Injection molding method, molding condition deriving device, and computer-readable storage medium

Abstract

The method comprises a step (S1200) for constructing a predictive model on the basis of input parameters including molding conditions for a molded article (500) and target variable values including quality values that quantify the required quality of the molded article (500) with respect to the input parameters, a step (S1210) for inferring a predictive distribution of target variable values with respect to the input parameters using the predictive model, and a step (S1220) for deriving molding conditions that satisfy the required quality of the molded article through Bayesian optimization using a regression model for seeking the input parameters that yield the highest quality value over an initial quality value in evaluation of the target variable values according to the predictive distribution.

Description

[Technical field]

[0001] The present application relates to an injection molding method, a molding condition deriving device, and a computer-readable storage medium. [Background technology]

[0002] Injection molding is a method of molding resin parts by injecting molten resin material into a mold, and is widely used. In injection molding, in order to mold high-quality resin parts (hereinafter referred to as molded products) that meet the required quality, it is essential to determine the appropriate molding conditions. However, since the appropriate molding conditions differ depending on the shape of the molded product and the properties of the resin used, the work of determining these molding conditions is carried out by skilled workers with a wealth of knowledge and experience.

[0003] Also, as a conventional technique, a method for optimizing the quality of molded products using a neural network has been proposed as a method for supporting molding by an injection molding machine. To construct this neural network, molding conditions are used as input parameters, and quality values ​​obtained by measuring the quality of molded products are used as output items (hereinafter referred to as objective variables) (for example, see Patent Document 1 below). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2008-110486 A Summary of the Invention [Problem to be solved by the invention]

[0005] The task of determining appropriate molding conditions that satisfy the required quality of the molded product is generally performed by skilled workers with ample knowledge and experience. On the other hand, when workers with little knowledge and experience determine molding conditions, they have to repeat a lot of trial and error, and there is a problem that it takes a very long time to determine the molding conditions.

[0006] Furthermore, when using a neural network as in the conventional technology, there is a problem in that a large amount of learning data, on the order of hundreds to tens of thousands, is required to construct a prediction function for optimizing molding conditions.

[0007] Furthermore, in the conventional technology, measured quality values ​​(product weight, warpage, dimensions, etc.) are utilized to optimize molding conditions. However, in the case of quality values ​​that require measurement with a high-resolution measuring instrument (sink marks, flow marks, etc.), problems arise such as the difficulty of preparing the measuring instrument being expensive, the need to cut out measurement samples, and the inability to measure easily.

[0008] The present application discloses technology for solving the problems described above, and aims to provide an injection molding method, a molding condition derivation device, and a computer-readable storage medium that can easily obtain appropriate molding conditions that satisfy the quality required of the molded product, regardless of the technical level of the molding worker. [Means for solving the problem]

[0009] The injection molding method disclosed in the present application comprises: An objective variable value including input parameters including molding conditions of a molded product and a quality value that quantifies a required quality of the molded product for the input parameters, , the above Indirect quality value including feature values ​​converted from the appearance image of the molded product To building a predictive model based on the inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that are most highly evaluated to produce a desired quality value for the molded product by a Bayesian optimization method using a regression model that determines the input parameters that produce the highest quality value for the objective variable value compared to the initial quality value based on the predictive distribution. By using an optimization method that repeatedly evaluates the objective variable values ​​under the derived molding conditions and derives the molding conditions that are most highly evaluated to achieve the desired quality value, molding conditions that satisfy the desired required quality are derived. The molding condition derivation device disclosed in the present application is a molding condition derivation device that optimizes molding conditions for a molded product based on the above-mentioned injection molding method, A storage unit in which information on molding conditions and required quality of the molded product is stored in advance; A control processing unit is provided, The control processing unit: The above an indirect quality value processing unit for obtaining an indirect quality value including a feature amount converted from an appearance image of a molded product; The above the indirect quality value from the indirect quality value processing unit of The system has a molding condition optimization unit that uses the captured quality value, the molding conditions stored in the memory unit, and the required quality information stored in the memory unit to derive molding conditions that satisfy the optimal required quality of the molded product by a Bayesian optimization method using a regression model. The present application also discloses a computer-readable storage medium, A computer-readable storage medium having a computer program stored thereon, the computer program comprising, when executed by a processor, the steps of: An objective variable value including input parameters including molding conditions of a molded product and a quality value that quantifies a required quality of the molded product for the input parameters, , the above Indirect quality value including feature values ​​converted from the appearance image of the molded product To building a predictive model based on the inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that are most highly evaluated to produce a desired quality value for the molded product by a Bayesian optimization method using a regression model that determines the input parameters that produce the highest quality value for the objective variable value compared to the initial quality value based on the predictive distribution. The method derives molding conditions that satisfy the desired required quality by using an optimization method that repeatedly evaluates the target variable values ​​under the derived molding conditions and derives the molding conditions that are most highly evaluated to achieve the desired quality value. The injection molding method disclosed in the present application further comprises: A step of constructing a prediction model based on input parameters including molding conditions of a molded product, feature quantities of sensor values ​​of a sensor arranged in an injection molding machine corresponding to the input parameters, and objective variable values ​​including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value which is the sensor value when the molded product satisfies a required quality; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that finds input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of the initial sensor value based on the predictive distribution. Further, a molding condition deriving device disclosed in the present application is a molding condition deriving device that optimizes molding conditions for a molded product based on the above-mentioned injection molding method, A storage unit in which information on molding conditions and required quality of the molded product is stored in advance; A control processing unit is provided, The control processing unit: a sensor value feature processing unit that calculates a feature of the sensor value obtained from the sensor value and a similarity of the sensor value to the reference sensor value; The molding condition optimization unit imports the feature values ​​of the sensor value and the similarity to the reference sensor value from the feature value processing unit of the sensor value, and uses the imported feature values ​​of the sensor value and the similarity to the reference sensor value, as well as the molding conditions and required quality information stored in the memory unit, to derive molding conditions that satisfy the optimal required quality of the molded product by a Bayesian optimization method utilizing a regression model. The present application also discloses a computer-readable storage medium, A computer-readable storage medium having a computer program stored thereon, the computer program comprising, when executed by a processor, the steps of: A step of constructing a prediction model based on input parameters including molding conditions of a molded product, feature quantities of sensor values ​​of a sensor arranged in an injection molding machine corresponding to the input parameters, and objective variable values ​​including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value which is the sensor value when the molded product satisfies a required quality; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and a step of deriving molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that finds input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of the initial sensor value based on the predictive distribution. Effect of the Invention

[0010] According to the injection molding method, molding condition deriving device, and computer-readable storage medium disclosed in the present application, a prediction function for optimizing molding conditions can be constructed even with a small amount of data, and molding conditions that satisfy the required quality can be easily derived without depending on the technical level of the molding operator. [Brief description of the drawings]

