WO2024190332A1 - Appareil de prédiction de valeur de propriété physique de résine et procédé de prédiction de valeur de propriété physique de résine - Google Patents

Appareil de prédiction de valeur de propriété physique de résine et procédé de prédiction de valeur de propriété physique de résine Download PDF

Info

Publication number
WO2024190332A1
WO2024190332A1 PCT/JP2024/006175 JP2024006175W WO2024190332A1 WO 2024190332 A1 WO2024190332 A1 WO 2024190332A1 JP 2024006175 W JP2024006175 W JP 2024006175W WO 2024190332 A1 WO2024190332 A1 WO 2024190332A1
Authority
WO
WIPO (PCT)
Prior art keywords
resin
physical property
property value
process data
time
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Ceased
Application number
PCT/JP2024/006175
Other languages
English (en)
Japanese (ja)
Inventor
史也 浜田
一利 寺内
俊則 山地
弘昌 金子
柊斗 山影
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
DIC Corp
Meiji University
Original Assignee
DIC Corp
Meiji University
Dainippon Ink and Chemicals Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by DIC Corp, Meiji University, Dainippon Ink and Chemicals Co Ltd filed Critical DIC Corp
Publication of WO2024190332A1 publication Critical patent/WO2024190332A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B23/00—Testing or monitoring of control systems or parts thereof
    • G05B23/02—Electric testing or monitoring
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C60/00—Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation

Definitions

  • the present invention relates to a resin physical property value prediction device and a resin physical property value prediction method.
  • Patent Document 1 there was a technique for estimating the evaluation value of product quality in a multi-item batch production process in which a variety of products are produced using similar processes using specified equipment (see, for example, Patent Document 1).
  • process variables such as temperature, pressure, tensile speed, oxygen concentration, pH, and humidity are used as explanatory variables, and objective variables such as pressure resistance, thickness, component ratio of the substance, temperature, brightness, viscosity, and molecular weight are estimated.
  • the present invention aims to provide a resin property value prediction device and a resin property value prediction method that can accurately predict resin property values when predicting resin property values using process data whose values change over time.
  • One aspect of the present invention is a resin property value prediction device that predicts resin property values from information about reactant raw materials and time-series process data during a polymerization reaction in a process in which a resin polymerization reaction is performed in a batch-type reaction tank, and that includes a storage unit that stores information about the reactant raw materials and time-series process data during the polymerization reaction in a past production in association with the resin property values, which are corresponding measured data; a preprocessing unit that performs preprocessing on the time-series process data; a model creation unit that creates a learning model using the preprocessed data as training data; and a calculation unit that calculates resin property values using the learned model created by the model creation unit based on the raw material blend amounts and process data in the reaction currently in progress.
  • the preprocessing performed by the preprocessing unit is a process of extracting the mutual similarity between multiple pieces of the time-series process data as a feature.
  • the similarity is extracted using the DTW (Dynamic Time Warping) method.
  • the pre-processing unit extracts the similarity as a feature after smoothing the time-series process data.
  • the model creation unit selects a method that reduces the difference from the measured property value from among any of the following algorithms: Lasso regression (Least Absolute Shrinkage and Selection Operator), Ridge regression, or Partial Least Squares Regression (PLS regression, Partial Least Squares Regression), and updates the regression parameters within the algorithm during the selection process.
  • Lasso regression Least Absolute Shrinkage and Selection Operator
  • Ridge regression Ridge regression
  • Partial Least Squares Regression PLS regression, Partial Least Squares Regression
  • the resin composition for which the calculation unit calculates the resin property values is an acrylic resin
  • the resin property values calculated by the calculation unit include at least one of viscosity, amount of non-volatile components, and amount of residual monomer.
  • one aspect of the present invention is a resin property value prediction device according to any one of (1) to (6) above, further comprising a candidate generation unit that generates process data from the current time onward in a resin polymerization reaction that is in progress, and a future property value calculation unit that calculates resin property values by inputting the raw material blend amounts and the process data generated by the candidate generation unit into a learning model created by the model creation unit.
  • the future physical property value calculation unit calculates a plurality of resin physical property values
  • the candidate generation unit generates future process data candidates that are candidates for future process data based on the plurality of resin physical property values calculated by the future physical property value calculation unit
  • the device further includes a process data determination unit that determines whether or not a resin physical property value set as a target value can be realized based on the progress of the physical property values calculated based on the future process data candidates generated by the candidate generation unit.
  • one aspect of the present invention is a method for predicting resin physical property values from information on reaction raw materials and time-series process data during a polymerization reaction in a process of performing a resin polymerization reaction in a batch-type reaction tank, the method comprising: a storage step for storing information on the reaction raw materials and time-series process data during the polymerization reaction in a past production in association with the resin physical property values, which are the corresponding measured data; a preprocessing step for preprocessing the time-series process data; a model creation step for creating a learning model using the preprocessed data as training data; and a calculation step for calculating the resin physical property values using the learned model created by the model creation step, based on the raw material blend amounts and process data in the reaction currently in progress.
  • the present invention provides a resin property value prediction device and a resin property value prediction method that can accurately predict resin property values when predicting resin property values using process data whose values change over time.
  • FIG. 2 is a functional configuration diagram showing an example of a functional configuration of a resin property value prediction device according to the first embodiment.
  • FIG. 4 is a diagram for explaining an example of sampling timing during a resin polymerization reaction according to the first embodiment.
  • FIG. 4 is a diagram illustrating an example of a change over time in time-series process data according to the first embodiment.
  • 6A and 6B are diagrams illustrating changes in a dropping speed when the smoothing process according to the first embodiment is applied to the dropping speed.
  • 4 is a diagram showing an example of conditions for acquiring teacher data in a learning stage of the resin property prediction device according to the first embodiment.
  • FIG. FIG. 4 is a first diagram showing a prediction result of the resin property value prediction device according to the first embodiment.
  • FIG. 4 is a second diagram showing the prediction result of the resin property value prediction device according to the first embodiment.
  • FIG. 11 is a functional configuration diagram showing an example of a functional configuration of a resin property value prediction device according to a second embodiment.
  • the resin physical property value prediction device and the resin physical property value prediction method according to the present embodiment are used to measure the properties of a resin composition in a resin polymerization process.
  • the resin composition that is the subject of the resin property value prediction device and resin property value prediction method according to this embodiment broadly includes polymers such as homopolymers and copolymers.
  • the resin composition may be a thermoplastic resin or a thermosetting resin.
  • Thermoplastic resins are not particularly limited, but examples include polypropylene (PP), polyethylene (PE), ABS resin, polyvinyl chloride (PVC), acrylic resin, polyester resin, polystyrene resin (PS), urethane resin (PU), polyphenylene sulfide resin (PPS), etc.
  • the resin composition is an acrylic resin, and a solution containing the acrylic resin (hereinafter, the form of the solution containing the resin composition is also referred to as a "resin product”) is used to perform the prediction.
  • the resin property value prediction device and resin property value prediction method according to this embodiment are applied to a process in which a resin polymerization reaction is performed in a batch-type reaction tank.
  • the resin property value prediction device and resin property value prediction method according to this embodiment predict resin property values from information on reaction raw materials and time-series process data during the polymerization reaction.
  • reaction raw materials used in a process in which a resin polymerization reaction is performed in a batch-type reaction tank include monomers, initiators, and solvents.
