Method and system for operating injection filling prediction simulation service, and computer-readable recording medium having, recorded thereon, program for operating service

Abstract

The present invention relates to: a method and a system for operating an injection filling prediction simulation service capable of predicting a filling result in injection molding by using an artificial intelligence model; and a computer-readable recording medium having the method recorded thereon. The present invention may comprise: an injection CAE analysis information extraction program for generating filling time distribution pattern vectors of objects being analyzed; a feature value extraction program for generating feature values between the objects being analyzed; a filling prediction model generation program for generating a filling prediction model; a condition information input program for receiving injection molding filling prediction simulation condition information; a filling time distribution pattern prediction program for generating filling time distribution pattern prediction information by using the feature values, extracted by the feature value extraction program, and the filling prediction model; and a filling time distribution pattern prediction information transmission program for transmitting the filling time distribution pattern prediction information to a user terminal.

Description

A computer-readable recording medium recording a method and system for operating an injection molding filling prediction simulation service and a program for operating the service.

[0001] The present invention relates to a method and a system for operating an injection molding filling prediction simulation service and a computer-readable recording medium recording a program for operating the service, and more particularly, to a method and a system for operating an injection molding filling prediction simulation service that can predict a filling result in injection molding using an artificial intelligence model, thereby enabling a user who is not an analysis expert to optimally design an injection molded product, an injection mold, and an injection process, and a computer-readable recording medium recording a program for operating the service.

[0002] The present invention refers to the research project of the Smart Manufacturing Innovation Technology Development Project (Project Unique Number: 1425171599, Project Number: RS-2022-00140364) carried out with support from the Small and Medium Business Technology Information Promotion Agency and funds from the Ministry of SMEs and Startups.

[0003] Among the plastic manufacturing industries, the manufacturing industry utilizing injection molding is an industry that is based on injection molding machines and consists of six major root technologies, the mold industry, and post-processing for surface treatment of injection molded products. In particular, plastic injection products are an industry that serves as a major component of all manufacturing industries, and are widely used in mobile devices, home appliances, automobiles, construction materials, etc., so the industrial linkages and ripple effects are very large.

[0004] In addition, CAE (Computer Aided Engineering) analysis simulation of injection molding is an essential element for improving the quality of injection molded products and activating smart factories / digital twins, but there are limitations in its application.

[0005] In the past, the participation of interpretation experts was essential, and at the same time, long interpretation times and high costs were required, which limited its application to industrial sites.

[0006] Therefore, to improve this, it is necessary to develop a simulator that is accessible to general engineers rather than analysis experts and that produces results in a short period of time.

[0007] The present invention has been proposed to solve these conventional problems, and can develop an AI-based standalone simulator that converts CAE analysis simulation into AI-based simulation, and as a simulator that uses CAE analysis simulation result data as AI learning data to create an AI model and derives the same results as CAE analysis simulation, it is novel compared to existing technologies that use AI as a tool for design of experiments (DOE) to optimize CAE analysis simulation or as a tool for applying ROM (Reduced Order Modeling, or Reduced Order Method), and can develop an AI-based injection molding product simulator that predicts injection molding results (injection molding quality) based on related data (product design CAD model, mold conditions, process conditions) in the field of injection molding, and can predict injection molding results in the product life cycle from design to product production by simulating injection molded products based on product CAD models, mold characteristics, and process conditions rather than a method of predicting quality based on process condition data collected during the injection molding process, and injection molding It is possible to develop a simulator that can accommodate geometric shapes and process conditions in simulation, and it is an AI-based injection molding product simulator that is created based on various CAE analysis results, and it is unique in that it breaks down the wall of constraints (type of product, process conditions, etc.) that simulation is possible, and it is possible to develop an independently operated simulator that conducts simulations with AI itself, rather than AI as an auxiliary / support means of CAE analysis, and by implementing the function of CAE analysis simulation as an independent AI-based simulator, it has the advantage of not only saving time and cost, but also enabling simulation application by non-experts, and simulation is not the exclusive domain of specific analysis specialist engineers, but many engineers can utilize simulation at various stages,Rather than predicting injection molding quality using the results of the injection molding process monitoring, a simulator can be developed that predicts injection molding results using product design data, mold conditions, process conditions, etc. as input conditions, and while the existing application of AI in the injection molding field has mainly focused on injection process monitoring and control, the injection molding simulation can be applied under various conditions and at various stages. The present invention provides a method and system for operating an injection filling prediction simulation service, and a computer-readable recording medium recording a program for operating this service. However, these tasks are exemplary and the scope of the present invention is not limited thereby.

[0008] A method for operating an injection molding filling prediction simulation service according to the idea of ​​the present invention for solving the above problem is provided by using a computer system having at least one user terminal and a server computer connected through a network, wherein the server computer comprises: an injection CAE analysis information extraction program for extracting injection analysis result information obtained by CAE injection analysis and generating a filling time distribution pattern vector of an analysis object connected by elements and nodes; a feature value extraction program for generating feature values ​​between the analysis objects; a filling prediction model generation program for generating an artificial intelligence-based filling prediction model by performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program for receiving injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a filling time distribution pattern prediction program for generating filling time distribution pattern prediction information by using the feature values ​​extracted from the feature value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions; A charging time distribution pattern prediction information transmission program that transmits the charging time distribution pattern prediction information to the user terminal, a simulation condition information database in which the injection molding charging prediction simulation condition information is stored, and a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, comprising: (a) a step of extracting injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program, and generating a charging time distribution pattern vector of an analysis target connected by elements and nodes; (b) a step of generating feature values ​​between the analysis targets by the feature value extraction program;(c) a step of performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the charging time distribution pattern vector and the characteristic value by the charging prediction model generation program to generate an artificial intelligence-based charging prediction model; (d) a step of receiving injection molding charging prediction simulation condition information for performing charging prediction from the user terminal by the condition information input program; (e) a step of generating charging time distribution pattern prediction information by the charging time distribution pattern prediction program using the characteristic value extracted by the characteristic value extraction program and the charging prediction model according to the input injection molding charging prediction simulation condition; and (f) a step of transmitting the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program.