[0011] [Figure 1] FIG. 2 is a diagram showing an example of an apparatus configuration required to realize the injection molding method according to the first embodiment. [Diagram 2] FIG. 2 is a diagram showing an example of an apparatus configuration required to realize the injection molding method according to the first embodiment. [Diagram 3] FIG. 13 is a schematic diagram showing an example of setting range information of molding conditions. [Figure 4] FIG. 13 is a schematic diagram showing an example of the item influence degree of molding conditions. [Diagram 5] FIG. 4 is a schematic diagram showing an example of molded product information. [Figure 6] FIG. 13 is a characteristic diagram showing an example of a correlation between a defined sink mark feature amount and a measured sink mark amount. [Figure 7] 7A, 7B, and 7C are explanatory diagrams showing an example of an image of a molded product that has been subjected to image processing. [Figure 8] FIG. 1 is a diagram showing a schematic diagram of a method for determining the next search conditions (molding conditions) using the EI value. [Figure 9] FIG. 13 is a diagram showing a schematic diagram of a method for determining the next search condition (molding condition) using the EI value. [Figure 10] 4 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of the present application. [Figure 11] 4 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of the present application. [Figure 12] 4 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of the present application. [Figure 13] 4 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of the present application. [Figure 14] 1 is an explanatory diagram showing an example of a screen of a molding condition derivation program operated by a molding condition derivation device. FIG. [Figure 15] 11 is an explanatory diagram showing an example of a screen of the molding condition derivation program in which molding of the initial number of data has been completed and quality values ​​have been input. FIG. [Figure 16] FIG. 13 is an explanatory diagram showing an example of a screen of a molding condition derivation program after the first optimization has been performed. [Figure 17] FIG. 2 is a diagram illustrating an example of a hardware configuration of a control processing unit of the present application. [Figure 18] 3 is a block diagram showing details of a control processing unit of the molding condition deriving device according to the embodiment 1. FIG. [Figure 19] 4 is a flowchart showing steps executed by a control processing unit of the molding condition deriving device according to the first embodiment. [Figure 20] FIG. 11 is a diagram showing an example of an apparatus configuration required to realize an injection molding method according to a second embodiment. [Figure 21] FIG. 11 is a diagram showing an example of an apparatus configuration required to realize an injection molding method according to a second embodiment. [Figure 22] 5A and 5B are diagrams illustrating an example of feature amounts in the X direction and the Y direction of a sensor value acquired in a control processing unit of the present application. [Diagram 23] 13 is a diagram showing an example of a calculation result in which the degree of similarity to a reference sensor value is low in the control processing unit of the present application; FIG. [Figure 24] 13 is a diagram showing an example of a calculation result in which the control processing unit of the present application has a high similarity to a reference sensor value; FIG. [Diagram 25] 11 is a flowchart showing an example of a series of processing procedures for making advance preparations for optimizing molding conditions in the injection molding method according to the second embodiment. [Figure 26] 11 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method according to the second embodiment. [Figure 27] 11 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method according to the second embodiment. [Figure 28] 11 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method according to the second embodiment. [Figure 29] 11 is a flowchart showing an example of a series of processing procedures for acquiring a feature amount of a sensor value in the injection molding method according to the second embodiment. [Diagram 30] FIG. 11 is an explanatory diagram showing an example of a screen of a molding condition derivation program in the second embodiment. [Diagram 31] FIG. 11 is an explanatory diagram showing an example of a screen of the molding condition derivation program in which molding of the initial number of data items has been completed and feature amounts of sensor values ​​have been input in the second embodiment. [Diagram 32] FIG. 11 is an explanatory diagram showing an example of a screen of a molding condition derivation program in which an initial optimization has been executed in the second embodiment. [Diagram 33] 11 is a block diagram showing details of a control processing unit of a molding condition deriving device according to embodiment 2. FIG. [Diagram 34]10 is a flowchart showing steps executed by a control processing unit of a molding condition deriving device according to the second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0012] The present application relates to an injection molding method and a molding condition deriving device that utilize a regression model that can obtain a posterior distribution of an output relative to an input. Hereinafter, a detailed description will be given based on an embodiment.

[0013] Embodiment 1 [Configuration for realizing the injection molding method] First, a configuration for carrying out the injection molding method of the present application will be described with reference to FIGS. 1 and 2 are diagrams showing an example of an apparatus configuration required to realize the injection molding method according to the first embodiment. 1, an injection molding machine 200 is equipped with a mold 210 for molding a molded product 211, and various sensors 212 are attached to the mold 210. Data measured by the sensor 212 is input via a measurement amplifier 220 to a control processing unit 120 of the molding condition derivation device 100, which will be described later. Meanwhile, the molded product 211 molded in the metal mold 210 is taken out by a take-out robot 300. The taken-out molded product 500 is placed on a transport conveyor 400, and then measured by a shape measuring device 600, and its appearance is photographed by a camera 700. The measurement results by the shape measuring device 600 and the appearance photograph taken by the camera 700 are input into a control processing unit 120 of the molding condition deriving device 100, which will be described later. The configuration of FIG. 1 will be explained in detail later.

[0014] As shown in FIG. 2, the molding condition deriving device 100 of the first embodiment includes a communication unit 110, a control processing unit 120, a display input unit 130, and a storage unit 140.

[0015] In this case, the molding condition derivation device 100 may be a single device, or may be composed of multiple devices or a system connected by a network such as a WAN (Wide Area Network) or a LAN (Local Area Network). Furthermore, the molding condition derivation device 100 may be realized by a system using distributed computing or cloud computing, or by multiple computer devices.

[0016] The communication unit 110 includes, for example, a communication interface such as a network interface card (NIC) and a direct memory access (DMA) controller. The communication unit 110 can communicate with the injection molding machine 200 via a network such as a WAN or a LAN.

[0017] The control processing unit 120 includes a molding condition next condition output unit 121, a molding condition optimization unit 122, an indirect quality value processing unit 123, and a direct quality value processing unit 124. The control processing unit 120 is configured, for example, as shown in FIG. 17, as an example of a hardware configuration, of a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a storage device 1010 (a storage device 140 described later), and is realized by the processor 1000 executing a program stored in the storage device 1010 (a storage device 140 described later). The components of the control processing unit 120 may be realized by hardware such as an FPGA (Field Programmable Gate Array), or may be configured by both software and hardware.

[0018] The display input unit 130 includes a display device such as an LCD display, and can be used by a molding operator operating the molding condition derivation device 100 to grasp the progress of molding condition optimization, and to perform settings and operations via a GUI (Graphical User Interface).

[0019] The storage unit 140 includes, for example, a hard disk drive (HDD), a solid state drive (SSD), a read only memory (ROM), a random access memory (RAM), etc. The storage unit 140 stores various programs for deriving molding conditions, such as firmware and application programs, as well as setting range information 141 of molding conditions, item influence degree 142 of molding conditions, and molded product information 143, which will be described later.

[0020] As shown in FIG. 1, an injection molding machine 200 includes a mold 210 as a tool for molding a molded product 211 that is the target of molding condition optimization, and various sensors 212 are attached to this mold 210.

[0021] The sensor 212 includes a strain or piezoelectric pressure sensor that measures the resin pressure, a strain gauge that measures the amount of strain in the mold 210, a thermocouple or infrared temperature sensor that measures the temperature of the mold 210 and the resin temperature, an AE (Acoustic Emission) sensor that detects acoustic radiation within the mold 210, and the like, and any type of sensor can be used as long as it can be attached to the mold 210. Various types of sensor data measured by the sensor 212 are input into the control processing unit 120 via the measurement amplifier 220, and are converted into a quality value for evaluating the quality of the molded product in an indirect quality value processing unit 123.

[0022] The molded product 211 formed in the metal mold 210 is removed by a removal robot 300. The removed molded product 500 is placed on a transport conveyor 400, and then its shape data such as flatness and dimensions are measured by a shape measuring device 600. The shape measuring device 600 may be a measuring device such as a vernier caliper or a height gauge, or may be a contact or non-contact three-dimensional measuring device.

[0023] The external appearance of the removed molded article 500 is photographed by a camera 700. In order to photograph the external appearance of the molded article 500 with the camera 700, lighting, a blackout curtain, and a jig may be added as necessary. Also, a single camera 700 may be used, or multiple cameras 700 may be used. The photographed external appearance photo of the molded article 500 is taken into the control processing unit 120, and is converted into a quality value for evaluating the quality of the molded article 211 in an indirect quality value processing unit 123.

[0024] FIG. 3 is a schematic diagram showing an example of set width information 141 of molding conditions stored in advance in storage unit 140. As shown in FIG. The molding condition setting range information 141 is preset with information on settable upper and lower limit values ​​for the setting items (input parameters) of the molding conditions for each type of molded product 211. The upper and lower limit values ​​may be set based on experience and knowledge, or may be set based on the results of a resin flow analysis. For example, the injection temperature may be set within the temperature range for each resin material recommended by the resin material manufacturer, or may be set after confirming in advance whether or not resin can be filled into the mold 210 by a resin flow analysis. Alternatively, the upper and lower limit values ​​may be set after actually performing a provisional molding to confirm whether or not there are any problems in molding.

