Injection molding method, molding condition deriving device, and computer-readable storage medium
Through Bayesian optimization method and regression model, injection molding conditions are automatically derived, which solves the problems of high technical standards and difficult data preparation in the existing technology, and achieves efficient and accurate molding conditions optimization.
Patent Information
- Application Number
- CN202380017303.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-01-25
- Filing Date
- 2023-01-18
- Publication Date
- 2024-12-06
AI Technical Summary
In the existing injection molding methods, the molding conditions that meet the required quality of the molded product need to rely on skilled workers with high technical standards, and require a large amount of learning data and expensive high-resolution measuring machines, resulting in large time consumption and difficulty in preparing data.
Using Bayesian optimization method of regression model, by constructing a predictive model to quantify the required quality values of molded products, using sensor data and appearance image feature quantities, automatically derive the molding conditions that meet the required quality, and reduce the technical dependence on the operator and Data requirements.
Even under the conditions of less data, molding conditions that meet the required quality of molded products can be effectively derived, which improves the efficiency and accuracy of the molding process, reduces the dependence on high-tech operators and the demand for measuring equipment.
Smart Images

Figure CN119095712A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to an injection molding method, a molding condition derivation device, and a computer-readable storage medium. Background Art
[0002] Injection molding is a widely used method for molding resin parts by injecting molten resin material into a mold. In injection molding, in order to mold a good quality resin part (hereinafter referred to as a molded product) that meets the required quality, it is essential to derive appropriate molding conditions. However, appropriate molding conditions vary depending on the shape of the molded product, the properties of the resin used, etc., so the operation of deriving the molding conditions is performed by a skilled operator with knowledge and experience.
[0003] In addition, as a prior art, a method for optimizing the quality of molded products using a neural network has been proposed as a method for assisting an injection molding machine in molding. In order to construct the neural network, the input parameters are molding conditions, and the output items (hereinafter referred to as target variables) are the quality values obtained by measuring good molded products (for example, refer to the following patent document 1).
[0004] Prior art literature
[0005] Patent Literature
[0006] Patent Document 1: Japanese Patent Application Publication No. 2008-110486 Summary of the invention
[0007] Problems to be solved by the invention
[0008] The work of deriving appropriate molding conditions that meet the required quality of molded products is generally performed by skilled operators with knowledge and rich experience. On the other hand, when operators who lack knowledge and experience derive molding conditions, they repeat a lot of trial and error, and there is a problem that deriving molding conditions takes a lot of time.
[0009] Furthermore, when a neural network in the related art is used, there is a problem that a large amount of learning data of about several hundred to several tens of thousands is required in order to construct a prediction function for optimizing molding conditions.
[0010] Furthermore, in the prior art, in order to optimize the molding conditions, the measured quality values (product weight, warpage, size, etc.) are used. However, in the case of quality values (sink marks, flow marks, etc.) that need to be measured by a high-resolution measuring machine, since the measuring machine is expensive, there are problems such as being difficult to prepare, or having to cut out the measurement sample, or being unable to measure easily.
[0011] The present application discloses a technology for solving the above-mentioned problems, and its purpose is to provide an injection molding method, a molding condition derivation device and a computer-readable storage medium, which can easily obtain appropriate molding conditions that meet the quality required for the molded product without relying on the technical level of the molding operator.
[0012] Means for solving problems
[0013] The injection molding method disclosed in the present application has the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including quality values obtained by quantifying the required quality of the molded product; using the prediction model to infer the predicted distribution of the target variable values for the input parameters; and based on the predicted distribution, deriving molding conditions that meet the required quality of the molded product by using a Bayesian optimization method that utilizes a regression model, the regression model obtains the input parameter for which the evaluation of the target variable value reaches the highest quality value compared to the initial quality value.
[0014] In addition, the molding condition derivation device disclosed in the present application optimizes the molding conditions for a molded product based on the above-mentioned injection molding method, wherein the molding condition derivation device has: a storage unit, which pre-stores information on the molding conditions and required quality of the molded product; and a control processing unit, which has: a direct quality value processing unit, which directly measures the molded product to obtain a direct quality value; an indirect quality value processing unit, which obtains an indirect quality value including a feature quantity converted based on data of a sensor set in a mold of an injection molding machine or an appearance image of the molded product; and a molding condition optimization unit, which takes in at least one of the direct quality value from the direct quality value processing unit and the indirect quality value from the indirect quality value processing unit as a quality value, and uses the taken-in quality value and the molding conditions and the information on the required quality stored in the storage unit to derive molding conditions that meet the optimal required quality of the molded product through a Bayesian optimization method that makes use of a regression model.
[0015] In addition, the computer-readable storage medium disclosed in the present application stores a computer program, and when the computer program is executed by a processor, the following steps are performed: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including quality values obtained by quantifying the required quality of the molded product; using the prediction model to infer the predicted distribution of the target variable values for the input parameters; and based on the predicted distribution, deriving molding conditions that meet the required quality of the molded product by using a Bayesian optimization method that utilizes a regression model, the regression model obtains the input parameter for which the evaluation of the target variable value is the highest quality value compared to the initial quality value.
[0016] In addition, the injection molding method disclosed in the present application has the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including feature quantities of sensor values of a sensor configured in an injection molding machine and similarity of sensor values when the molding conditions of the molded product are changed relative to a reference sensor value, the reference sensor value being the sensor value when the molded product meets the required quality; using the prediction model to infer a predicted distribution of the target variable values for the input parameters; and deriving molding conditions that meet the required quality of the molded product based on the predicted distribution by a Bayesian optimization method that utilizes a regression model, the regression model obtaining the input parameter whose evaluation value of the target variable value is closer to the feature quantity of the reference sensor value than the feature quantity of the initial sensor value.
[0017] In addition, the molding condition derivation device disclosed in the present application optimizes the molding conditions for a molded product based on the above-mentioned injection molding method, wherein the molding condition derivation device has: a storage unit, which pre-stores information on the molding conditions and required quality of the molded product; and a control processing unit, which has: a processing unit for characteristic quantities of sensor values, which calculates the characteristic quantities of the sensor values obtained based on the sensor values and the similarity of the sensor values to the reference sensor values; and a molding condition optimization unit, which takes in the characteristic quantities of the sensor values and the similarity to the reference sensor values from the processing unit for characteristic quantities of the sensor values, and uses the characteristic quantities of the sensor values and the similarity to the reference sensor values taken in, as well as the molding conditions and the required quality information stored in the storage unit, to derive molding conditions that meet the optimal required quality of the molded product by a Bayesian optimization method that makes use of a regression model.
[0018] In addition, the computer-readable storage medium disclosed in the present application stores a computer program, and when the computer program is executed by a processor, the following steps are performed: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including feature quantities of sensor values of a sensor configured in an injection molding machine and similarity of sensor values when the molding conditions of the molded product are changed relative to a reference sensor value, the reference sensor value being the sensor value when the molded product meets the required quality; using the prediction model to infer a predicted distribution of the target variable values for the input parameters; and based on the predicted distribution, deriving molding conditions that meet the required quality of the molded product by a Bayesian optimization method that utilizes a regression model, the regression model obtains the input parameter whose evaluation ratio of the target variable value is closer to the feature quantity of the reference sensor value than the feature quantity of the initial sensor value.
[0019] Effects of the Invention
[0020] According to the injection molding method, molding condition derivation device, and computer-readable storage medium disclosed in the present application, a prediction function for optimizing molding conditions can be constructed even with a relatively small amount of data. Therefore, molding conditions that meet the required quality can be easily derived without relying on the technical level of the molding operator. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] [ Figure 1 ] is a diagram showing an example of an apparatus structure required to implement the injection molding method of embodiment 1.
[0022] [ Figure 2 ] is a diagram showing an example of an apparatus structure required to implement the injection molding method of embodiment 1.
[0023] [ Figure 3 ] is a schematic diagram showing an example of setting range information of molding conditions.
[0024] [ Figure 4 ] is a schematic diagram showing an example of the influence of items of molding conditions.
[0025] [ Figure 5 ] is a schematic diagram showing an example of molded product information.
[0026] [ Figure 6 ] is a characteristic diagram showing an example of the correlation between the defined sink mark feature quantity and the measured sink mark amount.
[0027] [ Figure 7 ] Figure 7 A. Figure 7 B. Figure 7C is an explanatory diagram showing an example in which image processing is performed on an image of a molded product.
[0028] [ Figure 8 ] is a diagram schematically showing a method of determining the next search condition (molding condition) using the EI value.
[0029] [ Fig. 9 ] is a diagram schematically showing a method of determining the next search condition (molding condition) using the EI value.
[0030] [ Fig.10 ] 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.