  • Examples of time-series process data during the polymerization reaction include reaction liquid temperature, heat medium temperature, raw material input amount, pressure, etc. It is preferable that the time-series process data during the polymerization reaction be data that can be easily obtained from manufacturing equipment.
  • the properties of the resin composition that are the subject of the resin property value prediction device and resin property value prediction method according to this embodiment may specifically be polymer property values.
  • Polymer property values broadly include the amount of non-volatile components (NV; Non-Volatile), viscosity (Gardner viscosity), and the amount of residual monomer (residual monomer concentration), and it is preferable that at least one of these is included.
  • the physical properties of the resin composition according to this embodiment are not limited to the above-mentioned example, and may be measured by a method other than the measurement method described below.
  • the method for measuring the amount of non-volatile components is explained below.
  • a sample of the resin composition is placed in an aluminum dish whose weight has been measured beforehand, and the weight is measured.
  • the resin composition is then diluted to 20 wt% (weight percent) by adding a dilution solvent containing toluene and methanol, and heated in an oven heated to 108°C for one hour to remove the volatile components.
  • the weight of the resin composition residue remaining on the aluminum dish after heating is measured, and the amount of non-volatile components is calculated from the difference with the weight before heating.
  • the viscosity measurement method is explained below. First, a certain amount of the sampled resin composition is added to a Gardner bubble viscosity tube and sealed. The temperature of the resin composition is adjusted by immersing the sealed viscosity tube in a water bath controlled at 25°C. After that, the bubble rise speed is compared to determine the closest standard viscosity sample, and the value is converted into a numerical value using a standard viscosity conversion table.
  • the sampled resin composition is diluted to 0.5 wt% using N,N-dimethylformamide to prepare a measurement sample.
  • the amount of monomer in the resin composition is measured using gas chromatography with an FID detector.
  • the amount of residual monomer is measured from the total amount of all monomers.
  • FIG. 1 a resin property prediction device 1 according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 9.
  • FIG. 1 a resin property prediction device 1 according to a first embodiment of the present invention will be described with reference to FIGS. 1 to 9.
  • FIG. 1 is a functional configuration diagram showing an example of the functional configuration of a resin property prediction device according to embodiment 1.
  • the resin property prediction device 1 comprises a storage unit 10, a preprocessing unit 20, a model creation unit 30, an in-progress data acquisition unit 40, and a calculation unit 50.
  • Each of these functional units is realized, for example, using electronic circuits.
  • each functional unit may include internal storage means such as a semiconductor memory or a magnetic hard disk device as necessary.
  • each function may be realized by a computer and software.
  • the resin property value prediction device and resin property value prediction method according to this embodiment predict resin property values using machine learning.
  • the configuration used in the learning stage and the configuration used in the inference stage are respectively explained.
  • the configuration used in the learning stage is explained.
  • a memory unit 10 In the learning stage, a memory unit 10, a preprocessing unit 20, and a model creation unit 30 are used.
  • the memory unit 10 stores performance data of data measured in the past.
  • Performance data is data in which explanatory variables X and corresponding objective variables Y are associated with each other.
  • the explanatory variables X are process data used in the production of the resin composition, and specifically, examples of the explanatory variables X include information on the reacting raw materials used in the past production and time-series process data during the polymerization reaction.
  • the time-series process data during the polymerization reaction may be the reaction liquid temperature, the heat transfer medium temperature, and the amount of raw materials input.
  • the objective variables Y are resin physical property values that are actual measurement data measured from samples sampled when the resin composition is produced using the corresponding process data, and specifically, examples of the non-volatile component amount, viscosity, and amount of residual monomer are given.
  • the preprocessing unit 20 performs preprocessing on the data input to the machine learning algorithm. Preprocessing is performed both in the learning stage and in the inference stage. In the learning stage, the preprocessing unit 20 may perform preprocessing on either the explanatory variable X or the target variable Y.
  • the preprocessing according to this embodiment is applied to at least the time-series process data contained in the explanatory variable X.
  • the preprocessing according to this embodiment also includes at least a synchronization process.
  • the time required for production of the time-series process data may differ for each batch process.
  • the sampling time, reaction time, raw material dripping time, etc. may also differ for each batch process.
  • the synchronization process synchronizes the time-series process data by performing dynamic time warping.
  • An example of the synchronization process may be a process of extracting similarity between multiple pieces of time-series process data as a feature.
  • the pre-processing performed by the pre-processing unit 20 may be a process of extracting mutual similarity between multiple pieces of time-series process data as a feature.
  • the extracted similarity is passed to the model creation unit 30.
  • the similarity may be between the pieces of data, or may be similarity with a predetermined sample data (golden sample).
  • the similarity extracted by the synchronization process may be extracted by dynamic time warping (DTW).
  • DTW dynamic time warping
  • Dynamic time warping is a method used to measure the similarity between time series data, and a detailed explanation of this method will be omitted.
  • the pre-processing according to this embodiment may also include a smoothing process.
  • the smoothing process is applied to the time-series process data included in the explanatory variable X among the data input to the machine learning algorithm.
  • the dripping speed of the raw material among the time-series process data is a value calculated based on a value read visually by an operator who measures the dripping amount every 5 minutes using a balance. Therefore, the dripping speed of the raw material is obtained manually by the operator and may include errors.
  • the reaction liquid temperature may include minute fluctuations.
  • the smoothing process is performed to reduce the influence on the measurement results based on such errors and fluctuations.
  • the smoothing process may be performed by calculating a weighted moving average, for example.
  • the smoothing process does not necessarily have to be performed. For example, it may be better to perform smoothing process on the non-volatile component amount and the residual monomer amount among the explanatory variables X, but it may be better not to perform smoothing process on the viscosity. Therefore, the pre-processing unit 20 may perform smoothing process on a predetermined explanatory variable X. When performing smoothing process, the pre-processing unit 20 may extract the similarity as a feature after smoothing process on the time-series process data.
  • the preprocessing according to this embodiment may also include logarithmic conversion processing.
  • Logarithmic conversion may be performed on the objective variable Y, for example. Depending on the type of objective variable Y, it may be better to perform logarithmic conversion.
  • performing logarithmic conversion processing on the resin physical property values may increase the prediction error of the resin physical property values. In other words, performing logarithmic conversion processing on the resin physical property values may or may not improve accuracy. Therefore, whether or not to apply logarithmic conversion processing may be determined depending on the type of physical property value.