[0009] In addition, according to the present invention, in the step (a), the analysis object may be formed by selecting at least one of a cooling channel, an injection mold, an injection molded product, or a combination thereof.

[0010] In addition, according to the present invention, in the step (b), the characteristic value includes at least one of the number of peripheral nodes, the geometric distance between each node, the geometric characteristic value of each node, the characteristic physical property of each node, and combinations thereof, and the characteristic physical property may be a flow-related physical property including at least the viscosity of the resin.

[0011] In addition, according to the present invention, the geometric distance between the nodes may be a distance between surrounding nodes within a specific range, and the geometric characteristic value of the node may be a ratio of the length and thickness of the flow path from the gate to the corresponding node (L / t ratio).

[0012] In addition, according to the present invention, in the step (a), the injection analysis result information is generated by CAE injection analysis based on at least one of injection molded product model information, injection mold condition information, injection material condition information, and injection process condition information, and combinations thereof, and the injection process condition information may include temperature and flow rate information of a cooling water injected into a cooling channel, temperature information of a resin injected into an injection mold, and a ram speed profile that determines a speed of a resin injected into an injection mold.

[0013] In addition, according to the present invention, in the step (c), the filling prediction model may be an artificial intelligence model that predicts the filling time distribution by selecting at least one of an injection mold, an injection product, or a combination thereof.

[0014] In addition, according to the present invention, in the step (b), the characteristic value may include at least one of thickness, the ratio of the flow path length to the node and the thickness (L / t ratio), the common logarithm value of the ratio of the length and the thickness (L / t ratio(log)), the temperature value of the mold (Mold Temperature), and combinations thereof.

[0015] In addition, according to the present invention, in the step (d), the filling prediction simulation condition information may include any one of injection molding product model information, injection mold information, injection process condition information, and combinations thereof, which are simulation prediction targets.

[0016] In addition, according to the present invention, in the step (e), the charging time distribution pattern prediction information can verify the difference value between the result predicted by the charging prediction model and the CAE injection analysis value by using CAE charging analysis information that displays the charging time distribution of the analysis target in color, and display the charging analysis information by displaying the difference value and the charging time distribution of the analysis target in color on the analysis target.

[0017] Meanwhile, a system for operating an injection molding filling prediction simulation service according to the idea of ​​the present invention for solving the above problem is a system for operating an injection molding filling prediction simulation service using a computer system having at least one user terminal and a server computer connected through a network, wherein the server computer includes: an injection CAE analysis information extraction program for extracting injection analysis result information obtained by CAE injection analysis and generating a filling time distribution pattern vector of an analysis object connected by elements and nodes; a feature value extraction program for generating feature values ​​between the analysis objects; a filling prediction model generation program for generating an artificial intelligence-based filling prediction model by performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program for receiving injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a filling time distribution pattern prediction program for generating filling time distribution pattern prediction information using the feature values ​​extracted from the feature value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions. A program, a charging time distribution pattern prediction information transmission program that transmits the charging time distribution pattern prediction information to the user terminal, a simulation condition information database in which the injection molding charging prediction simulation condition information is stored, and a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, wherein, by the injection CAE analysis information extraction program, injection analysis result information obtained by CAE injection analysis is extracted to generate a charging time distribution pattern vector of an analysis object connected by elements and nodes, and by the feature value extraction program, feature values ​​between the analysis objects are generated, and by the charging prediction model generation program,The system may include a control unit programmed to perform ANN (Artificial Neural Network) learning to find out a functional relationship according to the charging time distribution pattern vector and the characteristic value, thereby generating an artificial intelligence-based charging prediction model, and receiving injection molding charging prediction simulation condition information for performing charging prediction from the user terminal by the condition information input program, and generating charging time distribution pattern prediction information by using the characteristic value extracted from the characteristic value extraction program and the charging prediction model according to the input injection molding charging prediction simulation condition by the charging time distribution pattern prediction program, and transmitting the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program.

[0018] Meanwhile, a computer-readable recording medium recording a program for operating an injection filling prediction simulation service according to the idea of ​​the present invention for solving the above problem is a computer-readable recording medium recording a method for operating an injection filling prediction simulation service using a computer system having at least one user terminal and a server computer connected through a network, wherein the server computer includes: an injection CAE analysis information extraction program for extracting injection analysis result information obtained by CAE injection analysis and generating a filling time distribution pattern vector of an analysis target connected by elements and nodes; a feature value extraction program for generating feature values ​​between the analysis targets; a filling prediction model generation program for generating an artificial intelligence-based filling prediction model by performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program for receiving injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a feature value extracted from the feature value extraction program according to the input injection molding filling prediction simulation condition and the A charging time distribution pattern prediction program that generates charging time distribution pattern prediction information using a charging prediction model, a charging time distribution pattern prediction information transmission program that transmits the charging time distribution pattern prediction information to the user terminal, a simulation condition information database in which the injection molding charging prediction simulation condition information is stored, and a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, comprising: (a) a step of extracting injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program, and generating a charging time distribution pattern vector of an analysis target connected by nodes;(b) a step of generating feature values ​​between the analysis objects by the feature value extraction program; (c) a step of generating an artificial intelligence-based charging prediction model by performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the charging time distribution pattern vector and the feature value by the charging prediction model generation program; (d) a step of receiving injection molding charging prediction simulation condition information for performing charging prediction from the user terminal by the condition information input program; (e) a step of generating charging time distribution pattern prediction information by using the feature values ​​extracted by the feature value extraction program and the charging prediction model according to the input injection molding charging prediction simulation conditions by the charging time distribution pattern prediction program; and (f) a step of transmitting the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program.