[0031] [ Fig.11 ] 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.
[0032] [ Fig.12 ] 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.
[0033] [ Fig.13 ] 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.
[0034] [ Fig.14 ] is an explanatory diagram showing an example of a screen of a molding condition derivation program operating in a molding condition derivation device.
[0035] [ Fig.15 ] is an explanatory diagram showing an example of a screen of a molding condition export program in which molding of an initial data number is completed and a quality value is input.
[0036] [ Fig.16 ] is an explanatory diagram showing an example of a screen in which the initial optimization molding condition derivation program is executed.
[0037] [ Fig.17 ] is a diagram showing an example of the hardware structure of the control processing unit of the present application.
[0038] [ Fig.18 ] is a block diagram showing the details of the control processing unit of the molding condition derivation device of embodiment 1.
[0039] [ Fig.19 ] is a flowchart showing the steps performed by the control processing unit of the molding condition derivation device of embodiment 1.
[0040] [ Fig. 20] is a diagram showing an example of a device structure required to implement the injection molding method of embodiment 2.
[0041] [ Fig.21 ] is a diagram showing an example of a device structure required to implement the injection molding method of embodiment 2.
[0042] [ Fig. 22 ] is a diagram showing an example of feature quantities in the X direction and the Y direction of the sensor value obtained in the control processing unit of the present application.
[0043] [ Fig.23 ] is a diagram showing an example of a calculation result of a low similarity with respect to a reference sensor value in the control processing unit of the present application.
[0044] [ Fig.24 ] is a diagram showing an example of a calculation result with a high similarity to a reference sensor value in the control processing unit of the present application.
[0045] [ Fig.25 ] is a flowchart showing an example of a series of processing steps for preparatory optimization of molding conditions in the injection molding method of embodiment 2.
[0046] [ Fig.26 ] is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of embodiment 2.
[0047] [ Fig. 27 ] is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of embodiment 2.
[0048] [ Fig.28 ] is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of embodiment 2.
[0049] [ Fig.29 ] is a flowchart showing an example of a series of processing steps for obtaining characteristic quantities of sensor values in the injection molding method of embodiment 2.
[0050] [ Fig.30 ] is an explanatory diagram showing an example of a screen of a molding condition export program in Implementation Example 2.
[0051] [ Fig.31 ] is an explanatory diagram showing an example of a screen of a molding condition derivation program in which molding of the initial data number in Implementation Example 2 is completed and feature quantities of sensor values are input.
[0052] [ Fig.32] is an explanatory diagram showing an example of a screen in which the initial optimized molding condition derivation program in Implementation Example 2 is executed.
[0053] [ Fig.33 ] is a block diagram showing the details of the control processing unit of the molding condition derivation device of embodiment 2.
[0054] [ Fig.34 ] is a flowchart showing the steps performed by the control processing unit of the molding condition derivation device of embodiment 2. DETAILED DESCRIPTION
[0055] The present application relates to an injection molding method and a molding condition derivation device using a regression model, wherein the regression model can obtain a posterior distribution of an output with respect to an input.
[0056] Implementation Method 1
[0057] [Structure for realizing injection molding method]
[0058] First, refer to Figure 1 and Figure 2 The structure for carrying out the injection molding method of the present application will be described.
[0059] Figure 1 and Figure 2 This is a diagram showing an example of a device configuration required to realize the injection molding method of the first embodiment.
[0060] like Figure 1 As shown, the injection molding machine 200 has 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 taken into the control processing unit 120 of the molding condition derivation device 100 described later via a measuring amplifier 220.
[0061] On the other hand, the molded product 211 molded in the mold 210 is taken out by the take-out robot 300. After being placed on the transport conveyor 400, the taken-out molded product 500 is measured by the shape measuring device 600, and a photo of its appearance is taken by the camera 700. The measurement results of the shape measuring device 600 and the appearance photo taken by the camera 700 are taken into the control processing unit 120 of the molding condition derivation device 100 described later.
[0062] in addition, Figure 1 The structure is described in detail later.
[0063] like Figure 2 As shown, the molding condition derivation 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 .
[0064] In this case, the molding condition derivation device 100 may be a single device, or may be composed of a plurality of devices or systems 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 also be implemented by a system or a plurality of computer devices using distributed computing or cloud computing.
[0065] The communication unit 110 includes, for example, a communication interface such as a NIC (Network Interface Card) and a DMA (Direct Memory Access) controller. The communication unit 110 can communicate with the injection molding machine 200 via a network such as a WAN or a LAN.
[0066] The control processing unit 120 includes a next condition output unit 121 for molding conditions, an optimization unit 122 for molding conditions, an indirect quality value processing unit 123, and a direct quality value processing unit 124. Fig.17 As an example of a hardware configuration, the control processing unit 120 is composed of a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a storage device 1010 (storage unit 140 described later), and is implemented by the processor 1000 executing a program stored in the storage device 1010 (storage unit 140 described later). In addition, the structural elements of the control processing unit 120 can be implemented by hardware such as an FPGA (Field Programmable Gate Array), or can be composed of both software and hardware.
[0067] The display input unit 130 includes a display device such as a liquid crystal display, and can be used by a molding operator who operates the molding condition derivation device 100 to understand the progress of molding condition optimization and perform settings and operations through a GUI (Graphical User Interface).
[0068] The storage unit 140 includes, for example, a HDD (Hard Disc Drive), an SSD (Solid State Drive), a ROM (Read Only Memory), and a RAM (Random Access Memory). In addition to storing various programs for exporting molding conditions such as firmware and application programs, the storage unit 140 also stores molding condition setting range information 141, molding condition item influence 142, and molded product information 143, which will be described later.
[0069] like Figure 1 As shown, the injection molding machine 200 includes a mold 210 as a tool for molding a molded product 211 to which molding conditions are optimized, and various sensors 212 are attached to the mold 210 .
[0070] The sensor 212 includes a strain gauge or piezoelectric pressure sensor for measuring resin pressure, a strain gauge for measuring the strain of the mold 210, a thermocouple or infrared temperature sensor for measuring the temperature of the mold 210 and the resin temperature, and an AE (Acoustic Emission) sensor for detecting the emission of sound in the mold 210, etc. In addition, any type of sensor can be installed in the mold 210. Various sensor data measured by the sensor 212 are taken into the control processing unit 120 via the measuring amplifier 220, and are converted into a quality value for evaluating the quality of the molded product in the indirect quality value processing unit 123.
[0071] The molded product 211 formed in the mold 210 is taken out by the take-out robot 300. After the taken-out molded product 500 is placed on the transport conveyor 400, shape data such as flatness and size are measured by the 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 machine.
[0072] In addition, the taken-out molded product 500 is photographed with the camera 700. In order to use the camera 700 to photograph the appearance of the molded product 500, lighting, a dark curtain, and a fixture may be added as needed. In addition, the camera 700 may be one or more. The photographed appearance of the molded product 500 is taken into the control processing unit 120, and is converted into a quality value for evaluating the quality of the molded product 211 in the indirect quality value processing unit 123.
[0073] Figure 3 1 is a schematic diagram showing an example of setting width information 141 of molding conditions pre-stored in the storage unit 140 .
[0074] The setting range information 141 of the molding condition is information on the upper and lower limits that can be set for the setting items (input parameters) of the molding condition according to various molded products 211. The upper and lower limits can be set based on experience and knowledge, or can also be set based on the results of resin flow analysis. For example, if it is an injection temperature, it can be set within the temperature range of each resin material recommended by the resin material manufacturer, or it can be set after confirming in advance through resin flow analysis whether the mold 210 can be filled with resin. Alternatively, temporary molding can be actually performed and the setting can be made after confirming whether there are no problems in molding.
[0075] Here, yes Figure 3 The following describes each item.
[0076] Mold temperature (movable) is a parameter used to control the temperature of the mold. "Moveable" refers to the mold that moves (opens) when the molded product is removed from the mold. The mold is provided with a pipe for the flow of fluids such as warm water, cold water or oil. The temperature of the fluid flowing in the pipe is controlled by a thermostat, thereby controlling the mold temperature.
[0077] Mold temperature (fixed) is the same as above, and is a parameter used to control the temperature of the mold. "Fixed" means the mold on the side that does not move when the molded product is taken out of the mold. In addition, it is generally set to give a temperature difference between the movable side and the fixed side of the mold (for example: 50°C on the movable side and 30°C on the fixed side).
[0078] Injection temperature 1 to 5 are parameters for controlling the temperature at which the resin pellets (resin pellets) injected into the mold are melted. In the heating cylinder of the injection molding machine, thermocouples are set at 4 to 6 positions in sequence from the end of the injection unit, and the numbers "1 to 5" indicate the locations of the thermocouples. In addition, the injection molding machine controls the heater of the heating cylinder so that the thermocouples become the set temperature value. Regarding the set value, the range recommended by the resin material manufacturer is used as a standard, and the setting is performed while confirming the quality of the molded product or the production cycle time.