[0025] Here, each item in FIG. 3 will be explained. Mold temperature (movable) is a parameter for controlling the temperature of the mold, and "movable" refers to the side of the mold that moves (opens) when the molded product is removed from the mold. The mold has pipes for the flow of hot water, cold water, oil, or other fluids, and the mold temperature is controlled by controlling the temperature of the fluid flowing through the pipes with a temperature regulator. Mold temperature (fixed), as mentioned above, is a parameter for controlling the mold temperature, and "fixed" refers to the side of the mold that does not move when the molded product is removed from the mold. Note that it is common to set a temperature difference between the movable and fixed sides of the mold (e.g. 50°C on the movable side, 30°C on the fixed side). Injection temperatures 1 to 5 are parameters that control the temperature at which the resin pellets (resin particles) injected into the mold are melted. The numbers "1 to 5" indicate the locations of the thermocouples, which are installed in 4 to 6 places in the heating barrel of the injection molding machine, in order from the tip of the injection unit. The injection molding machine also controls the heater of the heating barrel so that the thermocouple reaches the set temperature value. The set value is set within the range recommended by the resin material manufacturer, while checking the quality of the molded product or the production cycle time. Injection positions 1 to 4 are parameters that control the screw position of the injection molding machine, which switches the speed when the molten resin is injected into the mold. The flow of the injected resin is controlled by the combination of injection speed and injection position. The numbers "1 to 4" indicate the positions at which the injection speed is switched. (Example: Up to an injection position (screw position) of 100 mm, the injection speed is 50 mm / s. From injection position 100 mm to 40 mm, the injection speed is 30 mm / s.) Speed ​​and pressure switching is the setting of the screw position that switches the control of the injection molding machine's screw from injection speed control to holding pressure (holding pressure) control when molten resin is injected into a mold. Injection speeds 1 to 4 are parameters that control the speed at which molten resin is injected into the mold, and the flow of the injected resin is controlled by the combination of injection speed and injection position. The numbers "1 to 4" indicate how much the injection speed is changed relative to the injection position. ·Dwell pressures 1 to 3 are parameters that set the magnitude of the hold pressure control after injection speed control when molten resin is injected into a mold. Generally, in injection molding, after the resin is filled into the mold by injection speed control, shrinkage occurs due to the change in state of the resin (liquid to solid), so to compensate for this shrinkage, additional molten resin is filled into the mold by holding pressure control. The numbers "1 to 3" indicate how many stages and to what extent the hold pressure is changed when holding pressure is applied in multiple stages. Also, each hold pressure is controlled by time (for example, hold pressure 1 is 50Mpa for 3s, hold pressure 2 is 30Mpa for 4s). The cooling time is the time it takes to cool and solidify the molten resin after applying holding pressure to the resin inside the mold. Generally, no external pressure is applied to the resin during the cooling time, and only heat exchange occurs between the resin and the mold until the molten resin solidifies. If the cooling time is too short, it can cause warpage or problems releasing the resin from the mold, and if it is too long, it can cause an increase in production cycle time (increasing the cost of molded products) or problems releasing the resin from the mold.

[0026] FIG. 4 is a schematic diagram showing item influence degrees 142 of molding conditions prestored in the storage unit 140. As shown in FIG. The molding condition item influence 142 is set to the magnitude of influence of each molding condition item (e.g., temperature, pressure, injection speed, etc.) on the quality value (e.g., warp, sink marks) of the target molded product 211. In this case, the influence may be set based on experience and knowledge, or based on the results of a resin flow analysis. Alternatively, it may be set based on the results of an actual temporary molding. For example, an orthogonal table is created with the molding condition items as control factors, a resin flow analysis is performed under each condition, and a characteristic value for the quality value of the molded product is calculated. There is also a method of setting the molding condition item influence based on the characteristic value. Even when temporary molding is performed, the influence may be set using an orthogonal table.

[0027] FIG. 5 is a schematic diagram showing molded product information 143 pre-stored in storage unit 140. As shown in FIG. In this molded product information 143, an ID number for individual identification, the resin molding material used, and information on the required quality of the molded product 211 that needs to be optimized are set individually for each type of molded product 211. Any number of pieces of quality information for the molded product 211 can be set as the quality information for the molded product 211. For example, for molded product A in Fig. 5, the warp, sink mark, and dimensions that occur in molded product A are set as quality information. In this case, for example, the warp (information 1) is set as the flatness of an arbitrary measurement point, the sink mark (information 2) is set as the amount of dent at an arbitrary measurement point, and the dimensions (information 3) are set as the dimensional value and dimensional tolerance.

[0028] In implementing the configuration of the present application, the sensor 212, the measurement amplifier 220, the shape measuring device 600, and the camera 700 are devices for quantifying the quality of the molded product 211, and any one of them may be used. By realizing the above configuration, the injection molding method of the present application can be carried out.

[0029] [Quantification of molded product quality] To find optimal molding conditions that satisfy the required quality of the molded product 211 (hereinafter referred to as molding condition parameter optimization), it is necessary to consider this as an optimization problem of optimizing the relationship between input and output. In this embodiment 1, the input is the molding condition value, and the output is the required quality of the molded product. The input molding condition value is a quantitative value, but the required quality of the molded product, which is the output, may be expressed by the molding worker's visual inspection or feeling, and may not be defined as a quantitative value. Therefore, first, a method for acquiring a quality value that quantifies the required quality of the molded product 211 will be described.

[0030] In this embodiment 1, quality values ​​that are easy to measure, such as dimensions and warpage, are defined as direct quality values. On the other hand, quality values ​​that are difficult to measure, such as sink marks and flow marks, are quantified by replacing them with feature values ​​extracted from the values ​​of the sensor 212 installed in the mold 210 or feature values ​​from the images captured by the camera 700, and these feature values ​​are defined as indirect quality values. To use sink marks and flow marks as direct quality values, measurements using a high-resolution measuring instrument are required, but this makes it difficult to prepare the measuring instrument because it is expensive, and requires the work of cutting out measurement samples, making it difficult to measure, and therefore these are considered indirect quality values.

[0031] The direct quality value is found by directly measuring the dimensions, flatness, etc. of the molded product 211. For example, if the required quality is that any dimension falls within a dimensional tolerance, the dimensions are measured using a shape measuring device 600 (a measuring device such as a vernier caliper or a three-dimensional measuring machine), and the measured value is used as the quality value for the required quality. In addition, if the required quality is for warpage, the quality value can be found by measuring the geometric tolerances such as the flatness and squareness of any surface.

[0032] The measured quality value is passed to the display input unit 130 by transmission through a network or by input by the molding operator through a GUI (Graphical User Interface). After that, in the control processing unit 120, the direct quality value processing unit 124 performs preprocessing (processing such as combining the data with other quality values ​​and converting it into sequence information) for Bayesian optimization, which will be described later.

[0033] On the other hand, the indirect quality value is calculated by converting the sensor values ​​and images acquired by the sensor 212 and the camera 700 for the molded product 211 into a quality value. For example, if the required quality is to minimize the amount of sink marks, the sink mark feature amount (defined as the logarithm of the value obtained by dividing the time integral value of the temperature sensor installed in the mold 210 by the time integral value of the pressure sensor) becomes the quality value.

[0034] As shown in Fig. 6, this sink mark feature amount is directly correlated with the sink mark amount (sink mark measurement amount) measured by a high-resolution measuring device, and the smaller the sink mark feature amount, the smaller the sink mark amount (sink mark measurement amount). In addition, if the required quality is to minimize flow marks, the image of the removed molded product 500 taken by the camera 700 is subjected to black and white binarization processing by applying grayscaling, trimming, smoothing the image by passing it through various low-pass filters to blur the image, and the like.

[0035] FIG. 7 is an explanatory diagram showing an example of the result of image processing of an image of a molded product. In Fig. 7, the cases where the flow marks are large (Fig. 7A), small (Fig. 7B), and absent (Fig. 7C) are arranged from left to right. The ratio of the white area in the image changes depending on the size of the flow marks, so this ratio of the white area can be used as a quality value of the flow marks.

[0036] As described above, by using quality values ​​(direct quality value and indirect quality value) that quantify the required quality of the molded product 211, the input can be treated as the value of the molding conditions, and the output can be treated as an optimization problem of the required quality (quality value) of the molded product.

[0037] [Bayesian optimization method] Before describing the series of processes in the method for optimizing the molding conditions, an overview of the Bayesian optimization method used to derive molding conditions that satisfy the required quality of the molded product 211 will be given.