[0019] According to the various embodiments of the present invention, as described above, even general users who are not experts in CAE analysis can apply the method to the entire cycle of injection molded products, from product design, mold design, and determination of mass production conditions, thereby achieving the effect of improving the quality of injection molded products. Of course, the scope of the present invention is not limited by these effects.

[0020] FIG. 1 is a conceptual diagram illustrating a system for operating an injection filling prediction simulation service according to some embodiments of the present invention.

[0021] Figure 2 is a conceptual diagram showing the relationship between an operator and a user who operate the injection filling prediction simulation service of the present invention.

[0022] FIG. 3 is a flowchart illustrating a method of operating an injection filling prediction simulation service according to some embodiments of the present invention.

[0023] FIG. 4 is a conceptual diagram illustrating a system for operating an injection filling prediction simulation service according to some other embodiments of the present invention.

[0024] Fig. 5 is a conceptual diagram showing an AI simulator of a system that operates the injection filling prediction simulation service of Fig. 4.

[0025] FIG. 6 is a flowchart illustrating a method of operating an injection filling prediction simulation service according to some other embodiments of the present invention.

[0026] FIG. 7 is a diagram showing several examples of characteristic values ​​applied in the method of operating the injection filling prediction simulation service of FIG. 6.

[0027] Fig. 8 is a screen showing an example of an AI model applied in the method of operating the injection filling prediction simulation service of Fig. 6.

[0028] Fig. 9 is a screen showing an example of a graph that color-codes the filling time distribution of CAE analysis results used for learning in the method of operating the injection filling prediction simulation service of Fig. 6.

[0029] Fig. 10 is a screen showing an example of a graph that color-codes the AI ​​prediction result charging time distribution used for learning in the method of operating the injection charging prediction simulation service of Fig. 6.

[0030] Fig. 11 is a screen showing an example of a graph that displays the charging time difference distribution in color, which represents the difference value between the CAE analysis result of Fig. 9 and the AI ​​prediction result of Fig. 10.

[0031] Hereinafter, embodiments of the present invention will be described in detail with reference to the attached drawings. However, the present invention is not limited to the embodiments disclosed below, but can be implemented in various different forms. The following embodiments are provided to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention. Furthermore, for convenience of explanation, the sizes of components in the drawings may be exaggerated or reduced.

[0032] Hereinafter, a method and system for operating an injection filling prediction simulation service according to various embodiments of the present invention and a computer-readable recording medium recording a program for operating the service are as illustrated in the drawings.

[0033] FIG. 1 is a conceptual diagram illustrating a system for operating an injection filling prediction simulation service according to some embodiments of the present invention.

[0034] First, as illustrated in FIG. 1, a system for operating an injection filling prediction simulation service according to some embodiments of the present invention may include at least one user terminal (10), an operator terminal (90), and a server computer (60) connected via a network (50).

[0035] Here, the network (50) can be connected to various information terminals such as various smart terminals, personal terminals, company terminals, other server computers, or the Internet of Things (IOT) in addition to the terminals (10) (90) described above.

[0036] The server computer (60) is programmed to extract injection analysis result information obtained by CAE injection analysis, generate a filling time distribution pattern vector of an analysis object connected by elements and nodes, generate feature values ​​between the analysis objects, and perform ANN (Artificial Neural Network) learning to find out a functional relationship according to the filling time distribution pattern vector and the feature values ​​to generate an artificial intelligence-based filling prediction model, receive injection molding filling prediction simulation condition information for performing filling prediction from the user terminal (10), and generate filling time distribution pattern prediction information using the feature values ​​extracted from the feature value extraction program according to the input injection molding filling prediction simulation conditions and the filling prediction model, and transmit the filling time distribution pattern prediction information to the user terminal (10), so that an enterprise, factory, business, organization, data center, or the like that can operate an injection filling prediction simulation service through the network (50) It could be a computer located at the headquarters, branch, sales office, dealership, navigation terminal manufacturer, navigation operator, or data management company.

[0037] In addition, for example, when the user terminal (10) inputs injection molding filling prediction simulation condition information for performing filling prediction, the terminal can be applied to a wide variety of users, including not only the general public who wants to quickly receive filling time distribution pattern prediction information predicted by a learned filling prediction model, but also designers who want to design products using the filling time distribution pattern, simulation analysis personnel, simulation companies, etc.

[0038] More specifically, for example, the user terminal (10) can be applied to all information terminal devices capable of processing various types of information, for example, not only the various smart phones described above, but also various wearable devices, smart sensors, smart pads, mobile terminals, PDAs, notebooks, laptop computers, smart cameras, smart camcorders, e-books, smart scanners, personal computers, etc.

[0039] In addition, for example, the operator terminal (90) is a terminal or computer of an individual, company, factory, business, central control center, headquarters, branch, sales office, or IT manager who manages and operates the server computer (60), and is not necessarily limited to a computer or smart phone, and can be applied to various information terminals, wearable terminals, PDAs, smart watches, smart pads, cameras, camcorders, notebooks, laptop computers, e-books, personal computers, other server computers, etc. that can receive various text information, numeric information, or image information and select various commands.

