Injection molding system based on artificial intelligence and molding condition generation method in injection molding system

Abstract

According to one aspect of the present invention, an artificial intelligence-based injection molding system is capable of providing molding conditions with high accuracy in a short time, comprising: a specification data extraction unit, which extracts target specification data of a product produced by a mold from mold information about a mold to which a first molding material in a molten state is supplied; a molding condition output unit, which inputs the extracted target specification data into a pre-learned molding condition generation model and outputs the molding conditions; an injection molding machine, which supplies the first molding material to the mold according to the molding conditions to produce the product; and a judgment unit, which compares the production specification data of the produced product with the target specification data to determine whether the molding conditions are suitable. If the judgment unit (250) determines that the molding conditions are not suitable, the molding condition output unit (220) generates the production specification data and the molding conditions into a feedback data set, and uses the feedback data set to enable the molding condition generation model (230) to learn.

Description

Technical Field

[0001] The present invention relates to an injection molding system. Background Art

[0002] Injection molding is the most widely used manufacturing method for plastic products. For example, various parts including covers and shells of products such as televisions, mobile phones, and PDAs can be manufactured through injection molding.

[0003] Generally, the following processes are used to manufacture products by injection molding. First, the molding material to which pigments, stabilizers, plasticizers, fillers, etc. are added is put into a hopper to make it molten. Then, the molten molding material is injected into the mold and solidified by cooling. Then, the solidified molding material is taken out of the mold and the unnecessary parts are removed. Products of various types and sizes are manufactured by these processes.

[0004] As an apparatus for performing such injection molding, an injection molding machine is used. The injection molding machine includes an injection device for supplying a molding material in a molten state, and a clamping device for solidifying the molding material in a molten state by cooling.

[0005] In order to manufacture products with injection molding machines, personnel are required to directly set various variables such as temperature, speed, pressure, time, etc. Therefore, there is a problem as follows: the site can only rely heavily on experts. Even if experts set various variables, the process conditions will vary greatly depending on the person who sets them, resulting in unstable product quality.

[0006] In order to solve this problem, simulation technology has been proposed. However, simulation technology takes about 30 minutes to two hours in a normal computing environment, so there is a problem of long time required, and it cannot accurately simulate the actual experiment, so there is a problem of low accuracy. Summary of the invention

[0007] Problems to be solved by the invention

[0008] The present invention aims to solve the above-mentioned problems. The technical problem to be solved is to provide an injection molding system based on artificial intelligence and a molding condition generating method in the injection molding system, which can provide molding conditions with high accuracy in a short time.

[0009] The technical problem to be solved by the present invention is to provide an injection molding system based on artificial intelligence and a molding condition generation method in the injection molding system, which can generate molding conditions using a molding condition generation model based on deep learning.

[0010] The technical problem to be solved by the present invention is to provide an injection molding system based on artificial intelligence and a molding condition generation method in the injection molding system, which can provide optimal molding conditions by additional learning of molding conditions erroneously output from a molding condition generation model.

[0011] Technical solutions to the problem

[0012] According to one aspect of the present invention for achieving the above-mentioned purpose, an artificial intelligence-based injection molding system is characterized in that it includes: a specification data extraction unit 210, which extracts target specification data of a product produced by a mold from mold information about a mold to which a first molding material in a molten state is supplied; a molding condition output unit 220, which inputs the extracted target specification data into a pre-learned molding condition generation model 230 and outputs molding conditions; an injection molding machine 100, which supplies the first molding material to the mold according to the molding conditions to produce the product; and a judgment unit 250, which compares the production specification data of the produced product with the target specification data to determine whether the molding conditions are suitable. If the judgment unit 250 determines that the molding conditions are not suitable, the molding condition output unit 220 generates the production specification data and the molding conditions into a feedback data set, and uses the feedback data set to enable the molding condition generation model 230 to learn.

