Prediction System and Prediction Method

Abstract

To provide a prediction system, a prediction method and a program that are able to predict a defect caused by the shape of a product.SOLUTION: A prediction system 10 for predicting a defect of a target product, according to an embodiment of the present invention, includes a first learned model 103 learned based on a defect characteristic value indicating a defect associated with a portion of an existing product, an amount of feature of a three-dimensional shape of the existing product, and condition information indicating a manufacturing condition of the existing product. When the amount of feature of a three-dimensional shape of the target product is input, the first learned model 103 outputs a defect characteristic value indicating a defect associated with a portion of the target product.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present invention relates to a prediction system, a prediction method, and a program for predicting product defects.

Background Art

[0002] Conventionally, in various industries such as the automotive industry, technologies for predicting product defects have been utilized. As an example of such a technology, the characteristic prediction device disclosed in Patent Document 1 uses a neural network that outputs characteristic values of aluminum products manufactured under the manufacturing conditions when parameters indicating the manufacturing conditions of the aluminum products are input, to predict the characteristics of the aluminum products.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since the characteristic prediction device disclosed in Patent Document 1 predicts product defects only based on the manufacturing conditions of the products, there is a problem that it cannot predict defects caused by the shape of the products.

[0005] The present invention is for solving such problems, and an object thereof is to provide a prediction system, a prediction method, and a program capable of predicting defects caused by the shape of a product.

Means for Solving the Problems

[0006] A prediction system for predicting defects of a target product according to an aspect of the present invention includes a first learned model learned based on defect characteristic values indicating defects associated with parts of existing products, feature amounts of the three-dimensional shapes of existing products, and condition information indicating the manufacturing conditions of existing products. When the feature amount of the three-dimensional shape of the target product is input, the first learned model outputs a defect characteristic value indicating a defect associated with a part of the target product.

[0007] When the product is a casting, the product defects indicated by the defect characteristic value may include at least one of product seizure, shrinkage cavity, surface roughness, erosion, rough material deformation, mold cracking, and entrainment.

[0008] The defect characteristic value may include a value representing the degree of product defect.

[0009] When the product is a casting, the first learned model may be further learned by at least one of the mold volume, casting volume, casting surface area, and plate thickness of the casting.

[0010] The prediction system further includes a second learned model that outputs the feature amount of the three-dimensional shape of the existing product when the shape information indicating the three-dimensional shape of the existing product is input. The first learned model can be learned using the feature amount output by the second learned model.

[0011] When the product is a casting, the manufacturing conditions can include at least one of molten metal type, molten metal temperature, internal cooling temperature, water passing time, mold temperature, mold surface treatment, cycle time, die time, mold opening sequence, spray coating amount, spray time, and air blow sequence.

[0012] The prediction system includes a display device. The defect characteristic value indicating a defect associated with a part of the target product is displayed on the display device.

[0013] A prediction method for predicting defects of a target product according to an aspect of the present invention includes a computer For a first learned model learned based on a defect characteristic value indicating a defect associated with a part of an existing product, shape information indicating the three-dimensional shape of the existing product, and condition information indicating the manufacturing conditions of the existing product, a feature amount of the three-dimensional shape of a target product is input, and a defect characteristic value indicating a defect associated with a part of the target product is output.

[0014] Also, the computer inputs shape information indicating the three-dimensional shape of the existing product to a second learned model to output a feature amount of the three-dimensional shape of the existing product, and the first learned model can be learned using the feature amount output by the second learned model.

[0015] A program that is a learned model for predicting defects of a target product according to an aspect of the present invention, The learned model is learned based on a defect characteristic value indicating a defect associated with a part of an existing product, a feature amount of the three-dimensional shape of the existing product, and condition information indicating the manufacturing conditions of the existing product, When a feature amount of the three-dimensional shape of the target product is input, the learned model outputs a defect characteristic value indicating a defect associated with a part of the target product.

