JP-2026074436-A - Characterization system and characterization method

JP2026074436AJP 2026074436 AJP2026074436 AJP 2026074436AJP-2026074436-A

Abstract

[Problem] To provide a property evaluation system that appropriately evaluates and manages the properties of resins and resin products, while reducing the time and burden required to build a system and learning model for evaluating properties such as viscosity, thermal properties, and molecular weight of resins and resin products. [Solution] A characterization system 3 predicts the viscosity, thermal properties, and molecular weight of a resin based on the fluorescence properties of the resin, comprising a computing device that executes a program and a storage device that stores the program, wherein the storage device stores a database of the fluorescence properties of each of a plurality of products mainly composed of a first resin, and measured values of any of the viscosity, thermal properties, and molecular weight corresponding to the fluorescence properties, and the computing device receives the fluorescence properties of the product to be predicted, mainly composed of the first resin, as input, predicts any of the viscosity, thermal properties, and molecular weight of the product to be predicted based on the fluorescence properties of the product to be predicted and the database, and outputs the viscosity, thermal properties, or molecular weight of the product to be predicted. [Selection Diagram] Figure 1A

Inventors

森俊介
堀込純
岩佐真行
中尾上歩
鈴木啓幸

Assignees

株式会社日立ハイテク

Dates

Publication Date: 20260507
Application Date: 20241021

Claims (18)

It comprises an arithmetic unit for executing a program and a memory device for storing the program, The memory device stores a database of the fluorescence properties of each of several products mainly composed of resin, and a database of measured values of viscosity, thermal properties, or molecular weight corresponding to the fluorescence properties, or a predictive model learned based on the database. The calculation device receives the fluorescence properties of the product to be predicted, which mainly consists of the resin, as input. Based on the fluorescence properties of the product to be predicted and the database or prediction model, the viscosity, thermal properties, or molecular weight of the product to be predicted is predicted. Outputting the viscosity, thermal properties, or molecular weight of the product to be predicted. A characteristic evaluation system characterized by the following:
In the characteristic evaluation system of claim 1, A characteristic evaluation system characterized in that the viscosity is intrinsic viscosity.
In the characteristic evaluation system of claim 1, A property evaluation system characterized in that the thermal property is one of the glass transition temperature, crystallization temperature, or melting point.
In the characteristic evaluation system of claim 1, A characterization system characterized in that the molecular weight is one of the weight-average molecular weight, number-average molecular weight, or polydispersity.
In the characteristic evaluation system of claim 1, A characterization system characterized in that the fluorescence characteristics are any of the three-dimensional data of excitation wavelength, fluorescence wavelength, and fluorescence intensity.
In the characteristic evaluation system of claim 1, A characterization system characterized in that the wavelength range of the excitation light for the fluorescence properties is 250 to 600 nm, and the wavelength range of the spectral intensities for fluorescence and scattering/absorption is 200 to 700 nm.
In the characteristic evaluation system of claim 6, A characterization system characterized in that the wavelength range of the excitation light for the fluorescence properties is 250 to 450 nm, and the wavelength range of the spectral intensities for fluorescence and scattering/absorption is 250 to 600 nm.
In the characteristic evaluation system of claim 1, The aforementioned computing device is a characterization system characterized by generating a predictive model based on a database of fluorescence properties for each of a plurality of products mainly composed of resin, and measured values of viscosity, thermal properties, or molecular weight corresponding to the fluorescence properties.
In the characteristic evaluation system of claim 8, The aforementioned prediction model is characterized by being created using one of the following methods: simple regression, multiple regression, Lasso regression, PLS regression, ridge regression, support vector regression, elastic network, decision tree, random forest, or gradient boosting decision tree.
In the characteristic evaluation system of claim 2, A characteristic evaluation system characterized in that the intrinsic viscosity range is 0.46 to 0.99 dL/g.
In the characteristic evaluation system of claim 1, A characterization system characterized in that the form of the product is one of a bottle, container, film, sheet, pellet, flake, or strand.
In the characteristic evaluation system of claim 1, The property evaluation system is characterized in that the resin is a polyester resin.
In the characteristic evaluation system of claim 1, The resin is characterized by having ester bonds in the property evaluation system.
In the characteristic evaluation system of claim 12, The property evaluation system is characterized in that the resin is one of the following: polyethylene terephthalate, polycarbonate, polybutylene terephthalate, polyethylene naphthalate, polylactic acid, or a copolymer synthetic resin of polycarbonate and acrylonitrile butadiene styrene.
In the characteristic evaluation system of claim 8, The storage device has a predictive model for each product, with the first resin as the main component. The calculation device receives the fluorescence properties and product information of the product to be predicted, which mainly consists of the resin, as input. Based on the aforementioned product information, select a predictive model to use for the prediction. Based on the selected prediction model and the fluorescence properties of the product to be predicted, the viscosity, thermal properties, and molecular weight of the product to be predicted are predicted. A characteristic evaluation system characterized by the following:
The characteristic evaluation system of claim 1 is installed in accordance with the transport position of the product to be predicted, The storage device stores a predictive model learned based on a plurality of products containing the same main component as the product transported to the transport position, The calculation device acquires the fluorescence properties of the product that has been transported to the transport position. Based on the fluorescence properties and the database, the viscosity, thermal properties, or molecular weight of the product transported to the transport position is predicted. A characteristic evaluation system characterized by the following:
It comprises an arithmetic unit for executing a program and a memory device for storing the program, The aforementioned computing device accepts as input a database of fluorescence properties for each of several products mainly composed of resin, and measured values of viscosity, thermal properties, or molecular weight corresponding to the fluorescence properties. A characterization system characterized by taking the aforementioned fluorescence properties as input, learning the viscosity, thermal properties, or molecular weight as output, and generating a predictive model.
A property evaluation method for predicting the viscosity and thermal properties of a resin based on its fluorescence properties, A fluorescence property measurement step for measuring the fluorescence properties of a resin, A viscosity measurement step for measuring the viscosity of the resin, A thermal properties measurement step for measuring the thermal properties of a resin, A molecular weight measurement step for measuring the molecular weight of the resin, A predictive model creation step involves creating a predictive model that takes the first fluorescence property measured from the first resin as input and outputs one of the first viscosity, first thermal property, or first molecular weight. A prediction calculation step that predicts either the second thermal properties, second thermal properties, or second molecular weight relating to the second resin based on the second fluorescence properties measured from the second resin and the prediction model, A method for evaluating characteristics, characterized by comprising the following:

