Method for calculating the mixing ratio of resin compositions, resin compositions, and methods for mixing resin compositions

Abstract

The present invention provides a method for easily calculating the mixing ratio of resin compositions to ensure that the physical properties of the resin compositions are within specifications, a resin composition mixed so that its physical properties are within specifications, and a method for easily mixing resin compositions whose physical properties are within specifications. [Solution] The method for calculating the mixing ratio of a resin composition includes calculating the mixing ratio of a resin composition in order to bring the physical properties of a resin composition that is outside the specifications into specifications by mixing it with an arbitrary stock item, based on inventory information that associates the physical properties of the resin composition with the lot.

Description

[Technical Field]

[0001] This disclosure relates to a method for calculating the mixing ratio of a resin composition, a resin composition, and a method for mixing a resin composition. [Background technology]

[0002] As disclosed in Patent Document 1, a method for predicting the physical properties of a specific fiber-reinforced resin composition from the blending ratio of raw materials is known. Also, as disclosed in Patent Document 2, a method for producing a resin composition using off-spec products is known. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-154911 Public Relations [Patent Document 2] Japanese Patent Publication No. 2022-081294 [Overview of the project] [Problems that the invention aims to solve]

[0004] In the manufacture of resin compositions by combining resins and compounding materials, the physical properties of the manufactured resin composition may deviate from predetermined specifications for various reasons. Resin compositions whose physical properties deviate from specifications are subject to disposal or other measures as substandard products.

[0005] Here, from the perspective of yield and other factors, it is necessary to effectively utilize resin compositions whose physical properties deviate from the specifications. For example, one possible approach is to perform a rework process to adjust the physical properties of off-spec products to within specifications by mixing them with products that have physical properties within specifications or with other off-spec products.

[0006] As part of rework processing, when mixing off-specification parts with standard parts or other off-specification parts, it is conceivable to mix them gradually while checking whether the physical properties fall within the specified range. However, if multiple physical properties of the off-specification parts subject to rework processing deviate from the specifications, it is difficult to mix them in such a way that all physical properties fall within the specifications.

[0007] In view of these circumstances, the purpose of this disclosure is to provide a method for calculating the mixing ratio of resin compositions that can easily calculate the mixing ratio of resin compositions to bring the physical properties within specifications, a resin composition mixed so that the physical properties are within specifications, and a method for mixing resin compositions that can easily mix resin compositions whose physical properties are within specifications. [Means for solving the problem]

[0008] (1) A method for calculating the mixing ratio of a resin composition according to one embodiment of the present disclosure includes calculating the mixing ratio of a resin composition to bring the physical properties of a resin composition that is outside the specifications into specifications by mixing it with an arbitrary stock item, based on inventory information that associates the physical properties of the resin composition with the lot.

[0009] (2) The method for calculating the mixing ratio of the resin composition described in (1) above may include calculating the mixing ratio using a mixing ratio calculation model.

[0010] (3) The method for calculating the mixing ratio of the resin composition described in (1) or (2) above may include predicting at least one of the physical properties of the resin composition using a predictive model.

[0011] (4) The method for calculating the mixing ratio of the resin composition described in (1) above may include predicting at least one of the physical properties of the resin composition as the transit characteristics of the resin composition from the infrared spectrum of the resin composition using a first prediction model as the prediction model, and predicting at least one of the physical properties of the resin composition as the target characteristics of the resin composition from the prediction result of the transit characteristics of the resin composition using a second prediction model as the prediction model.

[0012] (5) A resin composition according to one embodiment of the present disclosure is a resin composition mixed based on a mixing ratio calculated by performing the method for calculating the mixing ratio of the resin composition described in any one of (1) to (4) above.

[0013] (6) A method for mixing a resin composition according to one embodiment of the present disclosure includes mixing the resin composition based on a mixing ratio calculated by performing the method for calculating the mixing ratio of a resin composition described in any one of (1) to (4) above. [Effects of the Invention]

[0014] According to the method for calculating the mixing ratio of resin compositions in this disclosure, the mixing ratio of resin compositions to bring the physical properties within specifications can be easily calculated. The resin compositions in this disclosure are mixed so that their physical properties are within specifications. According to the method for mixing resin compositions in this disclosure, resin compositions with physical properties within specifications can be easily mixed. [Brief explanation of the drawing]

[0015] [Figure 1] This block diagram shows an example configuration of the mixing ratio calculation system related to this disclosure. [Figure 2] This block diagram shows an example of a mixing ratio calculation model that calculates the mixing ratio of resin compositions to bring physical properties within specifications by inputting inventory information. [Figure 3] This is a flowchart showing an example of an inventory management scheme procedure. [Figure 4] This flowchart shows an example of the procedure for generating a mixing ratio calculation model. [Figure 5] This flowchart shows an example procedure for calculating the mixing ratio of a resin composition to bring the physical properties within a specified range. [Figure 6] This block diagram shows an example configuration of the prediction system related to this disclosure. [Figure 7A] This block diagram shows an example of a predictive model that outputs the desired characteristics of a sample by inputting the sample's IR spectrum. [Figure 7B] FIG. 1 is a block diagram showing an example of a prediction model that generates intermediate characteristics in the process of outputting target characteristics of a sample by inputting an IR spectrum of the sample. [Figure 8A] FIG. 2 is a schematic diagram for explaining measurement of an IR spectrum by the ATR method. [Figure 8B] FIG. 3 is a schematic diagram for explaining measurement of an IR spectrum by the transmission method. [Figure 9] FIG. 4 is a flowchart showing an example of a procedure for generating a physical property prediction model. [Figure 10] FIG. 5 is a flowchart showing an example of a procedure for measuring an IR spectrum of a sample. [Figure 11] FIG. 6 is a flowchart showing an example of a procedure for predicting physical properties of a sample from an IR spectrum of the sample. [Figure 12] FIG. 7 is a graph showing an example of an IR spectrum obtained by measuring a press-processed sample by the ATR method. [Figure 13] FIG. 8 is a graph showing an example of an IR spectrum obtained by measuring a sample by the transmission method. DETAILED DESCRIPTION OF THE INVENTION

[0016] Hereinafter, embodiments according to the present disclosure will be described in detail. The embodiments described below are examples for explaining the configuration according to the present disclosure, and are not intended to limit the configuration according to the present disclosure to the following contents. The configuration according to the present disclosure can be appropriately modified and implemented within the scope of the gist.

[0017] (Configuration Example of Mixing Ratio Calculation System 5) As shown in FIG. 1, a mixing ratio calculation system 5 according to the present disclosure includes an inventory management device 40 and a mixing ratio calculation device 50. The inventory management device 40 manages the inventory of the resin composition. In the present disclosure, the inventory management device 40 executes an inventory management scheme described later. The mixing ratio calculation device 50 calculates the mixing ratio of the resin composition for adjusting the physical properties of the resin composition. In the present disclosure, the mixing ratio calculation device 50 executes a mixing ratio calculation scheme described later.

[0018] <Mixing ratio calculation device 50> The mixing ratio calculation device 50 comprises an acquisition unit 51, a calculation unit 52, a storage unit 53, and an output unit 54.

[0019] The acquisition unit 51 is connected to the inventory management device 40 to acquire inventory information from the inventory management device 40. The inventory information will be described later.

[0020] The acquisition unit 51 may be equipped with a communication interface for connecting to the inventory management device 40, etc., via wired or wireless means. The communication interface may be configured to communicate with the inventory management device 40, etc., based on a communication standard such as RS-232C or RS-485. The communication interface may be configured to communicate with the inventory management device 40, etc., based on a communication standard such as Bluetooth®. The communication interface may be configured to communicate with the inventory management device 40, etc., via a network such as a LAN (Local Area Network).

[0021] The calculation unit 52 may be configured to include one or more processors, such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The processors constituting the calculation unit 52 may realize the functions of the calculation unit 52 by reading and executing a program stored in the memory unit 53, which will be described later. The calculation unit 52 may be configured to include one or more dedicated circuits. The dedicated circuits may include, for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). The calculation unit 52 may be configured in combination with a processor and dedicated circuits.

[0022] The storage unit 53 stores various information or data processed by the mixing ratio calculation device 50. The storage unit 53 may store, for example, a program executed in the calculation unit 52, or data or processing results used in the processing executed in the calculation unit 52. The storage unit 53 may also function as the work memory of the calculation unit 52. The storage unit 53 may include, but is not limited to, a semiconductor memory. For example, the storage unit 53 may be configured as the internal memory of a processor used as the calculation unit 52, or as a hard disk drive (HDD) accessible from the calculation unit 52. The storage unit 53 may be configured as a non-temporary readable medium. The storage unit 53 may be configured integrally with the calculation unit 52, or as a separate unit from the calculation unit 52.

[0023] The output unit 54 outputs the calculated result of the mixing ratio of the resin composition. The output unit 54 may be equipped with a display device. The display device may include, for example, an LCD (Liquid Crystal Display), an organic EL (Electro-Luminescence) display or an inorganic EL display, or a PDP (Plasma Display Panel). The display device is not limited to these displays and may include various other types of displays. The output unit 54 may be configured to output the calculated result of the mixing ratio of the resin composition to an external device, for example, an apparatus that performs mixing of the resin composition. The output unit 54 may be equipped with a communication interface for wired or wireless communication with the external device. The communication interface may be configured in the same way as the communication interface of the acquisition unit 51. The output unit 54 and the acquisition unit 51 may use a common communication interface.