Advantages of the Invention

[0016] According to the present invention, it is possible to provide a prediction system, a prediction method, and a program capable of predicting defects caused by the shape of a product.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Mode for Carrying Out the Invention

[0018] Hereinafter, with reference to the drawings, one aspect of the present invention will be described. FIG. 1 is a block diagram showing the configuration of a prediction device 10 according to one aspect of the present invention. The prediction device 10 is a device that predicts defects in a target product. Specific examples of the prediction device 10 include, but are not limited to, information processing devices such as servers and PCs (Personal Computers). The prediction device 10 corresponds to a prediction system. The target product includes, for example, cast products used in vehicles such as automobiles.

[0019] The prediction device 10 includes an arithmetic device 100, a storage device 110, and a display device 120. The arithmetic device 100 is an arithmetic device such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). The arithmetic device 100 executes a prediction method for predicting defects in the target product by reading and executing a program stored in the storage device 110.

[0020] The storage device 110 is a storage device that stores various data such as programs executed by the arithmetic device 100, information on existing products, and information on target products. Specifically, the information on existing products includes defect characteristic values indicating defects associated with parts of the existing products, shape information indicating the three-dimensional shape of the existing products, and condition information indicating the manufacturing conditions of the existing products. The defect characteristic values and shape information of the existing products can be obtained, for example, by performing CAE (Computer-Aided Engineering) analysis on the existing products. The information on the target product includes shape information indicating the three-dimensional shape of the target product and condition information indicating the manufacturing conditions of the target product.

[0021] When the product is a casting, the defects of the product indicated by the defect characteristic values can be regarded as common casting defects. For example, specific examples of casting defects include burning, shrinkage cavity, surface ripple, erosion, rough material deformation, mold cracking, and entrainment. Rough material deformation means undesired deformation that occurs when the casting is formed in the casting process and then cooled to room temperature. Note that the defects of the product indicated by the defect characteristic values are not limited to these.

[0022] The defect characteristic value is a quantitative variable and is a value representing the degree of product defect. The defect characteristic value can represent the degree of product defect according to its magnitude.

[0023] When the product is a casting, the manufacturing conditions indicated by the condition information can be the conditions set in a general casting process. For example, specific examples of the manufacturing conditions indicated by the condition information include the type of molten metal, molten metal temperature, internal cooling temperature, water passing time, mold temperature, mold surface treatment, cycle time, die time, mold opening sequence, spray coating amount, spray time, and air blow sequence related to the production of castings. Note that the manufacturing conditions are not limited to these.

[0024] The type of molten metal is the type of molten metal. The molten metal temperature is the temperature of the molten metal. The internal cooling temperature is the temperature of the water passing through the inside of the mold to cool the casting. The mold temperature is the temperature of the mold when forming the casting. The water passing time is the time for water to pass through the inside of the mold. The mold surface treatment is heat treatment or the like performed to prevent wear on the mold surface.

[0025] The cycle time is the time required in the casting process when continuously producing castings. The cycle of the casting process consists of mold closing, pouring, solidification, mold opening, removal of the casting, application of the release agent, air blow, and mold closing.

[0026] Dwell time is one of the times that make up the cycle time, which is the time from the completion of pouring to mold opening. The mold opening sequence is the order in which molds composed of multiple molds are opened. The spray coating amount is the coating amount of the mold release agent used to facilitate the release of the casting from the mold. The spray time is the time for applying the mold release agent. The air blow sequence is the order in which the mold release agent remaining in the mold is removed by air blowing.

[0027] The program executed by the arithmetic unit 100 includes a division unit 101, a model control unit 102, a first model 103, a second model 104, a prediction accuracy determination unit 105, and a prediction unit 106. In other embodiments, integrated circuits such as FPGA (Field-Programmable Gate Array) and ASIC (Application Specific Integrated Circuit) may execute these programs. Servers, PCs, arithmetic units, and integrated circuits correspond to computers.

[0028] The division unit 101 is a program that acquires information on existing products from the storage device 110 and divides the information on existing products into defective characteristic values of existing products and shape information of existing products.

[0029] The first model 103 is a program that is learned based on defective characteristic values indicating defects associated with parts of existing products, feature amounts of the three-dimensional shapes of existing products, and condition information indicating manufacturing conditions of existing products. The first model 103 can be learned using machine learning such as deep learning. For example, in the case of deep learning, the first model 103 can be realized by a neural network. Note that machine learning is not limited to deep learning, and other methods can be adopted.

