KR-20260065354-A - Mold wear prediction system

KR20260065354AKR 20260065354 AKR20260065354 AKR 20260065354AKR-20260065354-A

Abstract

The present invention relates to a mold wear prediction system capable of predicting and managing wear of a progressive mold comprising a plurality of forming stages to enable a continuous forming process, comprising: a first load sensor installed on the punch holder side of a shearing stage that shear forms a material located between the progressive molds to measure a shear load value applied to a shear punch; an acceleration sensor installed on the die holder side of the shearing stage to measure an acceleration value caused by vibration occurring in a shear die; a second load sensor installed on the punch holder side of a bending stage that bends the material to measure a bending load value applied to a bending punch; and, during a continuous forming process of the material, receiving the shear load value, the acceleration value, and the bending load value in real time from the first load sensor, the acceleration sensor, and the second load sensor, respectively calculating the changes of the shear load value, the acceleration value, and the bending load value received in real time into pattern signal data patterned in the form of a graph, and the pattern signal data and the It may include a control unit that predicts the degree of wear of the progressive mold by comparing stored reference pattern signal data.

Inventors

배기현
송정한
이종섭
박남수
김민기

Assignees

한국생산기술연구원

Dates

Publication Date: 20260508
Application Date: 20241101

Claims (13)

A mold wear prediction system for predicting and managing wear of a progressive die comprising a plurality of forming stages to enable a continuous forming process, A first load sensor installed on the punch holder side of a shearing stage that shear-forms a material located between the above progressive molds, for measuring a shear load value applied to a shear punch; An acceleration sensor installed on the die holder side of the shearing stage above, which measures acceleration values caused by vibrations occurring in the shearing die; A second load sensor installed on the punch holder side of the bending stage that bends and forms the above material, for measuring a bending load value applied to the bending punch; and A control unit that, during a continuous forming process of the above material, receives the shear load value, the acceleration value, and the bending load value in real time from the first load sensor, the acceleration sensor, and the second load sensor, calculates the changes of the shear load value, the acceleration value, and the bending load value received in real time into pattern signal data patterned in the form of a graph, compares the pattern signal data with previously stored reference pattern signal data, and predicts the degree of wear of the progressive die; A mold wear prediction system including
In Article 1, The above control unit is, A shear load monitoring unit that pre-calculates and stores a reference shear load signal patterned in the form of a graph of the change over time of the shear load value when the shear punch is in a normal state without wear; A vibration monitoring unit that pre-calculates and stores a reference acceleration signal patterned in the form of a graph of the change over time of the acceleration value when the shear punch is in a normal state without wear; and A bending load monitoring unit that pre-calculates and stores a reference bending load signal patterned in the form of a graph of the change over time of the bending load value when the bending punch is in a normal state without wear; A mold wear prediction system including
In Article 2, The above shear load monitoring unit is, During the shear forming process of the above material, the shear load value is received in real time from the first load sensor, and the change of the shear load value received in real time over time is patterned into a graph form and calculated as shear load pattern signal data, and the degree of wear of the shear punch is predicted by comparing the shear load pattern signal data with the reference shear load signal. The above vibration monitoring unit is, During the shear forming process of the above material, the acceleration value is received in real time from the acceleration sensor, and the change of the acceleration value received in real time over time is patterned into a graph form to produce acceleration pattern signal data, and the degree of wear of the shear punch is predicted by comparing the acceleration pattern signal data with the reference acceleration signal. The above-mentioned banding load monitoring unit is, A mold wear prediction system that, during a bending forming process of the above material, receives the bending load value from the second load sensor in real time, calculates a bending load pattern signal data by patterning the change of the bending load value received in real time over time into a graph form, and predicts the degree of wear of the bending punch by comparing the bending load pattern signal data with the reference bending load signal.
In Paragraph 3, The above shear load monitoring unit is, In the shear load pattern signal data above, the maximum load value and the minimum load value are extracted as shear load pattern characteristics of the shear load pattern signal data, and compared with the shear load pattern characteristics of the reference shear load signal to predict the degree of wear of the shear punch, The above vibration monitoring unit is, In the above acceleration pattern signal data, the maximum acceleration value, the minimum acceleration value, and the vibration magnitude value are extracted as vibration pattern characteristics of the above acceleration pattern signal data, and compared with the vibration pattern characteristics of the above reference acceleration signal to predict the degree of wear of the above shear punch, The above-mentioned banding load monitoring unit is, A mold wear prediction system that, in the above-mentioned bending load pattern signal data, extracts the minimum load value of the contact region (Steady contact region) and the load value at the time of withdrawing as bending load pattern characteristics of the above-mentioned bending load pattern signal data, and compares them with the bending load pattern characteristics of the above-mentioned reference bending load signal to predict the degree of wear of the bending punch.