[0024] The mixing ratio calculation device 50 may execute a mixing ratio calculation scheme using the mixing ratio calculation model 70, as shown in Figure 2. The mixing ratio calculation model 70 is configured to output a calculation result of the mixing ratio of resin compositions to adjust the physical properties of the resin compositions to desired specifications when inventory information is input. Details of the mixing ratio calculation model 70 will be described later.

[0025] The acquisition unit 51 acquires inventory information from the inventory management device 40 and outputs it to the calculation unit 52. The calculation unit 52 inputs the inventory information into the mixing ratio calculation model 70 and outputs the calculation result of the mixing ratio output from the mixing ratio calculation model 70 to the output unit 54. The output unit 54 displays the calculation result of the mixing ratio or outputs it to an external device.

[0026] <Inventory Management Device 40> As shown in Figure 1, the inventory management device 40 comprises an acquisition unit 41, a management unit 42, a storage unit 43, and an output unit 44.

[0027] The acquisition unit 41 acquires information necessary for managing the inventory quantity of the resin composition. The acquisition unit 41 may be equipped with a communication interface for wired or wireless communication with the prediction device 10 (see Figure 6), measuring device 20 (see Figure 6), or mixing ratio calculation device 50, etc., which will be described later. The communication interface may be configured in the same way as the communication interface of the acquisition unit 51 of the mixing ratio calculation device 50.

[0028] The management unit 42 manages the inventory quantity of the resin composition and generates inventory information. The management unit 42 may be configured to include one or more processors or one or more dedicated circuits, similar to the calculation unit 52 of the mixing ratio calculation device 50.

[0029] The storage unit 43 stores various information or data processed by the inventory management device 40. The storage unit 43 may store, for example, a program executed in the management unit 42, or data or processing results used in the processing executed in the management unit 42. The storage unit 43 may also function as the work memory of the management unit 42. The storage unit 43 may be configured similarly to the storage unit 53 of the mixing ratio calculation device 50. The storage unit 43 may be configured integrally with the management unit 42, or it may be configured separately from the management unit 42.

[0030] The output unit 44 outputs inventory information to the mixing ratio calculation device 50. The output unit 44 may be equipped with a communication interface for wired or wireless communication with the mixing ratio calculation device 50. The communication interface may be configured similarly to the communication interface of the acquisition unit 51 or the output unit 54 of the mixing ratio calculation device 50.

[0031] (Example of operation of the mixing ratio calculation system 5) The following describes examples of operations in which the inventory management device 40 executes the inventory management scheme and examples of operations in which the mixing ratio calculation device 50 executes the mixing ratio calculation scheme.

[0032] <Inventory Management Scheme> The inventory management scheme is a scheme for managing the inventory of resin compositions, which include resin and compounding. Compounding is other materials compounded with the resin. In this disclosure, the resin may include polyamide (PA), etc. Compounding may include inorganic fillers such as glass fiber (GF). Compounding may include colorants such as carbon black (CB). The resin composition may be a mixture obtained by kneading the resin with the compounding.

[0033] Resin compositions may have different physical properties depending on the production unit. These properties may include, for example, mechanical properties such as tensile strength or flexural strength, fluidity, viscosity, or molecular weight. The numerical values ​​representing the properties of the resin composition are obtained by measuring or predicting the properties for each production unit, i.e., each lot.

[0034] In the inventory management scheme relating to this disclosure, resin compositions are identified by assigning a lot number or the like to each arbitrary production unit of the resin composition. The produced resin compositions are managed by associating the physical property information of the resin composition with inventory information for each production unit, i.e., each lot. The physical property information of the resin composition includes a physical property value representing at least one type of physical property of each lot of the resin composition. The inventory information of the resin composition includes the inventory quantity of each lot of the resin composition. The inventory quantity of the resin composition is managed for each combination of physical property values ​​included in the physical property information by being managed on a lot-by-lot basis.

[0035] The management unit 42 of the inventory management device 40 may execute an inventory management method including the steps of the flowchart illustrated in Figure 3. The inventory management method may be implemented as an inventory management program to be executed by the processor constituting the management unit 42. The inventory management program may be stored on a non-temporary computer-readable medium.

[0036] The management unit 42 classifies the resin composition inventory by production unit and assigns a lot number to each production unit (step S1). The resin composition inventory may be acquired by the acquisition unit 41. The acquisition unit 41 may acquire the resin composition inventory by production unit by receiving input from the person in charge of resin composition inventory management. The acquisition unit 41 may also acquire the resin composition inventory by production unit from the resin composition production equipment.

[0037] The management unit 42 acquires the physical properties of each lot (step S2). The physical properties of each lot may be acquired by the acquisition unit 41. The acquisition unit 41 may acquire the physical properties of each lot of the resin composition for each production unit by receiving input from the person in charge of inventory management of the resin composition. The acquisition unit 41 may also acquire predicted values ​​of the physical properties of each lot of the resin composition for each production unit from the prediction device 10 (see Figure 6, etc.), which will be described later. The acquisition unit 41 may also acquire measured values ​​of the physical properties of each lot of the resin composition for each production unit.

[0038] The management unit 42 generates inventory information that associates the inventory and physical properties of each lot of the resin composition (step S3). In the inventory information, the inventory and physical properties of each lot of the resin composition may be associated via the lot number. The management unit 42 may store the generated inventory information in the storage unit 43. The management unit 42 may output the generated inventory information from the output unit 44 to the mixing ratio calculation device 50, etc. After executing the procedure in step S3, the management unit 42 terminates the execution of the flowchart in Figure 3.

[0039] <Scheme for calculating mixing ratio> The mixing ratio calculation scheme is a scheme for calculating the mixing ratio of resin compositions to adjust the physical properties of the resin composition. For example, the quality of a resin composition may be judged as acceptable if the physical properties of the resin composition fall within a predetermined range of arbitrary numerical values, and unacceptable if the physical properties do not fall within that range. In this disclosure, the arbitrary range for determining the quality of a resin composition is also referred to as the standard for the physical properties of the resin composition. One or more standards for physical properties may be set for a single type of resin composition.

[0040] Whether the physical properties of the resin composition are within specifications may be determined by a prediction device 10 (see Figure 6, etc.), which will be described later. The specifications may be stored in the storage unit 15 of the prediction device 10. The data processing unit 12 of the prediction device 10 may obtain the specifications for the physical properties of the sample resin composition from the storage unit 15 to determine the quality of the sample if the predicted values ​​of the physical properties of the sample resin composition fall within specifications. The data processing unit 12 may output the evaluation result of the sample quality to the output unit 16. The output unit 16 may display the evaluation result of the sample quality for the user to recognize.

[0041] If a sample is determined to be of unacceptable quality, its physical properties can be adjusted to fall within a desired numerical range by mixing it with any other sample. For example, by mixing a resin composition whose physical properties fall below the lower limit of the standard with a resin composition whose physical properties are within the standard or above the upper limit of the standard, the physical properties of the mixed resin composition can be adjusted to fall within a desired numerical range. The mixing ratio calculation device 50 according to this disclosure calculates the mixing ratio of resin compositions so that a resin composition whose quality is determined to be unacceptable because its physical properties fall outside a desired numerical range can be mixed with any stock product to produce a resin composition whose physical properties fall within a desired numerical range. The mixing ratio calculation device 50 may calculate the mixing ratio using a mixing ratio calculation model 70.

[0042] <<Mixing Ratio Calculation Model 70>> The following describes a model 70 for calculating the mixing ratio of a resin composition containing resin and compounding agents. The mixing ratio calculation model 70 calculates the mixing ratio of two lots with different physical properties so that the physical properties of those lots meet the specifications when those lots are mixed. The mixing ratio calculation model 70 may also calculate the mixing ratio of three or more lots with different physical properties so that the physical properties of those lots meet the specifications when those lots are mixed. In this example, the mixing ratio calculation model 70 is configured to calculate the mixing ratio of two lots.

[0043] In the mixing ratio calculation model 70, it is assumed that the specifications of the physical properties are set in advance. The specifications of the physical properties set in the mixing ratio calculation model 70 are also called the set specifications. The set specifications may be a specification for one type of physical property, or a combination of two or more types of physical property specifications. The mixing ratio calculation model 70 may be prepared in advance for each set specification. The mixing ratio calculation device 50 may store the mixing ratio calculation model 70 for each set specification in the storage unit 53 and retrieve it from the storage unit 53 according to the set specification. The mixing ratio calculation device 50 may also retrieve the mixing ratio calculation model 70 from an external device.

[0044] The mixing ratio calculation model 70 is configured to output the result of calculating a mixing ratio to bring the physical properties of two lots included in the inventory information within the set specifications, and information to identify the two lots to be mixed. If the numerical range of the specifications has a width, the mixing ratio may be calculated within that range. If the mixing ratio calculation model 70 calculates a mixing ratio with a width, it may output a mixing ratio that maximizes the proportion of off-specification products in order to utilize as many off-specification products as possible.

[0045] The mixing ratio calculation model 70 may be configured to, when inventory information of resin compositions is input, extract any two lots from the lots included in the inventory information, and output the numbers of the two extracted lots and the calculated mixing ratio of the two extracted lots.

[0046] The mixing ratio calculation model 70 may be configured to take inventory information of a resin composition and information specifying one lot included in the inventory information as input, extract one arbitrary lot to be mixed with the specified lot, and output the lot number of the extracted lot and the calculated mixing ratio between the specified lot and the extracted lot.