[0038] The Bayesian optimization method is one of parameter optimization methods that can be applied even when the function to be optimized is unknown. First, a prediction model is constructed based on the input parameters (molding conditions in this embodiment 1) and the objective variable values ​​for the input parameters (quality values ​​that quantify the required quality of the molded product 211 described above in this embodiment 1). Then, by using this prediction model, a predictive distribution of the objective variable for the input parameters to be considered is inferred. Using this predictive distribution, the input parameters (molding conditions to be implemented next) that are most highly evaluated for achieving the desired value for the objective variable are presented. By repeating this process, the input parameters are optimized.

[0039] The prediction model used here is a model that can obtain the posterior distribution of the output for the input (in this embodiment, a Gaussian process regression model), but other regression models, such as a random forest regression model, can also be used.

[0040] [Gaussian process regression model] Gaussian process regression is a non-parametric regression method, and compared to neural network regression, it is possible to construct a prediction function even with a relatively small amount of data.

[0041] Gaussian process regression is a model that estimates an objective function y=f(x) from an input variable x to a real-valued objective variable y. Specifically, Data D 1:t ={x 1:t ,y 1:t}, the objective function f(x t+1 ) predictive distribution P(f(x t+1 )|D 1:t , x t+1 ) is calculated using the following formula (1).

[0042]

number

[0043] The right-hand side of equation (1) is the average value (expected value) μ xt+1|x1:t , variance is σ 2 xt+1|x1:t This is a normal distribution (Gaussian distribution). For example, in a Gaussian process regression model where y is standardized so that the mean is 0, if the covariance matrix K, which indicates the relationship between the input parameter x, is expressed as an arbitrary kernel function k(x, x'), the predicted mean and predicted variance can be calculated using the following formulas (2) and (3).

[0044]

number

[0045]

number

[0046] where k ** is k(x * ,x * ), and k * is (k(x * ,x1),k(x * ,x2),···,k(x * ,x n )) T , and the matrix K is k(x n ,x' n ), and the vector y is (y1,y2,...,y n ) vector. In this embodiment, the Gaussian kernel (Θ1, Θ2 are parameters that determine the properties of the kernel) of the following formula (4) is used for the kernel function k(x, x'), but any of the exponential kernel, periodic kernel, and Mater kernel may also be used.

[0047]

number

[0048] [Parameter search method using Bayesian optimization] In the Bayesian optimization method, in order to determine the input parameters (molding conditions) to be verified next, an acquisition function is used to evaluate the combination of input parameter candidates. For this acquisition function, for example, PI (Probability of Improvement) or EI (Expected Improvement) may be used. Alternatively, UCB (Upper Confidence Bound) or MI (Mutual Information) may be used.

[0049] In this embodiment 1, as an example, we will explain the case where an acquisition function called EI (Expected Improvement), which calculates an expected value of how much improvement will be achieved from the minimum (maximum) value of a target variable, is used in correspondence with multiple target variables.

[0050] In the case of a minimization problem of an objective function y=f(x), Bayesian optimization using the EI value determines the next search condition to be the search condition that maximizes the expected value of improvement shown in the following formula (5).

[0051]

number

[0052] A specific method for calculating the EI value is based on the following formula (6).

[0053]

number

[0054] f in formula (6) minrepresents the minimum value of the objective function in the current search iteration, μ(x) and σ(x) represent the predicted mean and predicted standard deviation output from the Gaussian process regression model, Φ represents the cumulative distribution function, and φ represents the probability density function.

[0055] A schematic diagram of the method for determining the next search conditions using this EI value is shown in FIG. The top graph in Figure 8 shows the predictive distribution of the objective function f(x), where the black dots are observed data points, μ(x) is the predicted mean, and CI Upper and CI Lower indicates the upper and lower limits of the confidence interval calculated from the forecast standard deviation. Moreover, the graph in the lower part of Figure 8 shows the calculated EI values, and the black points have already been observed, so the EI values ​​are smaller. In the schematic diagram of FIG. 8, x5, which has the largest EI value, becomes the next search condition.

[0056] The results after the next search condition is subsequently tried are shown in the upper graph of FIG. 9, and a schematic diagram of a method for determining the next search condition is shown in the lower graph of FIG. The predictive distribution of the next search condition, x5, becomes clear, and the EI value of x5 becomes smaller. As a result, the next search condition is determined to be the search condition with the higher EI value. As described above, as shown in Figs. 8 and 9, the next search conditions are determined based on the EI value, and repeated trials are performed to find optimal input parameters.

[0057] In this embodiment 1, since there may be multiple objective functions (for example, the amount of warpage and sink marks of a molded product, and the color difference in an external photograph), the acquisition function is unified by combining each evaluation value (EI value). Although a weighted linear sum method is used as a method for unifying the evaluation value, a simple sum or product method may also be used.

[0058] [How to optimize molding conditions] 10 to 13 are flow charts showing an example of a series of processing procedures for optimizing molding conditions in the injection molding method of the present application. In the following, a specific explanation will be given by taking as an example a case in which the optimization objective variables are warpage and sink marks. In the figures, the symbol S represents a step.

[0059] In optimizing the molding conditions, the initial data collection process is started (steps S100 and S101). For this, first, the molding condition derivation device 100 is started, and the previously described molding condition setting range information 141, molding condition item influence degree 142, and molded product information 143 are set. After that, the molding condition derivation program stored in the storage unit 140 is started.

[0060] As an example, FIG. 14 shows the startup screen of a molding condition derivation program for five input parameters (mold temperature_movable, mold temperature_fixed, injection speed fourth stage, holding pressure first stage, holding pressure second stage) and two objective variables (objective variable 1: warpage, objective variable 2: sink mark).

[0061] When the molding condition derivation program stored in the storage unit 140 is started, it reads the optimized quality information of the molded product information 143 as a target variable, selects molding condition items that are input parameters from the molding condition item influence degree 142, and creates an initial molding condition table as shown in FIG. 14 so that the items fall within the range of the molding condition setting range information 141. Then, the contents are displayed on the display input unit 130. Note that FIG. 14 shows a case where the combination of each molding condition is set to be output randomly, but it is also possible to output the combination as a two-level orthogonal table with each molding condition as a control factor. In addition, the molding operator may change the molding condition items that are input parameters selected by the molding condition derivation program to other molding condition items.

[0062] The molding operator performs molding operations according to the initial molding condition table of Fig. 14 displayed on the display input unit 130 of the molding condition derivation device 100. At this time, the molding conditions may be manually input by the molding operator, or may be automatically input via the communication unit 110 if the injection molding machine 200 is connected to a network. The molding operations are performed sequentially from the first row of the initial molding condition table of Fig. 14, and molding stability is confirmed (step S102).

[0063] In injection molding, when molding conditions are set and then changed, the mold temperature gradually rises or falls as the number of moldings increases. After a certain number of moldings, the temperature change gradually decreases, and molding can be continued at the same mold temperature. This state is used as an indicator to determine that molding has stabilized.

[0064] Specific confirmation methods include a method of checking the time-dependent change in the value of a temperature sensor attached to the mold, or a method of checking the time-dependent change in the pressure value converted from the load cell of the molding machine. In this embodiment 1, the condition for confirming molding stability is that the change in the temperature sensor value is within ±1°C for three consecutive moldings.

[0065] Next, in a state where molding stability has been confirmed, the molding quality value is acquired for the molded product 211 (steps S103 and S501). During molding, the time series data of temperature and pressure acquired by the sensor 212 and the measurement amplifier 220 are transmitted to the molding condition derivation device 100 (step S505). The transmitted time series data is sent to the indirect quality value processing unit 123 and converted to the above-mentioned sensor feature amount (sink mark feature amount) (steps S506 and S507).

[0066] After molding, the molded product 500 is measured for flatness of a predetermined surface by the shape measuring device 600 (steps S502 and S503). The measured flatness is sent to the direct quality value processing unit 124 of the molding condition derivation device 100 by the molding operator inputting the flatness into a GUI (Graphical User Interface) in the display input unit 130, or by directly transmitting the measured flatness when the shape measuring device 600 is on the same network, and is converted into a direct quality value (steps S503 and S504).

[0067] In the case of required quality related to the design of the molded product, such as flow marks, the external appearance of the molded product 500 is photographed by the camera 700 after molding, the image is acquired (step S508), and sent to the molding condition derivation device 100. After that, the indirect quality value processing unit 123 calculates the indirect quality value based on the image processing described above (steps S509 and S510). When the acquisition of the required quality values ​​is completed, the molding quality value acquisition process ends (step S511).

