Information processing system, information processing method, and program

Abstract

To facilitate the quality improvement of a molded article by an operator.SOLUTION: An apparatus includes an extraction unit configured to extract, from molding data that is one of time-series data of physical quantities acquired by a sensor in one shot of one or more molded articles molded by a molding machine or data indicating a spatial distribution of the physical quantities, a plurality of feature quantities relating to quality of the molded article, a determination unit configured to determine an evaluation index for evaluating the feature quantity contributing to improvement of the quality of the molded article for each of the extracted plurality of feature quantities, using machine learning, statistical data analysis, or a function using the plurality of feature quantities as an explanatory variable, and an output unit configured to output information for presenting the plurality of feature quantities in a comparable manner and / or information for presenting information relating to the feature quantities rearranged according to the evaluation index.SELECTED DRAWING: Figure 4

Description

[Technical Field]

[0001] The present invention relates to an information processing system, an information processing method, and a program. [Background technology]

[0002] In the field of molding (for example, injection molding), a technology has been developed that uses, for example, a machine learning model to assist in detecting defects in the appearance of molded products that are to be judged as pass / fail (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Patent No. 7481048 Summary of the Invention [Problem to be solved by the invention]

[0004] However, there is a drawback in that the basis for the predictions presented by the machine learning model is unclear. Because of this drawback, there is a possibility that experts in the field of molding processing will not trust the predictions presented by the machine learning model. As such, it is difficult to grasp information about feature quantities that contribute to the quality of molded products (e.g., feature quantities that can be used in the machine learning model) or feature quantities such as molding condition items, making it difficult for operators to improve the quality of molded products.

[0005] The present invention has been made in view of the above-mentioned problems, and has an object to provide an information processing system, an information processing method, and a program that make it easier for an operator to improve the quality of molded products. [Means for solving the problem]

[0006] An information processing system according to a first aspect of the present invention includes: an extraction unit that extracts multiple feature quantities related to the quality of one or multiple molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or multiple molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit that determines an evaluation index for evaluating a feature quantity that contributes to quality improvement of a molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, or a function that uses the extracted feature quantities as explanatory variables; and an output unit that outputs information for presenting the evaluation index for each of the plurality of feature amounts in a comparable manner, and / or information for presenting information about the feature amounts sorted according to the evaluation index for each of the plurality of feature amounts.

[0007] With this configuration, the user can compare the evaluation indexes for each of multiple feature quantities, or can grasp information about feature quantities sorted according to these evaluation indexes, making it easier to grasp information about feature quantities that contribute to the quality of molded products, and therefore making it easier for the operator to improve the quality of molded products.

[0008] An information processing system according to a second aspect of the present invention is the information processing system according to the first aspect, wherein the determination unit determines, as the evaluation index, a contribution of each of the extracted feature quantities to the quality of the molded product by using machine learning, statistical data analysis, or a function that uses the extracted feature quantities as explanatory variables, and The output unit outputs information for presenting the contribution degree of each of the plurality of feature amounts in a comparable manner and / or information for presenting information about the feature amounts sorted according to the contribution degree of each of the plurality of feature amounts.

[0009] With this configuration, the user can compare the contribution of each of multiple feature quantities, or can identify feature quantities with high contributions, making it easier to understand the feature quantities that contribute to the quality of the molded product, and therefore making it easier for the operator to improve the quality of the molded product.

[0010] An information processing system according to a third aspect of the present invention is the information processing system according to the second aspect, the machine learning is principal component analysis, The determination unit determines at least one principal component vector by performing principal component analysis on the molding data, and determines each element of the principal component vector as the contribution of each feature amount.

[0011] According to this configuration, by using principal component analysis, it is possible to improve the accuracy of estimating the contribution of each feature amount.

[0012] An information processing system according to a fourth aspect of the present invention is the information processing system according to the third aspect, when determining the contribution degrees, the determiner determines the contribution degree of each feature amount to a first principal component and the contribution degree of each feature amount to a second principal component; The output unit outputs information for presenting a contribution degree of each feature amount to the first principal component and a contribution degree of each feature amount to the second principal component in a comparable manner.

[0013] According to this configuration, the user can compare the contribution of each feature to the first principal component with the contribution of each feature to the second principal component, and find the feature that contributes more to the quality of the molded product.

[0014] An information processing system according to a fifth aspect of the present invention is the information processing system according to any one of the second to fourth aspects, The statistical data analysis is a multiple regression analysis, The contribution determination unit executes multiple regression analysis using the value representing the quality of the molded product as a response variable and each of the feature quantities as an explanatory variable, and determines each of the obtained regression coefficients as the contribution of the corresponding feature quantity.

[0015] According to this configuration, by using multiple regression analysis, it is possible to improve the accuracy of estimating the contribution of each feature amount.

[0016] An information processing system according to a sixth aspect of the present invention is the information processing system according to any one of the second to fifth aspects, The function using the plurality of feature quantities as explanatory variables is an objective function that includes in its formula a difference between corresponding components of a representative vector using, as its elements, representative values ​​of the plurality of feature quantities in one shot of a plurality of past non-defective molded products, and an explanatory variable vector including, as elements, the plurality of feature quantities as explanatory variables, The determination unit selects one of the plurality of feature quantities included in the explanatory variable vector, and determines the contribution of the one feature quantity to the quality of the molded product based on the magnitude of change in the objective function when the value of the one feature quantity is slightly changed.

[0017] According to this configuration, by using this objective function, it is possible to improve the accuracy of estimating the contribution of each feature amount.

[0018] An information processing system according to a seventh aspect of the present invention is the information processing system according to any one of the second to sixth aspects, When determining the contribution degree, the determination unit determines the contribution degree of each of a plurality of feature amounts for a non-defective molded product and the contribution degree of each of a plurality of feature amounts for a defective molded product; The output unit outputs information for presenting the contribution of each of the plurality of feature amounts in the non-defective molded product and the contribution of each of the plurality of feature amounts in the defective molded product in a comparable manner.

[0019] According to this configuration, the user can compare the contribution of each feature value between a good product and a defective product, and can therefore understand which feature value contributes to a good product or a defective product.

[0020] An information processing system according to an eighth aspect of the present invention is the information processing system according to any one of the first to seventh aspects, an acquisition unit that acquires selected feature amounts that are feature amounts selected by a user; a score determination unit that determines a score representing a difference between a representative vector determined from a plurality of vectors representing the selected feature amounts in one shot of a plurality of past non-defective molded products, and a target vector that is a vector of the selected feature amounts in the molding data of one shot of the molded product to be judged, based on the representative vector and the target vector; Equipped with The output unit outputs information for presenting at least one score when the molded product to be judged is a good product and at least one score when the molded product to be judged is a defective product in a manner that allows comparison.