[0040] Meanwhile, for example, the user terminal (10) and the operator terminal (90) are not necessarily independent of each other. For example, the user terminal (10) and the operator terminal (90) may be the same.

[0041] That is, since the operator can play the role of the user, it is also possible for the operator to input injection molding filling prediction simulation condition information for directly performing filling prediction and to quickly receive filling time distribution pattern prediction information predicted by the learned filling prediction model.

[0042] For example, the user terminal (10) and the operator terminal (90) of FIG. 1 are connected to each other through the network (50) by having various applications, apps, hybrid apps, programs, etc. installed thereon, and the terminals connected by the network (50) can utilize communication networks such as existing 2G, 3G, 4G, 5G, LTE, etc. mobile communication networks, WIFI communication networks, Bluetooth communication networks, cellular communication networks, CDMA communication networks, LTE communication networks, Ethernet communication networks, WiMAX communication networks, local area networks (LANs), wide area networks (WANs), RF communication networks, infrared communication networks, and optical communication networks, and may also have an Internet browser (Netscape, Internet Explorer, etc.) capable of displaying web content in the form of HTML, XML, HTML5, etc., or a protocol device for connecting to an in-house or external or short-distance / long-distance wired / wireless network.

[0043] Meanwhile, the server computer (60) may include a program control unit (PG) that controls a program and a database (DB) that stores various types of information, as shown in FIG. 1.

[0044] In particular, the program control unit (PG), as illustrated in FIG. 1, includes a main program (61) that operates the entire program, a member registration program (62) that receives unique information from the user terminal (10) and registers it as member information, a login program (63) that receives login information from the user terminal (10), an injection CAE analysis information extraction program (64) that extracts injection analysis result information obtained by CAE injection analysis and generates a filling time distribution pattern vector of an analysis target connected by elements and nodes, a feature value extraction program (65) that generates feature values ​​between the analysis targets, a filling prediction model generation program (66) that performs ANN (Artificial Neural Network) learning to find out a functional relationship according to the filling time distribution pattern vector and the feature values ​​to generate an artificial intelligence-based filling prediction model, a condition information input program (67) that receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal (10), and the input injection molding filling prediction It may include a charging time distribution pattern prediction program (68) that generates charging time distribution pattern prediction information using the characteristic values ​​extracted from the characteristic value extraction program (65) according to simulation conditions and the charging prediction model, a charging time distribution pattern prediction information transmission program (69) that transmits the charging time distribution pattern prediction information to the user terminal, and other programs (70) that perform various functions such as advertising, promotion, payment, and bulletin board.

[0045] Here, for example, the main program (61) operates the entire program, and can be expressed online in the form of an operator management homepage or the main screen of an application or program, and can be a program that can receive various information and command signals from the user terminal (10) or the operator terminal (90) and control all of the programs.

[0046] In addition, for example, the member registration program (62) is a program that receives unique information from the user terminal (10) and registers the user as a member, and can receive unique information of the information terminal from the above-described terminals and have the user agree to standard terms and conditions or terms and conditions for information collection and use, and can receive real-name verification, public i-PIN, ID, password, email, mobile phone, address, personal information, patient information, resident registration number information, unique number, phone number stored in a SIM card, etc., and can be a program that determines whether the entered registration application information matches the pre-stored registration condition information and registers the user terminal (10) if they match.

[0047] In addition, for example, the login program (63) is a program that receives login information from the user terminal (10), and through this, the online connection status and location of the user terminal (10) can be confirmed.

[0048] In addition, for example, the injection CAE analysis information extraction program (64) is a program that extracts injection analysis result information obtained by CAE injection analysis and generates a filling time distribution pattern vector of an analysis target connected by elements and nodes, and the analysis target can be formed by selecting at least one of a cooling channel, an injection mold, and an injection molded product, or a combination thereof.

[0049] In addition, the injection analysis result information is generated by CAE injection analysis based on at least one of injection molded product model information, injection mold condition information, injection material condition information, and injection process condition information, and combinations thereof, and the injection process condition information may include temperature and flow rate information of cooling water injected into a cooling channel, temperature information of resin injected into an injection mold, and a ram speed profile that determines the speed of resin injected into the mold.

[0050] For example, the injection analysis result information may be applied to simulation information or test information using existing product information (Product CAD Data), mold information (Mold Data), process information (Process Data), etc., and such injection analysis result information may be easily obtained through existing CAE injection analysis information or injection analysis information of already produced products.

[0051] The above injection analysis result information may require a pre-processing process such as data standardization or data classification.

[0052] In addition, for example, the feature value extraction program (65) is a program that generates feature values ​​between the analysis objects, wherein the feature values ​​can be created by extracting node-specific variables or intermediate variables, and include at least one of the number of surrounding nodes, the geometric distance between each node, the geometric feature value of each node, the feature property value of each node, and combinations thereof, and the feature property value can be a flow-related property value including at least the viscosity of the resin.

[0053] For example, the geometric distance between the nodes may be the distance between surrounding nodes within a specific range, and the geometric characteristic value of the node may be the ratio of the length and thickness of the flow path from the gate to the node (L / t ratio).

[0054] More specifically, for example, the characteristic value may include at least one of thickness, the ratio of the flow path length to the node and the thickness (L / t ratio), the common logarithm of the ratio of the length to the thickness (L / t ratio(log)), the mold temperature, and combinations thereof.

[0055] Here, the temperature value of the mold can be a temperature value of the mold obtained from CAE analysis or a temperature value of the mold predicted through an AI model obtained from the implementation of the applicant's prior invention ("Method and system for operating injection cooling prediction simulation service and computer-readable recording medium recording the method" Application No. 10-2023-0071422).