[0068] Next, the injection molding machine 200 is driven to perform molding under each molding condition in the injection molding initial molding condition table, and the process of obtaining the molding quality value, which is the target variable, is repeated the number of times equal to the number of initial data displayed in the initial molding condition table (step S104).

[0069] FIG. 15 is an explanatory diagram showing an example of a screen of the molding condition derivation program in which molding of the initial number of data has been completed and quality values ​​have been input. As shown in FIG. 15, when molding of the initial data number is completed (steps S105, S200), the molding condition optimization unit 122 starts a molding condition optimization step (step S300).

[0070] When this molding condition optimization process is started (steps S300 and S301), molding is repeated using the Bayesian optimization method described above to optimize the molding conditions. To achieve this, a Gaussian process regression model is created using the input parameters (molding condition values) and the objective variable (quality value) collected in the initial data collection process (step S302). That is, a prediction model (prediction function) is created using the input parameters (molding condition values) and the objective variable (quality value). A combination of molding conditions that has not yet been molded is input into the Gaussian process regression model created in this way, and the predicted average value and predicted standard deviation are obtained. After that, an evaluation value of the molding conditions input into the Gaussian process regression model is calculated using the acquisition function EI (Expected Improvement) (step S303).

[0071] The combination of molding conditions that have not yet been input here is all combinations of the arithmetic progression created for the set input parameters within the upper and lower limit ranges of the molding condition setting range information 141. The molding condition with the highest evaluation value is displayed on the GUI (Graphical User Interface) in the display input unit 130 as the molding condition to be tried next (step S304).

[0072] FIG. 16 is an explanatory diagram showing an example of a screen of the molding condition derivation program on which the molding conditions to be next tried are displayed. In FIG. 16, the molding conditions displayed on the last line are the molding conditions to be tried next.

[0073] The above process of displaying the next molding conditions can be executed by clicking the "Execute Bayesian optimization" button on the screen of the molding condition derivation program shown in Figs.

[0074] Thereafter, as in the initial data collection process, the injection molding machine 200 is driven to perform molding under the displayed molding conditions (step S305). Similarly, molding stability is confirmed (step S306) and molding quality values ​​are acquired (step S307). Then, if the acquired quality values ​​satisfy the required quality (step S308), the molding condition optimization process ends (steps S310, S400). On the other hand, if the required quality is not satisfied, an end determination is made (step S309).

[0075] In the first embodiment, the end judgment is set to perform 10 repeated moldings as the number of times, but the number of times may be set arbitrarily. If the end judgment is not the end, the process returns to the molding condition optimization process (step S301) and repeats the series of steps. On the other hand, if the end judgment is the end, the molding condition optimization process ends (steps S310, S400).

[0076] As described above, in this embodiment 1, a Bayesian optimization method using a regression model capable of finding the posterior distribution of the output for the input, particularly a Gaussian process regression model, is used to derive molding conditions that satisfy the required quality of the molded product, so that a prediction function for optimizing the molding conditions can be constructed even with a small amount of data. Therefore, appropriate molding conditions that satisfy the quality required of the molded product can be easily derived without depending on the technical level of the molding operator.

[0077] That is, the injection molding method of embodiment 1 has the following steps, and the computer-readable storage medium storing the computer program of embodiment 1 performs the following steps when the computer program is executed by a processor. That is, the step of executing includes a step of constructing a prediction model based on input parameters including molding conditions of a molded product and objective variable values ​​including quality values ​​that quantify the required quality of the molded product for the input parameters; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; This is a step of deriving molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that finds the input parameters that result in the evaluation of the objective variable value being the highest quality value compared to the initial quality value based on the predictive distribution.

[0078] The above-described injection molding method and computer-readable storage medium are executed by the molding condition optimization unit 122 in the control processing unit 120 of the molding condition deriving device 100 of FIG. 1, as shown in FIG. The molding condition optimization unit 122 shown in Figure 18 includes a predictive model construction unit 1200 that constructs a predictive model based on the input parameters including the molding conditions of the molded product and the objective variable values ​​including the quality values ​​that quantify the required quality of the molded product for the input parameters, a predictive distribution inference unit 1210 that uses the predictive model to infer a predictive distribution of the objective variable values ​​for the input parameters, and a molding condition derivation unit 1220 that derives molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method that utilizes a regression model to find the input parameters that will result in the evaluation of the objective variable value being the highest quality value compared to the initial quality value based on the predictive distribution.

[0079] Also, as shown in the flowchart of FIG. 19, the injection molding method of embodiment 1 includes the following steps, and the computer-readable storage medium storing a computer program executes the following steps when the computer program is executed by a processor. That is, as shown in FIG. 19, a prediction model is constructed based on input parameters including molding conditions of a molded product and objective variable values ​​including quality values ​​that quantify the required quality of the molded product for the input parameters (step S1200). Inferring a predictive distribution of the objective variable values ​​with respect to the input parameters using the predictive model (step S1210); Using the predictive distribution, a Bayesian optimization method is used to derive molding conditions that satisfy the required quality of the molded product (step S1220), utilizing a regression model that determines the input parameters that result in the evaluation of the objective variable value being the highest quality value compared to the initial quality value.

[0080] As described above, the control processing unit 120 is realized by, for example, a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a memory device 1010 (memory unit 140), as shown in FIG. 17, which is an example of a hardware configuration, and by the processor 1000 executing a program stored in the memory device 1010 (memory unit 140). Therefore, the processing of the predictive model construction unit 1200, predictive distribution inference unit 1210, and molding condition derivation unit 1220 of the molding condition optimization unit 122 shown in FIG. 18, as well as steps S1200, S1210, S1220 executed in the flowchart shown in FIG. 19, are realized by the processor 1000 executing a program stored in the memory device 1010 (memory unit 140).

[0081] Embodiment 2 [Utilizing the features of sensor values] In the second embodiment, an injection molding method for maintaining and restoring a non-defective product by utilizing the feature amount of the sensor value will be described. To find optimal molding conditions that satisfy the required quality of a molded product (called molding condition parameter optimization), it is necessary to consider this as an optimization problem of optimizing the relationship between input and output, similar to the first embodiment. In the second embodiment, for example, a sensor value in the mold when the product is good and satisfies the required quality (hereinafter referred to as the reference sensor value) is used as a reference, and the molding conditions are changed based on Bayesian optimization utilizing features extracted from the sensor value described below so as to match this reference sensor value, thereby making it possible to maintain the product in a good condition or return the product from a defective condition to a good condition. In the following description, a method for restoring a defective state to a non-defective state will be described as an example.

[0082] [Configuration for realizing the injection molding method] The configuration for carrying out the injection molding method of the second embodiment will be described with reference to Figs. 20 and 21, focusing on the differences from the first embodiment. 20 and 21 are diagrams showing an example of an apparatus configuration required to realize the injection molding method according to the second embodiment. 20, an injection molding machine 200 includes a mold 210 for molding a molded product 211, and various sensors 212 are attached to the mold 210. Data measured by the sensor 212 is input via a measurement amplifier 220 to a control processing unit 120A of the molding condition derivation device 100A, which will be described later. The configuration in FIG. 20 is similar to that in FIG. 1 of the first embodiment, and therefore a detailed description thereof will be omitted.

[0083] As shown in FIG. 21, the molding condition deriving device 100A includes a communication unit 110, a control processing unit 120A, a display input unit 130, and a storage unit 140. The control processing unit 120A includes a next molding condition output unit 121, a molding condition optimization unit 122A, an indirect quality value processing unit 123, a direct quality value processing unit 124, and a sensor value feature amount processing unit 125. The storage unit 140 includes molding condition setting range information 141 , molding condition item influence levels 142 , and molded product information 143 . The molding condition derivation device 100A of the second embodiment differs from the molding condition derivation device 100 of the first embodiment in that a control processing unit 120A includes a processing unit 125 for processing the feature amount of the sensor value. The sensor value feature amount processing unit 125 uses the sensor values ​​input to the control processing unit 120A via the measurement amplifier 220 to calculate the sensor value feature amounts, which will be described later, used for optimizing the molding conditions. With the above configuration, the injection molding method according to the second embodiment can be carried out.