[0021] According to this configuration, it is possible to compare the scores for each shot of a good molded product with the scores for each shot of a defective molded product, and it is possible to understand the influence that the selected feature has on a good or defective product from the difference in scores between the defective and good products.

[0022] An information processing system according to a ninth aspect of the present invention is the information processing system according to the eighth aspect, the representative vector is a vector representing the center of gravity of the plurality of vectors, The score is a value of an objective function whose formula includes the difference between each corresponding component of the representative vector and the target vector.

[0023] This configuration provides a suitable score for comparison.

[0024] An information processing system according to a tenth aspect of the present invention is the information processing system according to any one of the first to ninth aspects, The apparatus further includes an abnormality degree determining unit that determines the abnormality degree of the molding data in one shot of the molded product to be judged based on a comparison between at least one feature value in the molding data in one shot of the target molded product extracted by the extraction unit and a representative value of at least one feature value in the molding data in one shot of a plurality of past good molded products.

[0025] According to this configuration, the user can grasp the degree of abnormality in the molding data for one shot of the target molded product.

[0026] An information processing system according to an eleventh aspect of the present invention is the information processing system according to the tenth aspect, an acquisition unit that acquires a feature quantity selected by a user from the plurality of feature quantities extracted by the extraction unit; The abnormality degree determining unit determines the abnormality degree based on the feature amount selected by the user.

[0027] According to this configuration, the user can grasp the degree of abnormality in molding data for one shot of the target molded product from the viewpoint of the feature amount selected by the user.

[0028] An information processing system according to a twelfth aspect of the present invention is the information processing system according to any one of the second to eleventh aspects, When outputting the information, the output unit replaces the names of the plurality of feature quantities with characteristic names that indicate the characteristics to which the feature quantities relate, and outputs information for presenting the contribution degrees in a comparable manner.

[0029] According to this configuration, the user can understand what kind of feature amount it is just by looking at the property name.

[0030] An information processing system according to a thirteenth aspect of the present invention is the information processing system according to the first aspect, the determining unit determines a representative value of the feature amounts obtained from the molding data of a non-defective product by using statistical data analysis for the extracted plurality of feature amounts, and determines a difference between the feature amount obtained from the target molding data and the representative value of the feature amount obtained from the molding data of a non-defective product as the evaluation index; The output unit outputs information for presenting the differences between each of a plurality of feature amounts in a comparable manner, and / or information for presenting feature amounts or molding condition items rearranged according to the differences between each of a plurality of feature amounts.

[0031] With this configuration, it is possible to identify features with large differences in the target molding data, and / or to identify features with large differences or molding condition items corresponding to features with large differences, making it easier for the operator to take action to stabilize quality.

[0032] An information processing system according to a 14th aspect of the present invention is an information processing system according to the 13th aspect, wherein the output unit refers to a storage device in which the relationship between features and molding condition items is stored, reads out molding condition items corresponding to the multiple feature values ​​from the storage device, and outputs information for presenting molding condition items rearranged according to the differences between each of the multiple feature values.

[0033] According to this configuration, molding condition items with larger differences are presented with higher priority, making it easier for the operator to know which molding condition items to operate.

[0034] An information processing system according to a fifteenth aspect of the present invention is the information processing system according to the thirteenth or fourteenth aspect, The output unit refers to a storage device in which feature values, the direction of deviation of the feature values ​​from the representative values ​​of the feature values ​​obtained from molding data of good products, sensor positions, and molding condition items are stored in association with each other, reads out the molding condition items corresponding to the target feature values, the direction of deviation of the target feature values, and the target sensor positions, and outputs information for presenting the read molding condition items.

[0035] According to this configuration, molding condition items to be changed can be presented more accurately.

[0036] An information processing method according to a sixteenth aspect of the present invention comprises: an extraction unit extracting a plurality of feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or more molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit determining an evaluation index for evaluating a feature quantity that contributes to quality improvement of the molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, and a function that uses the extracted feature quantities as explanatory variables; The output unit has a procedure for outputting information for presenting the evaluation index for each of the multiple feature amounts in a comparable manner, and / or information for presenting information about the feature amounts sorted according to the evaluation index for each of the multiple feature amounts.

[0037] With this configuration, the user can compare the evaluation indexes for each of multiple feature quantities, or can grasp information about feature quantities sorted according to these evaluation indexes, making it easier to grasp information about feature quantities that contribute to the quality of molded products, and therefore making it easier for the operator to improve the quality of molded products.

[0038] A program according to a seventeenth aspect of the present invention includes the steps of: an extraction unit that extracts multiple feature quantities related to the quality of one or multiple molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or multiple molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit that determines an evaluation index for evaluating a feature quantity that contributes to improving the quality of a molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, and a function that uses the extracted feature quantities as explanatory variables; an output unit that outputs information for presenting the evaluation indexes of the plurality of feature amounts in a comparable manner, and / or information for presenting information on the feature amounts sorted according to the evaluation indexes of the plurality of feature amounts; This is a program that functions as a

[0039] With this configuration, the user can compare the evaluation indexes for each of multiple feature quantities, or can grasp information about feature quantities sorted according to these evaluation indexes, making it easier to grasp information about feature quantities that contribute to the quality of molded products, and therefore making it easier for the operator to improve the quality of molded products. [Effects of the Invention]

[0040] According to one aspect of the present invention, a user can compare the evaluation indices for each of multiple feature quantities, or can grasp information about feature quantities sorted according to these evaluation indices, thereby making it easier to grasp information about feature quantities that contribute to the quality of molded products, and thereby making it easier for operators to improve the quality of molded products. [Brief explanation of the drawings]

[0041] [Figure 1] 1 is a schematic configuration diagram of a molding system according to an embodiment of the present invention. [Figure 2] 1 is a block diagram showing a schematic configuration of an information processing system. [Figure 3] 10 is an example of a table stored in a storage unit. [Figure 4] 1 is an example of a radar chart displayed on a display device. [Figure 5] 10 shows an example of a radar chart showing the score of each characteristic amount when the molded product to be judged is a non-defective product, and an example of a radar chart showing the score of each characteristic amount when the molded product to be judged is a defective product. [Figure 6]10 is an example of a screen showing scores for each shot of a non-defective molded product and scores for each shot of a defective molded product for two feature amounts selected by a user. [Figure 7] 10 is an example of a screen showing whether each feature amount obtained from molding data to be visualized has shifted to the positive side or the negative side in a modified example. [Figure 8] 10 is an example of a table in which feature quantities and molding condition items are stored in association with each other in a modified example. [Figure 9] This is an example screen showing whether each feature obtained from the molding data to be visualized in a modified example has shifted to the positive or negative side, and in which molding condition items corresponding to each feature are displayed as an auxiliary. [Figure 10] 10 is an example of a table in which feature amounts, directions of deviation of the feature amounts, sensor positions, and molding condition items are stored in association with each other in a modified example. DETAILED DESCRIPTION OF THE INVENTION

[0042] Hereinafter, each embodiment will be described with reference to the drawings. However, unnecessary detailed description may be omitted. For example, detailed description of well-known matters or redundant description of substantially the same configuration may be omitted. This is to avoid unnecessary redundancy in the following description and to facilitate understanding by those skilled in the art.