[0047] The mixing ratio calculation model 70 may be configured to output a calculation result of the mixing ratio of two specified lots when it receives inventory information of the resin composition and information specifying the two lots to be mixed. If two lots to be mixed are specified, only the inventory information of the two specified lots may be input to the mixing ratio calculation model 70.

[0048] The mixing ratio calculation model 70 may output predicted physical properties of the resin composition mixed at the calculated mixing ratio, along with the calculated mixing ratio.

[0049] The physical properties of the resin composition to be adjusted in the mixing ratio calculation model 70 may include mechanical properties such as tensile strength or flexural strength, fluidity, viscosity, or molecular weight. The fluidity of the resin composition represents how easily the resin composition flows when it is in a molten state, such as MFR (melt flow rate), MVR (melt volume rate), or shear viscosity. The fluidity of the resin composition may be expressed as a numerical value. The viscosity of the resin composition represents the specific viscosity, relative viscosity, intrinsic viscosity, or reduced viscosity obtained from the solution viscosity of the resin composition.

[0050] The mixing ratio calculation model 70 may be generated by performing training using training data that associates the physical properties of the two resin compositions to be mixed with the correct mixing ratio values ​​required for the physical properties of the mixed resin composition to fall within the set specifications. In other words, the mixing ratio calculation model 70 may be a pre-trained model.

[0051] The mixing ratio calculation model 70 may be a regression analysis model obtained by performing a regression analysis with the physical properties of the two resin compositions to be mixed as explanatory variables and the mixing ratio of the two resin compositions whose physical properties fall within the set specifications as the dependent variable. The mixing ratio calculation model 70 may also be a calibration curve model generated using chemometric methods.

[0052] The mixing ratio calculation model 70 is not limited to the pre-trained model, regression analysis model, or calibration curve model described above, but may be configured as a model of various kinds.

[0053] When constructing a mixing ratio calculation model 70, using a large number of explanatory variables results in a vast number of possible combinations of hyperparameters and other options for the mixing ratio calculation model 70. As the number of combinations increases, it becomes more difficult to set the optimal parameters. Furthermore, the mixing ratio calculation model 70 may become overfitted, reducing the accuracy of calculating the mixing ratio for unknown data. In this case, by using machine learning techniques, it becomes possible to find the optimal parameters while preventing overfitting of the mixing ratio calculation model 70, even when using a large number of explanatory variables in its construction. As a result, the accuracy of calculating the mixing ratio improves.

[0054] The mixing ratio calculation model 70 may be generated by a device other than the mixing ratio calculation device 50. The mixing ratio calculation device 50 may obtain the mixing ratio calculation model 70 generated by another device.

[0055] The mixing ratio calculation model 70 may be generated by the mixing ratio calculation device 50. The calculation unit 52 of the mixing ratio calculation device 50 may generate the mixing ratio calculation model 70 by executing the steps of the flowchart illustrated in Figure 4.

[0056] The calculation unit 52 sets the specifications for the physical properties to be adjusted by mixing the two resin compositions (step S11). The calculation unit 52 may obtain the setting of the physical property specifications as input from a user, such as the person responsible for creating the mixing ratio calculation model 70.

[0057] The calculation unit 52 acquires training data that associates the physical properties of the two resin compositions to be mixed with the correct mixing ratio required for the physical properties of the mixed resin composition to fall within the set specifications (step S12). The calculation unit 52 may acquire training data from a database that associates the physical properties of the two resin compositions to be mixed with the mixing ratio and the physical properties of the mixed resin composition. The calculation unit 52 may also acquire training data through user input.

[0058] The calculation unit 52 generates a mixing ratio calculation model 70 (step S13). If the calculation unit 52 generates the mixing ratio calculation model 70 as a trained model, it performs training using training data. If the calculation unit 52 generates the mixing ratio calculation model 70 as a regression analysis model, it performs regression analysis using training data. If the calculation unit 52 generates the mixing ratio calculation model 70 as a calibration curve model, it performs chemometrics methods.

[0059] After executing the procedure in step S13, the calculation unit 52 terminates the execution of the flowchart in Figure 4. The calculation unit 52 may store the generated mixing ratio calculation model 70 in the storage unit 53.

[0060] The calculation unit 52 of the mixing ratio calculation device 50 may execute a mixing ratio calculation method that includes the steps of the flowchart illustrated in Figure 5. The mixing ratio calculation method may be implemented as a mixing ratio calculation program to be executed by the processor constituting the calculation unit 52. The mixing ratio calculation program may be stored on a non-temporary computer-readable medium.

[0061] The calculation unit 52 sets the specifications for physical properties to adjust the physical properties by mixing the two resin compositions (step S21). The calculation unit 52 may obtain the setting of the specifications for physical properties as input from a user, such as the person in charge of mixing the resin compositions.

[0062] The calculation unit 52 obtains a mixing ratio calculation model 70 corresponding to the standard set in step S21 (step S22).

[0063] The calculation unit 52 inputs the information necessary to calculate the mixing ratio into the mixing ratio calculation model 70 (step S23). When the mixing ratio calculation model 70 extracts any one or two lots from the lots included in the inventory information, the information necessary to calculate the mixing ratio includes the inventory information. When one or two of the two lots to be mixed are specified, the information necessary to calculate the mixing ratio includes information specifying the lots to be mixed.

[0064] The calculation unit 52 obtains the result of the mixing ratio calculation from the mixing ratio calculation model 70 (step S24). The calculation unit 52 may display the result of the mixing ratio calculation using the output unit 54 to allow users, such as those responsible for mixing the resin composition, to recognize it. The calculation unit 52 may also output the result of the mixing ratio calculation from the output unit 54 to the device that performs the mixing of the resin composition. After executing the procedure in step S24, the calculation unit 52 terminates the execution of the flowchart in Figure 5.

[0065] <Method for mixing samples> The mixing of the two resin compositions is performed using the mixing ratio calculated by the mixing ratio calculation model 70, so that the physical properties of the resulting resin composition fall within the specified standards. Examples of methods for mixing the two resin compositions include, but are not limited to, the following two methods. (1) A method of mixing using commonly used mixers, such as a Henschel mixer, tumbler, ribbon blender, etc. (2) A method of pre-mixing the ingredients in a Henschel mixer or the like, if necessary, and then melt-mixing them in a single-screw or twin-screw extruder.

[0066] <Examples of utilizing off-spec products by mixing resin compositions> The following describes an example of calculating the mixing ratio. As illustrated in Table 1, the inventory information includes information on the inventory of resin compositions in four production units, each assigned lot numbers from B-1 to B-4. Relative viscosity and tensile strength are measured and associated with each lot as physical properties of the resin composition. Tensile strength was measured according to ISO 527. Relative viscosity was measured according to ISO 307. Table 1 shows whether the viscosity and tensile strength of the resin composition are within or outside the specifications. Lots in which both viscosity and tensile strength are within the specifications are judged as acceptable. Lots in which at least one of viscosity or tensile strength is outside the specifications are judged as unacceptable.

[0067] [Table 1]

[0068] In this embodiment, in order to utilize the rejected lots B-1 to B-3 exemplified in Table 1, it is conceivable to mix one of the B-1 to B-3 lots with the approved lot B-4, thereby bringing the physical properties of the mixed resin composition within specifications. As exemplified in Table 2, the pass / fail status of the mixed resin composition is determined based on the measurement results of the relative viscosity and tensile strength of the resin composition obtained by mixing one of the B-1 to B-3 lots with lot B-4 at various mixing ratios. [Table 2]

[0069] In Table 2, for example, when B-1 and B-4 were mixed in a 50%:50% or 70%:30% ratio, both the relative viscosity and tensile strength of the mixed resin composition were within specifications, and the mixed resin composition was judged to be acceptable. On the other hand, when B-1 and B-4 were mixed in a 30%:70% ratio, the relative viscosity of the mixed resin composition was outside specifications, and the mixed resin composition was judged to be unacceptable. In other words, the mixing ratio of B-1 and B-4 to bring the physical properties of the mixed resin composition within specifications should be within the range of 50%:50% to 70%:30%.

[0070] Furthermore, when B-2 and B-4 were mixed in a 30%:70% or 50%:50% ratio, both the relative viscosity and tensile strength of the mixed resin composition were within specifications, and the mixed resin composition was judged to be acceptable. On the other hand, when B-2 and B-4 were mixed in a 70%:30% ratio, the relative viscosity of the mixed resin composition was outside specifications, and the mixed resin composition was judged to be unacceptable. In other words, the mixing ratio of B-2 and B-4 required to bring the physical properties of the mixed resin composition within specifications is within the range of 30%:70% to 50%:50%.

[0071] Furthermore, when B-3 and B-4 were mixed in a 50%:50% ratio, both the relative viscosity and tensile strength of the mixed resin composition were within specifications, and the mixed resin composition was judged to be acceptable. On the other hand, when B-3 and B-4 were mixed in a 30%:70% ratio or a 70%:30% ratio, the relative viscosity of the mixed resin composition was outside specifications, and the mixed resin composition was judged to be unacceptable. In other words, the mixing ratio of B-3 and B-4 required to bring the physical properties of the mixed resin composition within specifications is 50%:50%.

[0072] The mixing ratio may be expressed as a percentage, as shown in the example in Table 2. The mixing ratio may also be expressed as a ratio of relatively prime integers. When the mixing ratio is expressed as a ratio of relatively prime integers, a 50%:50% mixing ratio is expressed as 1:1. The mixing ratio may also be expressed as a ratio of real numbers that add up to 10. When the mixing ratio is expressed as a ratio of real numbers that add up to 10, a 50%:50% mixing ratio is expressed as 5:5.