[0056] In addition, in applying the ratio of the flow path length and thickness (L / t ratio), the length of the flow path can be calculated using the geodesic distance, which is a distance calculated along the product surface, or the distance connecting the midpoints between surfaces that form the thickness of an element, and in a flow path where the thickness continuously changes, the ratio of the flow path length and thickness (L / t ratio) can be calculated by applying the logarithmic average of the continuous thickness, and the ratio of the thickness (L / t ratio) can be calculated by applying the Dijkstra algorithm.

[0057] In addition, for example, the charging prediction model generation program (66) is a program that generates an artificial intelligence-based charging prediction model by performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the charging time distribution pattern vector and the characteristic value, wherein the charging prediction model may be an artificial intelligence model that predicts a charging time distribution by selecting at least one of an injection mold, an injection product, or a combination thereof.

[0058] This type of artificial intelligence learning method can train artificial intelligence models in various ways, such as hyper parameter tuning that adjusts epochs, learning rates, etc.

[0059] For example, the charging prediction model may include various modules such as a cooling module, a charging module, a pressure module, and a deformation module.

[0060] Using an AI-based prediction model completed based on these CAE analysis results, injection molding predictions can be made faster, easier, and more accurately than prediction methods based on existing CAE analysis.

[0061] In particular, the present invention builds a filling prediction model based on injection molding-related data such as design or process rather than process monitoring results, thereby enabling accurate prediction of injection molding failure throughout the entire product cycle and overcoming existing limitations by applying it to various geometric shapes and process conditions.

[0062] In addition, for example, the condition information input program (67) is a program that receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal (10), through which the user can transmit his / her filling prediction simulation condition for which he / she wishes to receive a filling time distribution pattern.

[0063] In addition, for example, the charging time distribution pattern prediction program (68) is a program that generates charging time distribution pattern prediction information using the characteristic values ​​extracted from the characteristic value extraction program (65) and the charging prediction model according to the input injection molding charging prediction simulation conditions, and can generate charging time distribution pattern prediction information according to the user's charging prediction simulation condition information using the already verified charging prediction model.

[0064] Here, the above charging time distribution pattern prediction information can be used to verify the difference between the result predicted by the charging prediction model and the CAE injection analysis value by using CAE charging analysis information that displays the charging time distribution of the analysis target in color, and display the charging analysis information by displaying the difference value and the charging time distribution of the analysis target in color on the analysis target.

[0065] In addition, for example, the charging time distribution pattern prediction information transmission program (69) is a program that transmits the charging time distribution pattern prediction information to the user terminal, and when the user inputs injection molding charging prediction simulation condition information for performing charging prediction, the charging time distribution pattern prediction information predicted by the learned charging prediction model can be transmitted very quickly, from several seconds to several minutes.

[0066] Here, the programs described above can be operated in a form linked to an execution program, a screen control program, or a user application downloaded or installed on various terminals such as the user terminal (10) or the operator terminal (90).

[0067] However, the programs described above are not necessarily limited to being linked with executable programs or smartphone applications, but can be linked with all types of terminals.

[0068] Meanwhile, as illustrated in FIG. 1, the database (DB) may include a member registration information database (71) in which the member registration information is stored, a login information database (72) in which the login information is input, a charging prediction model information database (73) in which charging prediction model information related to the charging prediction model is stored, a simulation condition information database (74) in which the injection molding charging prediction simulation condition information is stored, a charging time distribution pattern prediction information database (75) in which the charging time distribution pattern prediction information is stored, and an other information database (76) in which other information is stored.

[0069] Accordingly, the server computer (60) extracts injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program (64), generates a filling time distribution pattern vector of an analysis target connected by elements and nodes, generates a feature value between the analysis targets by the feature value extraction program (65), performs ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature value and the filling time distribution pattern vector by the filling prediction model generation program (66), generates an artificial intelligence-based filling prediction model, receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal (10) by the condition information input program (67), and, by the filling time distribution pattern prediction program (68), extracts the feature value extracted by the feature value extraction program (65) according to the input injection molding filling prediction simulation condition and the feature value and the An injection charging prediction simulation service can be performed by generating charging time distribution pattern prediction information using a charging prediction model and transmitting the charging time distribution pattern prediction information to the user terminal (10) by the charging time distribution pattern prediction information transmission program (69).

[0070] Figure 2 is a conceptual diagram showing the relationship between an operator and a user who operates the injection filling prediction simulation service of the present invention.

[0071] As illustrated in Fig. 2, users can receive advanced data such as highly accurate and reliable charging time distribution patterns within a short period of time (e.g., seconds to minutes), thereby saving significantly on time and labor costs compared to receiving conventional CAE analysis results over several days or weeks. In return, operators can receive various fees, membership fees, information usage fees, etc., thereby providing a useful business model (revenue-generating model) or computer management model that is beneficial to all.

[0072] FIG. 3 is a conceptual diagram illustrating a method for operating an injection filling prediction simulation service according to some embodiments of the present invention.