[0043] 1 is a schematic diagram of a molding system according to this embodiment. The molding system 100 according to this embodiment includes a molding machine 1, an instrumentation amplifier 2, an A / D converter 3 connected to the output of the instrumentation amplifier 2, and a control device 4 connected to the output of the A / D converter 3. The control device 4 controls molding conditions (e.g., the speed, pressure, temperature, etc., when injecting resin into a mold). The molding system 100 also includes an information processing system 5 connected to the control device 4 via a communication network CN, and a display device 6 connected to the output of the information processing system 5. The display device 6 displays an image based on a video signal output from the information processing system 5.

[0044] The molding machine 1 is, for example, an injection molding machine that produces various plastic products by pouring molten resin (plastic) into a mold, cooling and solidifying it, and then removing it. Note that the molding machine 1 is not limited to a resin injection molding machine as long as it injects a material into a mold, and this material may be metal or ceramic. For example, this metal may be an aluminum alloy, and the molding machine may be a die-casting machine. The molding machine 1 has an injection unit 10 that uses heat to melt the resin material and inject it into the mold, and a mold clamping unit 20 that opens and closes the mold.

[0045] As shown in Figure 1, the injection unit 10 includes a hollow cylinder 11, a hopper 12, a screw 13, a check valve 14, a heater 15, a nozzle 16, a motor 17, etc. The hopper 12 is an inlet for the molding material (for example, pellets, which are granular molding material). The screw 13 is disposed inside the cylinder 11 and is rotatable and movable in the axial direction of the cylinder 11. The nozzle 16 is an injection site provided at the tip of the cylinder 11, and supplies the molten resin inside the cylinder 11 to the cavity of the mold 22 by the axial movement of the screw 13.

[0046] The mold clamping unit 20 opens and closes the attached mold, and when the mold is clamped, prevents the mold from opening due to the pressure of the molten resin injected into the mold cavity (molded product portion). The mold clamping unit 20 includes a fixed platen, a movable platen, tie bars 21, a drive unit, etc. The mold has a first mold and a second mold on the fixed side, and the first mold on the fixed side is fixed to the fixed platen. The fixed platen is capable of contacting the nozzle 16 and directing the resin injected from the nozzle 16 into the mold cavity. The cavity is formed between the first mold and the second mold and is an area corresponding to the product shape. The movable second mold is fixed to the movable platen and can move toward and away from the fixed platen. The tie bars 21 support the movement of the movable platen. The drive unit is, for example, a cylinder device, and moves the movable platen. The first mold has a supply path from the nozzle to the cavity. Note that a three-plate mold or the like can also be applied.

[0047] The control device 4 controls the motor 17 of the injection unit 10 and the drive device of the mold clamping unit 20 based on command values ​​(parameters) related to molding conditions.

[0048] The mold clamping unit 20 of the molding machine 1 is equipped with a mold 22 equipped with a sensor 23 that acquires mold internal data such as mold internal pressure. The sensor 23 is, for example, a pressure sensor that detects mold internal pressure from the molten resin, such as a load cell. In this case, the type and data communication method of the sensor are not important as long as it can acquire internal cavity data such as mold internal pressure, and the installation location is also arbitrary. For example, when the mold is manufactured, a sensor can be embedded in the mold itself or installed in a manner that makes contact with the cavity, thereby acquiring internal cavity data such as mold internal pressure. The installation location of the sensor 23 is not particularly limited as long as it is possible to acquire internal mold data, and it can be any location, whether inside or outside the mold.

[0049] The sensor 23 sends sensing data to the control device 4 via an instrumentation amplifier 2, an A / D converter 3, and the like, which serve as a data acquisition unit.

[0050] Here, the molding data for one shot molded by the molding machine 1 is, for example, time series data of physical quantities measured by a sensor in one shot molded by the molding machine 1, or time series data of an image representing the spatial distribution of physical quantities measured by a sensor in one shot molded by the molding machine 1. In the present embodiment, as an example, the molding data will be described below as being, for example, time series data of physical quantities measured by a sensor in one shot molded by the molding machine 1.

[0051] 2 is a block diagram showing a schematic configuration of the information processing system 5. As shown in FIG. 2, the information processing system 5 includes an input unit 51, an output unit 52, a storage unit 53, a communication unit 54, and a processor 55. The input unit 51 receives data from a user or an external device of the information processing system 5, for example. The output unit 52 outputs a video signal to the display device 6, for example. The communication unit 54 communicates, for example, by wire or wirelessly.

[0052] The storage unit 53 stores a program that is read and executed by the processor 55. The storage unit 53 may also store information that configures a machine learning model that predicts the quality of a molded product in one shot molded by a molding machine. In this case, the input unit 51 may receive, as training data, data including molding data for one shot molded by the molding machine 1 and information on the quality of the molded product in that one shot. The processor 55 may then use this training data to train the machine learning model. This training data may include feature values.

[0053] Here, the feature quantity is a physical quantity itself or a processed physical quantity, such as a pressure increasing rate, a pressure injection area, etc. ), pressure holding area, maximum pressure, pressure time differential, pressure relaxation coefficient, etc. The name of this feature quantity alone does not allow one to understand what characteristic it represents. On the other hand, for example, the pressure change rate is related to resin viscosity, so a characteristic name indicating that it is related to resin viscosity (e.g., Simulated_Viscosity) is assigned. Therefore, in this embodiment, as shown in FIG. 3, the name of the feature quantity and the characteristic name indicating what characteristic the feature quantity relates to are associated and stored in the storage unit 53.

[0054] 3 is an example of a table stored in the storage unit. As shown in Table T1 in FIG. 3, the storage unit 53 stores, for example, the name of each feature, the property name of the feature, and a description of the feature, in association with each other, for each feature. As a result, when the processor 55, functioning as the output unit 553 described below, outputs information for presenting the contribution of each feature name to the quality of a molded product, it replaces each of the names of the multiple feature quantities with the property name indicating the property to which the feature quantity relates, and outputs information for presenting the contributions in a comparable manner. Therefore, the user can understand what kind of feature a given feature is simply by looking at the property name.