[0073] <Example of calculating the mixing ratio> In this embodiment, when mixing two resin compositions whose first and second physical properties have been measured or predicted, a mixing ratio is calculated to ensure that both the first and second physical properties are within specifications.

[0074] The two resin compositions to be mixed are assumed to be from lot 1 and lot 2. The first and second physical properties of lot 1 are represented as V11 and V12, respectively. The first and second physical properties of lot 2 are represented as V21 and V22, respectively. Furthermore, the setting standard for the first physical property is assumed to be P11 or greater and P12 or less. The setting standard for the second physical property is assumed to be P21 or greater and P22 or less.

[0075] When the resin composition of the first lot and the resin composition of the second lot are mixed in a mixing ratio of R(%):100-R(%), the first physical property of the mixed resin composition is calculated using the following formula (1). {V11×R+V12×(100-R)} / 100 (1) Furthermore, the second physical property of the mixed resin composition is calculated using the following formula (2). {V21×R+V22×(100-R)} / 100 (2)

[0076] The mixing ratio required to bring both the first and second physical properties of the mixed resin composition within the specified standards is calculated by solving the following system of inequalities (3) and (4). P11≦{V11×R+V12×(100-R)} / 100≦P12 (3) P21≦{V21×R+V22×(100-R)} / 100≦P22 (4)

[0077] The lower and upper limits of R / 100 can be calculated from the system of inequalities described above. The lower limit of R / 100 is the larger of the following two values. (P11×V22-P21×V12) / (V11×V22-V12×V21), or, {V11×(P21-V22)-V2×(P11×V12)} / (V12×V21-V11×V22) Furthermore, the upper limit of R / 100 is the smaller of the following two values. (P12×V22-P22×V12) / (V11×V22-V12×V21), or, {V11×(P22-V22)-V2×(P12×V12)} / (V12×V21-V11×V22)

[0078] As described above, the mixing ratio calculation model 70 can calculate a mixing ratio that brings the physical properties of the mixed resin composition within the set specifications. The mixing ratio calculation model 70 may be configured to calculate a mixing ratio in which both the first and second physical properties are within the specifications in a calibration curve created using equations (1) and (2) for calculating the first and second physical properties.

[0079] Here, the upper limit of the mixing ratio (R / 100) when mixing two lots may be smaller than the lower limit. This means that no matter what mixing ratio is used to mix the two lots, the physical properties will not fall within the set specifications. In this case, the mixing ratio calculation model 70 may output information indicating that the two lots cannot be mixed. This information that the two lots cannot be mixed may be output as text information, or it may be output as a mixing ratio of 0:0, which is not practically applicable to mixing, but is not limited to these options.

[0080] As described above, the mixing ratio calculation system 5 of this disclosure allows for the easy calculation of the mixing ratio when mixing another resin composition with a resin composition whose physical properties are outside of specifications.

[0081] As a comparative example, one possible method involves mixing a non-standard resin composition with another resin composition while measuring its physical properties. In this case, the resin compositions need to be mixed little by little. Therefore, the mixing process and the measurement of physical properties must be repeated. In contrast, the mixing ratio calculation system 5 according to this disclosure reduces the number of times the mixing process and the measurement of physical properties are repeated by mixing the resin compositions at the calculated mixing ratio. As a result, resin compositions with physical properties within specifications can be easily mixed.

[0082] (Overview of Prediction System 1) The physical properties used in the inventory information of the mixing ratio calculation system 5 described above may be obtained by actually measuring the resin composition. However, the labor involved in measuring the physical properties can be substantial. In this disclosure, predicted values ​​of the physical properties of the resin composition, obtained from the IR spectrum of the resin composition, may be used as the physical properties of the resin composition. Below, an example configuration of a prediction system 1 capable of predicting the physical properties of a resin composition will be described.

[0083] <Material Property Prediction Scheme> The following describes a property prediction scheme that can easily predict at least one of the physical properties of a resin composition containing a resin and a compound, or the characteristics of a compound contained in the resin composition. The property prediction scheme can be implemented using the prediction model 80 (see Figure 7A) described later.

[0084] The prediction model 80 is capable of calculating predicted values ​​for the physical properties of a sample or the properties of a compound. In other words, the predicted values ​​for the physical properties of a sample or the properties of a compound are numerical values ​​that represent the predicted results for the physical properties of a sample or the properties of a compound. The compound is other materials blended into the resin. In this disclosure, the resin may include polyamide (PA), etc. The compound may include inorganic fillers such as glass fibers (GF). The compound may include colorants such as carbon black (CB). The resin composition may be a mixture obtained by kneading the compound into the resin.

[0085] The physical properties of the sample predicted by prediction system 1 may include, for example, mechanical properties such as tensile strength or flexural strength of the resin composition, fluidity, or molecular weight. The physical properties of the sample are expressed numerically. The fluidity of the resin composition is expressed as the ease with which the resin composition flows when it is in a molten state, such as MFR (melt flow rate), MVR (melt volume rate), or shear viscosity, or as specific viscosity, relative viscosity, intrinsic viscosity, or reduced viscosity obtained from the solution viscosity of the resin composition. The ease with which the resin composition flows may also be expressed numerically. The characteristics of the compound predicted by prediction system 1 may include, for example, the glass fiber content or fiber length if the compound is glass fiber.

[0086] The accuracy of the predicted values ​​of the physical properties of the sample or the characteristics of the compound, and the accuracy of the prediction model 80 (see Figure 7A or Figure 7B), are evaluated using known evaluation indices such as the coefficient of determination (R²), mean squared error (MSE), root mean squared error (RMSE), mean absolute error (MAE), or mean absolute percent error (MAPE), based on the measured and predicted values ​​of the physical properties of the sample or the characteristics of the compound. Specifically, for the coefficient of determination, a value closer to 1 indicates higher prediction accuracy of the physical properties of the sample or the characteristics of the compound. For mean squared error (MSE), root mean squared error (RMSE), MAE, and mean absolute percent error (MAPE), a value closer to 0 indicates higher prediction accuracy. Furthermore, if cross-validation is performed when constructing the prediction model 80, the prediction accuracy may be evaluated using the average value of the evaluation indices obtained for each fold. The measured values ​​of the physical properties of the sample or the characteristics of the compound are numerical values ​​obtained by actually measuring the physical properties of the sample or the characteristics of the compound. As measured values ​​of the physical properties of the sample, mechanical properties such as tensile strength or flexural strength of the resin composition, fluidity, or molecular weight may be obtained by actually measuring them. As measured values ​​of the properties of the compound contained in the sample, for example, if the compound is glass fiber, values ​​obtained by actually measuring the glass fiber content or fiber length may be obtained.

[0087] (Example configuration of prediction system 1) As shown in Figure 6, a prediction system 1 according to one embodiment of the present disclosure comprises a prediction device 10 and a measuring device 20. The prediction system 1 uses a resin composition containing a resin and a compound as a sample to predict the physical properties of the sample or the properties of the compound. In the prediction system 1, the measuring device 20 measures the IR spectrum of the sample. The IR spectrum is represented as multidimensional data, which is a collection of light intensities for each wavelength included in at least a portion of the infrared light wavelength band. The prediction device 10 predicts at least one of the physical properties of the sample or the properties of the compound based on the measurement results of the sample's IR spectrum. In other words, the prediction device 10 calculates a predicted value for at least one of the physical properties of the sample or the properties of the compound based on the multidimensional data representing the IR spectrum.

[0088] <Prediction device 10> The prediction device 10 comprises a data acquisition unit 11, a data processing unit 12, a storage unit 15, and an output unit 16.

[0089] The data acquisition unit 11 acquires the IR spectrum of the sample measured by the measuring device 20 from the measuring device 20. The data acquisition unit 11 is not limited to the measuring device 20 and may acquire information or data from other external devices. For example, the data acquisition unit 11 may acquire measured values ​​of the physical properties of the sample.

[0090] The data acquisition unit 11 may be equipped with a communication interface for connecting to the measuring device 20 or external devices via wired or wireless means. The communication interface may be configured to communicate with the measuring device 20 or external devices based on a communication standard such as RS-232C or RS-485. The communication interface may be configured to communicate with the measuring device 20 based on a communication standard such as Bluetooth®. The communication interface may be configured to communicate with the measuring device 20 or external devices via a network such as a LAN (Local Area Network).

[0091] The data processing unit 12 predicts the physical properties of the sample based on the sample's IR spectrum. The data processing unit 12 may include a spectrum processing unit 13 that processes the sample's IR spectrum. The data processing unit 12 may also include a prediction unit 14 that predicts the physical properties of the sample based on the sample's IR spectrum.

[0092] The data processing unit 12 may be configured to include at least one processor, such as a CPU (Central Processing Unit) or a GPU (Graphics Processing Unit). The data processing unit 12 may be composed of one processor or multiple processors. The processors constituting the data processing unit 12 may realize the functions of the data processing unit 12 by reading and executing a program stored in the memory unit 15, which will be described later. The data processing unit 12 may be configured to include one or more dedicated circuits. The dedicated circuits may include, for example, an FPGA (Field-Programmable Gate Array) or an ASIC (Application Specific Integrated Circuit). The data processing unit 12 may be configured in combination of a processor and dedicated circuits.