[0030] The second model 104 is a program that outputs feature amounts of the three-dimensional shape of an existing product when shape information indicating the three-dimensional shape of the existing product is input. The second model 104 can be realized by a convolutional neural network. The feature amounts of the three-dimensional shape can be expressed in the form of a feature vector.

[0031] The model control unit 102 is a program that controls the first model 103 and the second model 104. The model control unit 102 can train the second model 104 by inputting shape information indicating the three-dimensional shape of an existing product into the second model 104. The model control unit 102 can train the first model 103 by using the feature amount output by the second trained model, the defective characteristic value of the existing product obtained by the division unit 101, and one or more condition information of the existing product stored in the storage device 110.

[0032] In another embodiment, the model control unit 102 can input other feature amounts of the existing product into the first model 103 in addition to the feature amount output by the second trained model. For example, when the existing product is a casting, other feature amounts include the mold volume, casting volume, casting surface area, plate thickness of the casting, and the like. In other words, the first model 103 can be further trained by at least one of the mold volume, casting volume, casting surface area, and plate thickness of the casting.

[0033] The prediction accuracy determination unit 105 is a program that compares the defective characteristic value output by the first model 103 with the defective characteristic value of the existing product, and determines whether the prediction accuracy of defects by the first model 103 is equal to or higher than a certain accuracy. In this determination, the defective characteristic value for each part of the existing product obtained by simulation such as CAE analysis can be used as the defective characteristic value of the existing product.

[0034] The defective characteristic value can represent the degree of product defect. When the difference between the defective characteristic value output by the first model 103 and the defective characteristic value of the existing product obtained by simulation is equal to or less than a predetermined value, the prediction accuracy determination unit 105 can determine that the prediction accuracy of defects by the first model 103 is equal to or higher than a certain accuracy.

[0035] The prediction unit 106 is a program that predicts defects in a target product using the first learned model 103. Specifically, the prediction unit 106 can input a feature amount indicating a characteristic amount of the three-dimensional shape of the target product to the first learned model 103 and predict defects in the target product. The prediction unit 106 displays on the display device 120 a defect characteristic value indicating a defect associated with a part of the target product based on the defect characteristic value of the target product output by the first learned model 103.

[0036] In another embodiment, the prediction unit 106 may input, to the first learned model 103, condition information indicating one or more manufacturing conditions of the target product in addition to the feature amount indicating the characteristic amount of the three-dimensional shape of the target product, and predict defects in the target product.

[0037] FIG. 4 is a diagram showing an example of an image indicating a prediction result of a defect in a target product. In the example shown in FIG. 4, defects in the target product are displayed for each part. In the example shown in FIG. 4, circles are used for convenience to represent defects, but defects in the target product can be represented for each part using color display or various shapes. In this case, the degree of defect in the target product can be expressed by the type of color. Also, the degree of defect in the target product can be expressed by the size of the shape indicating the defect.

[0038] FIG. 2 is a flowchart showing an example of a process for training the first model 103 and the second model 104. In step S101, the division unit 101 of the prediction device 10 divides the information of the existing product into defect characteristic values and shape information. In step S102, the model control unit 102 inputs the shape information of the existing product to the second model 104.

[0039] In step S103, the second model 104 uses the shape information of the existing product to perform a convolution process on the three-dimensional shape of the existing product and generate a feature amount of the three-dimensional shape of the existing product. In step S104, the second model 104 outputs the generated feature amount of the three-dimensional shape of the existing product.

[0040] In step S105, the model control unit 102 inputs the defective characteristic values of the existing product obtained in step S101, the feature amounts of the existing product output in step S104, and the condition information of the existing product into the first model 103.

[0041] In step S106, the first model 103 associates the defective characteristic values, feature amounts, and condition information of the existing product. In step S107, the first model 103 constructs a regression formula based on the association of the defective characteristic values, feature amounts, and condition information of the existing product. In step S108, the first model 103 outputs the defective characteristic values associated with the parts of the existing product.