  • the model creation unit 30 creates a learning model using the data that has been preprocessed on the data stored in the storage unit 10 as teacher data. In other words, the model creation unit 30 can also be said to create a learning model by supervised learning.
  • the model creation unit 30 may create a learning model using a method such as Lasso regression (Last Absolute Shrinkage and Selection Operator), Ridge regression, or Partial Least Squares Regression (PLS regression, Partial Least Squares Regression). It is preferable that the model creation unit 30 creates a learning model using a suitable method among Lasso regression, Ridge regression, and Partial Least Squares Regression. Specifically, the model creation unit 30 may select a method that reduces the difference from the measured physical property values from among any of the algorithms of Lasso regression, Ridge regression, and Partial Least Squares Regression, and update the regression parameters within the algorithm during the selection process.
  • Lasso regression Lasst Absolute Shrinkage and Selection Operator
  • Ridge regression or Partial Least Squares Regression
  • PLS regression Partial Least Squares Regression
  • the learning model created by the model creation unit 30 in the learning stage the ongoing data acquisition unit 40, and the calculation unit 50 are used.
  • the in-progress data acquisition unit 40 acquires process data of the currently in-progress manufacturing process during the inference stage.
  • Examples of the process data acquired by the in-progress data acquisition unit 40 include information on the reaction raw materials and time-series process data during the polymerization reaction.
  • the time-series process data during the polymerization reaction may include the reaction liquid temperature, the heat transfer medium temperature, and the amount of raw materials input.
  • the calculation unit 50 calculates the resin physical property values using the trained model created by the model creation unit 30 based on the amounts of raw materials mixed in the reaction currently in progress and the process data.
  • the resin composition for which the calculation unit 50 calculates the resin physical property values may specifically be an acrylic resin.
  • FIG. 2 is a diagram for explaining an example of sampling timing during the resin polymerization reaction according to embodiment 1.
  • FIG. 2 shows the change in temperature of the polymerization reaction tank over time, with the temperature of the polymerization reaction tank on the vertical axis and time on the horizontal axis.
  • the resin polymerization reaction is carried out using a polymerization reaction tank.
  • the polymerization reaction tank is, for example, a reaction vessel not shown in the figure, and the temperature of the sample stored in the reaction vessel is controlled by injecting steam or the like into the reaction vessel.
  • the temperature of the polymerization reaction tank refers to the temperature of the reaction vessel itself, or the temperature inside the reaction vessel.
  • the temperature of the sample stored in the reaction vessel may also be controlled by submerging the reaction vessel in a specified temperature-regulated oil bath.
  • the temperature inside the reaction vessel is increased.
  • the temperature reaches a predetermined reaction temperature T.
  • a dripping process is carried out in which the monomer and polymerization initiator are dripped. Since this polymerization reaction is an exothermic reaction, the temperature of the sample stored in the reaction vessel is maintained by adding cooling water or lowering the oil bath temperature as necessary. The dripping process may take, for example, five hours. The time required for the dripping process may differ for each batch. During the dripping process, for example, one sampling is carried out. In the example shown in the figure, the first sampling is carried out once during the dripping process.
  • the monomers used to produce the acrylic resin are not particularly limited, and any known monomers can be used.
  • Examples of monomers include acrylic acid, alkyl acrylate, methacrylic acid, and alkyl methacrylate.
  • the polymerization initiator used in producing the acrylic resin is not particularly limited, and any known initiator can be used.
  • examples of polymerization initiators that can be used include persulfates such as ammonium persulfate and potassium persulfate; hydrogen peroxide; azo compounds; and organic oxides such as (2-ethylhexanoyl)(tert-butyl)peroxide.
  • radicals are generated in the initiation reaction due to thermal decomposition of the polymerization initiator, etc.
  • the radicals generated in the above-mentioned initiation reaction react sequentially with monomers to generate polymer radicals, and the polymer radicals further react with monomers to generate acrylic resin as a resin composition.
  • a hold step is performed from time t13 to time t14, in which the temperature is maintained at a predetermined temperature.
  • This hold step is a period during which the remaining monomer and polymerization initiator are consumed.
  • the hold step may last, for example, four hours.
  • the time required for the hold step may vary from batch to batch.
  • multiple samplings are performed, for example, every hour. In the example shown in the figure, a total of six samplings are performed during the hold step. By performing multiple samplings during this hold step, the viscosity of the resin composition can be measured, for example.
  • a dilution and cooling step is carried out from time t14 to time t15, and the temperature inside the reaction vessel is cooled. Sampling after this dilution and cooling step is carried out to measure, for example, the NV value of the resin composition, thereby confirming whether the product meets the specifications.
  • FIG. 3 is a diagram showing an example of the change over time in the time series process data according to the first embodiment.
  • the horizontal axis represents time, and the change over time in each time series process data is shown.
  • the vertical axis represents temperature [°C] for the reaction liquid temperature, weight [g] for the cumulative amount of monomer-initiator mixture dripped, and weight per minute [g/min] for the amount of monomer-initiator mixture dripped.
  • Times t11 to t15 shown in the figure correspond to FIG. 2. That is, from time t11 to time t12 is the process of increasing the temperature inside the reaction vessel, from time t12 to time t13 is the dripping process, from time t13 to time t14 is the hold process, and from time t14 to time t14 is the dilution and cooling process.
  • the amount of monomer-initiator mixture dripped and the cumulative amount of monomer-initiator mixture dripped are time-series process data that change only in the dripping process.
  • the amount of monomer/initiator mixture dripped is the average value of the amount dripped every 5 minutes measured by the operator on a balance.
  • the drip amount is measured by the operator visually reading the scale. Therefore, the amount of monomer/initiator mixture dripped may have a time error due to the measurement being taken every 5 minutes, and a measurement error due to the operator visually reading the scale. Therefore, in this embodiment, a smoothing process is performed by performing preprocessing using the preprocessing unit 20.
  • the sampling timing is indicated by a circle in the figure. In the example shown in the figure, sampling is performed seven times at irregular intervals. The sampling timing differs for each batch. Furthermore, the reaction time and the raw material dropping time also differ for each batch. Therefore, in this embodiment, the time-series process data is synchronized by performing preprocessing using the preprocessing unit 20.
  • FIG. 4 is a diagram showing the change when the smoothing process according to the first embodiment is applied to the drip rate.
  • the smoothing process is performed to remove the error factors described above (time error based on the measurement cycle and measurement error caused by the operator performing the measurement). The error factors can be reduced by the smoothing process.
  • FIG. 4(A) and FIG. 4(B) show the change in drip rate with the horizontal axis representing time [min] and the vertical axis representing the drip rate per minute [g/min].