[0093] The storage unit 15 stores various information or data processed by the prediction device 10. The storage unit 15 may store, for example, a program executed in the data processing unit 12, or data or processing results used in the processing executed in the data processing unit 12. The storage unit 15 may also function as the work memory of the data processing unit 12. The storage unit 15 may include, but is not limited to, semiconductor memory. For example, the storage unit 15 may be configured as the internal memory of a processor used as the data processing unit 12, or as a hard disk drive (HDD) accessible from the data processing unit 12. The storage unit 15 may be configured as a non-temporary readable medium. The storage unit 15 may be configured integrally with the data processing unit 12, or as a separate unit from the data processing unit 12.

[0094] The output unit 16 outputs the predicted results of the physical properties of the sample. The output unit 16 may include a display device. The display device may include, for example, an LCD (Liquid Crystal Display), an organic EL (Electro-Luminescence) display or an inorganic EL display, or a PDP (Plasma Display Panel). The display device is not limited to these displays and may include various other types of displays. The output unit 16 may be configured to output the predicted results of the physical properties of the sample to an external device.

[0095] The prediction device 10 may predict the physical properties of a sample using a prediction model 80, as illustrated in Figure 7A. The prediction model 80 is implemented by the prediction unit 14. The prediction unit 14 inputs the IR spectrum of the sample acquired by the data acquisition unit 11 to the prediction model 80. Based on the input IR spectrum of the sample, the prediction model 80 outputs the target characteristics of the sample. The target characteristics are the physical properties of the sample that correspond to the target properties for prediction. In other words, the prediction model 80 is configured to output the target characteristics of the sample when the IR spectrum of the sample is input.

[0096] The prediction device 10 may predict the physical properties of a sample using a prediction model 80 that includes a first prediction model 81 and a second prediction model 82, as illustrated in Figure 7B.

[0097] The prediction unit 14 inputs the IR spectrum acquired by the data acquisition unit 11 to the first prediction model 81. Based on the input IR spectrum, the first prediction model 81 outputs the transit characteristics of the sample. Transit characteristics are intermediate characteristics predicted in order to predict the target characteristics. For example, the content or size of compounding compounds blended into the resin composition may be predicted as transit characteristics. In other words, the first prediction model 81 is configured to output the transit characteristics of the sample when the IR spectrum of the sample is input.

[0098] The prediction unit 14 inputs the transit characteristics to the second prediction model 82. The second prediction model 82 outputs the target characteristics of the sample based on the input transit characteristics. In other words, the second prediction model 82 is configured to output the target characteristics of the sample when the transit characteristics of the sample are input.

[0099] The prediction model 80, which includes the first prediction model 81 and the second prediction model 82, is configured to output the target characteristics of a sample when the IR spectrum of the sample is input.

[0100] The prediction model 80 may be a trained model generated by performing training using data that associates the IR spectrum of a sample with the target characteristics of the sample as training data. The prediction model 80 may be a trained model generated by performing training using the IR spectrum of a sample as training data. The prediction model 80 may also be a regression analysis model obtained by performing regression analysis with the IR spectrum of the sample as the explanatory variable and the target characteristics of the sample as the dependent variable. The prediction model 80 may also be a calibration curve model generated by a chemometrics method.

[0101] The first prediction model 81 may be a trained model generated by performing training using data that associates the IR spectrum of a sample with the transit characteristics of the sample as training data. The first prediction model 81 may be a trained model generated by performing training using the IR spectrum of a sample as training data. The first prediction model 81 may be a regression analysis model obtained by performing regression analysis with the IR spectrum of a sample as the explanatory variable and the transit characteristics of the sample as the dependent variable. The first prediction model 81 may be a calibration curve model generated by chemometric methods.

[0102] The second prediction model 82 may be a trained model generated by performing training using data that associates the transit characteristics of the sample with the target characteristics of the sample as training data. The second prediction model 82 may be a trained model generated by performing training using the transit characteristics of the sample as training data. The second prediction model 82 may be a regression analysis model obtained by performing regression analysis with the transit characteristics of the sample as explanatory variables and the target characteristics of the sample as the dependent variable. The second prediction model 82 may be a calibration curve model generated by chemometric methods.

[0103] The prediction model 80, the first prediction model 81, or the second prediction model 82 are not limited to the trained model, regression analysis model, or calibration curve model described above, but may be configured as various types of models.

[0104] As described above, when predicting the physical properties of a sample using the prediction model 80, the more explanatory variable parameters that can be input into the prediction model 80, the more complex variations in the physical properties can be handled. In other words, even when the spectral variations are complex and cannot be predicted by regression analysis using one or a few explanatory variables, a prediction model 80 based on a large number of explanatory variables can be constructed. As a result of constructing the prediction model 80 for more complex spectral variations, the prediction accuracy improves. When the IR spectrum of a sample is input into the prediction model 80 as an explanatory variable, the parameters of the explanatory variable are the light intensity of each wavelength in the IR spectrum. The higher the resolution of the measured IR spectrum, the more explanatory variable parameters can be added, and the more complex the spectral variations that can be used to construct the prediction model 80.

[0105] When constructing a predictive model 80, using a large number of explanatory variables results in a vast number of possible combinations of hyperparameters for the predictive model 80 or parameters used in spectral preprocessing. This can make it difficult to set optimal parameters. Furthermore, the likelihood of the predictive model 80 overfitting increases, potentially reducing prediction accuracy for unknown data. In this case, by using machine learning techniques, it is possible to find optimal parameters while preventing overfitting of the predictive model 80, even when using a large number of explanatory variables, thereby improving prediction accuracy.

[0106] The prediction device 10 may be implemented using a cloud service or in an on-premises environment. Alternatively, the prediction device 10 may be implemented in a hybrid form combining a cloud service and an on-premises environment. If the prediction device 10 is implemented using a cloud service, for example, the output unit 16 may be configured as a display terminal for user use.

[0107] <Measuring device 20> The measuring device 20 is assumed to be an infrared spectrometer capable of measuring IR spectra. The measuring device 20 may be configured to measure IR spectra in a wavelength band that includes at least one of near-infrared (NIR), mid-infrared (MIR), or far-infrared (FIR), or in a wavelength band that includes all of them.

[0108] The measuring device 20 may be configured to measure the IR spectrum using the ATR (Attenuated Total Reflection) method. When measuring the IR spectrum using the ATR method, as shown in Figure 8A, the measuring device 20 includes a light source 21, a detector 22, and a prism 23. The measuring device 20 is configured so that the prism 23 can be placed in close contact with the sample 30. The measuring device 20 emits light containing infrared light from the light source 21 and causes the light to enter the prism 23. The light that enters the prism 23 is also called incident light. The light that enters the prism 23 undergoes total internal reflection at the interface where the prism 23 and the sample 30 are in close contact. The light that undergoes total internal reflection is also called reflected light. The reflected light passes through the prism 23 and enters the detector 22. The detector 22 detects the reflected light. The measuring device 20 measures the difference between the spectrum of the incident light emitted from the light source 21 and the spectrum of the reflected light detected by the detector 22 as the reflected IR spectrum of the sample 30.

[0109] When measuring the IR spectrum of sample 30 using the ATR method, sample 30 may be processed to adhere closely to the prism 23 in a solid state and placed in the measuring device 20. Sample 30 may be processed such that the compound is present in the range (evanescent wave leakage depth) where totally reflected light seeps into sample 30 on the surface in contact with the prism 23. Sample 30 may be processed so that the compound is exposed on the surface in contact with the prism 23.

[0110] Sample 30 may be, for example, a solid pellet. If sample 30 is a solid pellet, the solid pellet may be pressed before being measured by the ATR method. Pressing the solid pellet makes the pressed surface smooth. Pressing the solid pellet also improves the adhesion of the ATR to the prism 23. Pressing the solid pellet also makes the compound contained in the solid pellet more easily exposed on the surface that is in contact with the prism 23. Pressing the solid pellet also makes the compound contained in the solid pellet more easily present in the evanescent wave seepage depth on the surface that is in contact with the prism 23. The solid pellet may be pressed in such a way that the compound is exposed on the surface that is in contact with the prism 23 or present in the evanescent wave seepage depth.

[0111] In the process of pressing solid pellets, the pressing conditions are not particularly limited. From the viewpoint of minimizing the influence of heat on the composition, the pressing temperature may be set to 300°C or less, preferably 200°C or less, more preferably 100°C or less, and even more preferably 50°C or less. Furthermore, when pressing pellets that have a cut surface during production, such as pellets pelletized by strand cut or underwater cut, the composition is adequately exposed to the pressing surface by pressing so that the cut surface and the pressing surface of the press machine are parallel. Adequate exposure of the composition to the pressing surface improves the accuracy of predicting physical properties.

[0112] The sample 30 may be placed in the measuring device 20 in a molten state. When the sample 30 is in a molten state, it is more likely to adhere closely to the prism 23. Also, when the sample 30 is in a molten state, the compound is more likely to be exposed on the surface of the sample 30 that is in contact with the prism 23. By configuring the measuring device 20 to accommodate a molten sample 30, the measuring device 20 can be used in line to measure the resin composition being manufactured as a sample 30 during the resin composition manufacturing process.

