Technology for predicting dynamic characteristics of cutting processing system based on artificial intelligence

Abstract

An embodiment of the present disclosure provides a method for predicting dynamic characteristics of a tool included in a machine tool including a tool for processing a workpiece, a tool holder for supporting the tool, a spindle for transferring power to the tool and the tool holder, and an ATC in which the tool and the tool holder are received, the method including the steps in which: (a) a camera (C, C') photographs the tool and the tool holder and tool information and tool holder information are obtained using a photographed image; (b) information including the tool information and the tool holder information is input to a learning information generation module (10) and a frequency response function (FRF) of the tool is generated by the learning information generation module (10); (c) a learning module (12) learns a prediction model by using the information input to the learning information generation module (10) and the FRF generated by the learning information generation module (10); and (d) when the tool information and the tool holder information are input to the prediction model, the FRF is output, wherein the tool information includes a tool length, a tool diameter, the number of tool blades, a blade length, and a tool physical property, and the tool holder information includes a tool holder length, a tool holder diameter, and a tool holder physical property.

Description

AI-based cutting processing system dynamic characteristic prediction technology

[0001] The present disclosure relates to a method for predicting tool dynamic characteristics, and more particularly, to a device and method for predicting the dynamic characteristics of a cutting processing system including a tool by measuring the numerical values ​​of a tool and a tool holder using artificial intelligence, and based on the numerical values, in order to predict the dynamic characteristics of a spindle rotating at high speed.

[0002]

[0003] The material described in this section merely provides background information for the present disclosure and does not constitute prior art.

[0004] To process a workpiece, a machine tool is used. At this time, the machine tool includes a tool that directly contacts the workpiece to perform processing, a tool holder that supports the tool, a spindle that transmits power to the tool, and an ATC configured to accommodate the tool-tool holder assembly.

[0005] A tool is a part that directly contacts the workpiece and applies external force to it. Therefore, its dynamic characteristics must be accurately understood to ensure safe machining. To this end, experiments that directly measure the dynamic characteristics of the tool, i.e., impact examinations, have been performed in the past. Referring to Figure 1, a tool located at the end of a machine tool is illustrated. After striking the tool with an instrumented hammer, the acceleration of the tool is measured using an accelerometer attached to the tool. The measured acceleration is a result in the time domain, and when this is Fourier transformed, a result in the frequency domain can be obtained. The result in the frequency domain obtained in this way is called a frequency response feature (FRF). There is a frequency range where the FRF magnitude has a peak value at a specific frequency, and the spindle must be operated while avoiding this peak frequency range to ensure safe use of the machine tool.

[0006] However, the reliability of these impact tests is largely dependent on the skill of the operator performing the test.

[0007] Accordingly, a numerical analysis model utilizing the finite element method (FEM) has been proposed to replace impact testing. Using this model to calculate dynamic characteristics requires direct input of tool and tool holder dimensions. This requires the operator to measure the tool and tool holder dimensions themselves, which can lead to inaccurate measurements.

[0008] Accordingly, there is a growing demand for a method that can accurately and quickly predict the dynamic characteristics of tools.

[0009]

[0010] Accordingly, the present disclosure aims to provide a prediction device and a prediction method capable of accurately and quickly predicting the dynamic characteristics of a tool by automatically and accurately acquiring the numerical values ​​of the tool and tool holder.

[0011] The problems to be solved by the present invention are not limited to the problems mentioned above, and other problems not mentioned will be clearly understood by those skilled in the art from the description below.

[0012]

[0013] According to embodiments of the present disclosure, a method for predicting dynamic characteristics of a tool included in a machine tool is provided, which includes a tool for processing a workpiece, a tool holder for supporting the tool, a spindle for transmitting power to the tool and the tool holder, and an ATC for storing the tool and the tool holder, comprising: (a) a step in which, when a camera (C, C') photographs the tool and the tool holder, an image processor acquires tool information and tool holder information using the photographed image; (b) a step in which information including the tool information and the tool holder information is input to a learning information generation module (10), and a frequency response function (FRF) of the tool is generated by the learning information generation module (10); (c) a step in which a learning module (12) learns a prediction model using the information input to the learning information generation module (10) and the FRF generated by the learning information generation module (10); (d) a step of outputting FRF when tool information and tool holder information are input into the prediction model; wherein the tool information includes tool length, tool diameter, number of tool teeth, tooth length, and tool properties, and the tool holder information includes tool holder length, tool holder diameter, and tool holder properties. A prediction method is provided.

[0014] Also, preferably, when the tool and the tool holder according to one embodiment of the present disclosure are in a working position, the camera (C) photographs the tool and the tool holder.

[0015] Also, preferably, when the tool and the tool holder according to another embodiment of the present disclosure are stored in the ATC, the camera (C') photographs the tool and the tool holder.