[0073] As shown in FIGS. 1 to 3, a method for operating an injection molding filling prediction simulation service according to some embodiments of the present invention is sequentially shown. The method for operating an injection molding filling prediction simulation service according to some embodiments of the present invention comprises: (a) a step of extracting injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program (64) and generating a filling time distribution pattern vector of an analysis object connected by elements and nodes; (b) a step of generating feature values ​​between the analysis objects by the feature value extraction program (65); (c) a step of performing ANN (Artificial Neural Network) learning to find out a functional relationship according to the filling time distribution pattern vector and the feature values ​​by the filling prediction model generation program (66) and generating an artificial intelligence-based filling prediction model; and (d) a step of inputting injection molding filling prediction simulation conditions for performing filling prediction from the user terminal (10) by the condition information input program (67). It may include a step of inputting information, (e) a step of generating charging time distribution pattern prediction information using the characteristic values ​​extracted from the characteristic value extraction program (65) and the charging prediction model according to the input injection molding charging prediction simulation conditions by the charging time distribution pattern prediction program (68), and (f) a step of transmitting the charging time distribution pattern prediction information to the user terminal (10) by the charging time distribution pattern prediction information transmission program (69).

[0074] However, the present invention is not necessarily limited to the drawings, and various other steps may be additionally included.

[0075] FIG. 4 is a conceptual diagram illustrating a system for operating an injection filling prediction simulation service according to some other embodiments of the present invention.

[0076] As illustrated in FIG. 4, a system for operating an injection filling prediction simulation service according to some other embodiments of the present invention includes a type of AI Simulator. When various product information (Product CAD Data), mold information (Mold Data), or process information (Process Data) that have already been simulated and tested are input, the AI ​​Simulator of the present invention stores this in a database through a data management program, and uses this to repeat processes such as "Pre-Processing," "Feature Selection," and "Traning" to change internal weights while training an AI model for injection molding composed of a cooling module, a filling module, a pressure module, a deformation module, etc. in a direction with high accuracy, and then performs post-processing to complete the AI ​​model.

[0077] Therefore, when a user inputs simulation conditions, the AI ​​model can provide the predicted simulation results to the user through the User Interface System.

[0078] Therefore, users can utilize it for product design, mold design, process design, injection production, etc., and even general engineers who are not analysis experts can easily and quickly obtain prediction results.

[0079] Fig. 5 is a conceptual diagram showing an AI simulator of a system that operates the injection filling prediction simulation service of Fig. 4.

[0080] As illustrated in FIG. 5, the AI ​​Simulator of the present invention can perform a pre-processing process such as data standardization or data classification and a feature selection process using result data such as simulation information or test information using existing product information (Product CAD Data), mold information (Mold Data), process information (Process Data), etc.

[0081] These characteristic values ​​can be obtained by extracting node-specific variables such as distance to gate, thickness, distance to surface, and L / t ratio, or by extracting intermediate variables such as modified fill time. Through this, artificial intelligence models can be trained through a feedback process in various ways, such as hyperparameter tuning that adjusts epochs, learning rates, etc.

[0082] FIG. 6 is a flowchart illustrating a method of operating an injection filling prediction simulation service according to some other embodiments of the present invention.

[0083] As illustrated in FIG. 6, a method of operating an injection filling prediction simulation service according to some other embodiments of the present invention will be sequentially described. First, injection analysis result information obtained by CAE injection analysis is extracted, and then a filling time distribution pattern vector can be created and feature values ​​between analysis objects can be created.

[0084] At this time, the analysis target is a cooling channel, an injection mold, or an injection product composed of nodes, and the characteristic values ​​between the analysis target may be composed of a geometric distance between the constituent nodes, a geometric characteristic value of each node, and a representative material property of each node. The representative material property may be a flow-related material property such as viscosity, and the geometric distance between nodes may be a distance between surrounding nodes within a specific range, and the geometric characteristic value of a node may be a ratio of the flow path length and thickness from the gate to the corresponding node.

[0085] Next, ANN (Artificial Neural Network) learning is performed to find out the functional relationship between the charging time distribution pattern vector and the above-mentioned characteristic value, and an artificial intelligence-based charging analysis model can be created and stored.

[0086] Next, when the user inputs the condition values ​​of the injection molding process for which he or she wants to perform AI-based filling analysis, the input condition values ​​are applied to the filling analysis model, and an accurate filling time distribution pattern prediction result can be provided very quickly, in seconds to minutes.

[0087] Therefore, users can use these charging time distribution pattern prediction results in a variety of ways, such as changing their own designs or reflecting them in other designs.

[0088] FIG. 7 is a diagram showing several examples of characteristic values ​​applied in the method of operating the injection filling prediction simulation service of FIG. 6.

[0089] As shown in FIG. 7, the characteristic values ​​are thickness (Thickness), square root of thickness (SQRT), square of thickness, volume of element located at half the length of product size (Near Half), volume of element located at 1 / 4 the length of product size (Near Quarter), sum of the volume of each coolant element divided by the distance (Cool over dist), value calculated only from those located in the direction of the surface where the corresponding node is located when calculating Cool over dist, sum of the volume of each coolant element divided by the square of the distance (Cool over dist 2), sum of the volume of each coolant element located in the direction of the surface where the corresponding node is located when calculating Cool over dist divided by the square of the distance (Cool over dist on 2), diameter of the coolant channel composed of MODEL_BEAM elements (Diameter), volume of product element adjacent to the coolant channel divided by the distance (Part Near over dist), volume of product element adjacent to the coolant channel divided by the square of the distance (Part Near over dist2), crystallization resin (0: amorphous, 1: crystalline), transition temperature of the resin, specific heat (Cp, Specific heat), thermal conductivity (Terminal conductivity), solid state density (Density), product of solid state density and specific heat, product of solid state density, specific heat and thickness, product of solid state density, specific heat and thickness squared, melt temperature (Melt Temperature), cooling time (Cool Time), cooling channel temperature (Cool Temp), ratio of flow path length to thickness to node (L / t ratio), common logarithm of length to thickness ratio (L / t ratio(log)), mold temperature (Mold Temperature), etc. can be applied.

[0090] Fig. 8 is a screen showing an example of an AI model applied in the method of operating the injection filling prediction simulation service of Fig. 6.