[0079] Injection positions 1 to 4 are parameters for controlling the screw position of the injection molding machine when switching the speed of injecting molten resin into the mold. The flow of the resin to be injected is controlled by the combination of injection speed and injection position. The numbers "1 to 4" represent the number of positions where the injection speed is switched. (For example: until the injection position (screw position) is 100mm, the injection speed is 50mm / s. The injection position is 100mm to 40mm, and the injection speed is 30mm / s.)
[0080] Speed-pressure switching is the setting of the screw position that switches the control of the screw of the injection molding machine from injection speed control to holding pressure (holding pressure) control when injecting molten resin into the mold.
[0081] Injection speed 1 to 4 are parameters for controlling the speed of injecting molten resin 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 extent to which the injection speed is changed according to the injection position.
[0082] Holding pressure 1 to 3 are parameters for setting the size of the holding pressure control after the injection speed control when injecting molten resin into the mold. Generally speaking, in injection molding, after the resin is filled into the mold by injection speed control, shrinkage occurs with the change of the state of the resin (liquid → solid). Therefore, in order to compensate for the shrinkage of the resin, the mold is additionally filled with molten resin by holding pressure control. The numbers "1 to 3" indicate how much the size changes in multiple stages when the holding pressure is applied in multiple stages. In addition, the control of each holding pressure becomes time control (for example: holding pressure 1 is 50Mpa and 3s, and holding pressure 2 is 30Mpa and 4s).
[0083] The cooling time is the time it takes for the molten resin to cool and solidify after the resin is kept under pressure in the mold. Generally speaking, during the cooling time, no external pressure is applied to the resin, and only heat exchange between the resin and the mold occurs until the molten resin solidifies. If the cooling time is too short, warping or poor demolding from the mold may occur. If the cooling time is too long, the production cycle time may increase (increase the cost of the molded product) or poor demolding from the mold may occur.
[0084] Figure 4 1 is a schematic diagram showing the item influence degree 142 of the molding conditions pre-stored in the storage unit 140 .
[0085] The item influence degree 142 of the molding condition sets the influence degree of each item of the molding condition (such as temperature, pressure, injection speed, etc.) on the quality value (such as warpage, shrinkage mark) of the target molded product 211. Regarding the setting of the influence degree in this case, it can be set according to experience and knowledge, or it can be set according to the results of resin flow analysis. Alternatively, it can also be set according to the results of actual temporary molding. For example, an orthogonal table is made in which the items of the molding conditions are set as control factors, and resin flow analysis is performed under various conditions to calculate the characteristic value for the quality value of the molded product. There is also a method for setting the item influence degree of the molding condition according to the characteristic value. In addition, in the case of temporary molding, an orthogonal table can also be used to set the influence degree.
[0086] Figure 5 1 is a schematic diagram showing molded product information 143 pre-stored in the storage unit 140 .
[0087] The molded product information 143 sets the ID number for individual identification, the resin molding material to be used, and the information on the required quality of the molded product 211 to be optimized, etc., for each molded product 211. The quality information of the molded product 211 can be set to any number from the required quality of the molded product 211. For example, Figure 5 The molded product A is an example of setting the warpage, sink mark, and dimension generated in the molded product A as quality information. In this case, for example, the warpage (information 1) is set using the flatness of an arbitrary measurement location, the sink mark (information 2) is set using the amount of depression of an arbitrary measurement location, and the dimension (information 3) is set using the dimension value and the dimension tolerance.
[0088] In terms of realizing the configuration of the present application, the sensor 212 , the measuring amplifier 220 , the shape measuring device 600 , and the camera 700 are devices for quantifying the quality of the molded product 211 , and any of them may be used.
[0089] By realizing the above structure, the injection molding method of the present application can be implemented.
[0090] [Quantification of the quality of molded products]
[0091] In order to find the best molding conditions that meet the required quality of the molded product 211 (hereinafter referred to as molding condition parameter optimization), it is necessary to consider it as an optimization problem that optimizes the relationship between input and output. According to this embodiment 1, the input becomes the value of the molding condition, and the output becomes the required quality of the molded product. The value of the molding condition that becomes the input is a quantitative value, but the required quality of the molded product that becomes the output is sometimes expressed by the visual results or feelings of the molding operator, and further, sometimes it is not defined as a quantitative value. Therefore, first, a method for obtaining a quality value obtained by quantifying the required quality of the molded product 211 is described.
[0092] In the first embodiment, quality values such as dimensions and warpage that are easy to measure are defined as direct quality values. On the other hand, quality values such as sink marks and flow marks that are difficult to measure are replaced with feature quantities extracted from the values of the sensor 212 provided in the mold 210 or feature quantities extracted from the image captured by the camera 700, and are quantified, and these feature quantities are defined as indirect quality values. In order to set sink marks and flow marks as direct quality values, they need to be measured by a high-resolution measuring machine, but in this case, since the measuring machine is expensive, it is difficult to prepare, or the cutting operation of the measurement sample is required and the measurement cannot be easily performed, so they are set as indirect quality values.
[0093] The method of obtaining the direct quality value is to directly measure the dimensions, flatness, etc. of the molded product 211. For example, if the quality is required when any dimension falls within the dimension tolerance, the dimension is measured using the shape measuring device 600 (a measuring device such as a vernier caliper or a three-dimensional measuring machine), and the measured value is set as the quality value for the required quality. In addition, if the quality is required for warpage, the quality value can be obtained by measuring the geometric tolerances such as the flatness and perpendicularity of any surface.
[0094] The measured quality value is transmitted via the network or inputted by the molding operator through the GUI (Graphical User Interface) and is sent to the display input unit 130. Then, in the control processing unit 120, the direct quality value processing unit 124 performs preprocessing (data combination with other quality values and conversion into arrangement information, etc.) for performing the Bayesian optimization described later.
[0095] On the other hand, the indirect quality value is obtained by converting the sensor value and image obtained 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-integrated value of the temperature sensor provided in the mold 210 by the time-integrated value of the pressure sensor) becomes the quality value.
[0096] like Figure 6 As shown in FIG. 1 , the shrinkage mark feature quantity and the shrinkage mark amount (shrinkage mark measurement value) measured by a high-resolution measuring machine have a positive correlation, and the following relationship is formed: when the shrinkage mark feature quantity is small, the shrinkage mark amount (shrinkage mark measurement value) also becomes small. In addition, if the required quality is to minimize the flow mark, the image of the taken out molded product 500 is photographed by the camera 700, grayscale, cropping, image smoothing by various low-pass filters for blurring the image, etc. are applied to perform black and white binary processing.
[0097] Figure 7 This is an explanatory diagram showing an example of the result of image processing performed on an image of a molded product.
[0098] exist Figure 7 In the figure, the case where the flow mark is large is shown in parallel from left to right ( Figure 7 A) When the flow mark is small ( Figure 7 B) No flow marks ( Figure 7 C) The ratio of the white area in the image varies depending on the size of the flow mark, and therefore the ratio of the white area can be used as the quality value of the flow mark.
[0099] As described above, by using the quality values (direct quality value and indirect quality value) obtained by quantifying the required quality of the molded product 211, it is possible to process the problem of optimizing the required quality (quality value) of the molded product with the input being the value of the molding condition and the output being the value of the molding condition.
[0100] [Bayesian optimization method]
[0101] Before describing a series of processes of the molding condition optimization method, an overview of the Bayesian optimization method used to derive molding conditions that satisfy the required quality of the molded product 211 will be described.
[0102] The Bayesian optimization method is one of the parameter optimization methods that can be applied even when the optimization object function is unknown. First, a prediction model is constructed based on the input parameters (molding conditions in this embodiment 1) and the target variable values for the input parameters (quality values obtained by quantifying the required quality of the molded product 211 in this embodiment 1). Then, by using the prediction model, the predicted distribution of the target variable for the input parameter to be studied is inferred. Using this predicted distribution, the value of the target variable is prompted to become the input parameter with the highest evaluation of the expected value (the molding condition that should be implemented next). By repeating this process, the input parameters are adapted.
[0103] The prediction model used here adopts a model that can calculate the posterior distribution of the output with respect to the input (a Gaussian process regression model in this embodiment 1), but other regression models such as a random forest regression model can also be used.
[0104] [Gaussian process regression model]
[0105] Gaussian process regression is one of the non-parametric regression methods. Compared with neural network regression, it is able to construct a prediction function even with relatively less data.