[0055] The processor 55 reads and executes the programs stored in the memory unit 53, thereby functioning as an extraction unit 551, a contribution determination unit 552, an output unit 553, an acquisition unit 554, a score determination unit 555, and an abnormality determination unit 556.

[0056] The extraction unit 551 extracts multiple feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor (e.g., sensor 23) in one shot of one or more molded products molded by the molding machine 1 or data representing the spatial distribution of the physical quantities (e.g., image data in which one pixel value representing one physical quantity is assigned to each pixel). Here, the feature quantities may be the physical quantities (e.g., pressure, temperature, etc.) themselves, statistical values ​​of the physical quantities (e.g., average, median, maximum, minimum, etc.), or values ​​obtained by processing the physical quantities over time (e.g., time rate of change, differential value, integral value, etc.). The molding data may be data obtained in the past or data obtained when extracting the feature quantities. Furthermore, the molding data may be data obtained only from good molded products, data obtained from both good and defective molded products, or data obtained only from defective molded products.

[0057] The contribution degree determining unit 552 determines the contribution degree of each of the extracted feature quantities to the quality of the molded product by using machine learning, statistical data analysis, or a function that uses the extracted feature quantities as explanatory variables. Here, the contribution degree determining unit 552 may function as an example of a specific function of the determining unit 550.

[0058] Furthermore, for example, the above-mentioned "statistical data analysis" may be a multiple regression analysis. In this case, the contribution determination unit 552 may execute a multiple regression analysis using the value representing the quality of the molded product as a response variable and each feature as an explanatory variable, and determine each of the obtained regression coefficients as the contribution of the corresponding feature. With this configuration, the use of multiple regression analysis can improve the estimation accuracy of the contribution of each feature.

[0059] Furthermore, for example, the above-mentioned "function using multiple feature quantities as explanatory variables" may be an objective function that includes the difference between corresponding components of a representative vector whose elements are representative values ​​(e.g., center of gravity) of multiple feature quantities in one shot of multiple past non-defective molded products, and an explanatory variable vector whose elements include the multiple feature quantities as explanatory variables. In this case, the contribution determination unit 552 may select one of the multiple feature quantities included in the explanatory variable vector and determine the contribution of the selected feature quantity to the quality of the molded product based on the magnitude of change in the objective function when the value of the selected feature quantity is slightly changed. With this configuration, using this objective function can improve the accuracy of estimating the contribution of each feature quantity.

[0060] Specifically, for example, the contribution determination unit 552 may determine that the greater the change in the position of the center of gravity of the feature vector of the group of good products when one feature is changed (for example, when changed by a small amount) for all of the past multiple good products, the greater the contribution of the one feature to the quality of the molded product.

[0061] Here, this objective function may be the distance between a representative vector whose elements are representative values ​​(e.g., center of gravity) of multiple feature quantities in one shot of multiple past non-defective molded products, and an explanatory variable vector whose elements include the multiple feature quantities as explanatory variables. This distance may be Euclidean distance, but is not limited to Euclidean distance and may be a distance used in statistics. In this case, for example, the contribution degree determiner 552 may determine the contribution degree of a single feature quantity to the quality of a molded product so that the greater the change in the distance when the single feature quantity is changed (e.g., when changed by a small amount) for all of the past non-defective products, the greater the contribution of the single feature quantity to the quality of the molded product.

[0062] The output unit 553 outputs information for presenting the contribution degree of each of a plurality of feature amounts in a comparable manner.

[0063] For example, the above-mentioned "machine learning" may be principal component analysis. In this case, the contribution determination unit 552 may determine at least one principal component vector by performing principal component analysis on the formed data, and determine each element of the principal component vector as the contribution of each feature amount. With this configuration, the use of principal component analysis can improve the estimation accuracy of the contribution of each feature amount.

[0064] <First example of display format> Specifically, for example, when determining the contributions, the contribution determination unit 552 may determine the contribution of each feature amount to a first principal component and the contribution of each feature amount to a second principal component. In this case, the output unit 553 may output information for presenting the contribution of each feature amount to the first principal component and the contribution of each feature amount to the second principal component in a comparable manner. This allows the user to compare the contribution of each feature amount to the first principal component and the contribution of each feature amount to the second principal component, thereby enabling the user to find a feature amount that contributes more to the quality of the molded product. Specifically, for example, as shown in FIG. 4, the output unit 553 may output information for presenting the contribution of each feature amount to the first principal component and the contribution of each feature amount to the second principal component in a radar chart.

[0065] Fig. 4 is an example of a radar chart displayed on a display device. In Fig. 4, the contribution of each feature to the first principal component and the contribution of each feature to the second principal component are displayed in a radar chart. For example, the characteristic names stored in table T1 in Fig. 3 are displayed as features in this radar chart.

[0066] <Explanation of the process for calculating contribution using principal component analysis> The following describes the process by which the contribution determination unit 552 calculates the contribution of each feature in principal component analysis. In principal component analysis (PCA), the coefficient for each feature refers to a weighting coefficient when each principal component is expressed as a linear combination of the original feature. Specifically, each principal component is defined as follows:

[0067]

number

[0068] where w1, w2, …, w k is a coefficient for each feature (k is a natural number representing an index), and represents the direction and importance of the principal component. Principal component analysis finds the principal component that maximizes the variance of the original data, so it can be assumed that the feature with a larger coefficient plays an important role in that principal component. The sign of the coefficient also indicates whether the feature has a positive or negative influence on the principal component.

[0069] (Step 1) Center the data For each feature, we center the data by subtracting the mean of that feature from each value of that feature, which adjusts the mean of each feature to 0.

[0070]

number

[0071]

number

[0072] (Step 2) Calculate the covariance matrix Calculate the covariance matrix from the centered data.

[0073]

number

[0074] where X is the centered data matrix and X' is its transpose.

[0075] (Step 3) Calculate the eigenvalues ​​and eigenvectors Calculate the eigenvalues ​​and eigenvectors of the covariance matrix C. The eigenvalues ​​are scale factors that explain the variance of the data, and the eigenvectors indicate its direction.

[0076]

number

[0077] where λ is an eigenvalue and ν is an eigenvector.

[0078] (Step 4) Selecting principal components Principal components are selected based on the magnitude of the eigenvalues. Typically, the eigenvector corresponding to the largest eigenvalue becomes the first principal component, and the eigenvector corresponding to the next largest eigenvalue becomes the second principal component.

[0079] (Step 5) Determine the weights of the principal components as contributions Principal component weights (w1, w2, …, w k ) are the elements of the corresponding eigenvectors.