[0091] As illustrated in Fig. 8, the AI ​​model applied in the method of operating the injection filling prediction simulation service can also set conditions by adjusting Feature Sets, Excluded Features, Results Apply, Ai Model Layer Setting, Optimizer, etc.

[0092] Fig. 9 is a screen showing an example of a graph that color-codes the CAE analysis results used for learning in the method of operating the injection filling prediction simulation service of Fig. 6. Fig. 10 is a screen showing an example of a graph that color-codes the filling time distribution, which is an AI prediction result used for learning in the method of operating the injection filling prediction simulation service of Fig. 6. Fig. 11 is a screen showing an example of a graph that color-codes the difference value between the CAE analysis results of Fig. 9 and the AI ​​prediction results of Fig. 10.

[0093] Accordingly, by comparing the graph representing the charging time distribution as a result of the existing CAE analysis of Fig. 9 in color with the graph representing the charging time distribution as a result of the AI ​​prediction of Fig. 10 in color, a graph representing the difference values ​​as a result of the difference values ​​of Fig. 11 in color can be obtained, and by repeating learning in the direction of minimizing the difference values ​​of this difference value graph and determining the weights, a highly accurate AI model can be trained.

[0094] Therefore, according to the present invention, an AI-based standalone simulator that converts CAE analysis simulation into AI-based simulation can be developed, and an AI model is created by utilizing CAE analysis simulation result data as AI learning data, and as a simulator that derives the same results as CAE analysis simulation, it has novelty compared to existing technologies that use AI as an optimization tool or ROM application tool for CAE analysis simulation, and an AI-based injection molding product simulator that predicts injection molding results (injection molding quality) based on related data (product design CAD model, mold conditions, process conditions) in the injection molding field can be developed, and instead of a method of predicting quality based on process condition data collected during the injection molding process, injection molding results can be predicted throughout the product life cycle from design to product production by simulating injection molded products based on product CAD models, mold characteristics, and process conditions, and a simulator that can accommodate geometric shapes and process conditions in injection molding simulation can be developed, and an AI-based injection molding product simulator created based on various CAE analysis results, which can be simulated. It is original in that it breaks down the wall of constraints (type of product, process conditions, etc.), and it is possible to develop an independently operated simulator that performs simulations with AI itself, rather than AI as an auxiliary / support means of CAE analysis, and by implementing the function of CAE analysis simulation as an independent AI-based simulator, it has the advantage of not only saving time and cost, but also enabling simulation application by non-experts, and simulation is not the exclusive domain of specific analysis specialist engineers, but can be utilized by many engineers at various stages, and it is possible to develop a simulator that predicts injection molding results with product design data, mold conditions, process conditions, etc. as input conditions, rather than predicting injection molding quality using the results of injection molding process monitoring.While existing AI applications in the injection molding field have focused primarily on monitoring and controlling the injection process, the difference lies in the fact that injection molding simulation can be applied at various stages under various conditions.

[0095] In particular, according to the present invention, the accuracy with the CAE simulation results is 90 percent or more, the time can be significantly shortened to about 1 / 50, and the accuracy with the actual flow is also 90 percent or more, so even non-experts can obtain quick and accurate results with the help of the AI ​​model of the present invention.

[0096] That is, although recent simulation technology is the basic technology of digital twin / CPS and is emerging as an important technology from the design and production process, in the past, it required the specialized skills of an analysis expert, which significantly increased time and cost. However, with the AI ​​model of the present invention, not only is it fast and low-cost, but even general designers and manufacturers can easily use it as a simulation tool to achieve real-time quality improvement in various fields such as materials, electronics, and automobiles.

[0097] While the present invention has been described with reference to one embodiment illustrated in the drawings, this is merely exemplary, and those skilled in the art will understand that various modifications and equivalent alternative embodiments are possible. Therefore, the true technical protection scope of the present invention should be determined by the technical spirit of the appended claims.

Claims

1. A method for operating an injection filling prediction simulation service using a computer system having at least one user terminal and a server computer connected through a network, The server computer comprises: an injection CAE analysis information extraction program which extracts injection analysis result information obtained by CAE injection analysis and generates a filling time distribution pattern vector of an analysis target connected by elements and nodes; a feature value extraction program which generates feature values ​​between the analysis targets; a filling prediction model generation program which performs ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program which receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a filling time distribution pattern prediction program which generates filling time distribution pattern prediction information using the feature values ​​extracted by the feature value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions; a filling time distribution pattern prediction information transmission program which transmits the filling time distribution pattern prediction information to the user terminal; a simulation condition information database in which the injection molding filling prediction simulation condition information is stored; Includes a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, (a) a step of extracting injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program, and generating a filling time distribution pattern vector of an analysis target connected by elements and nodes; (b) a step of generating feature values ​​between the analysis targets by the feature value extraction program; (c) a step of generating an artificial intelligence-based charging prediction model by performing ANN (Artificial Neural Network) learning to find out the functional relationship according to the charging time distribution pattern vector and the characteristic value by the charging prediction model generation program; (d) a step of receiving injection molding filling prediction simulation condition information for performing filling prediction from the user terminal by the condition information input program; (e) a step of generating filling time distribution pattern prediction information by using the characteristic value extracted by the characteristic value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions by the filling time distribution pattern prediction program; and (f) a step of transmitting the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program; A method for operating an injection molding filling prediction simulation service, comprising:

2. In paragraph 1, In step (a) above, The above interpretation object is a method for operating an injection filling prediction simulation service by selecting at least one of a cooling channel, an injection mold, an injection molded product, or a combination thereof.