[0106] Gaussian process regression is one of the models for estimating a target function y=f(x) from an input variable x to a real value y as a target variable.
[0107] Specifically, the model is as follows: Given the data D 1:t ={x 1:t ,y 1:t}, the search object objective function f(x t+1 )’s prediction distribution P(f(x t+1 )|D 1:t ,x t+1 ).
[0108] [Formula 1]
[0109]
[0110] The right side of equation (1) is the average value (expected value) set to μ xt+1|x1:t , and set the variance to σ 2 xt+1|x1:t Normal distribution (Gaussian distribution).
[0111] For example, in a Gaussian process regression model in which y is standardized so that its mean becomes 0, when the covariance matrix K representing the relationship of the input parameter x is expressed using an arbitrary kernel function k(x,x'), the predicted mean and predicted variance can be calculated using the following formulas (2) and (3).
[0112] [Formula 2]
[0113]
[0114] [Formula 3]
[0115]
[0116] Here, k ** represents k(x * ,x * ), k * represents (k(x * ,x1),k(x * ,x2),…,k(x * ,x n )) T vector, matrix K represents k(x n ,x' n ) is set to the N×N covariance matrix of the elements, and the vector y represents (y1,y2,…,y n ) vector. In the first embodiment, a Gaussian kernel of the following formula (4) is used in the kernel function k(x, x') (θ1, θ2 are parameters that determine the properties of the kernel), but in addition, any of an exponential kernel, a periodic kernel, and a Marten kernel may be used.
[0117] [Formula 4]
[0118]
[0119] [Parameter search method based on Bayesian optimization]
[0120] In order to determine the input parameters (molding conditions) to be verified next by the Bayesian optimization method, an acquisition function for evaluating the combination of input parameter candidates is used. This acquisition function can use, for example, PI (Probability of Improvement) or EI (Expected Improvement). Alternatively, UCB (Upper Confidence Bound) or MI (Mutual Information) can also be used.
[0121] In the first embodiment, as an example, a case where an acquisition function called EI (Expected Improvement) is used in correspondence with a plurality of target variables to calculate expected values of improvement from the minimum (maximum) value of the target variable.
[0122] In the case of a minimization problem of the objective function y=f(x), the Bayesian optimization using the EI value determines the search condition with the maximum expected value of improvement expressed by the following equation (5) as the next search condition.
[0123] [Formula 5]
[0124] E[I(x)]=E[max(fmin-·Y,0)]·…(5)
[0125] As a specific method for calculating the EI value, there is a method of calculating according to the following formula (6).
[0126] [Formula 6]
[0127]
[0128] f in equation (6) min represents the minimum value of the objective function in the current search number, μ(x) and σ(x) represent the predicted mean and predicted standard deviation output from the Gaussian process regression model. In addition, Ф represents the cumulative distribution function, represents the probability density function.
[0129] Figure 8 A schematic diagram showing a method of determining the next search condition using the EI value is shown.
[0130] Figure 8 The upper curve is the predicted distribution of the objective function f(x), the black points represent the observed data points, μ(x) represents the predicted average, and CI Upper and CI Lower Represents the upper and lower limits of the confidence interval calculated based on the predicted standard deviation.
[0131] also, Figure 8 The lower graph is the calculated EI value. The black points are observed, so the EI value becomes smaller.
[0132] In this Figure 8 In the schematic diagram, x5 with the largest EI value becomes the next search condition.
[0133] then, Fig. 9 The upper graph shows the results of trying the next search condition. Fig. 9 The lower graph is a schematic diagram showing a method of determining the next search condition.
[0134] The predicted distribution of x5, which is the next search condition, becomes clear, and the EI value of x5 becomes smaller.
[0135] As a result, the next search condition is determined to be another search condition with a high EI value.
[0136] As mentioned above, Figure 8 and Fig. 9 As shown, the next search condition is determined according to the EI value, and repeated attempts are made to find the optimal input parameters.
[0137] In this first embodiment, there are sometimes multiple objective functions (for example, the warpage amount of the molded product, the amount of shrinkage, and the color difference of the appearance photograph), so the acquisition function is simplified by synthesizing the evaluation values (EI values). As a method for simplifying the evaluation values, a weighted linear sum method is used, but a simple sum or product method may also be used.
[0138] [Method for optimizing molding conditions]
[0139] Figure 10 to Figure 13 1 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. In addition, the following is a specific description of the case where the target variables for optimization are warpage and sink marks. In addition, in the figure, the symbol S represents a step.
[0140] Whenever the molding conditions are optimized, the initial data collection process (step S100, step S101) is started. Here, first, the molding condition derivation device 100 is started to set the molding condition setting range information 141, the molding condition item influence 142, and the molded product information 143 described above. Then, the molding condition derivation program stored in the storage unit 140 is started.
[0141] exist Fig.14As an example, the startup screen of the molding condition export program is shown with five input parameters (mold temperature_movable, mold temperature_fixed, injection speed level 4, holding pressure level 1, holding pressure level 2) and two target variables (target variable 1: warpage, target variable 2: sink mark).
[0142] When the molding condition derivation program stored in the storage unit 140 is started, the optimized quality information of the molded product information 143 is read as the target variable, and the molding condition items that become input parameters are selected from the molding condition item influence degree 142, and the molding condition items are generated in a manner that converges within the range of the molding condition setting range information 141. Fig.14 Then, the content is displayed in the display input unit 130. In addition, Fig.14 The example shows a case where the combination of molding conditions is randomly output, but it can also be output as a bi-level orthogonal table with each molding condition as a control factor. In addition, the molding operator can also change the molding condition item selected as the input parameter by the molding condition export program to another molding condition item.
[0143] The molding operator follows the display input unit 130 of the molding condition derivation device 100. Fig.14 The molding operation is performed according to the initial molding condition table. At this time, the molding conditions can be directly manually input by the molding operator, or if the injection molding machine 200 is connected via a network, they can also be automatically input via the communication unit 110. Fig.14 The molding operation is performed sequentially from the first row of the initial molding condition table to confirm the molding stability (step S102).
[0144] In injection molding, when molding is started after the molding conditions are set and changed, the mold temperature gradually rises or falls as the number of moldings increases. Then, when a certain number of moldings are performed, the temperature change gradually decreases, and molding can be performed at the same mold temperature state. This state is used as an indicator to determine whether the molding is stable.
[0145] As a specific confirmation method, there is a method of confirming the transition of the time change of the value of the temperature sensor installed in the mold, or a method of confirming the transition of the change of the pressure value converted from the load cell of the molding machine. In this embodiment 1, the change of the value of the temperature sensor is converged within ±1°C in three consecutive moldings as a confirmation condition for molding stability.
[0146] Next, after confirming the molding stability, the molding quality value is obtained for the molded product 211 (step S103, step S501). During molding, the time series data of temperature and pressure obtained by the sensor 212 and the measuring amplifier 220 are sent to the molding condition derivation device 100 (step S505). The sent time series data is sent to the indirect quality value processing unit 123 and converted into the sensor feature value (sink mark feature value) (step S506, step S507).
[0147] The flatness of a predetermined surface of the molded product 500 after molding is measured by the shape measuring device 600 (step S502, step S503). When the molding operator inputs the GUI (Graphical User Interface) located in the display input unit 130, or the shape measuring device 600 is located on the same network, the measured flatness is directly sent, and thus, the measured flatness is sent to the direct quality value processing unit 124 of the molding condition derivation device 100 and converted into a direct quality value (step S503, step S504).
[0148] In addition, in the case of required quality related to the design of the molded product such as flow marks, after molding, the appearance of the molded product 500 is photographed by the camera 700, and its image is obtained (step S508), and it is sent to the molding condition derivation device 100. Then, in the indirect quality value processing unit 123, the indirect quality value is obtained based on the image processing (step S509, step S510). After the necessary quality value is obtained, the molding quality value acquisition process is terminated (step S511).
[0149] Next, the following process is repeated for the number of initial data shown in the initial molding condition table: the injection molding machine 200 is driven, molding is performed under each molding condition in the injection molding initial molding condition table, and a molding quality value that becomes a target variable is obtained (step S104).
[0150] Fig.15 This is an explanatory diagram showing an example of a screen of a molding condition derivation program in which molding of the initial data number is completed and a quality value is input.
[0151] like Fig.15 As shown, after the molding of the initial data number is completed (step S105, step S200), in the molding condition optimization unit 122, the molding condition optimization process is started (step S300).
