore, the dimming unit adaptively adjusts the light solid parameters of a single slice layer including:

[0145] Set the maximum print thickness The corresponding exposure time is ;

[0146] Determine the actual Slice Layer The actual Layer thickness , get the Slice Layer No. Exposure duration ;

[0147] Further determine the actual Slice Layer The actual Layer thickness , get the actual Slice Layer No. Curing depth ;

[0148] Based on Curing depth Calculate the actual Slice Layer No. Light intensity distribution ,in, and They are the curing threshold and longitudinal curing coefficient respectively, which are only related to the photosensitive resin used for printing and can be regarded as constants.

[0149] Furthermore, the first communication unit constructs the hierarchical parameter table including:

[0150] Determine the actual Slice Layer The actual Layer thickness ;

[0151] Get the actual Slice Layer of Order NURBS curve equation ;

[0152] Get the Exposure duration and Light intensity distribution ;

[0153] Corresponding to the actual storage Layer thickness , No. Slice Layer of Order NURBS curve equation , No. Exposure duration and Light intensity distribution To the layered parameter table.

[0154] Further, the model manufacturing module includes a printing unit, a light curing unit, an adjustment unit and a second communication unit;

[0155] Print unit lock actual Slice Layer , get the actual first Layer thickness and Slice Layer of Order NURBS curve equation , and respectively as the additive thickness and boundary contour of the nozzle, and 3D printing is performed with preset printing parameters, the preset printing parameters including nozzle heating temperature and nozzle moving speed;

[0156] The light-fixing unit locks the actual Slice Layer , get the first Exposure duration and Light intensity distribution , set the UV laser curing parameters, and in the actual Slice Layer After printing, light curing is performed;

[0157] The adjustment unit locks the abnormal slice layer based on the adjustment instruction, and adjusts the preset printing parameters and / or exposure time of the abnormal slice layer based on the adjustment scheme in the adjustment instruction;

[0158] The second communication unit receives the adjustment instruction fed back by the scanning and inspection module.

[0159] Further, the scanning and inspection module includes a scanning unit, a detection and analysis unit, a text generation unit and a third communication unit;

[0160] The scanning unit obtains the resin STL model by scanning the jewelry resin model;

[0161] The detection and analysis unit determines whether the resin STL model has any abnormality. If there is no abnormality, it is determined to be a qualified jewelry resin model. If there is an abnormality, a problem label is generated based on the defect recognition network. The problem label is used to indicate the defect type of the jewelry resin model, which includes but is not limited to incomplete fusion, pores, unfused particles, spheroidization, fatigue cracks, and discontinuous solidification.

[0162] The text generation unit searches the label correspondence table, obtains the adjustment plan corresponding to the problem label, and obtains the abnormal sequence The actual slice layer where the triangle patch is located is defined as the abnormal slice layer, and the adjustment instruction is generated;

[0163] The third communication unit calls the jewelry STL model and label correspondence table in the data sharing module, and feeds back the adjustment instruction to the model manufacturing module.

[0164] Furthermore, the detection and analysis unit determines whether there is an abnormality in the resin STL model including:

[0165] Get the resin STL model triangular facets, sorting them in the same way as the analysis and parameter adjustment module;

[0166] Get jewelry STL model triangles, traverse a single triangle according to the sorting order;

[0167] Get the resin STL model Triangles The coordinates of the three vertices , get the jewelry STL model Triangles The coordinates of the three vertices , calculate the triangle patch deviation ,in, is the vertex coordinate and vertex coordinates The Euclidean distance of

[0168] Determine whether the deviation of the triangle face is greater than the deviation threshold. If so, it is considered as non-overlapping. Record the triangle patch number in , otherwise, it is regarded as a triangular face and triangular dough coincide;

[0169] After the triangle patch traversal is completed, the abnormal sequence is counted The number of numbers in , if the ratio of the number of numbers to the number of triangles If the ratio is less than or equal to the ratio threshold, the resin STL model is considered to be normal and is determined to be a qualified jewelry resin model. Otherwise, the resin STL model is considered to be abnormal.