[0013] According to another aspect of the present invention for achieving the above-mentioned purpose, a molding condition generating method in an injection molding system is characterized in that it includes: a step of extracting target specification data as a specification of a product from mold information about a mold to which a molding material in a molten state is supplied; a step of inputting the extracted target specification data into a molding condition generating model that has been learned in advance and outputting molding conditions; a step of supplying the molding material to the mold to produce a product according to the molding conditions; a step of measuring production specification data of the produced product; a step of comparing the measured production specification data with the target specification data to determine whether the molding conditions are suitable; and a step of using the unsuitable molding conditions and the production specification data as a feedback data set to enable the molding condition generating model to learn if the molding conditions are judged to be unsuitable based on the judgment result of whether the molding conditions are suitable.

[0014] Effects of the Invention

[0015] According to the present invention, there is an effect that molding conditions with high accuracy can be provided in a short time without skilled experts.

[0016] According to the present invention, since molding conditions can be generated using a molding condition generation model based on deep learning, there is an effect that the performance of the molding condition generation model can be guaranteed.

[0017] The present invention can gradually improve the performance of the molding condition generation model by additionally learning the molding conditions erroneously output from the molding condition generation model, thereby having the effect of being able to generate the optimal molding conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 is a diagram showing an artificial intelligence-based injection molding system according to an embodiment of the present invention.

[0019] Figure 2 1 is a diagram showing the configuration of an injection molding machine according to an embodiment of the present invention.

[0020] Figure 3 It is a figure which shows the situation when a fixed mold and a movable mold are opened.

[0021] Figure 4 This is a diagram showing a state when a fixed mold and a movable mold are clamped together by a moving part.

[0022] Figure 5 1 is a diagram showing the configuration of a molding condition generating device according to an embodiment of the present invention.

[0023] Figure 6 is a flowchart showing a method for generating molding conditions in an injection molding system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] Embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Figure 1 is a diagram showing an artificial intelligence-based injection molding system according to an embodiment of the present invention.

[0026] The artificial intelligence-based injection molding system 10 (hereinafter referred to as "injection molding system") according to the present invention utilizes molding materials to produce products according to optimal molding conditions. Figure 1 As shown, the injection molding system 10 includes an injection molding machine 100 and a molding condition generating device 200 .

[0027] The injection molding machine 100 according to the present invention manufactures a product by performing injection molding.

[0028] Figure 2 FIG. 1 is a diagram showing the structure of an injection molding machine 100 according to an embodiment of the present invention. Figure 1 and Figure 2 , the injection molding machine 100 will be described in more detail.

[0029] like Figure 1 and Figure 2 As shown, the injection molding machine 100 according to the present invention includes an injection device 102 and a clamping device 103 .

[0030] The injection device 102 supplies the molding material in a molten state to the clamping device 103. The injection device 102 may include a barrel 121, an injection screw 122 disposed inside the barrel 121, and an injection drive 123 for driving the injection screw 122. The barrel 121 may be configured to be parallel to the first axis direction (X axis direction). The first axis direction (X axis direction) may be parallel to the direction in which the injection device 102 and the clamping device 103 are spaced from each other. When the molding material is supplied to the inside of the barrel 121, the injection drive 123 rotates the injection screw 122, thereby moving the molding material supplied to the inside of the barrel 121 in the first direction (FD arrow direction). In this process, the molding material may be melted due to friction and heating. The first direction (FD arrow direction) is a direction from the injection device 102 toward the clamping device 103, and may be a direction parallel to the first axis direction (X axis direction). When the molten molding material is located in the first direction (FD arrow direction) relative to the injection screw 122, the injection drive unit 123 can move the injection screw 122 in the first direction (FD arrow direction). Thus, the molten molding material can be supplied from the cylinder 121 to the clamping device 103.

[0031] The mold clamping device 103 solidifies the molten molding material by cooling. The mold clamping device 103 may include a fixed mold plate 131 combined with a fixed mold 150, a movable mold plate 132 combined with a movable mold 160, and a moving part 133 that moves the movable mold plate 132 along a first axis direction (X axis direction).

[0032] Figure 3 and Figure 4 This is a diagram showing a state when a fixed mold and a movable mold are clamped together by a moving part.

[0033] The moving part 133 moves the moving template 132 in the second direction (the SD arrow direction) so that the moving mold 160 and the fixed mold 150 are closed, and the injection device 102 supplies the molten molding material to the inside of the moving mold 160 and the fixed mold 150. The second direction (the SD arrow direction) is parallel to the first axis direction (the X axis direction) and opposite to the first direction (the FD arrow direction). Then, the clamping device 103 solidifies the molten molding material filled in the inside of the moving mold 160 and the fixed mold 150 by cooling, and the moving part 133 moves the moving template 132 in the first direction (the FD arrow direction) so that the moving mold 160 and the fixed mold 150 are opened.