  • FIG. 4(A) shows the drip rate per minute [g/min] before the smoothing process is performed
  • FIG. 4(B) shows the drip rate per minute [g/min] after the smoothing process is performed.
  • the smoothing process may be performed by calculating a moving average.
  • the moving average may be, for example, a weighted moving average.
  • the smoothing process may use other methods such as a simple moving average as long as it can reduce error factors.
  • the smoothing process may also be called a smoothing process. As shown in the figure, it can be seen that the time series data is smoothed before and after the smoothing process.
  • the time-series process data explained with reference to Figure 3 differs for each batch (lot). Each batch is operated to meet the target temperature and drip rate, but the process time differs slightly. Synchronization processing adjusts process data of different lengths like this.
  • the DTW method dynamic time warping method
  • the DTW method (dynamic time warping method) allows the similarity between different batches of time-series process data to be calculated. If the two pieces of time-series process data are identical, the similarity is 0, and the more different the shapes of the time-series process data are, the greater the similarity.
  • the model creation unit 30 uses the similarity calculated from the time-series process data in this way as part of the explanatory variable X to perform learning and inference.
  • FIG. 5 is a diagram showing an example of conditions for acquiring teacher data in the learning stage of the resin property prediction device according to embodiment 1.
  • basic conditions data for eight lots, Run1 to Run8, is acquired.
  • the reaction is carried out while maintaining a predetermined temperature T in both the dripping process and the holding process.
  • the dripping process lasts for five hours, and the holding process lasts for four hours.
  • Run 9 to Run 12 are conditions where the reaction liquid temperature is increased; specifically, the temperature is increased by 5°C from the predetermined temperature T in both the dripping process and the hold process.
  • Run 10 to Run 12 are conditions where the reaction liquid temperature is decreased. Specifically, in Run 10, the temperature is decreased by 5°C from the predetermined temperature T in both the dripping process and the hold process. In Run 11 and Run 12, the temperature is decreased by 10°C from the predetermined temperature T in both the dripping process and the hold process.
  • the resin property value prediction device 1 learns using the data acquired under such basic conditions and changed conditions as teacher data.
  • FIG. 6 is a first diagram showing the prediction results of the resin physical property prediction device according to the first embodiment.
  • FIG. 7 is a second diagram showing the prediction results of the resin physical property prediction device according to the first embodiment.
  • the horizontal axis shows the actual measured value
  • the vertical axis shows the predicted value by the resin physical property prediction device 1
  • the error and the coefficient of determination R2 are calculated.
  • the machine learning algorithm for determining the NV value and viscosity is lasso
  • the machine learning algorithm for determining the residual monomer concentration is ElasticNet. Note that these machine learning algorithms may be configured so that a suitable one is appropriately selected by the model creation unit 30.
  • FIGS. 6(A) to 6(C) show an example of inference using a learning model trained based only on the basic conditions described above.
  • FIGS. 6(D) to 6(F) show an example of inference using a learning model trained based on the basic conditions described above and also on modified conditions that include temperature.
  • Figures 6(A) and 6(D) show the results for the NV value
  • Figures 6(B) and 6(E) show the results for the residual monomer concentration
  • Figures 6(C) and 6(F) show the results for the viscosity.
  • the average error rate is ⁇ 0.83 [%], and the coefficient of determination R2 is 0.99.
  • the average error rate is ⁇ 1.33 [%], and the coefficient of determination R2 is 0.99.
  • the residual monomer concentration shown in FIG. 6(B) the average error rate is ⁇ 7.4 [%], and the coefficient of determination R2 is 0.94.
  • the residual monomer concentration shown in FIG. 6(E) the average error rate is ⁇ 8.6 [%], and the coefficient of determination R2 is 0.94.
  • the viscosity shown in FIG. 6(C) the average error rate is ⁇ 15.4 [%], and the coefficient of determination R2 is 0.91.
  • the average error rate is ⁇ 18.6 [%], and the coefficient of determination R2 is 0.83.
  • Figure 7 shows the results of an experiment to see the change in prediction accuracy when smoothing processing is performed or not.
  • Figures 7(A) and 7(D) show the results of the NV value
  • Figures 7(B) and 7(E) show the results of the residual monomer concentration
  • Figures 7(C) and 7(F) show the results of the viscosity.
  • Figures 7(A) to 7(C) show the results when smoothing processing is performed.
  • Figures 7(D) to 7(F) show the results when smoothing processing is not performed.
  • Figures 7(A) to 7(C) use the same results as Figures 6(A) to 6(C).
  • the average error rate is ⁇ 1.00 [%], and the coefficient of determination R2 is 0.99. Therefore, it can be said that the prediction accuracy of the NV value in FIG. 7(A) where smoothing processing was performed is higher than that in FIG. 7(D) where smoothing processing was not performed.
  • the residual monomer concentration shown in FIG. 7(E) the average error rate is ⁇ 13.3 [%], and the coefficient of determination R2 is 0.87. Therefore, it can be said that the prediction accuracy of the residual monomer concentration in FIG. 7(B) where smoothing processing was performed is higher than that in FIG. 7(E) where smoothing processing was not performed. Looking at the viscosity shown in FIG.
  • the prediction accuracy of the NV value and residual monomer concentration is higher when smoothing processing is performed. Furthermore, the prediction accuracy of the viscosity is higher when smoothing processing is not performed. Therefore, whether or not to perform preprocessing by the preprocessing unit 20 may be determined depending on the type of resin property value to be measured.
  • the resin property value prediction device 1 predicts the resin property value from information on the reacting raw materials and time-series process data during the polymerization reaction in a process of performing a resin polymerization reaction in a batch-type reaction tank.
  • the resin property value prediction device 1 includes a storage unit 10, and stores information on the reacting raw materials when manufactured in the past and time-series process data during the polymerization reaction in association with the resin property value, which is the corresponding actual measurement data.
  • the resin property value prediction device 1 also includes a pre-processing unit 20, and performs pre-processing on the time-series process data.
  • the pre-processing includes at least a synchronization process of the time-series process data.
  • the resin property value prediction device 1 also includes a model creation unit 30, and creates a learning model using the pre-processed data as teacher data.
  • the resin property value prediction device 1 also includes a calculation unit 50, and calculates the resin property value using the learned model created by the model creation unit 30 based on the blending amount of the raw materials in the reaction currently in progress and the process data. That is, according to this embodiment, after the synchronization process of the time-series process data, learning and inference are performed. Therefore, according to the resin property value prediction device 1, when predicting a resin property value using process data whose values change over time, the resin property value can be predicted with high accuracy.
  • the preprocessing performed by the preprocessing unit 20 is a process of extracting the mutual similarity between multiple pieces of time-series process data as a feature. Furthermore, according to the embodiment described above, the similarity is extracted by the DTW method. In other words, according to the resin physical property value prediction device 1, since the time-series process data is learned based on the similarity, even if the sizes of the multiple pieces of process data are different, the resin physical property value can be predicted with high accuracy.