[0016] In addition, preferably, the images captured by the cameras (C, C') of the embodiments of the present disclosure are image processed through a tool edge search technique or a tool material prediction technique using color analysis.

[0017] In addition, preferably, in a prediction method according to another embodiment of the present disclosure, the learning information generation module (10) receives tool information and tool holder information from a tool and tool holder DB (3) in which the type of tool, tool information matching the type of the tool, the type of tool holder, and tool holder information matching the type of the tool holder are stored.

[0018] In addition, preferably, the tool information acquired in step (a) of the embodiments of the present disclosure includes the overall length of the tool, the fluted length, the in-hold length, the cutting diameter, the fluted diameter, the number of tool blades, the length of the blades, the modulus of elasticity, the Poisson's ratio, the structural damping ratio, and the volumetric density.

[0019] In addition, preferably, the tool holder information acquired in step (a) of the embodiments of the present disclosure includes an overhang length of the tool holder, a tapered length, a clamping length, an overhang diameter, a larger diameter of taper, a smaller diameter of taper, a diameter of an adjustable screw, a modulus of elasticity, a Poisson's ratio, a structural damping ratio, and a volumetric density.

[0020]

[0021] As described above, according to the embodiments of the present disclosure, since tool information and tool holder information can be automatically calculated through images captured by a camera, there is no need for a user to individually measure tool information or tool holder information, and thus, there is an effect in that consistent and objective data is obtained regardless of the user's skill level.

[0022] In addition, according to one embodiment of the present disclosure, when a camera (C) is placed at a work position, the status of the tool and tool holder before, during, and after machining the workpiece can all be photographed, so that when information on the tool and tool holder is changed, it is possible to know this in real time.

[0023] Additionally, according to another embodiment of the present disclosure, the camera is positioned in a space away from the working position, thereby preventing the camera from being exposed to contaminants generated by the processing operation.

[0024]

[0025] Figure 1 illustrates a conventional impact test method of the present disclosure.

[0026] FIG. 2 is a block diagram of a prediction device according to one embodiment of the present disclosure.

[0027] FIG. 3 is a block diagram of a learning information generation module according to one embodiment of the present disclosure.

[0028] FIG. 4 is for explaining input data according to one embodiment of the present disclosure.

[0029] Figure 5 illustrates the relationship between input data and output data according to one embodiment of the present disclosure.

[0030] FIG. 6 is for explaining output data according to one embodiment of the present disclosure.

[0031] FIG. 7 is a flowchart of a prediction method according to one embodiment of the present disclosure.

[0032] FIG. 8 is an example of a method for a prediction device according to one embodiment of the present disclosure to acquire tool information and tool holder information.

[0033] FIG. 9 is an example of a method for a prediction device according to another embodiment of the present disclosure to acquire tool information and tool holder information.

[0034] FIG. 10 is an example of another method by which a prediction device according to one embodiment of the present disclosure acquires tool information and tool holder information.

[0035]

[0036] Hereinafter, some embodiments of the present disclosure will be described in detail with reference to exemplary drawings. When designating components in each drawing, it should be noted that, where possible, identical components are given identical reference numerals, even if they appear in different drawings. Furthermore, in describing the present disclosure, detailed descriptions of related known structures or functions will be omitted if they are deemed to obscure the gist of the present disclosure.

[0037] In describing components of embodiments according to the present disclosure, symbols such as first, second, i), ii), a), b) may be used. These symbols are only for distinguishing the components from other components, and the nature, order, or sequence of the components are not limited by the symbols. When a part in the specification is said to "include" or "have" a component, this does not mean that other components are excluded, but rather that other components may be included, unless explicitly stated otherwise.

[0038]

[0039] FIG. 2 is a block diagram of a prediction device according to one embodiment of the present disclosure.

[0040] Referring to FIG. 2, a prediction device (1) according to one embodiment of the present disclosure includes a learning information generation module (10) and a learning module (12).

[0041] The learning information generation module (10) is configured to input information used for learning by the learning module (12) or to generate new information using the input information. To this end, the learning information generation module (10) includes all or part of a tool information input unit (100), a tool holder information input unit (102), a spindle information input unit (104), an FRF generation unit (106), and a learning information database (108).

[0042] The tool information input unit (100) is configured to input tool-related information. Specific examples of tool information are described in detail in FIGS. 3 and 4.

[0043] The tool holder information input unit (102) is configured to input information related to the tool holder. Specific examples of the tool holder information are described in detail in FIGS. 3 and 4.

[0044] The method of obtaining tool information and tool holder information is described in detail in FIGS. 8 to 10.