[0152] When starting the molding condition optimization process (step S300, step S301), repeated molding based on the Bayesian optimization method is performed to achieve molding condition optimization. To this end, first, a Gaussian process regression model is created using the input parameters (values of molding conditions) and target variables (quality values) collected in the initial data collection process (step S302). That is, a prediction model (prediction function) is created using the input parameters (values of molding conditions) and target variables (quality values). A combination of molding conditions that have not yet been molded is input to the Gaussian process regression model created in this way, and the predicted mean and predicted standard deviation are calculated. Then, the evaluation value of the molding conditions input to the Gaussian process regression model is calculated by the acquisition function EI (Expected Improvement) (step S303).
[0153] In addition, the combination of molding conditions that have not yet been molded is all the combinations of arithmetic progressions created for the set input parameters within the upper and lower limits of the setting range information 141 of the molding conditions. The molding condition with the highest evaluation value is displayed as the molding condition to be tried next on the GUI (Graphical User Interface) located in the display input unit 130 (step S304).
[0154] Fig.16 This is an explanatory diagram showing an example of a screen of a molding condition derivation program that displays molding conditions to be tried next.
[0155] exist Fig.16 The molding conditions displayed in the last row become the molding conditions to be tried next.
[0156] By clicking on Figure 14 to Figure 16 The "Execute Bayesian Optimization" button in the molding condition export program screen shown above can execute the above-mentioned process of displaying the next molding condition.
[0157] Then, similarly to the initial data collection process, the injection molding machine 200 is driven to perform molding under the displayed molding conditions (step S305). Similarly, the molding stability is confirmed (step S306) and the molding quality value is obtained (step S307). Then, if the obtained quality value meets the required quality (step S308), the molding condition optimization process ends (step S310, step S400). On the other hand, if the required quality is not met, the end judgment is made (step S309).
[0158] In the first embodiment, the end determination is set as the number of times specified to repeat the molding 10 times, but the number of times can also be set arbitrarily. If the end determination is not the end, the molding condition optimization step (step S301) is returned again to repeat a series of processes. On the other hand, if the end determination is the end, the molding condition optimization step ends (step S310, step S400).
[0159] As described above, in the first embodiment, the Bayesian optimization method using a regression model (particularly a Gaussian process regression model) that can obtain the posterior distribution of the output with respect to the input is used to derive the molding conditions that meet the required quality of the molded product, so that even with a small number of data, a prediction function for optimizing the molding conditions can be constructed. Therefore, it is possible to easily derive the appropriate molding conditions that meet the required quality of the molded product without relying on the technical level of the molding operator.
[0160] That is, the injection molding method of the first embodiment includes the following steps. Furthermore, the computer-readable storage medium storing the computer program of the first embodiment executes the following steps when the computer program is executed by a processor.
[0161] That is, the executed steps are the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including a quality value obtained by quantifying the required quality of the molded product; using the prediction model to infer the predicted distribution of the target variable values for the input parameters; and deriving molding conditions that meet the required quality of the molded product based on the predicted distribution by using a Bayesian optimization method that utilizes a regression model, the regression model obtaining the input parameter for which the evaluation of the target variable value is the highest quality value compared to the initial quality value.
[0162] like Fig.18 As shown, the above injection molding method and computer readable storage medium are Figure 1 The molding condition optimization unit 122 in the control processing unit 120 of the molding condition derivation device 100 is executed.
[0163] and, Fig.18The molding condition adaptation unit 122 shown includes: a prediction model construction unit 1200, which constructs a prediction model based on the input parameters including the molding conditions of the molded product and the target variable value for the input parameters, the target variable value including the quality value obtained by quantifying the required quality of the molded product; a prediction distribution inference unit 1210, which uses the prediction model to infer the prediction distribution of the target variable value for the input parameter; and a molding condition derivation unit 1220, which, based on the prediction distribution, derives the molding conditions that meet the required quality of the molded product by using a Bayesian optimization method that utilizes a regression model, the regression model obtains the input parameter for which the evaluation of the target variable value is the highest quality value compared to the initial quality value.
[0164] In addition, if Fig.19 As shown in the flowchart of , the injection molding method of embodiment 1 executes the following steps, and the computer-readable storage medium storing the computer program executes the following steps when the processor executes the computer program.
[0165] That is, Fig.19 As shown, a prediction model is constructed based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including quality values obtained by quantifying the required quality of the molded product (step S1200); the prediction distribution of the target variable values for the input parameters is inferred using the prediction model (step S1210); and based on the prediction distribution, molding conditions that meet the required quality of the molded product are derived by using a Bayesian optimization method that utilizes a regression model, the regression model obtains the input parameter for which the evaluation of the target variable value is the highest quality value compared to the initial quality value (step S1220).
[0166] As mentioned above, for example, Fig.17 As an example of the hardware structure is shown, the control processing unit 120 is composed of a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a storage device 1010 (storage unit 140), and is implemented by the processor 1000 executing a program stored in the storage device 1010 (storage unit 140).
[0167] therefore, Fig.18 The processing of the prediction model construction unit 1200, the prediction distribution inference unit 1210 and the molding condition derivation unit 1220 of the molding condition optimization unit 122 shown in FIG. Fig.19The steps (steps S1200 , S1210 , and S1220 ) executed in the flowchart are realized by the processor 1000 executing a program stored in the storage device 1010 (storage unit 140 ).
[0168] Implementation Method 2
[0169] [Utilization of characteristic quantities of sensor values]
[0170] In the second embodiment, an injection molding method for maintaining and restoring good products using feature quantities of sensor values will be described.
[0171] In order to obtain the best molding conditions that satisfy the required quality of the molded product (referred to as optimization of molding condition parameters), it is necessary to consider it as an optimization problem for optimizing the relationship between input and output, as in the first embodiment.
[0172] In embodiment 2, for example, the sensor value in the mold when a good product meets the required quality (hereinafter referred to as the reference sensor value) is used as a reference, and the molding conditions are changed in a manner consistent with the reference sensor value based on Bayesian optimization that utilizes feature quantities extracted from the sensor values described later, thereby maintaining a good product state or returning from a bad state to a good product state.
[0173] In the following description, a method of returning from a defective state to a good state is described as an example.
[0174] [Structure for realizing injection molding method]
[0175] Regarding the structure for performing the injection molding method of Embodiment 2, using Fig. 20 and Fig.21 , the description will focus on the differences from Implementation 1.
[0176] Fig. 20 and Fig.21 This is a diagram showing an example of an apparatus configuration required to realize the injection molding method of the second embodiment.
[0177] like Fig. 20 As shown, the injection molding machine 200 has 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 taken into the control processing unit 120A of the molding condition derivation device 100A described later via a measuring amplifier 220.
[0178] in addition, Fig. 20 The structure and implementation mode 1 Figure 1 The same, so the detailed description is omitted.
[0179] like Fig.21As shown, the molding condition derivation device 100A includes a communication unit 110 , a control processing unit 120A, a display input unit 130 , and a storage unit 140 .
[0180] The control processing unit 120A includes a next condition output unit 121 for molding conditions, an optimization unit 122A for molding conditions, an indirect quality value processing unit 123, a direct quality value processing unit 124, and a feature value processing unit 125 for sensor values.
[0181] The storage unit 140 has setting range information 141 of molding conditions, item influence degree 142 of molding conditions, and molded product information 143 .
[0182] The molding condition derivation device 100A according to the second embodiment is different from the molding condition derivation device 100 according to the first embodiment in that a processing unit 125 for a feature amount of a sensor value is provided in a control processing unit 120A.
[0183] The sensor value feature amount processing unit 125 calculates a sensor value feature amount to be used for optimization of molding conditions, which will be described later, using the sensor value input to the control processing unit 120A via the measuring amplifier 220 .
[0184] With the above configuration, the injection molding method of the second embodiment can be implemented.
[0185] [Feature quantity extracted from sensor value]
[0186] In the second embodiment, the input is the value of the molding condition, and the output is the feature amount extracted from the measurement value of the sensor 212 in the mold.
[0187] In this example, there are three feature quantities extracted from the sensor values in the mold as output. The first and second feature quantities extracted from the sensor values are Fig. 22 The characteristic quantities of the sensor values in the X direction and the Y direction (specifically, the maximum value of the sensor value or the time when the maximum value is reached, the sensor value at the filling completion time or the time when the sensor value is reached, etc.) are shown.
[0188] exist Fig. 22 In the example, the pressure sensor value is used as the sensor value, the characteristic quantity of the sensor value in the X direction is the time when the injection pressure value reaches the maximum value, and the characteristic quantity of the sensor value in the Y direction is the maximum injection pressure value. In addition, the combination of the sensor value in the X direction and the sensor value in the Y direction as the first and second characteristic quantities is arbitrary. In addition, Fig. 22 As shown, there may be two feature quantities in the X direction (x1, x2) or two feature quantities in the Y direction (y1, y2).