[0080] (1) Weight of the first principal component The first principal component is the eigenvector corresponding to the largest eigenvalue. Each element of this eigenvector is the weight of the first principal component (w1, w2, …, w k The contribution determining section 552 determines this weight as the contribution to the first principal component.

[0081] (2) Weight of the second principal component The second principal component is an eigenvector corresponding to the next largest eigenvalue, and each element of the eigenvector is the weight of the second principal component. The contribution determining unit 552 determines this weight as the contribution to the second principal component.

[0082] For example, if there are three features (k=3), the eigenvectors of the covariance matrix can be obtained as follows:

[0083]

number

[0084] where ν1 is the eigenvector of the first principal component, and its element ν 11 , ν 12 , ν 13 are the weights (w1, w2, w3) of the first principal component, respectively.

[0085] <Second example of display format> When determining the contributions, the contribution determination unit 552 may determine the contribution of each of a plurality of feature amounts for a good molded product and the contribution of each of a plurality of feature amounts for a defective molded product. In this case, the output unit 553 may output information for presenting the contribution of each of the plurality of feature amounts for the good molded product and the contribution of each of the plurality of feature amounts for the defective molded product in a comparable manner. Specifically, for example, as shown in FIG. 5, the output unit 553 may display, in parallel, a radar chart showing the contribution of each feature amount when the molded product to be evaluated is good and a radar chart showing the contribution of each feature amount when the molded product to be evaluated is defective.

[0086] Figure 5 shows an example of a radar chart showing the contribution of each feature when the molded product being evaluated is a good product, and an example of a radar chart showing the contribution of each feature when the molded product being evaluated is a defective product. The radar chart shows the score of each feature. Here, each feature is represented by the characteristic name in Figure 3. This allows the user to compare the contribution of each feature between a good product and a defective product, and therefore understand which feature contributes to a good or defective product.

[0087] The display mode is not limited to this, and these two radar charts may be displayed overlapping each other. As another mode of displaying the contribution degree, the value of the contribution degree of each feature amount may be displayed as a bar graph. Specifically, for example, in one bar graph, the bars representing the contribution degree of the feature amounts of good products and defective products may be displayed next to each other, or may be displayed overlapping each other. Alternatively, a graph representing the contribution degree of each feature amount of good products and a graph representing the contribution degree of each feature amount of defective products may be displayed side by side.

[0088] <Comparison of scores between good and bad products for selected features> It may also be possible to compare scores between non-defective products and defective products for selected features, which are features selected by a user, where the score represents the difference between a representative vector (e.g., a vector representing the center of gravity of the plurality of vectors) determined from a plurality of vectors representing the selected features in one shot of a plurality of past non-defective molded products, and a target vector, which is a vector of the selected features in the molding data of one shot of the molded product to be judged.

[0089] For example, when two feature quantities are selected by the user, a screen such as that shown in Fig. 6 is displayed on the display device 6. Fig. 6 is an example of a screen showing the scores for each shot of a good molded product and the scores for each shot of a defective molded product for the two feature quantities selected by the user. After the feature quantities are selected by the user in this way, the following processing may be performed while this screen is displayed.

[0090] The acquisition unit 554 acquires, for example, selected features selected by a user. In this case, the score determination unit 555 may determine a score representing the difference between a representative vector (e.g., a vector representing the center of gravity of a plurality of vectors) determined from a plurality of vectors representing the selected features in one shot of a plurality of past non-defective molded products, and a target vector, which is a vector of the selected features in the molding data of one shot of the molded product to be evaluated, based on the representative vector and the target vector. Here, the score may be, for example, the value of an objective function including the difference between corresponding components of the representative vector and the target vector. Specifically, this objective function may be, for example, the distance between the representative vector and the target vector. Note that this distance may be, but is not limited to, the Euclidean distance.

[0091] In this case, the output unit 553 may output information for presenting at least one score when the molded product to be judged is a good product and at least one score when the molded product to be judged is a defective product in a comparable manner. As a result, for example, as shown in Fig. 6, the score for each shot of the good molded product and the score for each shot of the defective molded product are displayed. This makes it possible to compare the score for each shot of the good molded product with the score for each shot of the defective molded product, and the influence of the selected feature on whether the product is good or defective can be understood from the difference in scores between the defective and good products.

[0092] The abnormality degree determining unit 556 may determine the abnormality degree of the molding data for one shot of the molded product to be evaluated based on a comparison between at least one feature amount in the molding data for one shot of the target molded product extracted by the extracting unit 551 and a representative value (e.g., average value, median, value extracted from a group of feature amounts) of at least one feature amount in the molding data for one shot of multiple past good molded products. Specifically, for example, the abnormality degree determining unit 556 may determine, as the abnormality degree, the distance between the vector of at least one feature amount in the molding data for one shot of the target molded product and the vector of the representative value (e.g., average value, median, value extracted from a group of feature amounts) of the feature amount in the molding data for one shot of multiple past good molded products. This allows the user to grasp the abnormality degree of the molding data for one shot of the target molded product.

[0093] Here, the acquiring unit 554 may acquire a feature selected by a user from among a plurality of feature amounts using the extracting unit 551. In this case, the abnormality degree determining unit 556 may determine the degree of abnormality based on the feature selected by the user. Specifically, for example, the abnormality degree determining unit 556 may determine, as the degree of abnormality, a difference between the selected feature in the molding data of one shot of the target molded product and a representative value (e.g., an average value, a median value, a value extracted from a group of feature amounts, etc.) of the selected feature in the molding data of one shot of a plurality of past non-defective molded products. This allows the user to grasp the degree of abnormality of the molding data of one shot of the target molded product from the perspective of the feature selected by the user.

[0094] As described above, the information processing system 5 according to this embodiment includes an extraction unit 551 that extracts multiple feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or more molded products molded by the molding machine 1 or data representing the spatial distribution of the physical quantities. The information processing system 5 includes a contribution determination unit 552 that determines the contribution of each of the extracted multiple feature quantities to the quality of the molded product by using machine learning, statistical data analysis, or a function that uses the extracted multiple feature quantities as explanatory variables. The information processing system 5 includes an output unit 553 that outputs information for presenting the contribution degree of each of the plurality of feature amounts in a comparable manner.

[0095] According to this configuration, the user can compare the contribution of each of a plurality of feature quantities, which makes it easier to understand the feature quantities that contribute to the quality of the molded product.