[0042] In step S109, the prediction accuracy determination unit 105 compares the defective characteristic values output by the first model 103 with the defective characteristic values of the existing product obtained by simulation, and determines whether the prediction accuracy of defects by the first model 103 is equal to or higher than a certain accuracy. If the prediction accuracy of defects by the first model 103 is less than a certain accuracy (NO), the process returns to step S102, and the learning of the first model 103 and the second model 104 is repeated. On the other hand, if the prediction accuracy of defects by the first model 103 is equal to or higher than a certain accuracy (YES), the process in FIG. 2 ends.

[0043] FIG. 3 is a flowchart showing an example of a process for predicting defects of a target product. In step S201, the prediction unit 106 of the prediction device 10 inputs a feature amount indicating the feature of the three-dimensional shape of the target product to the first learned model 103 learned by the process shown in FIG. 2. In step S202, the first learned model 103 outputs a defective characteristic value indicating a defect associated with a part of the target product. In step S203, the prediction unit 106 displays the defective characteristic value of the target product output by the first learned model 103 on the display device 120, and the process in FIG. 3 ends.

[0044] In the above-described embodiment, the first model 103 is learned based on defect characteristic values indicating defects associated with parts of an existing product, feature amounts of the three-dimensional shape of the existing product, and condition information indicating the manufacturing conditions of the existing product. When the feature amounts of the three-dimensional shape of the target product and the condition information indicating the manufacturing conditions of the target product are input, the first learned model 103 outputs defect characteristic values indicating defects associated with parts of the target product.

[0045] The defect characteristic values of the existing product are correlated with the defect characteristic values of the target product. Also, the three-dimensional shape of the product is correlated with the defects of the product. Furthermore, the manufacturing conditions of the product are correlated with the defects of the product. Therefore, by using the first learned model 103 learned based on the defect characteristic values of the existing product, the feature amounts of the three-dimensional shape, and the condition information indicating the manufacturing conditions, it is possible to predict defects related to the three-dimensional shape of the target product. Accordingly, it is possible to predict, for each part, defects caused by the shape of the target product such as a new product.

[0046] When the product is a casting, the defects of the product indicated by the defect characteristic values include at least one of product seizure, shrinkage cavity, surface roughness, erosion, rough material deformation, mold cracking, and entrainment. Thereby, it is possible to predict seizure, shrinkage cavity, surface roughness, erosion, rough material deformation, mold cracking, and entrainment of the target product. In particular, it is possible to predict seizure, shrinkage cavity, surface roughness, erosion, rough material deformation, and entrainment for each part of the target product.

[0047] Also, the defect characteristic values include values representing the degree of defects of the target product. Therefore, it is possible to predict the degree of defects for each part of the target product.

[0048] When the product is a casting, the first learned model 103 can be further learned by at least one of the mold volume, casting volume, casting surface area, and plate thickness of the casting. Thereby, the first learned model 103 can predict defects considering the mold volume, casting volume, casting surface area, and plate thickness of the casting.

[0049] When the product is a casting, the manufacturing conditions include at least one of the molten metal type, molten metal temperature, internal cooling temperature, water passing time, mold temperature, mold surface treatment, cycle time, die time, mold opening sequence, spray coating amount, spray time, and air blow sequence. Thereby, it is possible to predict defects for each part of the target product based on these various manufacturing conditions.

[0050] In the above example, when the program is loaded into a computer, it includes a set of instructions (or software code) for causing the computer to perform one or more functions described in the embodiment. The program may be stored in a non-transitory computer-readable medium or a tangible storage medium. By way of example and not limitation, the computer-readable medium or tangible storage medium includes random-access memory (RAM), read-only memory (ROM), flash memory, solid-state drive (SSD) or other memory technologies, CD-ROM, digital versatile disk (DVD), Blu-ray (registered trademark) disk or other optical disk storage, magnetic cassette, magnetic tape, magnetic disk storage or other magnetic storage devices. The program may be transmitted on a transitory computer-readable medium or a communication medium. By way of example and not limitation, the transitory computer-readable medium or communication medium includes electrical, optical, acoustic, or other forms of propagated signals.