  • the pre-processing unit 20 smoothes the time-series process data and then extracts the similarity as a feature. Therefore, according to this embodiment, even if the time-series process data contains errors, it is possible to accurately predict the resin property values.
  • the model creation unit 30 selects a method that reduces the difference from the measured physical property value from among any of the algorithms of Lasso regression, Ridge regression, and partial least squares regression, and updates the regression parameters within the algorithm during the selection process. Therefore, the resin physical property value prediction device 1 can perform inference using a learning model created using a more suitable machine learning algorithm. Therefore, the resin physical property value prediction device 1 can accurately predict the resin physical property value when predicting the resin physical property value using process data whose values change over time.
  • the resin composition for which the calculation unit 50 calculates the resin physical property values is an acrylic resin
  • the resin physical property values calculated by the calculation unit include at least one of the viscosity, the amount of non-volatile components, and the amount of remaining monomers. Therefore, according to this embodiment, when predicting resin physical property values such as viscosity, the amount of non-volatile components, and the amount of remaining monomers using process data whose values change over time, it is possible to accurately predict the resin physical property values.
  • a second embodiment will be described with reference to FIG. 8.
  • future (e.g., 30 minutes later) physical property values are predicted. That is, the second embodiment differs from the first embodiment in that future physical property values are predicted from a resin polymerization reaction currently in progress.
  • a plurality of objective variables Y are calculated when the parameters of the explanatory variable X are varied, and it is determined how the explanatory variable X should be set to obtain more suitable physical property values, and feedback may be provided to use suitable parameters.
  • FIG. 8 is a functional configuration diagram showing an example of the functional configuration of a resin property value prediction device according to embodiment 2.
  • the resin property value prediction device 1A differs from the resin property value prediction device 1 in that it further includes a candidate generation unit 60, a future property value calculation unit 70, and a process data determination unit 80.
  • components similar to those of the resin property value prediction device 1 may be denoted by similar reference numerals and description thereof may be omitted.
  • the candidate generation unit 60 generates process data from the current time onward in the resin polymerization reaction that is in progress.
  • the process data from the current time onward generated by the candidate generation unit 60 becomes candidates for future process data. Therefore, the process data from the current time onward generated by the candidate generation unit 60 can also be called future process data candidates.
  • the candidate generation unit 60 can also generate multiple future process data candidates.
  • the multiple future process data candidates generated by the candidate generation unit 60 are preprocessed by the preprocessing unit 20, and inference is performed using the learning model created by the model creation unit 30.
  • the future physical property value calculation unit 70 calculates the resin physical property value by inputting the blending amount of the raw materials and the process data generated by the candidate generation unit 60 into the learning model created by the model creation unit.
  • the resin physical property value calculated by the future physical property value calculation unit 70 can be said to be the resin physical property value when a resin polymerization reaction is performed using the future process data candidate.
  • the future physical property value calculation unit 70 calculates multiple resin physical property values.
  • the candidate generation unit 60 may further generate future process data candidates based on the multiple resin physical property values calculated by the future physical property value calculation unit 70.
  • the future process data candidates generated based on the multiple resin physical property values calculated by the future physical property value calculation unit 70 may be those that can obtain more suitable resin physical property values.
  • the resin physical property value prediction device 1A further includes a process data determination unit 80, which determines whether or not the resin physical property value set as the target value can be realized. Specifically, the process data determination unit 80 determines whether or not the resin physical property value set as the target value can be realized based on the trend of the physical property value calculated based on the future process data candidates generated by the candidate generation unit 60.
  • the target value may be determined in advance and stored in the storage unit 10, or may be calculated based on the current resin physical property value.
  • the candidate generation unit 60 may further raise or lower the process data based on the trend of the physical property value obtained as a result of raising or lowering the process data.
  • the resin property value prediction device 1A includes the candidate generation unit 60, and generates process data from the current time onward in a resin polymerization reaction in progress.
  • the resin property value prediction device 1A also includes the future property value calculation unit 70, and calculates resin property values by inputting the blending amounts of raw materials and the process data generated by the candidate generation unit 60 into a learning model created by the model creation unit 30.
  • the resin property value prediction device 1A can predict future (e.g., 30 minutes later) property values.
  • the future physical property value calculation unit 70 calculates multiple resin physical property values
  • the candidate generation unit 60 generates future process data candidates, which are candidates for future process data, based on the multiple resin physical property values calculated by the future physical property value calculation unit 70.
  • the resin physical property value prediction device 1A determines whether or not a resin physical property value set as a target value can be realized based on the progress of the physical property values calculated based on the future process data candidates generated by the candidate generation unit 60. Therefore, according to this embodiment, it is possible to swing the future process data candidates up and down and select a more suitable candidate, thereby achieving the desired resin physical property value.
  • each part of the resin property prediction device 1 in the above-mentioned embodiment may be realized by recording a program for realizing these functions on a computer-readable recording medium, and reading and executing the program recorded on the recording medium into a computer system.
  • the "computer system” here includes hardware such as the OS and peripheral devices.
  • Computer-readable recording media refers to portable media such as optical magnetic disks, ROMs, and CD-ROMs, as well as storage units such as hard disks built into computer systems.
  • “computer-readable recording media” may also include devices that dynamically store programs for a short period of time, such as communication lines when transmitting programs over a network such as the Internet, and devices that store programs for a fixed period of time, such as volatile memory within a computer system that serves as a server or client in such cases.
  • the above programs may also be ones that realize some of the functions described above, or may be ones that can realize the functions described above in combination with programs already recorded in the computer system.
  • 1...Resin property value prediction device 10...Memory unit, 20...Preprocessing unit, 30...Model creation unit, 40...In-progress data acquisition unit, 50...Calculation unit, 60...Candidate generation unit, 70...Future property value calculation unit, 80...Process data judgment unit