[0113] The measuring device 20 may be configured to measure the IR spectrum using the transmission method. When measuring the IR spectrum using the transmission method, the measuring device 20 includes a light source 21 and a detector 22, as shown in Figure 8B. The measuring device 20 emits light containing infrared light from the light source 21 and causes the light to incident on the sample 30. The light incident on the sample 30 is also called incident light. The light incident on the sample 30 is transmitted through the sample 30. The light that has been transmitted through the sample 30 is also called transmitted light. The transmitted light is incident on the detector 22. The detector 22 detects the transmitted light. The measuring device 20 measures the difference between the spectrum of the incident light emitted from the light source 21 and the spectrum of the transmitted light detected by the detector 22 as the transmitted IR spectrum of the sample 30.

[0114] The shape of the infrared spectrum of the resin composition is not particularly limited, but it is preferable that, when expressed in terms of absorbance, the maximum absorbance is 0.3 or more and 2.0 or less, the minimum absorbance is 1.0 or less, and the difference between the maximum and minimum absorbance is 0.03 or more. More preferably, the shape of the infrared spectrum of the resin composition may be one that satisfies the conditions that the maximum absorbance is 0.5 or more and 1.7 or less, the minimum absorbance is 0.5 or less, and the difference between the maximum and minimum absorbance is 0.05 or more.

[0115] Preferably, the maximum or minimum absorbance of the infrared spectrum is determined based on the raw data of the measured infrared spectrum.

[0116] By having a spectral shape that satisfies the above-mentioned conditions, the effects of noise or measurement variability are reduced. This reduction in the effects of noise or measurement variability improves prediction accuracy.

[0117] When measuring the IR spectrum of sample 30 using the transmission method, sample 30 may be processed into a shape that easily transmits light, such as a sheet or thin plate. Sample 30 may be processed so that its thickness in the direction of light transmission is, for example, 1 mm or less, preferably 0.5 mm or less, and more preferably 0.2 mm or less. By processing sample 30 so that its thickness in the direction of light transmission is within the above range, the shape of the spectrum becomes appropriate. An appropriate spectral shape improves prediction accuracy.

[0118] (Example of operation of prediction system 1) The following provides a detailed explanation of how prediction system 1 works.

[0119] <Model generation> In prediction system 1, the prediction device 10 may predict at least one of the physical properties of a sample or the properties of a compound by inputting the IR spectrum of the sample into the prediction model 80 and obtaining a predicted value of at least one of the physical properties of the sample or the properties of the compound output from the prediction model 80, as described above. The prediction device 10 may use a pre-prepared model obtained from an external device as the prediction model 80. The prediction device 10 may also generate and use the prediction model 80 itself. An example of operation when the prediction device 10 generates the prediction model 80 will be described below. The target of prediction system 1 is at least one of the physical properties of the sample or the properties of a compound contained in the sample. In the following description, when it is stated that the physical properties of a sample are to be predicted, it also includes predicting the properties of a compound contained in the sample.

[0120] The data processing unit 12 of the prediction device 10 may execute a model generation method, including the steps of the flowchart illustrated in Figure 9, to generate a prediction model 80. The model generation method may be implemented as a model generation program to be executed by the processor constituting the data processing unit 12. The model generation program may be stored on a non-temporary computer-readable medium.

[0121] The data processing unit 12 acquires the IR spectrum of the sample (step S31). The IR spectrum of the sample is measured by the measuring device 20 and acquired from the measuring device 20 by the data acquisition unit 11.

[0122] The IR spectrum of the sample resin composition preferably includes a wavenumber range in which the absorbance changes in response to changes in the properties of the compound contained in the resin composition. For example, the IR spectrum of the resin composition may include a wavenumber range in which the absorption peak of the compound contained in the resin composition appears, or a wavenumber range in which the height or slope of the baseline absorbance changes in response to changes in the content or size of the compound contained in the resin composition. The IR spectrum of the resin composition may also include a wavenumber range in which the absorption peak originating from the resin component contained in the resin composition changes in response to changes in the properties of the compound contained in the resin composition. Furthermore, the IR spectrum of the resin composition may also include a wavenumber range in which the absorption peak originating from the resin component does not change in response to changes in the properties of the compound contained in the resin composition. In this way, by using an IR spectrum that includes a wavenumber region containing absorbance originating from the resin component in addition to a wavenumber range in which the absorbance changes in response to changes in the properties of the compound contained in the resin composition, the accuracy of predicting the physical properties of the resin composition and the properties of the compound is improved.

[0123] The data processing unit 12 acquires measured values ​​of the physical properties of the sample (step S32). The measured values ​​of the physical properties of the sample are measured by a device for measuring the physical properties of the sample. The device for measuring the physical properties of the sample may include, for example, a device for measuring mechanical properties such as the tensile strength or flexural strength of the sample, a device for measuring the fluidity of the sample, or a device for measuring the molecular weight of the sample.

[0124] The prediction system 1 may include a device for measuring the physical properties of a sample, designated as a measuring device 20. If the prediction system 1 includes a device for measuring the physical properties of a sample, the data processing unit 12 can acquire the measured values ​​of the physical properties of the sample within the prediction system 1.

[0125] The prediction system 1 does not necessarily need to have a device for measuring the physical properties of the sample. In other words, the physical properties of the sample may be measured by an external device. When the physical properties of the sample are measured by an external device, the data acquisition unit 11 of the prediction device 10 may be configured to communicate with the external device so that the physical properties of the sample can be acquired from the external device.

[0126] The data processing unit 12 associates the sample's IR spectrum with the measured physical properties of the sample (step S33). The data processing unit 12 generates a predictive model 80 by performing learning using the data, which associates the sample's IR spectrum with the measured physical properties of the sample, as training data (step S34). The data processing unit 12 may store the generated predictive model 80 in the storage unit 15. After executing the procedure in step S34, the data processing unit 12 terminates the execution of the procedure in the flowchart of Figure 9.

[0127] The method for generating the prediction model 80 from the above training data is not particularly limited, but in order to describe the relationship between the multidimensional data of the spectrum and the physical properties and to obtain higher prediction accuracy, it is preferable to generate the prediction model 80 using machine learning techniques. As the prediction model 80 generated using machine learning techniques, known models such as linear regression models or nonlinear regression models can be used. Examples of linear regression models include multiple regression analysis, PLS (Partial Least Squares), PCR (Principal Component Regression), ridge regression, Lasso regression, or elastic networks. In addition, tree models such as regression trees, random forests, or boosting trees, kernel methods such as support vector regression, kernel ridge regression, or Gaussian process regression, or neural networks such as recurrent networks or Long Short-Term Memory can also be used as the prediction model 80.

[0128] The acquired spectra may be processed by performing spectral processing such as smoothing, baseline correction, normalization, SNV (Standard Normal Variate) transformation, or differential processing before being used for model generation. Furthermore, when generating models using machine learning, the parameters for the spectral processing described above may be determined to improve the predictive accuracy of the generated model.

[0129] When generating a prediction model 80 using machine learning techniques, it is preferable that the IR spectra used as training data include a wavenumber range in which the absorbance changes in accordance with changes in the properties of the compound contained in the resin composition, which is the sample from which the IR spectra were measured. Furthermore, it is preferable that the degree of change in absorbance in accordance with changes in the properties of the compound is within a specific range. Specifically, it is preferable that the IR spectra used as training data include a wavenumber range that satisfies the condition that the difference between the absorbance of the IR spectrum corresponding to the resin composition where the compound's properties, such as the content or size of the compound, are at their minimum and the absorbance of the IR spectrum corresponding to the resin composition where the compound's properties are at their maximum is 0.05 or more and 2.0 or less. More preferably, the IR spectra used as training data may include a wavenumber range that satisfies the condition that the difference in absorbance of the IR spectra is 0.1 or more and 1.5 or less. By including a wavenumber range in which the change in absorbance of the IR spectra used as training data satisfies the above conditions, the influence of noise or measurement variability is reduced. As a result, prediction accuracy is improved.

[0130] The wavenumber range in which absorbance changes in response to changes in the properties of the compound contained in the resin composition, as described above, may be identified based on the raw IR spectrum data obtained by measuring the resin composition. Furthermore, when performing machine learning using pre-processed IR spectra as training data to generate the prediction model 80, it is preferable that the wavenumber range in which absorbance changes in response to changes in the properties of the compound contained in the resin composition, as described above, is identified based on the IR spectrum after pre-processing.

[0131] In particular, the wavenumber region in which the height or slope of the baseline absorbance changes depending on the content or size of the compound is preferably identified based on the spectrum before baseline correction is performed in the IR spectrum pretreatment. Alternatively, the wavenumber region in which the height or slope of the baseline absorbance changes depending on the content or size of the compound may be identified based on the IR spectrum after pretreatment other than baseline correction has been performed.

[0132] The wavenumber region of the IR spectrum used as training data or training data for generating the prediction model 80 is preferably the same as the wavenumber region of the IR spectrum input to the prediction model 80 when predicting the physical properties of the resin composition or the properties of the compound contained in the resin composition using the prediction model 80.

[0133] The data processing unit 12 does not have to perform the procedure in step S32 for acquiring measured values ​​of the physical properties of the sample, or the procedure in step S33 for associating the IR spectrum of the sample with the measured values ​​of the physical properties of the sample. In this case, the data processing unit 12 may generate the predictive model 80 in step S34 by performing training using the IR spectrum of the sample as training data.