[0189] The third feature quantity extracted from the sensor value is a feature quantity that uses the similarity of the sensor value when the molding condition is changed with respect to the reference sensor value.
[0190] In the second embodiment, as an example, the case where the Euclidean distance is used as the similarity is shown. The Euclidean distance is the shortest distance between two points in any dimension. Regarding the 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 Euclidean distance calculated, Fig.23 Shows the case where the similarity is low, Fig.24 Indicates the case where the similarity is high. Fig.23 and Fig.24 In FIG. 1 , the solid line shows the waveform of the reference sensor value, and the dashed-dotted line shows the waveform of the sensor value when the molding conditions are changed.
[0191] In addition, as the similarity between sensor values, in addition to the Euclidean distance, any of the Manhattan distance, the cosine similarity, and the time-integrated value of the sensor value may be used.
[0192] [Formula 7]
[0193] V1=(a1,a2,…,a n )…(7)
[0194] [Formula 8]
[0195] V2=(b1,b2,…,b n )…(8)
[0196] [Formula 9]
[0197]
[0198] [Parameter search method based on Bayesian optimization]
[0199] In the second embodiment, similarly to the first embodiment, in order to determine the input parameters (molding conditions) to be verified next by the Bayesian optimization method, an acquisition function for evaluating the combination of input parameter candidates is used. This acquisition function can use, for example, PI (Probability of Improvement) or EI (Expected Improvement). Alternatively, UCB (Upper Confidence Bound) or MI (Mutual Information) can also be used.
[0200] In the second embodiment, as an example, the acquisition function called EI (Expected Improvement) described in the first embodiment, which calculates the expected value of improvement from the minimum (or maximum) value of the target variable, is used in correspondence with the three target variables, which are the feature quantities extracted from the sensor values, so as to match the reference sensor value. As a specific method, the case where the acquisition function EI (Expected Improvement) is a minimization algorithm (Formula (5) and Formula (6)) is described.
[0201] In order to be consistent with the reference sensor value, it is necessary to combine different optimization methods for the three feature quantities of the sensor value that becomes the target variable. The feature quantities in the X and Y directions of the sensor value set the range of ±3% of the feature quantities in the X and Y directions of the reference sensor value as the optimization target, and optimize with the purpose of converging to this range. In order to use the minimization algorithm to make the feature quantities in the X and Y directions of the sensor value as the target variable converge to the range of the optimization target, by using the value μ' calculated by applying the predicted average μ output from the Gaussian process regression model to formula (10), it is possible to achieve optimization that converges to the target range. In addition, RLupper in formula (10) represents the +3% value of the reference sensor value in the X and Y directions, and RLlower represents the -3% value of the reference sensor value in the X and Y directions. In addition, the ratio of the upper and lower limits to the reference sensor value can also be arbitrarily set within the range of ±10%.
[0202] [Formula 10]
[0203]
[0204] In addition, the similarity with respect to the reference sensor value is optimized in such a way that the similarity is maximized. In order to use the minimization algorithm to perform processing that maximizes the similarity as the target variable, the target variable can be maximized by multiplying the predicted average μ output from the Gaussian process regression model by -1 to reverse the positive and negative values.
[0205] In this second embodiment, the above-mentioned processing is performed on the three target variables, and then the acquisition function is simplified by synthesizing the evaluation value (EI value) based on the EI (Expected Improvement) described in the first embodiment. As a method for simplifying the evaluation value, a method of taking a weighted linear sum is used, but a method of simply taking a sum or a product may also be used.
[0206] In this example, the pressure sensor value is used as the sensor value. However, the temperature sensor value, AE (Acoustic Emission) sensor value, mold strain value measured by a strain gauge, etc. may be used as the sensor value and the same processing as described above may be performed.
[0207] In addition, in this example, the feature quantity in the X direction and the feature quantity in the Y direction of the two-dimensional coordinates are illustrated as sensor values, but the feature quantity in the X direction, the feature quantity in the Y direction, and the feature quantity in the Z direction of the three-dimensional coordinates may be used as sensor values, and the feature quantity in the x1 direction, the feature quantity in the x2 direction, ..., the feature quantity in the xN direction of N-dimensional coordinates (N is an integer greater than 2) may be used as sensor values, and the same processing as described above may be performed.
[0208] [Method for optimizing molding conditions]
[0209] Figure 25 to Figure 28 1 is a flowchart showing an example of a series of processing steps for optimizing molding conditions in the injection molding method of Embodiment 2. In addition, the following specifically describes the case where a molded product is returned to a good state when a defective molded product is generated due to external disturbances such as temperature changes and batch changes of resin materials. In addition, in the figure, the symbol S represents a step.
[0210] Fig.25 This is a flowchart showing an example of a series of processing steps for performing preliminary preparations for optimizing molding conditions in the injection molding method according to the second embodiment.
[0211] First, before optimizing the molding conditions for returning to a good product state, Fig.25 As a preliminary preparation, the process of obtaining the reference sensor value is started (step S601).
[0212] Then, first, the molding stability is checked (step S602 ). As a specific method for checking the molding stability, in this second embodiment, the molding stability checking condition is that the change in the value of the temperature sensor is within ±1° C. during three consecutive moldings.
[0213] After confirming the molding stability, the molding quality value acquisition step described in the first embodiment (see Fig.13 ) in the step of confirming that the molding quality value meets the required quality (step S603, step S604, step S501 to step S511).
[0214] Assuming that the state in which the molding quality value does not satisfy the required quality continues, a molding condition that enables good products to be molded using the method of Embodiment 1 is derived. Then, the value of the sensor 212 (in this example, the pressure sensor) when the required quality is satisfied is stored in the storage unit 140 from the measuring amplifier 220 via the display input unit 130 (step S605). In this case, the sensor value is stored for a predetermined set number of times (step S606). In this example, for example, as the number of times the sensor value is stored, i.e., the set number of times, molding is repeated 30 times, but the set number of times can also be set arbitrarily.
[0215] After the sensor values are acquired, the reference sensor values are created using the stored sensor values (step S607). As a specific method for creating reference sensor values, there are methods of calculating the average value, central value, and weighted average value of sensor values after performing preprocessing such as removing or replacing missing values and noise, but in this example, the average value is used. At this time, the reference sensor values are acquired from the created reference sensor values. Fig. 22 The characteristic quantity of the sensor value in the X direction (in this example, the time when the maximum injection pressure value is reached) and the characteristic quantity in the Y direction (in this example, the maximum injection pressure value) are stored in the storage unit 140. After the reference sensor value is prepared and stored, the reference sensor value acquisition process (step S608) is completed. This advance preparation can be performed not only when the molding conditions of the mold are derived, but also in the mass production of molded products.
[0216] Next, after the reference sensor value of the target molded product is produced, the molding conditions are optimized for returning the product to a good state.
[0217] That is, according to Figure 26 to Figure 28 The initial data collection process (step S100, step S101) is started as shown in the flowchart. Here, first, the molding condition derivation device 100A is started to set the setting range information 141 of the molding condition, the item influence degree 142 of the molding condition, and the molded product information 143 described in the first embodiment. Then, the molding condition derivation program stored in the storage unit 140 is started.
[0218] exist Fig.30 As an example, the startup screen of the molding condition export program is shown with four input parameters (resin temperature level 1, injection speed level 3, injection speed level 4, holding pressure level 1) and three target variables (target variable 1: feature value of the sensor value in the X direction, target variable 2: feature value of the sensor value in the Y direction, target variable 3: similarity of the sensor value to the reference waveform).
[0219] When the molding condition derivation program stored in the storage unit 140 is activated, the molding condition item that is the input parameter that has a greater influence on the change of the sensor value is selected from the molding condition item influence degree 142, and the molding condition item is generated so as to be within the range of the molding condition setting range information 141. Fig.30 Then, the content is displayed on the display input unit 130.
[0220] in addition, Fig.30 The situation in which the combination of molding conditions is randomly output is shown, but it is also possible to output it as a two-level orthogonal table in which each molding condition is set as a control factor, or to set the scalar value of [a matrix of randomly selected molding conditions of any number] × [the transposed matrix of the selected molding conditions] as the optimization benchmark and select the initial conditions that maximize the optimization benchmark.
[0221] The molding operator follows the display input unit 130 of the molding condition derivation device 100A. Fig.30 The molding operation is performed according to the initial molding condition table. At this time, the molding conditions can be directly manually input by the molding operator, or if the injection molding machine 200 is connected via a network, they can also be automatically input via the communication unit 110. Fig.30 The molding operation is performed sequentially from the first row of the initial molding condition table to confirm the molding stability (step S102).