[0096] In addition to the above, the output unit 553 may output information for presenting (for example, displaying, outputting audio, etc.) feature quantities sorted according to the contribution of each of the plurality of feature quantities (for example, feature quantities sorted so that feature quantities with higher contributions are ranked higher). As a result, the feature quantities are sorted and presented to the operator according to the degree of contribution, so the operator can understand that the molding condition item (for example, "injection speed") corresponding to the feature quantity with higher contribution (for example, peak pressure) should be operated with priority. Furthermore, if a database showing the relationship between feature quantities and molding condition items is stored in advance in the storage unit 23, the output unit 553 may output, instead of feature quantities, information for presenting (e.g., displaying, outputting audio, etc.) molding condition items rearranged according to the contribution degree of each of the plurality of feature quantities. Therefore, the output unit 553 may output information for presenting the contribution degree of each of the plurality of feature quantities in a comparative manner, and / or information for presenting information about the feature quantities rearranged according to the contribution degree of each of the plurality of feature quantities. Here, the information about the feature quantities may be the feature quantities themselves, or may be molding condition items corresponding to the feature quantities.

[0097] The above-mentioned contribution level is an example of an evaluation index for evaluating a feature value that contributes to improving the quality of a molded product. However, this evaluation index is not limited to the contribution level. As described in the following modified example, this evaluation index may be the difference (also referred to as deviation) between a feature value obtained from the molding data to be visualized and a representative value (e.g., average, median, etc.) of the feature value obtained from the molding data of a non-defective product. Here, the difference may be an absolute value or may include a positive or negative value. In this case, the determination unit 557 may determine a representative value (e.g., average, median, etc.) of the feature value obtained from the molding data of a non-defective product using statistical data analysis for the extracted multiple feature values, and may determine the difference between the feature value obtained from the molding data to be visualized and the representative value (e.g., average, median, etc.) of the feature value obtained from the molding data of a non-defective product. In this case, the output unit 553 may output information for presenting the difference between each of the multiple feature values ​​in a comparable manner and / or information for presenting the feature values ​​sorted according to the difference between each of the multiple feature values. Here, the information regarding the feature may be the feature value itself or a molding condition item corresponding to the feature value.

[0098] That is, the determination unit 550 determines an evaluation index for each of the extracted multiple feature quantities to evaluate the feature quantity's contribution to improving the quality of the molded product, using machine learning, statistical data analysis, or a function that uses the extracted multiple feature quantities as explanatory variables. That is, the output unit 553 may output information for presenting the evaluation indexes for each of a plurality of feature amounts in a comparable manner, and / or information for presenting information about the feature amounts sorted according to the evaluation indexes for each of the plurality of feature amounts. Here, the information about the feature amounts may be the feature amounts themselves, or may be molding condition items corresponding to the feature amounts.

[0099] <Modification> In the embodiment of Fig. 4 described above, there is a problem in that the operator of the molding machine 1 may find it difficult to understand what to do when looking at the contribution display. For example, in the radar chart of Fig. 4, the feature quantities contributing to the first and second principal components are different, and it may be difficult to understand at a glance which feature quantities contribute to quality. Furthermore, the operator must look at the contribution quantities to determine whether to increase or decrease the numerical value of a certain molding condition item, but even when looking at the radar chart of Fig. 4, there is a risk that the operator will not be able to determine whether to increase or decrease the numerical value.

[0100] (1) For example, molding data for multiple shots used to mold a non-defective product is stored in advance in the storage unit 53. The processor 55 of the information processing system 5, for example, compares the molding data for multiple shots used to mold a non-defective product stored in the storage unit 53 with the molding data for the shot to be visualized. Here, the molding data is either time-series data of a physical quantity (e.g., pressure, temperature, etc.) or data representing the spatial distribution of a physical quantity (e.g., pressure, temperature, etc.). Specifically, for example, the processor 55 compares a representative value (e.g., average, median, etc.) of a feature obtained from the molding data of a non-defective product with a feature obtained from the molding data of the visualization target. Then, the processor 55 may output information for presenting (for example, displaying, outputting audio, etc.) feature quantities obtained from the molding data of the visualization target, sorted according to deviations from the representative value (e.g., average, median, etc.) of the feature quantities obtained from the molding data of the non-defective product (for example, feature quantities sorted so that feature quantities with the largest deviations are ranked higher). This allows the operator to see which features have changed significantly in the molding data being visualized (for example, this is visually clear when displayed), making it easier for the operator to take action to stabilize quality.

[0101] (2) The processor 55 may output information for presenting (e.g., displaying, audio outputting, etc.) the deviation width (i.e., absolute value of deviation) and / or deviation direction (e.g., positive or negative) of the feature obtained from the molding data to be visualized relative to a representative value (e.g., average, median, etc.) of the feature obtained from the molding data of a non-defective product. This allows the operator to grasp the deviation width and / or deviation direction (specifically, whether the deviation is in the positive or negative direction). FIG. 7 is an example screen showing the deviation width and deviation direction of each feature obtained from the molding data to be visualized in a modified example. In FIG. 7, the feature values ​​are arranged from top to bottom in descending order of deviation width (i.e., ranked by deviation width), and the horizontal axis represents the deviation amount, including plus and minus. Since this deviation amount includes information on both the deviation width and the deviation direction, both the deviation width and the deviation direction are shown. Here, instead of just one sensor 23, eight pressure sensors are provided at different positions inside and / or outside the mold, and the suffixes ch1 to ch8 at the end of the feature values ​​in FIG. 7 indicate the channel numbers of the eight pressure sensors. "pressure_areas" is the pressure area (i.e., the time integral of pressure), "pressure_peaks" is the peak pressure, "pressure_increase_rates" is the pressure increase rate, "pressure_decrease_rates" is the pressure decrease rate, and "standup_time_diffs" is the rise time difference. As a result, as shown in Figure 7, for example, the deviation amount and / or deviation direction (specifically, whether it is a deviation in the positive or negative direction) of each feature amount is displayed on the display device 6. This makes it easier for the operator to determine whether to increase or decrease the numerical value of the molding condition item, and also makes it easier to determine how much to change the numerical value of the molding condition item from the deviation amount. Depending on the feature amount, if the deviation of the feature amount is on the negative side, the condition adjustment will be on the positive side, in which case the processor 55 may output information for presenting the direction of the condition adjustment (for example, the positive side). Also, depending on the feature amount, even if the deviation of the feature amount is on the negative side, the condition adjustment may be on the negative side, so if this relationship is known in advance, the processor 55 may output information for presenting the negative direction as the direction of the condition adjustment.