[0045] The spindle information input unit (104) is configured to input information related to the spindle. The information related to the spindle may be, for example, a spindle dynamic model. The spindle dynamic model is implemented by a dynamic characteristic simulation based on a physical formula, and preferably, it may be a spindle dynamic characteristic simulation according to the finite element method (FEM). Meanwhile, the spindle dynamic model is not necessarily limited to the FEM, and any model that can model the operation of the spindle may be selected.

[0046] The FRF generation unit (106) is configured to calculate the FRF (frequency response feature) of the tool using tool information, tool holder information, and spindle dynamic model.

[0047] The learning information database (108) is configured to receive and store tool information, tool holder information, and tool FRF. Furthermore, upon request from the learning module (12), it is configured to transmit all or part of the stored information to the learning module (12).

[0048]

[0049] Figure 3 is a block diagram of a learning information generation module according to one embodiment of the present disclosure. Figure 4 is for explaining input data according to one embodiment of the present disclosure.

[0050] A data processing method in a learning information generation module (10) according to one embodiment of the present disclosure will be described with reference to FIGS. 3 and 4.

[0051] Tool length, tool diameter, number of tool blades, blade length, and tool properties can be input into the tool information input section (100).

[0052] Tool length is the overall length of the individual tool (l in Fig. 4). t ), fluted length (l in Fig. 4) f) and tool insertion depth (in-hold length, l in Fig. 4 i ) includes the total length (l) t ) means the total length of the part marked in blue in (a) of Fig. 4. In addition, the groove length (l f ) means the depth at which the tool can actually be inserted into the workpiece. Also, the tool insertion depth (l i ) refers to the depth of the tool part inserted into the tool holder.

[0053] The tool diameter is the cutting diameter of the individual tool (d in Fig. 4). t ) and fluted diameter (d in Fig. 4) f ) includes the cutting diameter (d) t ) refers to the diameter of the part inserted into the actual workpiece. In addition, the groove diameter (d f ) refers to the diameter of the tool part inserted into the tool holder.

[0054] Tool properties include the modulus of elasticity, Poisson's ratio, structural damping ratio, and volumetric density of the individual tool.

[0055]

[0056] Tool holder length, tool holder diameter, and tool holder properties can be input into the tool holder information input section (102).

[0057] The tool holder length is the overhang length of the individual tool holder (l in Fig. 4). h ), tapered length (l in Fig. 4) tp ) and clamping length (l in Fig. 4 c ) includes the overhang length (l). h) means the length of the tool holder overhang, i.e., the length of the protruding red part in (a) of Fig. 4. In addition, the clamping part length (l c ) is formed inside the tool holder and means the maximum insertion depth into which the tool can be inserted, i.e., the depth of the hole with a larger diameter among the holes formed inside the red part in (a) of Fig. 4.

[0058] The tool holder diameter is the overhang diameter of the individual tool holder (d in Fig. 4). h ), larger diameter of taper (d in Fig. 4) tp1 ), smaller diameter of taper (d in Fig. 4) tp2 ) and adjustable screw diameter (diameter of adjustable screw, d in Fig. 4 a ) is included. The overhang diameter (dh) refers to the maximum inner diameter of the element overhanging the tool holder, i.e., the part indicated in gray in (a) of Fig. 4.

[0059] The material properties of the tool holder include the modulus of elasticity, Poisson's ratio, structural damping ratio, and volumetric density of the individual tool holder.

[0060]

[0061] A spindle dynamic model can be input into the spindle information input section (104). When tool information and tool holder information are input into the spindle dynamic model, the dynamic characteristics of the tool, specifically, the response characteristics in the frequency domain, are derived.

[0062]

[0063] The FRF generation unit (106) calculates the FRF of the tool using a preset method by using the tool information input into the tool information input unit (100), the tool holder information input into the tool holder information input unit (102), and the spindle dynamic model input into the spindle information input unit (104). Here, the preset method may mean a mathematical formula stored in the FRF generation unit (106).

[0064]

[0065] The learning information database (108) receives tool information input into the tool information input unit (100), tool holder information input into the tool holder information input unit (102), and FRF generated by the FRF generation unit (106), and stores the data as a dataset.

[0066] The learning information database (108) can transmit the stored data set to the learning module (12) when requested by the learning module (12).

[0067]

[0068] Fig. 5 illustrates the relationship between input data and output data according to one embodiment of the present disclosure. Fig. 6 is for explaining output data according to one embodiment of the present disclosure.

[0069] Referring to FIG. 5, the learning module (12) can learn a prediction model using tool information and tool holder information as input data and the FRF of the tool as output data among the data received from the learning information database (108). By simply inputting the tool information and tool holder information into the prediction model learned in this way, the prediction model can predict the FRF of the tool, i.e., the dynamic characteristics of the tool.

[0070] The prediction model may preferably be one of ANN (artificial neural network), CNN (convolutional neural network), RNN (recurrent neural networks), RL (reinforced learning), Bayesian network, and decision tree, but is not necessarily limited thereto, and any type of prediction model may be used.