[0134] In step S32, the data processing unit 12 may further acquire measured values ​​of the sample's transit characteristics. In this case, the data processing unit 12 may generate a first predictive model 81 by associating the sample's IR spectrum with the measured values ​​of the sample's transit characteristics in step S33, and performing learning in step S34 using the data obtained by associating the sample's IR spectrum with the measured values ​​of the sample's transit characteristics as training data. Alternatively, the data processing unit 12 may generate a second predictive model 82 by associating the sample's transit characteristics with the measured values ​​of the sample's physical properties in step S33, and performing learning in step S34 using the data obtained by associating the sample's transit characteristics with the measured values ​​of the sample's physical properties as training data.

[0135] The data processing unit 12 may include a model generation unit. If the data processing unit 12 includes a model generation unit, the model generation unit may execute a model generation method.

[0136] <Sample Measurement> In prediction system 1, the prediction device 10 inputs the IR spectrum of the sample into the prediction model 80, as described above. The IR spectrum of the sample is measured by the measuring device 20. The prediction device 10 acquires the IR spectrum of the sample measured by the measuring device 20. Below, an example of a spectral measurement procedure performed by a user of prediction system 1 to measure the IR spectrum of a sample will be explained with reference to the flowchart in Figure 10.

[0137] The user processes the sample (step S41). Specifically, the user processes the sample in accordance with the method used to measure the sample's IR spectrum. When measuring the sample's IR spectrum using the ATR method, the user may process the sample by pressing it so that the compounding contained in the resin composition (the sample) is present at the seepage depth of the evanescent wave. When measuring the sample's IR spectrum using the ATR method, the user may process the resin composition (the sample) into a molten state. When measuring the sample's IR spectrum using the transmission method, the user may process the sample into a sheet or plate shape so that it transmits light easily. The user may process the sample so that the thickness in the light-transmitting direction of the sample is, for example, 1 mm or less.

[0138] The user sets the processed sample in the measuring device 20 (step S42). The user measures the IR spectrum of the sample using the measuring device 20 (step S43). The measuring device 20 outputs the measurement result of the sample's IR spectrum to the prediction device 10. The prediction device 10 acquires the sample's IR spectrum using the data acquisition unit 11. The prediction device 10 may output the acquired sample's IR spectrum to the data processing unit 12 or store it in the storage unit 15. After performing the procedure in step S43, the user finishes performing the procedure in the flowchart of Figure 10. The user may measure the IR spectrum for each of multiple samples by performing the procedures from steps S41 to S43.

[0139] The prediction system 1 may be configured to automatically perform at least some of the steps of the spectral measurement method. For example, the prediction system 1 may include means for processing the sample, or means for setting the sample in the measuring device 20. The measuring device 20 may automatically start the measurement when the sample is set.

[0140] <Prediction of the physical properties of the sample> In prediction system 1, the prediction device 10 may predict the physical properties of a sample by inputting the IR spectrum of the sample to the prediction model 80 and obtaining the predicted values ​​of the sample's physical properties output from the prediction model 80, as described above. Below, an example of the operation when the prediction device 10 predicts the physical properties of a sample using the prediction model 80 will be described.

[0141] The data processing unit 12 of the prediction device 10 may execute a prediction method, including the steps of the flowchart illustrated in Figure 11, to predict the physical properties of a sample. The prediction method may be implemented as a prediction program to be executed by the processor constituting the data processing unit 12. The prediction program may be stored on a non-temporary computer-readable medium.

[0142] The data processing unit 12 acquires the IR spectrum of the sample (step S51). The data processing unit 12 may acquire the IR spectrum of the sample from the measuring device 20 through the data acquisition unit 11, or it may acquire the IR spectrum of the sample that has been previously acquired by the data acquisition unit 11 and stored in the storage unit 15.

[0143] The data processing unit 12 inputs the acquired sample's IR spectrum into the prediction model 80 (step S52). The data processing unit 12 may process the sample's IR spectrum using the spectrum processing unit 13 so that it can be input into the prediction model 80. The data processing unit 12 inputs the sample's IR spectrum into the prediction unit 14, which implements the prediction model 80. The data processing unit 12 may input the IR spectrum processed by the spectrum processing unit 13 into the prediction unit 14, or it may input the acquired IR spectrum, which has not been processed by the spectrum processing unit 13, into the prediction unit 14.

[0144] The IR spectrum is input to the prediction unit 14 and may be processed by performing spectral processing such as smoothing, baseline correction, normalization, SNV transformation, or differentiation before being used by the prediction model 80. The selection of the spectral processing method or the spectral processing conditions are preferably the same as the spectral processing performed when generating the prediction model 80.

[0145] The data processing unit 12 acquires the predicted value of the physical properties of the sample from the prediction model 80 (step S53). Specifically, the prediction model 80 outputs the predicted value of the physical properties corresponding to the input IR spectrum. The data processing unit 12 acquires the predicted value of the physical properties of the sample output from the prediction model 80. The data processing unit 12 may output the predicted value of the physical properties of the sample to the output unit 16. The output unit 16 may display the predicted value of the physical properties of the sample so that the user can recognize it. The data processing unit 12 may store the predicted value of the physical properties of the sample in the storage unit 15. The output unit 16 may display the predicted value of the physical properties of the sample stored in the storage unit 15 in response to an inquiry operation from the user. After executing the procedure of step S53, the data processing unit 12 ends the execution of the procedure of the flowchart in FIG. 11. The data processing unit 12 may execute the procedures from step S51 to S53 for each of the plurality of samples to predict the physical properties of each sample.

[0146] <Evaluation of Sample Quality> The prediction device 10 may evaluate the quality of the resin composition as a sample based on the prediction result of the physical properties of the resin composition as a sample, and output the evaluation result of the quality. That is, the prediction device 10 may execute a quality evaluation method including a procedure for evaluating the quality of the sample.

[0147] (Example) Hereinafter, a specific example will be described in which the prediction model 80 according to the present disclosure predicts the physical properties of a sample, calculates the mixing ratio with any other sample so that the physical properties fall within an arbitrary numerical range based on the predicted physical property values, and obtains a mixture of resin compositions.

[0148] <Prediction of Physical Properties Based on IR Spectrum in the Band of MIR Measured by ATR Method> In the prediction system 1, the measuring device 20 may measure the IR spectrum of a sample, which is a press-processed solid pellet, by the ATR method. The prediction device 10 may predict at least one of the physical properties of the resin composition as a sample or the characteristics of the formulation contained in the resin composition as a sample based on the IR spectrum of the sample, which is a solid pellet, measured by the ATR method.

[0149] As one example, the sample is assumed to be a resin composition in which glass fibers are compounded with polyamide. Both polyamide and glass fibers have absorption peaks in the mid-infrared (MIR) band. Therefore, the data processing unit 12 of the prediction device 10 acquires the IR spectrum of the sample in the mid-infrared (MIR) band, as illustrated in Figure 12. The horizontal axis of the graph in Figure 12 represents wavenumber. The vertical axis of the graph in Figure 12 represents absorbance at each wavenumber. The data processing unit 12 may also extract the MIR band using the spectrum processing unit 13.

[0150] The IR spectrum illustrated in Figure 12 was measured under the following conditions.

[0151] The samples for which IR spectra were measured were prepared as follows. First, a twin-screw extruder (Toshiba Machine TEM26SS) was set to a barrel temperature of 280°C, a screw rotation speed of 200-400 rpm, and a discharge rate of 20 kg / h. Melt-mixing of polyamide 66 (Asahi Kasei Corporation 1402S), glass fiber (Nippon Electric Glass ECS3T-275H), and carbon black was performed, and strand cutting was carried out to obtain pellets. The amount of carbon black added was 0.08% by weight. As shown in Table 3 below, 13 different pellet samples A-1 to A-13 were prepared by varying the glass fiber (GF) content and the screw rotation speed during mixing in 13 different ways.

[0152] [Table 3]

[0153] As described above, the 13 types of pellets obtained were pressed at room temperature using a "Quick Press" press manufactured by Chromat Science Co., Ltd. During the pressing process, the pellets were pressed so that the strand cut surface and the press surface of the press machine were parallel, in order to properly align the glass fibers on the press surface.

[0154] The IR spectra of samples A-1 to A-13, obtained by pressing pellets, were measured using the ATR method with an infrared spectrophotometer. Three pellets were used for each sample, and each pellet was measured once using the ATR method. In other words, three IR spectra were obtained for each of samples A-1 to A-13. The measurement conditions are shown below. Infrared spectrophotometer: Perkin-Elmer FT-IR Spectrum Two ATR Crystal: Diamond Number of reflections: 1 Measurement wave number: 450~3800cm -1 Wavenumber resolution: 2cm -1 Total number of times: 16 Detector: TGS detector

[0155] The graph of IR spectra shown in Figure 12 includes graphs of 13 IR spectra obtained for each of the 13 samples A-1 to A-13. Each IR spectrum was pre-treated so that the maximum absorbance was 1 and the minimum absorbance was 0. The IR spectra shown in Figure 12 are the pre-treated IR spectra.