[0102] (3) The storage unit 53 may pre-store a database showing the relationship between feature quantities and molding condition items, as shown in FIG. 8. FIG. 8 shows an example of a table in which feature quantities and molding condition items are stored in association with each other in a modified example. In this case, the processor 55 may, for example, read from the database molding condition items corresponding to the feature quantities (e.g., feature quantities ranked by deviation width) for which deviations are to be displayed, and output information for displaying (e.g., display, audio output, etc.) the read molding condition items. As a specific example of this output, the processor 55 may output information for auxiliary display of molding condition items on the screen. FIG. 9 shows an example of a screen in a modified example showing whether each feature quantity obtained from molding data to be visualized has deviated positively or negatively, with the molding condition items corresponding to each feature quantity displayed auxiliary. As a result, as shown in FIG. 9, the display device 6 displays molding condition items corresponding to the ranked feature quantities (e.g., the injection speed for the peak pressure feature quantity), making it easier for the operator to determine which molding condition items to operate. In this way, the output unit 553 may refer to a storage device that stores the relationship between feature amounts and molding condition items, read out molding condition items corresponding to the plurality of feature amounts from the storage device, and output information for presenting molding condition items rearranged according to the differences between the plurality of feature amounts. This allows molding condition items with larger differences to be presented preferentially, making it easier for the operator to understand which molding condition item to operate. In summary, the determining unit 550 may use statistical data analysis to determine a representative value (e.g., average, median, etc.) of the extracted feature values ​​obtained from the molding data of non-defective products, and determine the difference between the feature values ​​obtained from the target molding data and the representative value of the feature values ​​obtained from the molding data of non-defective products as the evaluation index. In this case, the output unit 553 may output information for presenting the differences between the multiple feature values ​​in a comparable manner, and / or information for presenting feature values ​​or molding condition items sorted according to the differences between the multiple feature values. This allows the operator to identify feature values ​​with large differences for the target molding data, and / or identify feature values ​​with large differences or molding condition items corresponding to feature values ​​with large differences, making it easier for the operator to take action to stabilize quality.

[0103] (4) In addition to at least one of (1) to (3), the processor 55 may calculate the contribution of a feature (contributing to the quality of a molded product) for each sensor position and output information for presenting (e.g., displaying, audio output, etc.) this contribution. In FIGS. 7 and 9, for example, channel 1 (Ch1) is the gate side of the mold, and channel 2 (Ch2) is the end of the mold. The processor 55 may calculate the contribution of each feature (e.g., peak pressure, pressure area, etc.) for each channel (i.e., sensor position) and output information for presenting (e.g., displaying, audio output, etc.) this contribution. As a specific example of this output, the processor 55 may output information for displaying the contribution of each feature for each sensor position. Since the contribution of each feature for each sensor position is presented to the operator, the operator can recognize that, for each sensor position, the operator should prioritize the operation of the molding condition item (e.g., injection speed) corresponding to the feature (e.g., peak pressure) with the highest contribution.

[0104] The processor 55 may output information for presenting (for example, by displaying or outputting audio) the degree of contribution along with the deviation amount for each feature amount. As a result, the degree of contribution is presented to the operator along with the deviation amount, so that the operator can understand that the molding condition item (for example, "injection speed") corresponding to the feature amount (for example, peak pressure) with a large deviation amount and high degree of contribution should be operated with priority.

[0105] Furthermore, as shown in FIG. 10, the database showing the relationship between feature values ​​and molding condition items stored in the storage unit 53 may further store sensor position information and the relationship between the feature value and the deviation direction (positive or negative). FIG. 10 shows an example of a table in which feature values, the deviation direction of the feature value, sensor position, and molding condition items are stored in association with each other in a modified example. In this case, the processor 55 may read from the database a molding condition item corresponding to a combination of a target feature value (e.g., a feature value ranked high), the deviation direction of the feature value, and the sensor position, and output information for presenting (e.g., displaying, outputting audio, etc.) the read molding condition item. As a specific example of this output, the processor 55 may output information for auxiliary display of the molding condition item on a screen. Specifically, for example, if the feature value is "peak pressure," and the sensor position is "mold end" (e.g., channel 2: Ch2) and the "deviation direction of the feature value" is negative, there is a high possibility of a filling defect. To address this situation, the processor 55 may, for example, refer to a database and look up a record in which the feature is "peak pressure," the sensor position is "mold end," and the deviation direction of the feature is "negative." The processor 55 then reads out the "injection speed" from the molding condition items in that record and outputs information recommending a change to the "injection speed." In this way, the output unit 553 may refer to the storage unit 53, which stores the feature, the deviation direction of the feature relative to the representative value of the feature obtained from molding data of non-defective products, the sensor position, and the molding condition item in association with each other, read out the molding condition item corresponding to the target feature, the deviation direction of the target feature, and the target sensor position, and output information for presenting the read molding condition item. This allows for more accurate presentation of the molding condition item to be changed.

[0106] At least a part of the information processing system 5 described in the above embodiment may be configured with hardware or software. When configured with software, a program that realizes at least a part of the functions of the information processing system 5 may be stored in a computer-readable recording medium and read and executed by a computer. The recording medium is not limited to removable recording media such as magnetic disks and optical disks, but may also be fixed recording media such as hard disk drives and memories.

[0107] In addition, a program that realizes at least a part of the functions of the information processing system 5 may be distributed via a communication line (including wireless communication) such as the Internet. Furthermore, the program may be encrypted, modulated, or compressed and distributed via a wired line or wireless line such as the Internet, or stored on a recording medium.

[0108] Furthermore, the information processing system 5 may be operated by one or more information devices. When multiple information devices are used, at least one of the devices may be a computer, and the computer may execute a predetermined program to realize the function as at least one means of the information processing system 5.

[0109] In the method invention, all processes (steps) may be realized by automatic control using a computer. Alternatively, each process may be performed by a computer, with progress control between processes being performed manually. Furthermore, at least some of the processes may be performed manually.

[0110] As described above, the present invention is not limited to the above-described embodiments, and the components can be modified and embodied in practice without departing from the spirit of the invention. Furthermore, various inventions can be formed by appropriately combining multiple components disclosed in the above-described embodiments. For example, some components may be omitted from all the components shown in the embodiments. Furthermore, components from different embodiments may be appropriately combined. [Explanation of symbols]

[0111] 1 Molding machine 10 Injection Unit 11 cylinders 12 Hopper 13 Screw 14 Check valve 15 Heater 16 nozzles 17 Motor 2 Instrumentation amplifier 20 Mold clamping unit 21 Tie bar 22 Mold 23 Sensors 3 A / D converter 4. Control device 5. Information Processing Systems 51 Input section 52 Output section 53 Memory section 54 Communications Department 55 processors 550 Decision Section 551 Extraction part 552 Contribution Determination Unit 553 Output section 554 Acquisition Department 555 Score Determination Section 556 Abnormality determination unit 6 Display device

Claims

1. an extraction unit that extracts a plurality of feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or more molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit that determines an evaluation index for evaluating a feature quantity that contributes to improving the quality of a molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, or a function that uses the extracted feature quantities as explanatory variables; and an output unit that outputs information for presenting the evaluation indexes of the plurality of feature amounts in a comparable manner and / or information for presenting information on the feature amounts sorted according to the evaluation indexes of the plurality of feature amounts; An information processing system comprising:

2. the determination unit determines, as the evaluation index, a contribution degree of each of the extracted feature quantities to the quality of the molded product by using machine learning, statistical data analysis, or a function that uses the extracted feature quantities as explanatory variables for the extracted feature quantities; The output unit outputs information for presenting the contribution degree of each of the plurality of feature amounts in a comparative manner and / or information for presenting information about the feature amounts sorted according to the contribution degree of each of the plurality of feature amounts. The information processing system according to claim 1 .