[0084] [Features extracted from sensor values] In the second embodiment, the input is the value of the molding condition, and the output is the feature value extracted from the measurement value of the sensor 212 in the mold. In this example, there are three feature amounts extracted from the sensor values ​​in the mold that are output. The first and second feature amounts extracted from the sensor values ​​are the feature amount in the X direction and the feature amount in the Y direction of the sensor value (specifically, the maximum value of the sensor value or the time when the maximum value is reached, the sensor value at the time of filling completion or the time when the maximum value is reached, etc.) as shown in Fig. 22. In Fig. 22, as an example, a pressure sensor value is used as the sensor value, the feature value of the sensor value in the X direction is the time when the maximum injection pressure value is reached, and the feature value of the sensor value in the Y direction is the maximum injection pressure value. Note that any combination of the X direction and Y direction sensor values ​​as the first and second feature values ​​is acceptable. Also, as shown in Fig. 22, there may be two feature values ​​in the X direction (x1, x2) or two feature values ​​in the Y direction (y1, y2).

[0085] The third feature extracted from the sensor value is the similarity between the reference sensor value and the sensor value when the molding conditions are changed. In the second embodiment, a case where Euclidean distance is used as the similarity is shown as an example. Euclidean distance is the shortest distance between two points in any dimension. As a specific calculation method, when formula (7) is the reference sensor value and formula (8) is the sensor value when the molding conditions are changed, the Euclidean distance d between the sensor values ​​is calculated by formula (9). As an example of the calculated Euclidean distance, a case where the similarity is low is shown in FIG. 23 and a case where the similarity is high is shown in FIG. 24. In FIG. 23 and FIG. 24, the solid line shows the waveform of the reference sensor value, and the dotted line shows the waveform of the sensor value when the molding conditions are changed. As the similarity between sensor values, in addition to the Euclidean distance, any of the Manhattan distance, cosine similarity, and the time integral value of the sensor values ​​may be used.

[0086]

number

[0087]

number

[0088]

number

[0089] [Parameter search method using Bayesian optimization] In the second embodiment, an acquisition function is used to evaluate combinations of input parameter candidates in order to determine the input parameters (molding conditions) to be verified next by the Bayesian optimization method as in the first embodiment. This acquisition function may be, for example, PI (Probability of Improvement) or EI (Expected Improvement). Alternatively, UCB (Upper Confidence Bound) or MI (Mutual Information) may be used.

[0090] In the second embodiment, as an example, an acquisition function called EI (Expected Improvement) described in the first embodiment, which calculates an expected value of how much improvement will be made from the minimum (or maximum) value of the objective variable, is used in correspondence with the three objective variables that are the feature quantities extracted from the sensor values ​​described above, to match the sensor values ​​with the reference sensor values. As a specific method, a case will be described in which the acquisition function EI (Expected Improvement) is a minimization algorithm (the above-mentioned formulas (5) and (6)).

[0091] In order to match the sensor value with the reference sensor value, it is necessary to combine different optimization methods for the three sensor value features described above, which are the objective variables. The optimization target for the sensor value features in the X and Y directions is a range of ±3% of the reference sensor value features in the X and Y directions, and optimization is performed to keep them within that range. In order to use a minimization algorithm to keep the sensor value features in the X and Y directions, which are the objective variables, within the optimization target range, the value μ' calculated by applying the predicted average μ output from the Gaussian process regression model described above to formula (10) can be used to realize optimization that keeps the sensor value within the target range. In formula (10), RLupper indicates a value of +3% in the X and Y directions of the reference sensor value, and RLlower indicates a value of -3% in the X and Y directions of the reference sensor value. The upper and lower limits of the reference sensor value may be set arbitrarily within the range of ±10%.

[0092]

number

[0093] In addition, the similarity to the reference sensor value is optimized so as to maximize the similarity. In order to perform processing to maximize the similarity, which is the objective variable, using the above-mentioned minimization algorithm, the objective variable can be maximized by multiplying the predicted mean μ output from the above-mentioned Gaussian process regression model by -1 and using a value obtained by inverting the positive and negative signs.

[0094] In this embodiment 2, the above-mentioned process is performed on three objective variables, and then the acquisition function is unified by combining the evaluation values ​​(EI values) according to EI (Expected Improvement) described in embodiment 1. Although a method of taking a weighted linear sum is used as a method of unifying the evaluation values, a simple sum or product method may also be used.

[0095] In this example, a case where a pressure sensor value is used as the sensor value is described, but the same processing as described above may also be performed using a temperature sensor value, an AE (Acoustic Emission) sensor value, a mold strain value measured by a strain gauge, etc. as the sensor value. In addition, in this example, the sensor values ​​are described taking the feature amount in the X direction and the feature amount in the Y direction of two-dimensional coordinates as an example. However, the sensor values ​​may focus on the feature amount in the X direction, the feature amount in the Y direction, and the feature amount in the Z direction of three-dimensional coordinates, or the sensor values ​​may focus on the feature amount in the x1 direction, the feature amount in the x2 direction, ..., the xN direction of N-dimensional coordinates (N is an integer of 2 or more) as the sensor values, and the same processing as described above may be performed.

[0096] [How to optimize molding conditions] 25 to 28 are flow charts showing an example of a series of processing procedures for optimizing molding conditions in the injection molding method of embodiment 2. In the following, a specific example will be given of a case where a molded product is returned to a non-defective state when a defect occurs in the molded product due to disturbances such as temperature change or a change in the lot of the resin material. In the figures, the symbol S represents a step.

[0097] FIG. 25 is a flowchart showing an example of a series of process steps for making advance preparations for optimizing molding conditions in the injection molding method according to the second embodiment. First, before optimizing the molding conditions to return the product to a non-defective state, a step of acquiring a reference sensor value is started as advance preparation according to the flow chart of FIG. 25 (step S601).

[0098] Then, first, molding stability is confirmed (step S602). As a specific method for confirming molding stability in this second embodiment, the condition for confirming molding stability is that the change in the temperature sensor value is within ±1°C after three consecutive moldings. After the molding stability has been confirmed, in the molding quality value acquisition process (see FIG. 13) described in the first embodiment, it is confirmed that the molding quality value satisfies the required quality (steps S603, S604, and S501 to S511).

[0099] If the molding quality value continues not to satisfy the required quality, molding conditions that enable molding of a non-defective product are derived using the method of embodiment 1. After that, the value of sensor 212 (pressure sensor in this example) when the required quality is satisfied is stored in memory unit 140 from measurement amplifier 220 via display input unit 130 (step S605). In this case, the sensor value is stored a predetermined set number of times (step S606). In this example, for example, molding is repeated 30 times as the set number of times that the sensor value is stored, but the set number of times may be set arbitrarily.

[0100] After the acquisition of the sensor value is completed, the stored sensor value is used to create a reference sensor value (step S607). A specific method for creating the reference sensor value is to perform preprocessing such as removing or replacing missing values ​​and noise, and then calculate the average value, median value, and weighted average value of the sensor value, but in this example, the average value is used. At this time, the feature value in the X direction of the sensor value (in this example, the time when the maximum injection pressure value is reached) and the feature value in the Y direction (in this example, the maximum injection pressure value) shown in FIG. 22 described above are obtained from the created reference sensor value, and are stored together in the storage unit 140. After the reference sensor value is created and stored, the process of acquiring the reference sensor value is terminated (step S608). This advance preparation can be performed not only when the molding conditions of the mold are derived, but also during mass production of molded products.

[0101] Next, a reference sensor value for the target molded product is created, and then the molding conditions are optimized to return the product to a non-defective state. That is, the initial data collection process is started (step S100, step S101) according to the flowcharts of Fig. 26 to Fig. 28. For this, first, the molding condition derivation device 100A is started, and the molding condition setting range information 141, molding condition item influence degree 142, and molded product information 143 described in the first embodiment are set. After that, the molding condition derivation program stored in the storage unit 140 is started.

[0102] FIG. 30 shows, as an example, the start-up screen of a molding condition derivation program in the case of four input parameters (first stage resin temperature, third stage injection speed, fourth stage injection speed, first stage holding pressure) and three objective variables (objective variable 1: X-direction feature value of the sensor value, objective variable 2: Y-direction feature value of the sensor value, objective variable 3: similarity to the sensor value of the reference waveform).