[0034] The clamping device 103 may include a pulling rod 134. The pulling rod 134 guides the movement of the movable template 132. The movable template 132 may be movably coupled to the pulling rod 134. The movable template 132 may move in a first axis direction (X axis direction) along the pulling rod 134. The pulling rod 134 may be configured to be parallel to the first axis direction (X axis direction). The pulling rod 134 may be inserted and coupled to the fixed template 131 and the movable template 132, respectively.

[0035] On the other hand, the injection molding machine 100 according to the present invention produces a product by supplying molding material to the clamped movable mold 160 and fixed mold 150 according to the molding conditions generated by the molding condition generating device 200. The movable mold 160 and fixed mold 150 are hereinafter referred to as molds.

[0036] The molding condition generating device 200 generates molding conditions and transmits them to the injection molding machine 100. At this time, in order to generate the optimal molding conditions, the molding condition generating device 200 determines whether the molding conditions are suitable by using the products produced according to the molding conditions.

[0037] Below, refer to Figure 5 , the molding condition generating device 200 according to the present invention is described in more detail.

[0038] Figure 5 FIG. 2 is a diagram showing the structure of a molding condition generating device 200 according to an embodiment of the present invention. Figure 5 As shown, the molding condition generating device 200 includes a specification data extracting unit 210 , a molding condition outputting unit 220 , a molding condition generating model 230 , and a determining unit 250 .

[0039] The specification data extraction unit 210 extracts target specification data as product specifications from the mold information. Specifically, the specification data extraction unit 210 extracts target specification data of the product produced by the mold from the mold information about the mold for receiving the first molding material in a molten state. At this time, the first molding material represents the molding material for the product to be produced.

[0040] In an embodiment, the specification data includes at least one of shape information and weight information of the product.

[0041] In one embodiment, the shape information may include the total volume of the product produced by the mold, the volume of the cavity of the mold, the number of cavities, the number of gates of the mold, the surface area of ​​the product, the surface area of ​​the cavity, the first projection area XY of the product, the second projection area YZ of the product, the third projection area ZX of the product, the maximum thickness of the product, the average thickness of the product, the standard deviation of the thickness of the product, the diameter of the gate, the maximum flow distance from the gate to the end of the product, and at least one of the ratio of the maximum flow distance to the average thickness of the product.

[0042] At this time, the first to third projection areas represent the vertical projection areas of the product on each axial plane (XY, YZ, ZX). The diameter of the gate represents the circular diameter or hydraulic diameter.

[0043] In one embodiment, the specification data extraction unit 210 can scan the mold used to produce the product to generate mold information, and extract the shape information of the product from the mold information. Different from this embodiment, the specification data extraction unit 210 can also receive the input mold image of the product to generate mold information, and thereby extract the shape information of the product.

[0044] In one embodiment, the specification data extraction unit 210 extracts the solid density of the first molding material among the plurality of molding materials from the material property database 215. In addition, the specification data extraction unit 210 may extract the weight of the product using the extracted solid density of the first molding material and the total volume of the product.

[0045] The material property database 215 stores the solid density of a plurality of molding materials. Figure 5 In the figure, for the sake of convenience, the molding condition generating device 200 is illustrated as including the material property database 215, but this is merely an example, and the material property database 215 may also be constructed as a structure independent of the molding condition generating device 200.

[0046] The molding condition output unit 220 inputs the target specification data extracted by the specification data extraction unit 210 into the pre-learned molding condition generation model 230, and outputs the molding condition. In one embodiment, the molding condition may include at least one of the temperature of the mold, the temperature of the barrel 121, the injection speed of the injection molding machine 100, the holding time of the injection molding machine 100, and the holding pressure of the injection molding machine 100.

[0047] The molding condition output unit 220 transmits the output molding condition to the injection molding machine 100. Thus, the injection molding machine 100 supplies the first molding material to the mold according to the molding condition, thereby producing a product.