In Paragraph 3, The above shear load monitoring unit is, The shear load pattern signal data in the form of time-series data is visualized as a shaded texture image by a Recurrence Plot (RP) algorithm to extract shear load pattern characteristics, and the degree of wear of the shear punch is predicted by comparing it with the shear load pattern characteristics of the reference shear load signal. The above vibration monitoring unit is, The acceleration pattern signal data in the form of time-series data is visualized as a textured image of shading by the above recurrence plot algorithm to extract vibration pattern characteristics, and the degree of wear of the shear punch is predicted by comparing it with the vibration pattern characteristics of the reference acceleration signal. The above-mentioned banding load monitoring unit is, A mold wear prediction system that predicts the degree of wear of a banding punch by imaging the banding load pattern signal data in the form of time series data into a shaded texture image using the recurrence plot algorithm, extracting the banding load pattern characteristics, and comparing them with the banding load pattern characteristics of the reference banding load signal.
In Article 4 or Article 5, The above control unit is, A shear punch wear learning unit that stores in advance a shear punch wear detection algorithm calculated by mechanically learning by artificial intelligence the shear load pattern characteristics of the reference shear load signal, the shear load pattern characteristics of the shear load pattern signal data that may occur differently depending on the degree of wear of the shear punch, the vibration pattern characteristics of the reference acceleration signal, and the vibration pattern characteristics of the acceleration pattern signal data that may occur differently depending on the degree of wear of the shear punch; and A banding punch wear learning unit that stores in advance a banding punch wear detection algorithm calculated by mechanically learning, by artificial intelligence, the banding load pattern characteristics of the reference banding load signal and the banding load pattern characteristics of the banding load pattern signal data that may occur differently depending on the degree of wear of the banding punch; A mold wear prediction system including further
In Article 6, The above shear load monitoring unit is, By applying the shear punch wear detection algorithm to the shear load pattern characteristics extracted from the shear load pattern signal data, the degree of wear of the shear punch is predicted, and The above vibration monitoring unit is, By applying the shear punch wear detection algorithm to the vibration pattern characteristics extracted from the acceleration pattern signal data, the degree of wear of the shear punch is predicted, and The above-mentioned banding load monitoring unit is, A mold wear prediction system that predicts the degree of wear of a banding punch by applying the banding punch wear detection algorithm to the banding load pattern characteristics extracted from the banding load pattern signal data.
In Article 6, The above shear punch wear detection algorithm is, The algorithm is configured such that, by sequentially executing a Prison Component Analysis (PCA) algorithm and a Support Vector Machine (SVM) algorithm, the dimensions of the shear load pattern characteristics of the reference shear load signal and the shear load pattern characteristics of the shear load pattern signal data that may vary depending on the degree of wear of the shear punch, as well as the vibration pattern characteristics of the reference acceleration signal and the vibration pattern characteristics of the acceleration pattern signal data that may vary depending on the degree of wear of the shear punch, are reduced, and the extracted shear load pattern characteristics and vibration pattern characteristics are classified and determined. The above banding punch wear detection algorithm is, A mold wear prediction system comprising an algorithm capable of performing the above principal component analysis algorithm and the above support vector machine algorithm sequentially, reducing the dimensionality of the bending load pattern characteristics of the above bending load pattern signal data that may occur differently depending on the bending load pattern characteristics of the reference bending load signal and the degree of wear of the above bending punch, and classifying and determining the extracted bending load pattern characteristics.
In Article 1, The first load sensor and the second load sensor are, A mold wear prediction system, which is a piezo washer sensor formed in the shape of a washer and inserted into the shear punch and the banding punch on the punch holder side.
In Paragraph 3, A vision sensor installed at the rear end of the shearing stage, based on the transfer direction of the material sequentially transferred in the progressive mold, for capturing the shearing surface of the material sheared in the shearing stage; A mold wear prediction system including further
In Article 10, The above control unit is, A shear surface quality determination unit that detects the heights of the rollover zone, shear zone, and rupture zone of the shear surface of the material in the shear surface image of the material applied from the vision sensor, and determines the quality of the shear surface of the material; A mold wear prediction system including further
In Article 11, The above-mentioned shear surface quality judgment unit is, A mold wear prediction system that determines the quality of the shear surface of the material as defective when the height of the indentation area on the shear surface of the material is detected to be greater than or equal to a predetermined reference value.
In Article 12, The above control unit is, A wear management unit that calculates an acceptable degree of wear of the shear punch by analyzing the correlation between the degree of wear of the shear punch predicted by the shear load monitoring unit and the quality of the shear surface of the material determined by the shear surface quality determination unit; A mold wear prediction system including further