[0170] like Figure 4 As shown, further, the detection and analysis unit assigns problem labels through the defect recognition network including:

[0171] Convert the jewelry STL model and the resin STL model into a first image and a second image of a specified pixel size;

[0172] The first image and the second image are simultaneously input into the downsampling subnetwork, the dimensions of the first image and the second image are reduced by pooling, the number of channels is increased, and then the deep features in the first image and the second image are further extracted by two convolutions, and the first feature and the second feature are output;

[0173] The first feature and the second feature are further input into the residual coding sub-network, reconstructed into the first reconstructed feature and the second reconstructed feature through the attention mechanism, and then the first feature and the second feature are superimposed through the residual structure to generate the first code and the second code;

[0174] The first code and the second code are input into the linear sub-network, and the aggregate code is generated by combining the linear fully connected layer and the ReLU activation function;

[0175] The aggregated code is input into the upsampling subnetwork, and upsampling decoding is performed through two deconvolutions to generate expanded-dimensional features;

[0176] The expanded dimension features are input into the multi-layer perception sub-network, linearly modulated by two linear fully connected layers, and the softmax function is used for nonlinear activation to generate a probability vector. The question label with the largest probability in the probability vector is selected and assigned to the resin STL model.

[0177] Further, the data sharing module includes a storage unit and a fourth communication unit;

[0178] The storage unit stores a label correspondence table in advance and stores the data uploaded by other units in real time;

[0179] The fourth communication unit receives the data uploaded by the remaining units, and responds to the calls of the remaining units to transmit the required data.

[0180] Furthermore, the mold construction module fixes the qualified jewelry resin model in the medium, and the qualified jewelry resin model must be fully immersed in the medium, and the heating temperature and heating time are set so that the heating temperature is greater than the melting temperature of the photosensitive resin and the hardening temperature of the medium, and is less than the melting temperature of the medium, and the heating is continued until the jewelry resin model is completely melted and the medium is completely hardened to obtain the jewelry mold.

[0181] Furthermore, the finished product quality inspection module includes an image acquisition unit, an image analysis unit, a parameter measurement unit and an evaluation unit;

[0182] The image acquisition unit captures the surface image of the finished jewelry product and transmits it to the image analysis unit;

[0183] The image analysis unit obtains the surface image of the finished jewelry product and determines whether there are defects on the surface of the finished jewelry product through image recognition technology. If there are no defects, the surface qualified label is transmitted to the evaluation unit;

[0184] The parameter measurement unit measures the size and quality of the finished jewelry, compares it with the pre-defined finished product size standard and quality standard, calculates the size error and quality error, and transmits the parameter compliance label to the assessment unit if the sum of the size error and the quality error is less than or equal to the preset error threshold;

[0185] The assessment unit will only judge the same finished jewelry product as a superior product if it obtains the surface qualified label and parameter compliance label. In all other cases, it will be judged as a inferior product.

[0186] The invention discloses a multi-device collaborative operation control system for jewelry 3D printing, comprising an analysis and adjustment module, a model manufacturing module, a scanning and inspection module, a data sharing module, a mold construction module, a finished product casting module, a finished product quality inspection module and a recycling module; the analysis and adjustment module acquires and sorts triangular facets for the complex structure of the jewelry STL model, establishes a triangular facet adjacency matrix, and uses adaptive stratification to adjust the stratification thickness of a single slice layer to avoid a step effect in printing; for a single slice layer, uses NURBS curve fitting to generate an accurate layer profile, and adjusts the light curing parameters in real time based on the stratification thickness of the single slice layer to generate a stratification parameter table; the model manufacturing module finely prints the jewelry resin model based on the stratification parameter table, generates a problem label of an unqualified model through a defect recognition network of the scanning and inspection module, autonomously adjusts the printing parameters, ensures the printing quality, and improves the stability and automation of the process flow of printing the jewelry resin model; the finished product quality inspection module combines image recognition technology and traditional measurement technology to monitor the quality of the finished jewelry product, melts inferior products through the recycling module, and re-acquires precious metal raw materials, thereby ensuring the quality of the finished jewelry 3D printed products and reducing the cost of the lost wax casting process.