[0103] The molding condition derivation program stored in the storage unit 140, when started, selects molding condition items that are input parameters that greatly affect the fluctuation of the sensor value from the molding condition item influence degree 142, and creates an initial molding condition table as shown in Fig. 30 so that it falls within the range of the molding condition setting range information 141. Then, the contents of the initial molding condition table are displayed on the display input unit 130. Note that Figure 30 shows a setting in which combinations of each molding condition are output randomly, but it is also possible to output each molding condition as a two-level orthogonal array with the control factor as each molding condition, or to set the scalar value of [matrix of molding conditions selected at random in any number] x [transposed matrix of selected molding conditions] as the optimization criterion and select the initial conditions that maximize this optimization criterion.

[0104] The molding operator performs molding operations according to the initial molding condition table of Fig. 30 displayed on the display input unit 130 of the molding condition derivation device 100A. At this time, the molding conditions may be manually input by the molding operator, or may be automatically input via the communication unit 110 if the injection molding machine 200 is connected to a network. The molding operations are performed sequentially from the first row of the initial molding condition table of Fig. 30, and molding stability is confirmed (step S102).

[0105] In injection molding, when molding conditions are set and then changed, the mold temperature gradually rises or falls as the number of moldings increases. After a certain number of moldings, the temperature change gradually decreases, and molding can be continued at the same mold temperature. This state is used as an indicator to determine that molding has stabilized. As a specific confirmation method, in this example, the condition for confirming molding stability was that the change in the temperature sensor value was within ±1° C. after three consecutive moldings, as in the example of embodiment 1.

[0106] Next, after the molding stability is confirmed, the sensor value feature amount is acquired for the molded product 211 according to the flow chart of FIG. 29 (step S1000, step S801). During molding, the pressure time series data acquired by the sensor 212 and the measurement amplifier 220 is transmitted to the molding condition derivation device 100A (step S802). The transmitted time series data is sent to the sensor value feature amount processing unit 125. The sensor value feature amount processing unit 125 calculates the similarity of the current sensor value to the reference sensor value (step S803), the sensor value feature amount in the X direction (step S804), and the sensor value feature amount in the Y direction (step S805), which are the aforementioned three sensor value feature amounts. When the acquisition of the necessary values ​​is completed, the sensor value feature amount acquisition process is terminated (step S806).

[0107] Next, the injection molding machine 200 is driven to perform molding under each molding condition in the initial molding condition table, and the process of acquiring the feature quantities of the sensor values ​​that become the target variables is repeated the number of times equal to the number of initial data displayed in the initial molding condition table (step S104). FIG. 31 is an explanatory diagram showing an example of a screen of the molding condition derivation program in which molding of the initial number of data has been completed and the feature amounts of the sensor values ​​have been input. As shown in FIG. 31, when molding of the initial data number is completed and the initial data collection step is ended (steps S105, S200), the molding condition optimization section 122A starts the molding condition optimization step (step S300).

[0108] When the molding condition optimization process is started (steps S300 and S301), molding is repeated using the Bayesian optimization method described above to optimize the molding conditions. To achieve this, a Gaussian process regression model is created using the input parameters (molding condition values) and the objective variables (feature values ​​of the sensor values) collected in the initial data collection process (step S302). That is, a prediction model (prediction function) is created using the input parameters (molding condition values) and the objective variables (feature values ​​of the sensor values). A combination of molding conditions that has not yet been molded is input into the Gaussian process regression model created in this way, and the predicted average value and predicted standard deviation are obtained. After that, numerical processing is performed to apply the minimization algorithm described above to the predicted average value, and an evaluation value of the molding conditions input into the Gaussian process regression model is calculated using the acquisition function EI (Expected Improvement) (step S303).

[0109] The combination of molding conditions that have not yet been input here is all combinations of the arithmetic progression created for the set input parameters within the upper and lower limit ranges of the molding condition setting range information 141. The molding condition with the highest evaluation value is displayed on the GUI (Graphical User Interface) in the display input unit 130 as the molding condition to be tried next (step S304).

[0110] FIG. 32 is an explanatory diagram showing an example of a screen of the molding condition derivation program on which the molding conditions to be next tried are displayed. In FIG. 32, the molding conditions displayed on the last line will be the molding conditions to be tried next. The above-mentioned process of displaying the next molding conditions can be executed by clicking the "Execute Bayesian optimization" button on the screen of the molding condition derivation program shown in Figs.

[0111] Thereafter, similar to the initial data collection process, the injection molding machine 200 is driven to perform molding under the displayed molding conditions (step S305). Furthermore, similar to the initial data collection process, molding stability is confirmed (step S306) and the feature quantities of the sensor values ​​are acquired (steps S2000, S801). Then, if the feature quantities of the acquired sensor values ​​satisfy the requirements (step S308), the molding condition optimization process ends (steps S310, S400). In this example, if the Euclidean distance adopted for the similarity to the reference sensor value is 0.7 or more, it is determined that the requirements are satisfied. On the other hand, if the requirements are not satisfied, an end determination (step S309) is performed.

[0112] In this example, the end judgment (step S309) is set to perform 10 repeated moldings as the number of times specified, but the number of times may be set arbitrarily. If the end judgment does not indicate the end, the process returns to the molding condition optimization process (step S301) again, and the series of steps is repeated. On the other hand, if the end judgment indicates the end, the molding condition optimization process ends (steps S310, S400).

[0113] As described above, in this second embodiment, molding conditions that satisfy the required quality of a molded product are derived using a regression model that can obtain the posterior distribution of the output for the input, particularly a Bayesian optimization method that utilizes a Gaussian process regression model, so that a prediction function for optimizing the molding conditions can be constructed even with a small amount of data. Therefore, when a molded product becomes defective due to disturbances such as temperature change or a change in the lot of resin material, appropriate molding conditions for returning the product to a non-defective state can be easily derived without depending on the skill level of the molding operator.

[0114] That is, the injection molding method of embodiment 2 has the following steps, and the computer-readable storage medium storing the computer program of embodiment 2 performs the following steps when the computer program is executed by a processor. That is, the step of executing includes a step of constructing a prediction model based on input parameters including molding conditions of the molded product, feature quantities of sensor values ​​of a sensor arranged in an injection molding machine corresponding to the input parameters, and a target variable value including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value, which is the sensor value when the molded product satisfies a required quality; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; A step of deriving molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that finds the input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of the initial sensor value based on the predictive distribution.

[0115] The above-described injection molding method and computer-readable storage medium are executed by a molding condition optimization unit 122A in a control processing unit 120A of the molding condition deriving device 100A in FIG. 21, as shown in FIG. The molding condition optimization unit 122A shown in FIG. 33 includes a predictive model construction unit 1200A that constructs a predictive model based on the input parameters including the molding conditions of the molded product, the feature values ​​of the sensor values ​​of the sensor arranged in the injection molding machine for the input parameters, and the objective variable values ​​including the similarity of the sensor value when the molding conditions of the molded product are changed to the reference sensor value, which is the sensor value when the molded product satisfies the required quality, a predictive distribution inference unit 1210A that infers a predictive distribution of the objective variable value for the input parameters using the predictive model, and a molding condition derivation unit 1220A that derives molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that obtains the input parameters such that the evaluation of the objective variable value is closer to the feature values ​​of the reference sensor value than the feature values ​​of the initial sensor value based on the predictive distribution.

[0116] Also, as shown in the flowchart of FIG. 34, the injection molding method of embodiment 1 includes the following steps, and a computer-readable storage medium having a computer program stored therein executes the following steps when the computer program is executed by a processor. That is, as shown in FIG. 34, a predictive model is constructed based on input parameters including molding conditions of the molded product, feature values ​​of the sensor values ​​of a sensor arranged in an injection molding machine for the input parameters, and an objective variable value including a similarity of the sensor value when the molding conditions of the molded product are changed to a reference sensor value which is the sensor value when the molded product satisfies the required quality (step S1200A). The predictive model is used to infer a predictive distribution of the objective variable value for the input parameters (step S1210A). Based on the predictive distribution, a Bayesian optimization method is used to find the input parameters for which the evaluation of the objective variable value is closer to the feature values ​​of the reference sensor value than the feature values ​​of the initial sensor value, thereby deriving molding conditions that satisfy the required quality of the molded product (step S1220A).