3. the machine learning is principal component analysis, The determination unit determines at least one principal component vector by performing principal component analysis on the formed data, and determines each element of the principal component vector as the contribution of each feature amount. The information processing system according to claim 2 .

4. when determining the contribution degrees, the determiner determines the contribution degree of each feature amount to a first principal component and the contribution degree of each feature amount to a second principal component; The output unit outputs information for presenting a contribution degree of each feature amount to the first principal component and a contribution degree of each feature amount to the second principal component in a comparable manner. The information processing system according to claim 3 .

5. The statistical data analysis is a multiple regression analysis, The determination unit executes a multiple regression analysis using the value representing the quality of the molded product as a response variable and each of the feature quantities as an explanatory variable, and determines each of the obtained regression coefficients as a contribution of the corresponding feature quantity. The information processing system according to claim 2 .

6. The function using the plurality of feature quantities as explanatory variables is an objective function that includes in its formula a difference between corresponding components of a representative vector using, as its elements, representative values ​​of the plurality of feature quantities in one shot of a plurality of past non-defective molded products, and an explanatory variable vector including, as elements, the plurality of feature quantities as explanatory variables, The determination unit selects one feature from the plurality of feature values ​​included in the explanatory variable vector and determines the degree of contribution of the one feature value to the quality of the molded product based on the magnitude of change in the objective function when the value of the one feature value is slightly changed. The information processing system according to claim 2 .

7. When determining the contribution degree, the determination unit determines the contribution degree of each of a plurality of feature amounts for a non-defective molded product and the contribution degree of each of a plurality of feature amounts for a defective molded product; The output unit outputs information for presenting a contribution degree of each of a plurality of feature amounts in the case of the non-defective molded product and a contribution degree of each of a plurality of feature amounts in the case of the defective molded product in a comparative manner.

4. The information processing system according to claim 2 or 3.

8. an acquisition unit that acquires selected feature amounts that are feature amounts selected by a user; a score determination unit that determines a score representing a difference between a representative vector determined from a plurality of vectors representing the selected feature amounts in one shot of a plurality of past non-defective molded products, and a target vector that is a vector of the selected feature amounts in the molding data of one shot of the molded product to be judged, based on the representative vector and the target vector; Equipped with The output unit outputs information for comparatively presenting at least one score when the molded product to be judged is a non-defective product and at least one score when the molded product to be judged is a defective product.

3. The information processing system according to claim 1.

9. the representative vector is a vector representing the center of gravity of the plurality of vectors, The score is a value of an objective function that includes the difference between each corresponding component of the representative vector and the target vector. The information processing system according to claim 8 .

10. The apparatus further includes an abnormality degree determining unit that determines the abnormality degree of the molding data in one shot of the molded product to be judged based on a comparison between at least one feature amount in the molding data in one shot of the target molded product extracted by the extracting unit and a representative value of at least one feature amount in the molding data in one shot of a plurality of past non-defective molded products.

3. The information processing system according to claim 1.

11. an acquisition unit that acquires a feature quantity selected by a user from the plurality of feature quantities extracted by the extraction unit; The abnormality degree determining unit determines the abnormality degree based on the feature amount selected by the user. The information processing system according to claim 10.

12. When outputting the information, the output unit replaces each name of the plurality of feature amounts with a property name indicating what property the feature amount relates to, and outputs information for presenting the contribution degree in a comparable manner.

4. The information processing system according to claim 2 or 3.

13. the determining unit determines a representative value of the feature amounts obtained from the molding data of a non-defective product by using statistical data analysis for the extracted plurality of feature amounts, and determines a difference between the feature amount obtained from the target molding data and the representative value of the feature amount obtained from the molding data of a non-defective product as the evaluation index; The output unit outputs information for presenting the differences between the plurality of feature amounts in a comparable manner, and / or information for presenting the feature amounts or molding condition items rearranged according to the differences between the plurality of feature amounts. The information processing system according to claim 1 .

14. The output unit refers to a storage device in which the relationships between feature amounts and molding condition items are stored, reads out molding condition items corresponding to the plurality of feature amounts from the storage device, and outputs information for presenting molding condition items rearranged according to the differences between the plurality of feature amounts. The information processing system according to claim 13.

15. The output unit refers to a storage device in which feature amounts, the direction of deviation of feature amounts from representative values ​​of feature amounts obtained from molding data of non-defective products, sensor positions, and molding condition items are stored in association with each other, reads out molding condition items corresponding to the target feature amounts, the direction of deviation of the target feature amounts, and the target sensor positions, and outputs information for presenting the read out molding condition items.

15. The information processing system according to claim 13 or 14.

16. an extraction unit extracting a plurality of feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or more molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit determining an evaluation index for evaluating a feature quantity that contributes to quality improvement of the molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, and a function that uses the extracted feature quantities as explanatory variables; an output unit outputs information for presenting the evaluation indexes of the plurality of feature amounts in a comparable manner, and / or information for presenting information on the feature amounts sorted according to the evaluation indexes of the plurality of feature amounts; An information processing method comprising:

17. On the computer, an extraction unit that extracts a plurality of feature quantities related to the quality of one or more molded products from molding data, which is either time-series data of physical quantities acquired by a sensor in one shot of the one or more molded products molded by the molding machine, or data representing the spatial distribution of the physical quantities; a determination unit that determines an evaluation index for evaluating a feature quantity that contributes to improving the quality of a molded product for each of the extracted feature quantities by using machine learning, statistical data analysis, and a function that uses the extracted feature quantities as explanatory variables; an output unit that outputs information for presenting the evaluation indexes of the plurality of feature amounts in a comparable manner, and / or information for presenting information on the feature amounts sorted according to the evaluation indexes of the plurality of feature amounts; A program to function as a

Images