[0048] In one embodiment, when the molding condition is judged to be inappropriate based on the result of producing the product using the output molding condition, the molding condition output unit 220 generates a feedback data set with the production specification data of the product produced using the inappropriate molding condition and the molding condition. In addition, the molding condition output unit 220 uses the feedback data set to enable the molding condition generation model 230 to learn.

[0049] In one embodiment, the molding condition output unit 220 may enable the molding condition generation model 230 to perform transfer learning using the feedback data set when learning.

[0050] When the molding condition generation model 230 completes learning using the feedback data set, the molding condition output unit 220 may input the target specification data into the molding condition generation model 230 and output the modified molding conditions.

[0051] As described above, the present invention combines the production specification data of products produced under inappropriate molding conditions with the molding conditions to form a data set, allowing the molding condition generation model 230 to learn. This can not only gradually improve the performance of the molding condition generation model 230, but can also automatically find the optimized molding conditions, thereby achieving the effect of producing the highest quality products without skilled experts.

[0052] When the molding condition generation model 230 is input to the target specification data through the molding condition output unit 220, the corresponding molding condition is generated. The molding condition generation model 230 can be learned through the molding condition output unit 220. In particular, when the molding condition is judged to be inappropriate based on the result of producing a product using the molding condition output by the molding condition output unit 220, the molding condition generation model 230 according to the present invention can perform additional learning using the production specification data of the product produced using the inappropriate molding condition and the molding condition as a feedback data set.

[0053] In one embodiment, the molding condition generation model 230 may be a neural network capable of outputting molding conditions based on a plurality of weights and a plurality of biases and according to target specification data. According to this embodiment, the molding condition generation model 230 may be implemented using an artificial neural network (ANN) algorithm.

[0054] The judging unit 250 compares the production specification data of the product produced using the molding conditions output by the molding condition output unit 220 with the target specification data extracted by the specification data extraction unit 210 to judge whether the molding conditions are suitable. Specifically, when the production specification data is beyond a predetermined reference range from the target specification data, the judging unit 250 judges the molding conditions as unsuitable. Furthermore, when the production specification data is within a predetermined reference range from the target specification data, the judging unit 250 judges the molding conditions as suitable.

[0055] For example, when the weight of the product included in the production specification data is measured to be 100g, the weight of the product included in the target specification data is extracted to be 90g, and the reference range is 5g, the judgment unit 250 judges that the molding conditions are inappropriate because the weight of the production specification data exceeds the reference range compared with the weight of the target specification data.

[0056] When the molding condition is judged to be unsuitable, the judging unit 250 transmits a stop command to the injection molding machine 100. Thus, the injection molding machine 100 stops the production of the product. Furthermore, when the molding condition is judged to be unsuitable, the judging unit 250 transmits a feedback learning command to the molding condition output unit 220. Thus, the molding condition output unit 220 uses the unsuitable molding condition and the production specification data as one feedback data set to make the molding condition generation model 230 learn.

[0057] On the other hand, according to the molding condition generating device 200 of the present invention, Figure 5 As shown in the figure, a specification data measuring unit 240 and a model generating unit 260 may be further included.

[0058] The specification data measuring unit 240 measures the production specification data of the product produced by the injection molding machine 100. Figure 5 As shown, the specification data measuring unit 240 includes a fetching unit 242 and a specification data generating unit 244 .

[0059] The taking-out unit 242 takes out the produced product from the mold. For example, the taking-out unit 242 may be a multi-joint taking-out robot.

[0060] The specification data generating unit 244 generates production specification data from the taken product. Specifically, the specification data generating unit 244 photographs the product to generate first shape information, measures the weight of the product to generate first weight information, and generates production specification data including the first shape information and the first weight information. At this time, the specification data generating unit 244 can be implemented as a visual system (not shown) to generate the first shape information.

[0061] In one embodiment, the first shape information of the product may include the total volume of the product, the volume of the part corresponding to the cavity of the mold, the number of parts corresponding to the cavity, the number of parts corresponding to the gate of the mold, the surface area of ​​the product, the first projection area XY of the product, the second projection area YZ of the product, the third projection area ZX of the product, the maximum thickness of the product, the average thickness of the product, the standard deviation of the thickness of the product, the diameter of the part corresponding to the gate, the maximum flow distance from the part corresponding to the gate to the end of the product, and at least one of the ratio of the maximum flow distance to the average thickness of the product.