[0222] In injection molding, when molding is started after the molding conditions are set and changed, the mold temperature gradually rises or falls as the number of moldings increases. Then, when a certain number of moldings are performed, the temperature change gradually decreases, so that molding can be performed at the same mold temperature state. This state is used as an indicator to determine whether the molding is stable.
[0223] As a specific confirmation method, in this example, similarly to the example of the first embodiment, the change in the value of the temperature sensor is set to be within ±1° C. in three consecutive moldings as a confirmation condition for molding stability.
[0224] Next, after confirming the molding stability, Fig.29According to the flowchart, the characteristic quantity of the sensor value is obtained for the molded product 211 (step S1000, step S801). During molding, the time series data of the pressure obtained by the sensor 212 and the measuring amplifier 220 is sent to the molding condition derivation device 100A (step S802). The sent time series data is sent to the processing unit 125 of the characteristic quantity of the sensor value. In the processing unit 125 of the characteristic quantity of the sensor value, the characteristic quantity of the three sensor values, namely, the similarity of the current sensor value with respect to the reference sensor value (step S803), the characteristic quantity of the sensor value in the X direction (step S804), and the characteristic quantity of the sensor value in the Y direction (step S805) are calculated. After the acquisition of the necessary values is completed, the process of acquiring the characteristic quantity of the sensor value is ended (step S806).
[0225] Next, the following process is repeated for the number of initial data shown in the initial molding condition table: the injection molding machine 200 is driven, molding is performed under each molding condition in the initial molding condition table, and the characteristic value of the sensor value that becomes the target variable is obtained (step S104).
[0226] Fig.31 This is an explanatory diagram showing an example of a screen of a molding condition derivation program in which molding of an initial number of data is completed and feature quantities of sensor values are input.
[0227] like Fig.31 As shown, after the molding of the initial data number is completed and the initial data collecting process is ended (step S105, step S200), the molding condition optimization process (step S300) is started in the molding condition optimization unit 122A.
[0228] When the molding condition optimization process (step S300, step S301) is started, repeated molding based on the Bayesian optimization method is performed to achieve molding condition optimization. To this end, first, a Gaussian process regression model is created using the input parameters (values of molding conditions) and target variables (feature quantities of sensor values) collected in the initial data collection process (step S302). That is, a prediction model (prediction function) is created using the input parameters (values of molding conditions) and target variables (feature quantities of sensor values). A combination of molding conditions that have not yet been molded is input to the Gaussian process regression model thus created, and the predicted mean value and predicted standard deviation are calculated. Then, numerical processing is performed to apply the minimization algorithm to the predicted mean value, and the evaluation value of the molding conditions input to the Gaussian process regression model is calculated by the acquisition function EI (Expected Improvement) (step S303).
[0229] In addition, the combination of molding conditions that have not yet been molded is all the combinations of arithmetic progressions created for the set input parameters within the upper and lower limits of the setting range information 141 of the molding conditions. The molding condition with the highest evaluation value is displayed as the molding condition to be tried next on the GUI (Graphical User Interface) located in the display input unit 130 (step S304).
[0230] Fig.32 This is an explanatory diagram showing an example of a screen of a molding condition derivation program that displays molding conditions to be tried next.
[0231] exist Fig.32 The molding conditions displayed in the last row become the molding conditions to be tried next.
[0232] By clicking on Figure 30 to Figure 32 The "Execute Bayesian Optimization" button in the molding condition export program screen shown above can execute the above-mentioned process of displaying the next molding condition.
[0233] Then, similarly to the initial data collection process, the injection molding machine 200 is driven to perform molding under the displayed molding conditions (step S305). Furthermore, similarly to the initial data collection process, the molding stability is confirmed (step S306) and the characteristic quantity of the sensor value is obtained (step S2000, step S801). Then, if the characteristic quantity of the acquired sensor value meets the requirements (step S308), the molding condition optimization process ends (step S310, step S400). In this example, when the Euclidean distance used in the similarity with respect to the reference sensor value is greater than 0.7, it is judged that the requirements are met. On the other hand, if the requirements are not met, the end judgment is made (step S309).
[0234] In this example, the end determination (step S309) is set as the number of times to repeat the molding 10 times, but the number of times can be set arbitrarily. If the end determination is not the end, the molding condition optimization process (step S301) is returned again to repeat a series of processes. On the other hand, if the end determination is the end, the molding condition optimization process ends (step S310, step S400).
[0235] As described above, in the second embodiment, the Bayesian optimization method using a regression model (particularly a Gaussian process regression model) that can obtain the posterior distribution of the output with respect to the input is used to derive molding conditions that meet the required quality of the molded product, so that even with a small number of data, a prediction function for optimizing the molding conditions can be constructed. Therefore, when a molded product is defective due to external interference such as temperature change or batch change of the resin material, appropriate molding conditions for returning to a good state can be easily derived without relying on the technical level of the molding operator.
[0236] That is, the injection molding method of the second embodiment includes the following steps. Furthermore, the computer-readable storage medium storing the computer program of the second embodiment executes the following steps when the computer program is executed by a processor.
[0237] That is, the executed steps are the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product, characteristic quantities of sensor values of a sensor configured on an injection molding machine for the input parameters, and target variable values, wherein the target variable value includes the similarity of the sensor value when the molding conditions of the molded product are changed relative to a reference sensor value, and the reference sensor value is the sensor value when the molded product meets the required quality; using the prediction model to infer the predicted distribution of the target variable value for the input parameters; and deriving the molding conditions that meet the required quality of the molded product based on the predicted distribution by using a Bayesian optimization method that utilizes a regression model, wherein the regression model obtains the input parameter whose evaluation ratio of the target variable value is closer to the characteristic quantity of the reference sensor value than the characteristic quantity of the initial sensor value.
[0238] like Fig.33 As shown, the above injection molding method and computer readable storage medium are Fig.21 The molding condition optimization unit 122A in the control processing unit 120A of the molding condition derivation device 100A is executed.
[0239] and, Fig.33The molding condition adaptation unit 122A shown has: a prediction model construction unit 1200A, which constructs a prediction model based on the input parameters including the molding conditions of the molded product, the characteristic values of the sensor values of the sensors configured on the injection molding machine for the input parameters, and the target variable value, wherein the target variable value includes the similarity of the sensor value when the molding conditions of the molded product are changed relative to the benchmark sensor value, and the benchmark sensor value is the sensor value when the molded product meets the required quality; a prediction distribution inference unit 1210A, which uses the prediction model to infer the prediction distribution of the target variable value for the input parameters; and a molding condition derivation unit 1220A, which derives the molding conditions that meet the required quality of the molded product based on the prediction distribution by using the Bayesian optimization method of the regression model, and the regression model obtains the input parameter whose evaluation value of the target variable value is closer to the characteristic value of the benchmark sensor value than the initial sensor value.
[0240] In addition, if Fig.34 As shown in the flowchart of , the injection molding method of embodiment 2 executes the following steps, and the computer-readable storage medium storing the computer program executes the following steps when the processor executes the computer program.
[0241] That is, Fig.34 As shown, a prediction model is constructed based on input parameters including molding conditions of a molded product, characteristic quantities of sensor values of a sensor configured on an injection molding machine for the input parameters, and a target variable value, wherein the target variable value includes a similarity between a sensor value when the molding conditions of the molded product are changed and a reference sensor value, wherein the reference sensor value is the sensor value when the molded product meets the required quality (step S1200A); the prediction distribution of the target variable value for the input parameters is inferred using the prediction model (step S1210A); and based on the prediction distribution, the molding conditions that meet the required quality of the molded product are derived by using a Bayesian optimization method that utilizes a regression model, wherein the regression model obtains the input parameter whose evaluation ratio of the target variable value is closer to the characteristic quantity of the reference sensor value than the initial sensor value (step S1220A).
[0242] In the second embodiment, similarly to the first embodiment, for example, Fig.17As an example of the hardware structure, the control processing unit 120A is composed of a processor 1000 such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit) and a storage device 1010 (storage unit 140), and is implemented by the processor 1000 executing a program stored in the storage device 1010 (storage unit 140).
[0243] therefore, Fig.33 The processing of the prediction model construction unit 1200A, the prediction distribution inference unit 1210A and the molding condition derivation unit 1220A of the molding condition optimization unit 122A shown in FIG. Fig.34 The steps (steps S1200A, S1210A, and S1220A) executed in the flowchart are realized by the processor 1000 executing a program stored in the storage device 1010 (storage unit 140 ).
[0244] In addition, although the present application describes exemplary embodiments, various features, modes, and functions described in the embodiments are not limited to application in specific embodiments, and can be applied to the embodiments alone or in various combinations.