[0156] The graphs of the IR spectra in Figure 12 show the IR spectra of 13 different resin compositions obtained by varying the glass fiber content and the screw rotation speed during mixing, as described above. Among the IR spectra, the wavenumbers are 900-1100 cm⁻¹. -1 Absorption peaks based on Si-O bonds in glass fibers appear in a wavenumber range of 900-1100 cm². -1 The absorbance of each spectrum differs within a certain wavenumber range. In particular, the absorbance tends to increase with increasing glass fiber content, and at 1000 cm⁻¹, the absorbance differs between the sample with 0% glass fiber content and the sample with 60% glass fiber content. -1A difference of 0.25 or more was observed in absorbance at this point. In other words, the difference in the combination of glass fiber content and the screw rotation speed during mixing was observed at wavenumbers of 900-1100 cm. -1 This affects the absorbance of each spectrum in a certain wavenumber range. Here, the screw rotation speed during mixing correlates with the fiber length of the glass fibers. Therefore, differences in the combination of glass fiber content and size affect the wavenumber range of 900-1100 cm. -1 This affects the absorbance of each spectrum in a certain wavenumber range. In other words, the IR spectrum illustrated in Figure 12 includes a wavenumber range in which the absorbance changes depending on changes in properties such as the content or size of the compound contained in the resin composition. Furthermore, the measured wavenumber range of IR spectra was used to construct the prediction model 80.

[0157] The data processing unit 12 inputs the IR spectrum of the MIR band of the resin composition as a sample into the prediction model 80 and obtains predicted values ​​of the physical properties of the resin composition or the properties of the compound contained in the resin composition output from the prediction model 80. The data processing unit 12 outputs the predicted values ​​of the physical properties of the resin composition or the properties of the compound contained in the resin composition to the output unit 16. The output unit 16 displays the predicted values ​​of the physical properties of the resin composition or the properties of the compound contained in the resin composition as prediction results, allowing the user to recognize the prediction results.

[0158] The prediction model 80 used in this embodiment was created using code written in the programming language Python. Using data combining measured physical properties of multiple samples and their IR spectra as training data, the prediction model 80 was created using PLS regression analysis, including a preprocessing step for the IR spectra. In this process, baseline correction, smoothing, normalization, or differentiation were set as the preprocessing steps for the IR spectra. The presence or absence of each preprocessing step, the various parameters of each preprocessing step, and the parameters in the PLS regression analysis performed after preprocessing were selected to optimize the prediction accuracy index in the cross-validation verification of the predictions.

[0159] In this embodiment, in order to evaluate the results of the data processing unit 12's prediction of the physical properties of the resin composition as a sample and the characteristics of the compound contained in the resin composition, the values ​​of the physical properties of the resin composition and the characteristics of the compound contained in the resin composition that were actually measured were compared with the predicted values ​​of the physical properties of the resin composition and the characteristics of the compound contained in the resin composition.

[0160] The physical properties of the resin composition were measured, including tensile strength, flexural modulus, and Charpy impact strength. Furthermore, the properties of the compound contained in the resin composition, specifically the glass fiber (GF) content (ash content), were measured. Tensile strength was measured according to ISO 527. Flexural modulus was measured according to ISO 178. Charpy impact strength was measured according to ISO 179. GF content (ash content) was measured according to ISO 3451-1.

[0161] The prediction accuracy was calculated based on a comparison between the actual measured values ​​of the physical properties of the resin composition and the properties of the compound contained in the resin composition, and the predicted values ​​of the physical properties of the resin composition and the properties of the compound contained in the resin composition. As an indicator of prediction accuracy, the average value of the coefficient of determination (R2CV) for each fold was calculated in the cross-validation performed when creating the prediction model 80. The prediction accuracy of the physical properties of the resin composition and the properties of the compound contained in the resin composition by the prediction model 80 described above is shown in Table 4 below.

[0162] [Table 4]

[0163] As shown in Table 4, the R2CV values ​​for each item predicted in this embodiment are 0.7 or higher. An R2CV value of 0.7 or higher is often considered to indicate good accuracy. Therefore, this embodiment demonstrates that the physical properties of the resin composition and the characteristics of the compounded materials contained in the resin composition can be predicted with high accuracy from the IR spectrum.

[0164] The prediction model 80 used in this embodiment may be a pre-trained model generated by performing training using training data obtained by extracting IR spectra in the MIR band.

[0165] The prediction model 80 may be divided into a first prediction model 81 and a second prediction model 82. In this embodiment, the first prediction model 81 may be configured to output the glass fiber content when an IR spectrum in the MIR band is input. The second prediction model 82 may be configured to output predicted values ​​of the physical properties of the resin composition when the glass fiber content is input.

[0166] As described above, by press-forming the sample to obtain an IR spectrum used to predict the physical properties of the sample, the accuracy of predicting the physical properties of the sample based solely on the IR spectrum can be improved.

[0167] In the measuring device 20, by processing the sample so that the formulation is exposed on the surface where the sample adheres to the prism 23 or the formulation exists within the penetration depth of the evanescent wave, the prediction accuracy of the physical properties of the sample can be further enhanced. In other words, by appropriately selecting the processing conditions for pressing the sample, the prediction accuracy of the physical properties of the sample can be further enhanced.

[0168] Further, by predicting the physical properties of the sample using the IR spectrum in the MIR band including the absorption peaks of polyamide and glass fiber, the prediction accuracy of the physical properties of the resin composition in which glass fiber is blended with polyamide can be enhanced.

[0169] Since the physical properties of the sample can be predicted with high accuracy only from the IR spectrum, there is no need to analyze the sample by other methods. As a result, the physical properties of the sample can be easily measured.

[0170] <IR spectrum measured by the transmission method> In the prediction system 1, the measuring device 20 may measure the IR spectrum of the sample processed into a sheet shape by the transmission method. The IR spectrum obtained by measuring a polyamide resin containing glass fiber (GF) as a sample by the transmission method is illustrated in FIG. 13. The horizontal axis of the graph in FIG. 13 represents the wave number. The vertical axis of the graph in FIG. 13 represents the absorbance at each wave number. Also, a sample with a high GF content and a sample with a low GF content were each measured. In the graph of FIG. 13, the IR spectrum with a large overall absorbance is the IR spectrum obtained by measuring a sample with a high GF content. Conversely, the IR spectrum with a small overall absorbance is the IR spectrum obtained by measuring a sample with a low GF content.

[0171] For the IR spectrum illustrated in FIG. 13, the wave number range where the wave number is 5000 to 5600 cm -1 and the wave number range where the wave number is 6000 to 7800 cm -1No clear absorption peaks are observed in any of the wavenumber ranges. However, within these wavenumber ranges, the baseline absorbance or slope changes depending on the GF content. The reason why the baseline absorbance or slope changes depending on the GF content is that as the GF content increases, the proportion of infrared light scattered by the glass fibers increases, the proportion of light transmitted through the sample decreases overall, and the baseline absorbance increases.

[0172] Furthermore, not only does the GF content differ, but the baseline absorbance or slope also changes depending on the fiber length of the glass fibers. This is because the rate at which infrared light is scattered by the glass fibers changes, causing the baseline absorbance to fluctuate. Also, the degree of infrared light scattering differs at each wavenumber depending on the fiber length of the glass fibers. As a result, the baseline slope changes depending on the fiber length of the glass fibers. Our investigation revealed that in the near-infrared (NIR) band, differences in glass fiber length tend to manifest as changes in baseline absorbance or slope. Therefore, it was discovered that the shape of the IR spectrum in wavenumber ranges where no clear absorption peaks are observed is important for predicting the physical properties of the sample or the characteristics of the compounds contained in the sample.

[0173] While embodiments of this disclosure have been described based on the drawings and examples, it should be noted that those skilled in the art can make various modifications or alterations based on this disclosure. Therefore, it should be noted that these modifications or alterations are included within the scope of this disclosure. For example, the functions included in each component or step can be rearranged in a logically consistent manner, and multiple components or steps can be combined into one or divided. Embodiments relating to this disclosure can also be realized as programs executed by a processor in the device or as storage media recording such programs. These should also be understood to be included within the scope of this disclosure. [Explanation of Symbols]

[0174] 1. Prediction System 5. Mixing Ratio Calculation System 10 Prediction device (11: data acquisition unit, 12: data processing unit, 13: spectrum analysis unit, 14: prediction unit, 15: storage unit, 16: output unit) 20 Measuring device (21: light source, 22: detection unit, 23: prism) 30 samples 40. Inventory management device (41: Acquisition unit, 42: Management unit, 43: Storage unit, 44: Output unit) 50 Mixing ratio calculation device (51: acquisition section, 52: management section, 53: storage section, 54: output section) 70 Mixing Ratio Calculation Model 80 Prediction Models (81: First Prediction Model, 82: Second Prediction Model)

Claims

1. A method for calculating the mixing ratio of a resin composition, which includes calculating the mixing ratio of a resin composition to bring the physical properties of a resin composition that are outside the specifications into specifications by mixing it with an arbitrary stock item, based on inventory information that associates the physical properties of the resin composition with the lot.

2. A method for calculating the mixing ratio of a resin composition according to claim 1, comprising calculating the mixing ratio using a mixing ratio calculation model.

3. A method for calculating the mixing ratio of a resin composition according to claim 1 or 2, comprising predicting at least one of the physical properties of the resin composition using a predictive model.

4. At least one of the physical properties of the resin composition is predicted as a transit characteristic of the resin composition from the infrared spectrum of the resin composition using the first prediction model as the prediction model. At least one of the physical properties of the resin composition is predicted as the target property of the resin composition from the prediction result of the transit properties of the resin composition, using the second prediction model as the prediction model. A method for calculating the mixing ratio of a resin composition according to claim 3, including the method described in claim 3.

5. A resin composition mixed based on a mixing ratio calculated by performing the method for calculating the mixing ratio of a resin composition according to claim 1 or 2.

6. A mixing method comprising mixing a resin composition based on a mixing ratio calculated by performing the method for calculating the mixing ratio of a resin composition described in claim 1 or 2.

Images