[0062] The specification data generating unit 244 transmits the generated production specification data to the determining unit 250 .

[0063] The model generation unit 260 generates the molding condition generation model 230. Specifically, the model generation unit 260 may generate the molding condition generation model 230 by using a plurality of learning data sets to allow the neural network to learn.

[0064] The model generation unit 260 integrates the plurality of learning molding conditions collected in advance with the learning specification data of the product produced according to each learning molding condition to generate a plurality of learning data sets. At this time, the learning molding condition may include at least one of the temperature of the mold, the temperature of the barrel 121, the injection speed of the injection molding machine 100, the holding time of the injection molding machine 100, and the holding pressure of the injection molding machine 100. The learning specification data may include at least one of shape information and product weight information.

[0065] The model generation unit 260 causes the neural network to learn using the generated plurality of learning data sets, thereby generating the molding condition generation model 230 .

[0066] As an example, the model generation unit 260 uses the learning data set to learn a neural network with a predetermined layer structure to construct a weight prediction system, and performs Min-Max normalization to convert the learning data set to the same value range. At this time, the learning data set can be divided into n-dimensional input data consisting of shape information and molding conditions, and one-dimensional output data consisting of product weight information. n can represent the number of information included in the shape information and molding conditions. For example, if the shape information includes 15 pieces of information and the molding conditions include 5 pieces of information, then n is 20.

[0067] In addition, the model generation unit 260 divides the input data and the output data into learning, verification, and testing data according to a predetermined ratio. In order to improve the accuracy of the molding condition generation model 230, the model generation unit 260 extracts the shape information related to the weight information of the product from the shape information, and uses it to generate the product weight prediction system. In one embodiment, the model generation unit 260 can perform sensitivity analysis to extract the shape information related to the weight information of the product from the shape information.

[0068] In one embodiment, the model generation unit 260 may perform a grid search or a random search to determine the hyperparameters of the neural network. In this case, the grid search may be applicable to the activation function, the optimization method, and the initialization method, while the random search may be applicable to the remaining hyperparameters.

[0069] The model generation unit 260 uses the generated weight prediction system to generate a molding condition generation model 230 that can reversely derive molding conditions corresponding to the weight when the weight is provided. Therefore, when the shape information and weight information are input, the molding condition generation model 230 can receive the weight information input according to the shape information, thereby deriving the corresponding molding conditions.

[0070] In one embodiment, the model generation unit 260 may generate the molding condition generation model 230 from the weight prediction system using particle swarm optimization or random search.

[0071] The present invention can guide users to find process conditions without professional knowledge about injection molding through the molding condition generation model 230 generated by the model generation unit 260, thereby reducing dependence on experts. By utilizing additional learning of feedback data, the molding condition generation model 230 can be improved, thereby ensuring higher accuracy and having the effect of building a smart factory in the injection molding field based on an unmanned injection molding system.

[0072] Below, refer to Figure 6 The method for generating molding conditions in the injection molding system according to the present invention is specifically described. At this time, the method for generating molding conditions in the injection molding system according to the present invention can be Figure 1 The illustrated injection molding system performs.

[0073] Figure 6 is a flowchart showing a method for generating molding conditions in an injection molding system according to an embodiment of the present invention.

[0074] The injection molding system 10 extracts target specification data as specifications of the product from the mold information S600. Specifically, the injection molding system 10 extracts target specification data of the product produced by the mold from the mold information about the mold supplied with the first molding material in a molten state. At this time, the first molding material represents the molding material for the product to be produced.

[0075] In an embodiment, the specification data includes at least one of shape information and weight information of the product.

[0076] In one embodiment, the shape information of the product may include the total volume of the product produced by the mold, the volume of the cavity of the mold, the number of cavities, the number of gates of the mold, the surface area of ​​the product, the surface area of ​​the cavity, the first projection area XY of the product, the second projection area YZ of the product, the third projection area ZX of the product, the maximum thickness of the product, the average thickness of the product, the standard deviation of the thickness of the product, the diameter of the gate, the maximum flow distance from the gate to the end of the product, and at least one of the ratio of the maximum flow distance to the average thickness of the product.