[0245] Therefore, numerous modifications not shown in the examples are conceivable within the technical scope disclosed in the present application, including, for example, modifying at least one structural element, adding at least one structural element, or omitting at least one structural element.
[0246] Description of symbols
[0247] 100, 100A: molding condition derivation device; 110: communication unit; 120, 120A: control processing unit; 121: next condition output unit of molding condition; 122, 122A: molding condition adaptation unit; 123: indirect quality value processing unit; 124: direct quality value processing unit; 125: sensor value feature value processing unit; 130: display input unit; 140: storage unit; 141: setting range information of molding condition; 142: item influence of molding condition degrees; 143: molded product information; 200: injection molding machine; 210: mold; 211: molded product (in the process of molding); 212: sensor; 220: measuring amplifier; 300: removal robot; 400: transport conveyor; 500: molded product (after molding); 600: shape measuring equipment; 700: camera; 1200, 1200A: prediction model building unit; 1210, 1210A: prediction distribution inference unit; 1220, 1220A: molding condition derivation unit.
Claims
1. An injection molding method, the injection molding method comprising the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including quality values obtained by quantifying required quality of the molded product; inferring a predicted distribution of the target variable values for the input parameters using the predictive model; and Based on the predicted distribution, a molding condition satisfying the required quality of the molded product is derived by a Bayesian optimization method using a regression model, wherein: The regression model obtains the input parameter that causes the target variable value to be evaluated to the highest quality value compared to the initial quality value.
2. The injection molding method according to claim 1, wherein: In the derivation of the molding conditions, at least one of a direct quality value and an indirect quality value is used. The direct quality value is obtained by directly measuring the molded product, and the indirect quality value includes a feature value converted from data of a sensor set in a mold of an injection molding machine or an appearance image of the molded product.
3. The injection molding method according to claim 1 or 2, wherein: The regression model is a Gaussian process regression model.
4. A molding condition derivation device for optimizing molding conditions for a molded product according to the injection molding method according to any one of claims 1 to 3, wherein: The molding condition deriving device comprises: a storage unit storing in advance information on molding conditions and required quality of the molded product; and Control Processing Unit, The control processing unit comprises: A direct quality value processing unit, which directly measures the molded product to obtain a direct quality value; an indirect quality value processing unit that obtains an indirect quality value including a feature quantity converted from data of a sensor provided in a mold of an injection molding machine or an appearance image of the molded product; as well as A molding condition optimization unit, which takes in at least one of the direct quality value from the direct quality value processing unit and the indirect quality value from the indirect quality value processing unit as a quality value, uses the taken-in quality value and the molding condition and the required quality information stored in the storage unit, and derives molding conditions that meet the optimal required quality of the molded product through a Bayesian optimization method that makes use of a regression model.
5. The molding condition derivation device according to claim 4, wherein: The molding condition optimization unit includes: a prediction model construction unit that constructs a prediction model based on the input parameters including molding conditions of the molded product and target variable values for the input parameters, the target variable values including the quality value obtained by quantifying the required quality of the molded product; a prediction distribution inference unit that uses the prediction model to infer the prediction distribution of the target variable value for the input parameter; as well as A molding condition derivation unit, which derives molding conditions that meet the required quality of the molded product based on the predicted distribution by using a Bayesian optimization method that utilizes a regression model, wherein the regression model obtains the input parameter that causes the evaluation of the target variable value to reach the highest quality value compared to an initial quality value.
6. A computer-readable storage medium storing a computer program, which, when executed by a processor, performs the following steps: constructing a prediction model based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including quality values obtained by quantifying required quality of the molded product; inferring a predicted distribution of the target variable values for the input parameters using the predictive model; and Based on the predicted distribution, a molding condition satisfying the required quality of the molded product is derived by a Bayesian optimization method using a regression model, wherein: The regression model obtains the input parameter that causes the target variable value to be evaluated to the highest quality value compared to the initial quality value.
7. The computer-readable storage medium according to claim 6, wherein: In the derivation of the molding conditions, at least one of a direct quality value and an indirect quality value is used. The direct quality value is obtained by directly measuring the molded product, and the indirect quality value includes a feature value converted from data of a sensor set in a mold of an injection molding machine or an appearance image of the molded product.
8. The computer-readable storage medium according to claim 6 or 7, wherein: The regression model is a Gaussian process regression model.
9. An injection molding method, comprising the steps of: A prediction model is constructed based on input parameters including molding conditions of a molded product and target variable values for the input parameters, wherein: The target variable value includes a feature value of a sensor value of a sensor disposed in the injection molding machine and a similarity of the sensor value when the molding condition of the molded product is changed to a reference sensor value, wherein the reference sensor value is the sensor value when the molded product satisfies the required quality; inferring a predicted distribution of the target variable value for the input parameter using the predictive model; as well as Based on the predicted distribution, the molding conditions that meet the required quality of the molded product are derived by utilizing the Bayesian optimization method of the regression model, and the regression model obtains the input parameters of the evaluation ratio of the target variable value that is closer to the feature quantity of the initial sensor value than the feature quantity of the reference sensor value.
10. The injection molding method according to claim 9, wherein: In deriving the molding conditions, the feature values in the x1 direction, the feature values in the x2 direction, ..., the feature values in the xN direction of the N-dimensional coordinates (N is an integer greater than 2) of the sensor value obtained based on the sensor value and the similarity of the sensor value to the reference sensor value are used.
11. The injection molding method according to claim 9 or 10, wherein: The regression model is a Gaussian process regression model.
12. A molding condition deriving device for optimizing molding conditions for a molded product according to the injection molding method according to any one of claims 9 to 11, wherein: The molding condition deriving device comprises: a storage unit storing in advance information on molding conditions and required quality of the molded product; and Control Processing Unit, The control processing unit comprises: a processing unit for processing a feature amount of a sensor value, which calculates a feature amount of the sensor value obtained based on the sensor value and a similarity of the sensor value to the reference sensor value; and A molding condition optimization unit, which takes in the feature quantity of the sensor value and the similarity with respect to the reference sensor value from a feature quantity processing unit of the sensor value, uses the taken-in feature quantity of the sensor value and the similarity with respect to the reference sensor value, and the information on the molding conditions and the required quality stored in the storage unit, and derives molding conditions that meet the optimal required quality of the molded product by a Bayesian optimization method that makes use of a regression model.
13. The molding condition derivation device according to claim 12, wherein: The molding condition optimization unit includes: a prediction model construction unit that constructs a prediction model based on the input parameters including the molding conditions of the molded product and target variable values for the input parameters, the target variable values including a feature amount of a sensor value of a sensor disposed in an injection molding machine and a similarity of the sensor value when the molding conditions of the molded product are changed to a reference sensor value, the reference sensor value being the sensor value when the molded product satisfies a required quality; a prediction distribution inference unit that uses the prediction model to infer the prediction distribution of the target variable value for the input parameter; as well as A molding condition derivation unit derives molding conditions that satisfy the required quality of the molded product based on the predicted distribution by utilizing a Bayesian optimization method using a regression model, wherein the regression model obtains the input parameter that is closer to the feature quantity of the reference sensor value than the feature quantity of the initial sensor value for the evaluation of the target variable value.
14. A computer-readable storage medium storing a computer program, which, when executed by a processor, performs the following steps: A prediction model is constructed based on input parameters including molding conditions of a molded product and target variable values for the input parameters, the target variable values including feature quantities of sensor values of a sensor disposed in an injection molding machine and similarity of the sensor values when the molding conditions of the molded product are changed to a reference sensor value, the reference sensor value being the sensor value when the molded product meets required quality; inferring a predicted distribution of the target variable value for the input parameter using the predictive model; as well as Based on the predicted distribution, the molding conditions that meet the required quality of the molded product are derived by utilizing the Bayesian optimization method of the regression model, and the regression model obtains the input parameters of the evaluation ratio of the target variable value that is closer to the feature quantity of the initial sensor value than the feature quantity of the reference sensor value.
15. The computer-readable storage medium according to claim 14, wherein: In deriving the molding conditions, the feature values in the x1 direction, the feature values in the x2 direction, ..., the feature values in the xN direction of the N-dimensional coordinates (N is an integer greater than 2) of the sensor value obtained based on the sensor value and the similarity of the sensor value to the reference sensor value are used.
16. The computer-readable storage medium according to claim 14 or 15, wherein: The regression model is a Gaussian process regression model.
Citation Information
Patent Citations
Support device of injection molding machine
JP2008110486A
Cited By
Intelligent production line for wood fiber injection molding
CN121403514A
Wood fiber injection molding intelligent production line
CN121403514B
Parameter adjustment method, electronic equipment and readable storage medium
CN121650208A
Parameter adjustment method, electronic device, and readable storage medium
CN121650208B