Landscapes

  • Engineering & Computer Science (AREA)
  • Computing Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Automation & Control Theory (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Testing And Monitoring For Control Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

Cet appareil de prédiction de valeur de propriété physique de résine prédit, lors d'un processus de réalisation d'une réaction de polymérisation de résine dans un récipient de réaction de type discontinu, une valeur de propriété physique de résine à partir d'information concernant une matière première de réaction et des données de processus chronologique pendant la réaction de polymérisation. L'appareil de prédiction de valeur de propriété physique de résine comprend: une unité de stockage dans laquelle une information concernant la matière première de réaction et des données de processus chronologique pendant une réaction de polymérisation au moment de la fabrication dans le passé sont stockées en association avec une valeur de propriété physique de résine constituant un centre de données de mesure correspondant; une unité de prétraitement qui effectue un prétraitement sur les données de processus chronologique; une unité de création de modèle qui crée un modèle d'apprentissage au moyen des données sur lesquelles le prétraitement a été effectué en tant que données d'enseignant; et une unité de calcul qui, sur la base d'une quantité composée de la matière première et des données de processus dans une réaction actuellement en cours, calcule une valeur de propriété physique de résine au moyen d'un modèle entraîné créé par l'unité de création de modèle.
PCT/JP2024/006175 2023-03-15 2024-02-21 Appareil de prédiction de valeur de propriété physique de résine et procédé de prédiction de valeur de propriété physique de résine Ceased WO2024190332A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2023-041016 2023-03-15
JP2023041016A JP7630538B2 (ja) 2023-03-15 2023-03-15 樹脂物性値予測装置及び樹脂物性値の予測方法

Publications (1)

Publication Number Publication Date
WO2024190332A1 true WO2024190332A1 (fr) 2024-09-19

Family

ID=92755490

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2024/006175 Ceased WO2024190332A1 (fr) 2023-03-15 2024-02-21 Appareil de prédiction de valeur de propriété physique de résine et procédé de prédiction de valeur de propriété physique de résine