[0077] At this time, the first to third projection areas represent the vertical projection areas of the product on the respective axial planes (XY, YZ, ZX), and the diameter of the gate represents the circular diameter or the hydraulic diameter.

[0078] The injection molding system 10 extracts the solid density of the first molding material among the plurality of molding materials from the material property database 215. In addition, the injection molding system 10 may extract the weight of the product using the extracted solid density of the first molding material and the total volume of the product.

[0079] Then, the injection molding system 10 inputs the extracted target specification data into the pre-learned molding condition generation model 230, and outputs the molding condition S610. In one embodiment, the molding condition may include at least one of the temperature of the mold, the temperature of the barrel 121, the injection speed of the injection molding machine 100, the holding time of the injection molding machine 100, and the holding pressure of the injection molding machine 100.

[0080] Then, the injection molding system 10 supplies the first molding material to the mold according to the molding conditions to produce a product S620.

[0081] Then, the injection molding system 10 measures production specification data of the produced product S630.

[0082] Then, the injection molding system 10 compares the measured production specification data with the target specification data to determine whether the molding conditions are suitable S640. Specifically, when the production specification data is within a predetermined reference range from the target specification data, the injection molding system 10 determines that the molding conditions are suitable S650. When the production specification data is beyond the predetermined reference range from the target specification data, the injection molding system 10 determines that the molding conditions are not suitable S660.

[0083] When the molding conditions are judged to be inappropriate, the injection molding system 10 stops the production of the product.

[0084] Then, if the molding condition is determined to be inappropriate, the injection molding system 10 uses the inappropriate molding condition and the production specification data as a feedback data set to enable the molding condition generation model 230 to learn S670.

[0085] In one embodiment, the injection molding system 10 may use the feedback data set to enable the molding condition generation model 230 to perform transfer learning.

[0086] Then, when the molding condition generation model 230 completes learning using the feedback data set, the injection molding system 10 can input the target specification data into the molding condition generation model 230 and output the modified molding conditions.

[0087] It is understandable that a person skilled in the art in the technical field to which the present invention belongs can implement the present invention described above in other specific ways without changing the technical idea or essential features.

[0088] Furthermore, at least a portion of the methods described herein may be implemented using one or more computer programs or components. The components may be provided in the form of a series of computer commands via a computer-readable medium or machine-readable medium including volatile and non-volatile memory. The commands may be provided in the form of software or firmware, or may be implemented in whole or in part in a hardware structure such as ASICs, FPGAs, DSPs, or other similar components. The commands may be executed by one or more processors or other hardware structures, and when the processors or other hardware structures execute the series of computer commands, all or part of the methods and steps disclosed herein are executed or can be executed.

[0089] Therefore, it should be understood that the embodiments described above are exemplary in all aspects and not restrictive. The scope of the present invention is presented by the following claims rather than the specification. The meaning and scope of the claims and all changes or variant forms derived from equivalent concepts should be interpreted as being included within the scope of the present invention.

Claims

1. An artificial intelligence-based injection molding system, It is characterized in that include: A specification data extraction unit (210) extracts target specification data of a product produced by the mold from mold information about a mold to which the first molding material in a molten state is supplied; A molding condition output unit (220) inputs the extracted target specification data into a pre-learned molding condition generation model (230) to output molding conditions; An injection molding machine (100) supplies the first molding material to the mold according to the molding conditions to produce the product; as well as The judging unit (250) compares the production specification data of the produced product with the target specification data to judge whether the molding conditions are suitable. When the production specification data exceeds a predetermined reference range compared to the target specification data, the judging unit (250) judges the molding conditions to be inappropriate; when the production specification data is within a predetermined reference range compared to the target specification data, the judging unit (250) judges the molding conditions to be appropriate. If the judging unit (250) judges that the molding conditions are not suitable, the judging unit 250 transmits a stop command to the injection molding machine (100) to stop producing the product, and the molding condition output unit (220) generates a feedback data set from the production specification data and the inappropriate molding conditions, and uses the feedback data set to enable the molding condition generation model (230) to learn. The specification data extraction unit (210) scans the mold or receives a mold image of the product to generate mold information; The specification data includes shape information and weight information of the product; The shape information includes at least one of a total volume of the product produced by the mold, a volume of a cavity of the mold, the number of cavities, the number of gates of the mold, a surface area of ​​the product, a surface area of ​​the cavity, a first projection area (XY) of the product, a second projection area (YZ) of the product, a third projection area (ZX) of the product, a maximum thickness of the product, an average thickness of the product, a standard deviation of the thickness of the product, a diameter of the gate, a maximum flow distance from the gate to an end of the product, and a ratio of the maximum flow distance to the average thickness of the product; The injection molding system also includes: a material property database (215) storing a plurality of solid densities of molding materials; The specification data extraction unit (210) extracts the solid density of the first molding material among a plurality of molding materials from the material property database (215), and calculates the weight of the product using the total volume of the product and the solid density of the first molding material.