[0187] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A multi-device collaborative operation control system for jewelry 3D printing, characterized in that: Including analysis and parameter adjustment module, model making module and scanning inspection module: The analysis and parameter adjustment module obtains and sorts the triangular facets of the jewelry STL model, adjusts the layer thickness based on the minimum angle between the normal vector of the triangular facet and the slicing direction in a single slice layer, uses NURBS curve fitting to obtain the layer contour by the intersection of the single slice layer and the triangular facet, and sets the light curing parameters based on the layer thickness to construct a layer parameter table; The model manufacturing module obtains the layered parameter table and sequentially prints and photocures a single slice layer to obtain a jewelry resin model, and adjusts the preset printing parameters and / or exposure time based on the adjustment instruction; The scanning inspection module compares the triangular facets and counts abnormal sequences to identify unqualified jewelry resin models, assigns problem labels to the unqualified jewelry resin models through a defect recognition network, and generates adjustment instructions.

2. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 1, characterized in that: The analysis and parameter adjustment module includes a parsing unit, an adaptive stratification unit, a track determination unit, a light adjustment unit and a first communication unit; The parsing unit obtains all triangular facets based on the jewelry STL model, sorts the triangular facets, and uses an adjacency matrix Construct an adjacency undirected graph in the form of ; The adaptive stratification unit adjusts the stratification thickness based on the minimum angle between the normal vector of the triangle face in a single slicing layer and the slicing direction to divide the slicing layer; The track determination unit uses NURBS curve fitting to obtain the layer profile by fitting the intersection of a single slice layer and a triangular facet; The dimming unit sets the light-fixing parameters of a single slice layer based on the layer thickness; The first communication unit constructs a layered parameter table based on layer thickness, layer profile and light-curing parameters.

3. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 2, characterized in that: The adaptive hierarchical unit divides the slice layer into: Get the minimum printing thickness, maximum printing thickness and total height of the jewelry STL model, along The starting tangent plane is translated in the positive direction of the axis Get the actual Slice layer and judge Whether the remaining model height is less than the maximum printing thickness; like The remaining model height is greater than or equal to the maximum printing thickness, and the maximum printing thickness is translated to obtain the initial Slice layer, get the same as the initial The set of triangles where the slice layers intersect; Traverse a single triangle in the triangle set, based on the normal vector of the single triangle and the layering direction Minimum angle calculation theory Layer thickness, select the theoretical The larger of the layer thickness and the minimum printing thickness is used as the actual Layer thickness, get the actual Slice layer; like The remaining model height is less than the maximum printing thickness and greater than or equal to the minimum printing thickness, then set the actual The layer thickness is The remaining model height and the actual Slice layer; like The remaining model height is less than the minimum printing thickness, then Separate a portion of the layer thickness to make the actual The layer thickness is the minimum printing thickness. Slice layer and actual Slice layer.

4. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 2, characterized in that: The orbit determination unit uses NURBS curve fitting to obtain the layer profile of a single slice layer, including: Get and actual The intersecting triangles of the slice layer are collected and the intersection coordinates are calculated. The starting triangle is selected based on the adjacency matrix. Construct a triangle face chain, select the traversal direction, and sort the intersection points; choose NURBS curves of order, based on basis functions The recursive formula determines the node interval of the intersection, adds the front supplementary node and the back supplementary node, and constructs the actual The node vector of the slice layer; Assign the values ​​of the front supplementary node and the rear supplementary node to 0 and 1 respectively, determine the values ​​of the remaining nodes in the node vector based on the length of the polyline segment constructed based on the intersection point, and obtain the actual Slice layer The NURBS curve equation of the order is used as the layer profile.

5. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 2, characterized in that: The parsing unit constructs an adjacency undirected graph include: Sort triangles and build point sets , construct a matrix of all zeros, defined as the adjacency matrix ; Lock triangles, get the The triangles have two identical vertices. triangular patches, define the The triangles and The triangles are adjacent and have edge relationships, set the adjacency matrix No. Line Column elements and Line The column element is 1; Traverse the triangle patch to obtain all edge relationships and build an edge set , with the adjacency matrix An adjacency undirected graph is established in the form of .

6. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 2, characterized in that: The dimming unit sets the light-fixing parameters of a single slice layer based on the layer thickness, including: Set the exposure time corresponding to the maximum printing thickness; According to the actual The ratio of layer thickness to maximum print thickness is set Exposure duration; Get more Layer thickness, calculate the Depth of cure; Based on Cure Depth Setting Light intensity distribution.

7. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 1, characterized in that: The scanning and inspection module includes a scanning unit, a detection and analysis unit, a text generation unit and a third communication unit; The scanning unit obtains the resin STL model by scanning the jewelry resin model; The detection and analysis unit compares the triangular facets of the resin STL model and the jewelry STL model, counts the abnormal sequences, and determines whether the resin STL model has an abnormality based on the number of abnormal sequence numbers. If an abnormality exists, a problem label is assigned based on the defect recognition network; The text generation unit searches the label correspondence table, obtains the adjustment plan corresponding to the problem label, obtains the abnormal slice layer based on the abnormal sequence, and generates the adjustment instruction; The third communication unit calls the jewelry STL model and the label correspondence table, and feeds back the adjustment instruction to the model manufacturing module.

8. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 7, characterized in that: The detection and analysis unit determines whether the resin STL model is abnormal including: Get all the triangles of the resin STL model and sort them; Traverse the single triangle face of the resin STL model and the jewelry STL model according to the sorting order; Get the resin STL model The coordinates of the three vertices of the triangle face are obtained to obtain the first The coordinates of the three vertices of a triangle patch are calculated and judged. Is the deviation of the triangles greater than the deviation threshold? If so, in the abnormal sequence Record the triangle patch number in ; After the traversal is completed, the abnormal sequence is counted If the ratio of the number of numbers to the number of triangles is less than or equal to the ratio threshold, the resin STL model is determined to be a qualified jewelry resin model, otherwise it is determined to be abnormal.

9. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 7 or 8, characterized in that: The detection and analysis unit assigns problem labels through the defect recognition network, including: Convert the jewelry STL model and the resin STL model into the first image and the second image and input them into the downsampling sub-network, and output the first feature and the second feature through pooling and multiple convolutions; The first feature and the second feature are input into the residual coding sub-network, and the first code and the second code are generated by superposition of the attention mechanism and the residual structure; The first code and the second code are input into the linear sub-network, and the aggregate code is generated by combining the linear fully connected layer and the ReLU activation function; The aggregated code is input into the upsampling sub-network, and the expanded dimension features are generated through multiple deconvolutions; The expanded dimension features are input into the multi-layer perceptron sub-network, and the linear modulation of the expanded dimension features is activated by the softmax function nonlinearly to generate a probability vector, and the question label is selected to be assigned to the resin STL model.

10. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 1, characterized in that: The model manufacturing module includes a printing unit, a light curing unit, an adjustment unit and a second communication unit; The printing unit sets the additive thickness and boundary profile of the nozzle based on the layered parameter table to perform 3D printing with preset printing parameters; The photocuring unit sets the photocuring parameters of the ultraviolet laser based on the layered parameter table to perform photocuring; The adjustment unit locks the abnormal slice layer based on the adjustment instruction, and adjusts the preset printing parameters and / or exposure time of the abnormal slice layer; The second communication unit is used to receive an adjustment instruction.

11. A multi-device collaborative operation control system for jewelry 3D printing as claimed in claim 1, characterized in that: It also includes data sharing module, mold building module, finished casting module, finished product quality inspection module and recycling module: The data sharing module has a built-in label correspondence table to store and share uploaded data from each module; The mold building module fixes the qualified jewelry resin model in the medium and heats it to make the jewelry mold; The finished product casting module is infused with precious metal raw materials to obtain finished jewelry products; The finished product quality inspection module obtains the surface image of the finished jewelry, identifies defects based on image recognition technology, measures the size and quality of the finished jewelry, and comprehensively determines whether the finished jewelry is a good product or a bad product; The recovery module melts inferior products and recovers precious metal raw materials.

Images