Country Status (2)

Country Link
JP (1) JP7630538B2 (fr)
WO (1) WO2024190332A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015509710A (ja) * 2012-02-16 2015-04-02 オックスフォード ナノポール テクノロジーズ リミテッド ポリマーの測定の解析
JP2018112903A (ja) * 2017-01-12 2018-07-19 横河電機株式会社 プラント運転支援装置、プラント運転支援方法、プラント運転支援プログラム及び記録媒体
JP2021043325A (ja) * 2019-09-11 2021-03-18 東京応化工業株式会社 情報処理システム、情報処理装置、情報処理方法及びプログラム
WO2022202978A1 (fr) * 2021-03-25 2022-09-29 日本ペイントコーポレートソリューションズ株式会社 Procédé et appareil de production de résine
JP2023507082A (ja) * 2019-12-18 2023-02-21 ダウ グローバル テクノロジーズ エルエルシー 分析データに基づく特徴付けのための機械視覚

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102816634B1 (ko) * 2021-08-24 2025-06-05 (주)브릭 쉐이프릿 기반 예측 모델의 학습 방법 및 쉐이프릿 기반 예측 모델을 이용한 예측 방법

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2015509710A (ja) * 2012-02-16 2015-04-02 オックスフォード ナノポール テクノロジーズ リミテッド ポリマーの測定の解析
JP2018112903A (ja) * 2017-01-12 2018-07-19 横河電機株式会社 プラント運転支援装置、プラント運転支援方法、プラント運転支援プログラム及び記録媒体
JP2021043325A (ja) * 2019-09-11 2021-03-18 東京応化工業株式会社 情報処理システム、情報処理装置、情報処理方法及びプログラム
JP2023507082A (ja) * 2019-12-18 2023-02-21 ダウ グローバル テクノロジーズ エルエルシー 分析データに基づく特徴付けのための機械視覚
WO2022202978A1 (fr) * 2021-03-25 2022-09-29 日本ペイントコーポレートソリューションズ株式会社 Procédé et appareil de production de résine

Also Published As

Publication number Publication date
JP2024131014A (ja) 2024-09-30
JP7630538B2 (ja) 2025-02-17

Similar Documents

Publication Publication Date Title
Tsen et al. Predictive control of quality in batch polymerization using hybrid ANN models
EP2019977A1 (fr) Procédé et dispositif destinés à optimiser la performance d'un contrôle avancé de procédé par commande par logique floue
Khayyam et al. Dynamic prediction models and optimization of polyacrylonitrile (PAN) stabilization processes for production of carbon fiber
US20230119336A1 (en) Prediction device, calculation device, manufacturing device, and manufacturing method
KR20220112681A (ko) 정보 처리 장치, 기록 매체 및 프로세스 조건 탐색 방법
WO2020175418A1 (fr) Dispositif de développement à commande de prédiction, procédé de développement à commande de prédiction et programme de développement à commande de prédiction
JP2005339498A (ja) 品質制御装置およびその制御方法、品質制御プログラム、並びに該プログラムを記録した記録媒体
JP7302134B2 (ja) 物性情報推測方法、物性情報推測モデル生成方法、物性情報推測装置、物性情報推測モデル生成装置、物性情報推測プログラム、及び物性情報推測モデル生成プログラム
CN113435106A (zh) 一种过渡模态操作故障检测方法和系统
JP7630538B2 (ja) 樹脂物性値予測装置及び樹脂物性値の予測方法
US12429378B2 (en) Resin property value prediction apparatus, learning apparatus, and method for predicting resin property value
Gerlinger et al. Dynamic Optimization and Non‐linear Model Predictive Control to Achieve Targeted Particle Morphologies
JP7669693B2 (ja) 化学反応における活性化エネルギーの学習装置及び予測装置
CN116225100B (zh) 一种封闭高温腔体内温度控制方法、装置和设备
Shi et al. Advances in calculation of kinetic parameters in free-radical polymerization by data-driven methods
CN119626418B (zh) 利用多模态大模型辅助材料合成的方法、系统及存储介质
CN119861190B (zh) 一种水性树脂质量在线检测方法
WO2022264885A1 (fr) Dispositif de prédiction, appareil de production et procédé de prédiction
TWI795155B (zh) 建立製程預測模型的方法及系統
Skålén et al. Nonlinear MPC for grade transitions in an industrial LDPE tubular reactor
JP7168118B1 (ja) 検量装置、検量方法および検量プログラム
Duran-Villalobos et al. Iterative learning modelling and control of batch fermentation processes
CN113724795B (zh) 萃余过氧化氢浓度计算方法、装置、存储介质及处理器
CN115860450A (zh) 一种基于状态空间模型的预测控制方法、装置及介质
WO2010049810A2 (fr) Procédé et appareil d'estimation et de prévision en ligne de la concentration d'espèces pendant une réaction

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 24770445

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 24770445

Country of ref document: EP

Kind code of ref document: A1