Description

This invention relates to a characterization system and method for evaluating the properties of resin materials based on fluorescent fingerprints. With the shift from a linear economy to a circular economy, quality control of resin materials and products is also shifting from management only during manufacturing and use to lifecycle management that includes recycling. Specific examples of quality control for resin materials and products include molecular weight, viscosity (solution viscosity, intrinsic viscosity, etc.), phase transition, glass transition temperature, crystallization temperature, and melting temperature. However, when considering recycling, the quality control of viscosity, thermal properties, and molecular weight is particularly important. Viscosity is a property that contributes to molding and processing; thermal properties contribute to both mechanical properties such as shape and appearance, as well as optical properties; and molecular weight and its polydispersity (related to molecular weight distribution) are physical property indicators that show the state of degradation and the state of regeneration during recycling. These properties are generally evaluated by destructive analysis, but in recent years, non-destructive analysis for characterization has also become widespread. For example, the abstract of Patent Document 1 describes "a method for quality control of a resin composition, comprising a measurement step of irradiating a resin composition in which TiO2 particles are dispersed in a base material mainly composed of silicone rubber with a laser and measuring a Raman spectrum, and a determination step of determining the concentration of the TiO2 particles in the resin composition based on the intensity of the fluorescence spectrum in the Raman spectrum." Paragraph 0073 further explains, "Furthermore, the resin composition quality control method according to the above embodiment, the cable or tube quality control method, the determination device and inspection system used in the cable or tube quality control method, etc., can also be applied to material development using materials informatics (MI), which analyzes data using machine learning and artificial intelligence (AI)." Thus, in this document, the quality of the resin composition is non-destructively analyzed using artificial intelligence, etc., based on the intensity of the fluorescence spectrum. Japanese Patent Publication No. 2023-044083 Functional block diagram of the evaluation system in Example 1.Functional block diagram of the evaluation system for a modified example of Example 1.A functional block diagram illustrating an example of the spectrofluorometer in Example 1.An example of fluorescent fingerprint data for a resin material.Flowchart of the predictive model creation process using the evaluation system in Example 1.Flowchart of the evaluation process using the evaluation system of Example 1.An example of a modeling database for Example 1.An example of the results of the multiple regression analysis performed in Example 1.An example of a prediction model database for Example 1.An example of the analysis database for Example 1.An example of a characteristic prediction database for Example 1.Relationship between measured and predicted intrinsic viscosity in the first characteristic evaluation of Example 1.Relationship between measured and predicted glass transition temperatures in the second characteristic evaluation of Example 1.Relationship between the measured fluorescence intensity at a predetermined excitation wavelength and the measured glass transition temperature in the second characteristic evaluation of Example 1.Relationship between measured and predicted crystallization peak temperatures in the third characteristic evaluation of Example 1.Relationship between measured and predicted melting points in the third characteristic evaluation of Example 1.Relationship between measured and predicted glass transition temperatures in the fourth characteristic evaluation of Example 1.A graph showing the wavelengths of the spectral intensity measured in the fifth characteristic evaluation of Example 1.Relationship between measured and predicted intrinsic viscosity in the fifth characteristic evaluation of Example 1.A graph showing the wavelengths of the spectral intensity measured in the sixth characteristic evaluation of Example 1.Relationship between measured and predicted intrinsic viscosity in the sixth characteristic evaluation of Example 1.Relationship between measured and predicted weight-average molecular weight in the seventh characteristic evaluation of Example 1.Relationship between measured and predicted values of the number-average molecular weight in the seventh characteristic evaluation of Example 1.Relationship between measured and predicted values of polydispersity in the seventh characteristic evaluation of Example 1.An example of the display image in the display unit of Example 2.An example of the