[0051] The present invention is not limited to the above-described embodiments, and can be appropriately changed without departing from the spirit of the present invention. For example, in the above-described embodiment, the prediction device 10, which is a single device, corresponds to the prediction system, but in other embodiments, the prediction system may be realized by a plurality of devices. For example, the division unit 101, the model control unit 102, the first model 103, the second model 104, the prediction accuracy determination unit 105, and the prediction unit 106 implemented in the arithmetic device 100 may be distributed and implemented in a plurality of devices.

Explanation of Reference Numerals

[0052] 10 Prediction Device, Prediction System 100 Computing Device 101 Division Unit 102 Model Control Unit 103 First Model 104 Second Model 105 Prediction Accuracy Judgment Unit 106 Prediction Unit 110 Storage Device 120 Display Device

Claims

1. A prediction system for predicting defects in a target product, comprising: a first trained model which is a program machine-learned based on defect characteristic values indicating defects associated with parts of an existing product, feature amounts of the three-dimensional shape of the existing product, and condition information indicating manufacturing conditions of the existing product; a second trained model which is a program that, when shape information indicating the three-dimensional shape of the existing product is input, executes convolution processing on the three-dimensional shape of the existing product using the shape information of the existing product to output feature amounts of the three-dimensional shape of the existing product; a prediction unit that uses the first trained model to predict defects in the target product, wherein the prediction unit inputs feature amounts of the three-dimensional shape of the target product to the first trained model to output defect characteristic values indicating defects associated with parts of the target product; wherein the first trained model is machine-learned using the feature amounts output by the second trained model; a prediction system.

2. The product is a casting, and the product defects indicated by the defect characteristic values may include at least one of burning, shrinkage cavity, surface depression, erosion, rough material deformation, mold cracking, and entrainment of the product. The prediction system according to claim 1.

3. The defect characteristic values include values representing the degree of product defects. The prediction system according to claim 1 or 2.

4. When the product is a casting, the first trained model may be further learned by at least one of the mold volume, casting volume, casting surface area, and plate thickness of the casting. The prediction system according to any one of claims 1 to 3.

5. When the product is a casting, the manufacturing conditions may include at least one of molten metal type, molten metal temperature, internal cooling temperature, water passing time, mold temperature, mold surface treatment, cycle time, die time, mold opening sequence, spray coating amount, spray time, and air blow sequence. The prediction system according to any one of claims 1 to 4.

6. The prediction system includes a display device, and displays defect characteristic values indicating defects associated with parts of the target product on the display device. The prediction system according to any one of claims 1 to 5.

7. A prediction method for predicting defects in a target product, wherein a computer For a first learned model which is a program machine-learned based on a defect characteristic value indicating a defect associated with a part of an existing product, a feature amount of a three-dimensional shape of the existing product, and condition information indicating manufacturing conditions of the existing product, input a feature amount of a three-dimensional shape of the target product and output a defect characteristic value indicating a defect associated with a part of the target product. The first learned model is machine-learned using feature amounts output by a second learned model. The second learned model is a program that, when shape information indicating the three-dimensional shape of the existing product is input, uses the shape information of the existing product to perform a convolution process on the three-dimensional shape of the existing product and output a feature amount of the three-dimensional shape of the existing product. Prediction method.

8. When the product is a casting, The product defect indicated by the defect characteristic value includes at least one of product seizure, shrinkage cavity, surface depression, erosion, rough material deformation, mold cracking, and entrainment. The prediction method according to claim 7.

9. The defect characteristic value includes a value representing the degree of product defect. The prediction method according to claim 7 or 8.

10. When the product is a casting, the first learned model can be further machine-learned by at least one of the mold volume, casting volume, casting surface area, and plate thickness of the casting. The prediction method according to any one of claims 7 to 9.

11. When the product is a casting, The manufacturing conditions include at least one of molten metal type, molten metal temperature, internal cooling temperature, water passing time, mold temperature, mold surface treatment, cycle time, die time, mold opening sequence, spray coating amount, spray time, and air blow sequence. The prediction method according to any one of claims 7 to 10.

12. Display a defect characteristic value indicating a defect associated with a part of the target product on a display device. The prediction method according to any one of claims 7 to 11.

Images