2. The artificial intelligence-based injection molding system according to claim 1, It is characterized in that The molding condition output unit (220) uses the feedback data set to enable the molding condition generation model (230) to perform transfer learning.

3. The artificial intelligence-based injection molding system according to claim 1, It is characterized in that If the molding condition generation model (230) completes learning, the molding condition output unit (220) inputs the target specification data into the molding condition generation model (230) and outputs the modified molding conditions.

4. The artificial intelligence-based injection molding system according to claim 1, It is characterized in that The molding condition generation model (230) is composed of a neural network, and the neural network is capable of outputting the molding condition based on a plurality of weights and a plurality of biases and according to the target specification data.

5. The artificial intelligence based injection molding system according to claim 1, It is characterized in that The method further comprises: a model generating unit (260) for generating the molding condition generating model (230), The model generation unit (260) integrates a plurality of learning molding conditions with learning specification data of products produced according to each learning molding condition to generate a plurality of learning data sets, and uses the plurality of learning data sets to enable the neural network to learn, thereby generating the molding condition generation model (230).

6. The artificial intelligence based injection molding system according to claim 1, It is characterized in that The molding conditions include at least one of a temperature of the mold, a temperature of a barrel, an injection speed of the injection molding machine, a holding time of the injection molding machine, and a holding pressure of the injection molding machine.

7. The artificial intelligence based injection molding system according to claim 1, It is characterized in that The system further comprises: a specification data measuring unit (240) for measuring the production specification data of the produced product. The specification data determination unit (240) includes: a taking-out unit (242) for taking out the produced product from the mold; and A specification data generating unit (244) generates the production specification data, wherein the production specification data includes first shape information generated by photographing the taken product and first weight information generated by measuring the weight of the product.

8. A method for generating molding conditions in an injection molding system, It is characterized in that include: A step of extracting target specification data as specifications of a product from mold information about a mold to which a molding material in a molten state is supplied; The step of inputting the extracted target specification data into a pre-learned molding condition generation model to output molding conditions; The step of supplying the molding material to the mold to produce a product according to the molding conditions; The step of determining the production specification data of the produced product; A step of comparing the measured production specification data with the target specification data to determine whether the molding conditions are suitable, wherein when the production specification data exceeds a predetermined reference range compared to the target specification data, the molding conditions are determined to be unsuitable, and when the production specification data is within a predetermined reference range compared to the target specification data, the molding conditions are determined to be suitable; and If the molding condition is judged to be unsuitable by the judgment result of whether the molding condition is suitable, the production of the product is stopped, and the unsuitable molding condition and the production specification data are used as a feedback data set to make the molding condition generation model learn the step, The mold information is generated by scanning the mold or receiving a mold image of the product; The specification data includes shape information and weight information of the product; The shape information includes at least one of a total volume of the product produced by the mold, a volume of a cavity of the mold, the number of cavities, the number of gates of the mold, a surface area of ​​the product, a surface area of ​​the cavity, a first projection area (XY) of the product, a second projection area (YZ) of the product, a third projection area (ZX) of the product, a maximum thickness of the product, an average thickness of the product, a standard deviation of the thickness of the product, a diameter of the gate, a maximum flow distance from the gate to an end of the product, and a ratio of the maximum flow distance to the average thickness of the product; The weight information of the product is calculated using the total volume of the product and the solid density of the product.

Images