[0117] In the second embodiment, as in the first embodiment, the control processing unit 120A is realized by, for example, a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a memory device 1010 (memory unit 140), as shown in FIG. 17, which is an example of a hardware configuration, and by the processor 1000 executing a program stored in the memory device 1010 (memory unit 140). Therefore, the processing of the predictive model construction unit 1200A, predictive distribution inference unit 1210A, and molding condition derivation unit 1220A of the molding condition optimization unit 122A shown in FIG. 33, as well as steps S1200A, S1210A, S1220A executed in the flowchart shown in FIG. 34, are realized by the processor 1000 executing a program stored in the memory device 1010 (memory unit 140).

[0118] Although exemplary embodiments are described in this application, the various features, aspects, and functions described in the embodiments are not limited to application to a particular embodiment, but may be applied to the embodiments alone or in various combinations.

[0119] Therefore, countless modifications not illustrated are conceivable within the scope of the technology disclosed in this application, including, for example, modifying, adding, or omitting at least one component. [Explanation of symbols]

[0120] 100,100A molding condition derivation device, 110 communication unit, 120,120A control processing unit, 121 molding condition next condition output unit, 122,122A molding condition optimization unit, 123 indirect quality value processing unit, 124 direct quality value processing unit, 125 sensor value feature amount processing unit, 130 display input unit, 140 memory unit, 141 molding condition setting range information, 142 molding condition item influence degree, 143 molded product information, 200 injection molding machine, 210 mold, 211 molded product (during molding), 212 sensor, 220 measurement amplifier, 300 take-out robot, 400 transport conveyor, 500 molded product (after molding), 600 shape measuring device, 700 camera, 1200,1200A prediction model construction unit, 1210,1210A Predictive distribution inference unit, 1220, 1220A molding condition derivation unit.

Claims

1. A step of constructing a prediction model based on input parameters including molding conditions of a molded product and an objective variable value including a quality value that quantifies a required quality of the molded product for the input parameters, the objective variable value including an indirect quality value that includes a feature amount converted from an appearance image of the molded product; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that are most highly evaluated to produce a desired quality value for the molded product by a Bayesian optimization method using a regression model that determines the input parameters that produce the highest quality value for the objective variable value compared to the initial quality value based on the predictive distribution. An injection molding method that derives molding conditions that satisfy the desired required quality by using an optimization method that repeatedly evaluates the objective variable values ​​under the derived molding conditions and derives the molding conditions that are most highly evaluated to achieve the desired quality value.

2. The injection molding method of claim 1 , wherein the regression model is a Gaussian process regression model.

3. A molding condition deriving device for optimizing molding conditions for a molded product based on the injection molding method according to claim 1 or 2, A storage unit in which information on molding conditions and required quality of the molded product is stored in advance; A control processing unit is provided, The control processing unit: an indirect quality value processing unit for obtaining an indirect quality value including a feature amount converted from the appearance image of the molded product; a molding condition derivation device having a molding condition optimization unit that imports the indirect quality value from the indirect quality value processing unit as a quality value, and uses the imported quality value and information on the molding conditions and the required quality stored in the memory unit to derive molding conditions that satisfy the optimal required quality of the molded product by a Bayesian optimization method utilizing a regression model.

4. The molding condition optimization section includes: a prediction model construction unit that constructs a prediction model based on the input parameters including molding conditions of the molded product and objective variable values ​​including the quality value that quantifies the required quality of the molded product for the input parameters; a predictive distribution inference unit that infers a predictive distribution of the objective variable value with respect to the input parameters using the predictive model; A molding condition derivation unit derives molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method using a regression model that obtains the input parameters that result in the evaluation of the objective variable value being the highest quality value compared to the initial quality value based on the predictive distribution. The molding condition deriving device according to claim 3.

5. A computer-readable storage medium having a computer program stored thereon, the computer program comprising, when executed by a processor, the steps of: A step of constructing a prediction model based on input parameters including molding conditions of a molded product and an objective variable value including a quality value that quantifies a required quality of the molded product for the input parameters, the objective variable value including an indirect quality value that includes a feature amount converted from an appearance image of the molded product; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that are most highly evaluated to produce a desired quality value for the molded product by a Bayesian optimization method using a regression model that determines the input parameters that produce the highest quality value for the objective variable value compared to the initial quality value based on the predictive distribution. A computer-readable storage medium that derives molding conditions that satisfy the desired required quality by using an optimization method that repeatedly evaluates the target variable values ​​under the derived molding conditions and derives the molding conditions that are most highly evaluated to achieve the desired quality value.

6. The computer-readable storage medium of claim 5 , wherein the regression model is a Gaussian process regression model.

7. A step of constructing a prediction model based on input parameters including molding conditions of a molded product, feature quantities of sensor values ​​of a sensor arranged in an injection molding machine corresponding to the input parameters, and objective variable values ​​including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value which is the sensor value when the molded product satisfies a required quality; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that satisfy a required quality of the molded product by a Bayesian optimization method utilizing a regression model that finds the input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of an initial sensor value based on the predictive distribution.

8. 8. The injection molding method according to claim 7, wherein the molding conditions are derived using a feature amount in an x1 direction, a feature amount in an x2 direction, ..., a feature amount in an xN direction of an N-dimensional coordinate (N is an integer of 2 or more) of the sensor value obtained from the sensor value, and a similarity of the sensor value to the reference sensor value.

9. The injection molding method according to claim 7 or claim 8, wherein the regression model is a Gaussian process regression model.

10. A molding condition deriving device for optimizing molding conditions for a molded product based on the injection molding method according to claim 7 or 8, A storage unit in which information on molding conditions and required quality of the molded product is stored in advance; A control processing unit is provided, The control processing unit: a sensor value feature processing unit that calculates a feature of the sensor value obtained from the sensor value and a similarity of the sensor value to the reference sensor value; a molding condition derivation device having a molding condition optimization unit that imports the feature values ​​of the sensor value and a similarity to the reference sensor value from a feature value processing unit of the sensor value, and derives molding conditions that satisfy the optimal required quality of the molded product by a Bayesian optimization method using a regression model using the imported feature values ​​of the sensor value and the similarity to the reference sensor value, and information on the molding conditions and the required quality stored in the memory unit.

11. The molding condition optimization section includes: a prediction model construction unit that constructs a prediction model based on the input parameters including the molding conditions of the molded product, feature quantities of sensor values ​​of a sensor disposed in an injection molding machine corresponding to the input parameters, and objective variable values ​​including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value that is the sensor value when the molded product satisfies a required quality; and a predictive distribution inference unit that infers a predictive distribution of the objective variable value with respect to the input parameters using the predictive model; and a molding condition derivation unit that derives molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that determines the input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of the initial sensor value based on the predictive distribution.

12. A computer-readable storage medium having a computer program stored thereon, the computer program comprising, when executed by a processor, the steps of: A step of constructing a prediction model based on input parameters including molding conditions of a molded product, feature quantities of sensor values ​​of a sensor arranged in an injection molding machine corresponding to the input parameters, and objective variable values ​​including a similarity of a sensor value when the molding conditions of the molded product are changed to a reference sensor value which is the sensor value when the molded product satisfies a required quality; inferring a predictive distribution of values ​​of the objective variable with respect to the input parameters using the predictive model; and deriving molding conditions that satisfy the required quality of the molded product by a Bayesian optimization method utilizing a regression model that determines the input parameters such that an evaluation of the objective variable value is closer to the feature value of the reference sensor value than to the feature value of the initial sensor value based on the predictive distribution.

13. The computer-readable storage medium according to claim 12, wherein the molding conditions are derived using a feature quantity in the x1 direction, a feature quantity in the x2 direction, ..., a feature quantity in the xN direction of the N-dimensional coordinates (N is an integer of 2 or more) of the sensor value obtained from the sensor value, as well as a similarity of the sensor value to the reference sensor value.

14. The computer-readable storage medium of claim 12 or 13, wherein the regression model is a Gaussian process regression model.

Images