3. In paragraph 1, In step (b) above, The above characteristic value is composed of at least one of the number of surrounding nodes, the geometric distance between each node, the geometric characteristic value of each node, the characteristic material property of each node, and combinations thereof. A method for operating an injection filling prediction simulation service, wherein the above characteristic properties are flow-related properties including at least the viscosity of the resin.

4. In paragraph 3, A method for operating an injection filling prediction simulation service, wherein the geometric distance between the nodes is the distance between surrounding nodes within a specific range, and the geometric characteristic value of the node is the ratio of the length and thickness of the flow path from the gate to the node (L / t ratio).

5. In paragraph 1, In step (a) above, The above injection analysis result information is generated by CAE injection analysis based on at least one of injection molded product model information, injection mold condition information, injection material condition information, and injection process condition information, and combinations thereof. A method for operating an injection filling prediction simulation service, wherein the above injection process condition information includes temperature and flow rate information of cooling water injected into a cooling channel, temperature information of resin injected into an injection mold, and a ram speed profile that determines the speed of resin injected into the injection mold.

6. In paragraph 1, In step (c) above, A method for operating an injection filling prediction simulation service, wherein the above filling prediction model is an artificial intelligence model that predicts a filling time distribution by selecting at least one of an injection mold, an injection product, or a combination thereof.

7. In paragraph 1, In step (b) above, A method for operating an injection filling prediction simulation service, wherein the above characteristic values ​​include at least one of thickness, a ratio of the flow path length to the node and the thickness (L / t ratio), a common logarithm of the ratio of the length to the thickness (L / t ratio(log)), a mold temperature value (Mold Temperature), and combinations thereof.

8. In paragraph 1, In step (e) above, When generating the above charging time distribution pattern prediction information, the difference between the result predicted by the charging prediction model and the CAE injection analysis value is verified using the charging analysis information that displays the charging time distribution of the analysis target in color. The above charging time distribution pattern prediction information is a method for operating an injection charging prediction simulation service that displays charging analysis information by representing the difference value and the charging time distribution of the analysis object in color on the analysis object.

9. In a system that operates an injection filling prediction simulation service using a computer system having at least one user terminal and a server computer connected through a network, The server computer comprises: an injection CAE analysis information extraction program which extracts injection analysis result information obtained by CAE injection analysis and generates a filling time distribution pattern vector of an analysis target connected by elements and nodes; a feature value extraction program which generates feature values ​​between the analysis targets; a filling prediction model generation program which performs ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program which receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a filling time distribution pattern prediction program which generates filling time distribution pattern prediction information using the feature values ​​extracted by the feature value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions; a filling time distribution pattern prediction information transmission program which transmits the filling time distribution pattern prediction information to the user terminal; a simulation condition information database in which the injection molding filling prediction simulation condition information is stored; Includes a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, A control unit programmed to extract injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program, generate a charging time distribution pattern vector of analysis objects connected by nodes, generate feature values ​​between the analysis objects by the feature value extraction program, perform ANN (Artificial Neural Network) learning to find out a functional relationship according to the charging time distribution pattern vector and the feature values ​​by the charging prediction model generation program, generate an artificial intelligence-based charging prediction model, receive injection molding charging prediction simulation condition information for performing charging prediction from the user terminal by the condition information input program, generate charging time distribution pattern prediction information using the feature values ​​extracted from the feature value extraction program and the charging prediction model according to the input injection molding charging prediction simulation conditions by the charging time distribution pattern prediction program, and transmit the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program. A system that operates an injection molding filling prediction simulation service.

10. A computer-readable recording medium having recorded thereon a program for operating an injection filling prediction simulation service using a computer system having at least one user terminal and a server computer connected through a network, The server computer comprises: an injection CAE analysis information extraction program which extracts injection analysis result information obtained by CAE injection analysis and generates a filling time distribution pattern vector of an analysis target connected by elements and nodes; a feature value extraction program which generates feature values ​​between the analysis targets; a filling prediction model generation program which performs ANN (Artificial Neural Network) learning to find out a functional relationship according to the feature values ​​and the filling time distribution pattern vector; a condition information input program which receives injection molding filling prediction simulation condition information for performing filling prediction from the user terminal; a filling time distribution pattern prediction program which generates filling time distribution pattern prediction information using the feature values ​​extracted by the feature value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions; a filling time distribution pattern prediction information transmission program which transmits the filling time distribution pattern prediction information to the user terminal; a simulation condition information database in which the injection molding filling prediction simulation condition information is stored; Includes a charging time distribution pattern prediction information database in which the charging time distribution pattern prediction information is stored, (a) a step of extracting injection analysis result information obtained by CAE injection analysis by the injection CAE analysis information extraction program and generating a filling time distribution pattern vector of an analysis target connected by nodes; (b) a step of generating feature values ​​between the analysis targets by the feature value extraction program; (c) a step of generating an artificial intelligence-based charging prediction model by performing ANN (Artificial Neural Network) learning to find out the functional relationship according to the charging time distribution pattern vector and the characteristic value by the charging prediction model generation program; (d) a step of receiving injection molding filling prediction simulation condition information for performing filling prediction from the user terminal by the condition information input program; (e) a step of generating filling time distribution pattern prediction information by using the characteristic value extracted by the characteristic value extraction program and the filling prediction model according to the input injection molding filling prediction simulation conditions by the filling time distribution pattern prediction program; and (f) a step of transmitting the charging time distribution pattern prediction information to the user terminal by the charging time distribution pattern prediction information transmission program; A computer-readable recording medium having recorded thereon a program for operating an injection filling prediction simulation service, which includes:

Images