Description

Mold wear prediction system The present invention relates to a mold wear prediction system, and more specifically, to a mold wear prediction system capable of predicting and managing wear of a progressive mold comprising a plurality of molding stages to enable a continuous molding process. Press processing is a type of plastic processing that uses a press forming device to cut or alter a sheet-shaped material to deform its shape, thereby forming it into a product having a predetermined shape. For example, it can be processed into various forms of material such as shearing, punching, trimming, bending, burring, curling, and drawing. In such a press forming device, raw material such as a sheet is placed between the upper and lower molds of a forming die mounted inside, and then the upper mold is pressed against the lower mold and pressure is applied to form the product. A progressive die is a press processing method that sequentially transfers multiple forming processes and performs continuous operations, and is a die capable of improving processing efficiency and productivity. In addition, it is a processing method that can significantly increase production volume and improve quality with stability by automating the process compared to manual production through multiple forming processes in a conventional single-step die, and is also referred to as a sequential transfer die or a continuous die. Due to the characteristics of such progressive molds, press forming is performed as a continuous forming process under identical conditions once mass production conditions are set; however, there was a problem where quality variation occurred in products mass-produced under these conditions due to wear on the mold (punch) occurring in real time during mass production. For example, conventional press forming processes using progressive dies faced the problem that it was impossible to check the degree of wear on the punch installed inside the die in real time during mass production. Consequently, punches had to be replaced at predetermined intervals regardless of wear, or defects had to be identified through visual inspection of the final product. Furthermore, if defects were identified based on the inspection of the final product, it was difficult to pinpoint the cause of the defect when it occurred, leading to significant time and cost consumption for re-establishing mass production conditions. FIG. 1 is a cross-sectional view schematically illustrating a mold wear prediction system according to one embodiment of the present invention. Figure 2 is an image showing a material that is continuously formed by each forming stage included in the progressive mold of Figure 1. FIG. 3 is a block diagram showing an example of the configuration of the control unit of the mold wear prediction system of FIG. 1. Figure 4 shows images of shear punches fabricated for wear simulation. Figure 5 is a graph showing shear load pattern signal data by the shear punches of Figure 4. Figure 6 is a graph showing the shear load pattern characteristics of the shear load pattern signal data of Figure 5. Figure 7 is a graph showing acceleration pattern signal data by the shear punches of Figure 4. Figure 8 is a graph showing the vibration pattern characteristics of the acceleration pattern signal data of Figure 7. Figure 9 shows images of banding punches fabricated for wear simulation. Figure 10 is a graph showing the bending load pattern signal data by the bending punches of Figure 9. Figures 11 and 12 are graphs showing the banding load characteristics of banding load pattern signal data. FIG. 13 is an image showing one example of a texture image imaged by a recurrence plot algorithm. FIG. 14 is a block diagram showing another embodiment of the configuration of the control unit of the mold wear prediction system of FIG. 1. FIGS. 15 to 17 are images showing the classification results of shear load pattern signal data, acceleration pattern signal data, and banding load pattern signal data, classified by a shear punch wear detection algorithm and a banding punch wear detection algorithm composed of a principal component analysis algorithm and a support vector machine algorithm. FIG. 18 is a cross-sectional view schematically illustrating a mold wear prediction system according to another embodiment of the present invention. Figure 19 is an image showing a material that is continuously formed by each forming stage included in the progressive mold of Figure 18. FIG. 20 is a block diagram showing an example of the configuration of the control unit of the mold wear prediction system of FIG. 18. FIG. 21 is an image and graph showing embodiments for determining the quality of a shear plane based on a shear plane image of a material applied from the vision sensor of FIG. 18. Hereinafter, several preferred embodiments of the present invention will be described in detail with reference to the attached drawings. The embodiments of the present invention are p