[0071] Fig. 6 illustrates the FRF predicted using the prediction model. Referring to Fig. 6 (a) to (c), it can be confirmed that the experimental values ​​(black solid lines in each graph of Fig. 6) and the predicted values ​​predicted by the learning model according to one embodiment of the present disclosure (red dots and solid lines in each graph of Fig. 6) show a high degree of agreement.

[0072]

[0073] FIG. 7 is a flowchart of a prediction method according to one embodiment of the present disclosure.

[0074] Referring to FIG. 7, a method for predicting tool dynamic characteristics using a prediction method according to one embodiment of the present disclosure is described. While steps S700 to S720 in the present disclosure are depicted as being performed sequentially, this is for the purpose of explaining the distinction between each step, and it should be noted that steps S700 to S720 may be input simultaneously or in a reversed order.

[0075] Tool information is entered into the tool information input section (100) (S700). At this time, the tool information includes tool length, tool diameter, number of tool blades, blade length, and tool properties.

[0076] Tool holder information is entered into the tool holder information input section (102) (S710). At this time, the tool holder information includes tool holder length, tool holder diameter, and tool holder properties.

[0077] A spindle dynamic model is input into the spindle information input section (104) (S720). At this time, the spindle dynamic model is a dynamic characteristic simulation based on physical formulas and is configured to simulate the operation of the spindle.

[0078] Tool information, tool holder information, and a spindle dynamic model are input into the FRF generation unit (106) (S730). The FRF generation unit (106) uses the input information and the spindle dynamic model to calculate the dynamic characteristics of the tool, i.e., the frequency response characteristics (FRF), using a preset method.

[0079] Tool information, tool holder information, and FRF are stored in the learning information database (108) (S740). At this time, it is preferable that the tool information, tool holder information, and FRF be stored as a single dataset. For example, the tool information, tool holder information, and FRF can be stored as a single vector, thereby forming a single dataset.

[0080] The learning module (12) sends a data transmission request to the learning information generation module (10), and the learning information generation module (10) that receives the data transmission request transmits the data stored in the learning information database (108) to the learning module (12). At this time, tool information and tool holder information are input data, and FRF is input as output data to the learning module (12) (S750).

[0081] The learning module (12) trains a prediction model configured to predict the dynamic characteristics of the tool, i.e., the frequency response characteristics, using the received data set (S760).

[0082]

[0083] FIG. 8 is an example of a method for a prediction device according to one embodiment of the present disclosure to acquire tool information and tool holder information.

[0084] Referring to FIG. 8, a method for a prediction device according to one embodiment of the present disclosure to acquire tool information and tool holder information is described.

[0085] The machine tool (2) includes all or part of a support (20), a worktable (21), a workpiece (22), an ATC (23), a tool (24), and a tool holder (25).

[0086] The support (20) is in contact with the ground and is configured to support the entire machine tool (2).

[0087] The worktable (21) is placed on the support (20), and is a space where the workpiece (22) is placed on the worktable (21) and processed. In the present disclosure, the shape of the worktable (21) is illustrated as having a groove formed in the center, but is not necessarily limited thereto, and any shape may be used.

[0088] An automatic tool changer (ATC) (23) is also referred to as an automatic tool exchanger. The ATC (23) is configured to accommodate one or more tool and tool holder assemblies and automatically exchange used tool and tool holder assemblies with other tool and tool holder assemblies. In the present disclosure, the ATC is illustrated as having a drum shape, but a chain shape may be adopted, and various other shapes are possible.

[0089] The tool (24) is configured to come into direct contact with the workpiece and process the workpiece.

[0090] The tool holder (25) is configured to have at least a portion of the tool (24) inserted therein and to fix the tool (24).

[0091] The camera (C) of the prediction device (1) according to one embodiment of the present disclosure is arranged to photograph the tool (24) and the tool holder (25) when the tool (24) and the tool holder (25) are placed at a position for performing processing, i.e., a working position.

[0092] The image captured by the camera (C) can be transmitted to an image processor (not shown) for processing. The image processor uses the image to calculate tool information and tool holder information. Meanwhile, the image processor may be a component included in a separate computing device, or may be a part of the prediction device (1) according to an embodiment of the present disclosure.

[0093] Since tool and tool holder information can be calculated from images captured by the camera (C), users do not need to individually measure tool or tool holder information. Therefore, consistent and objective data is obtained regardless of the user's skill level.

[0094] Meanwhile, when a camera (C) is placed adjacent to a tool (24) and a tool holder (25) placed at a work position, the state of the tool (24) and the tool holder (25) before machining the workpiece (22), the state of the tool (24) and the tool holder (25) during machining, and the state of the tool (24) and the tool holder (25) after machining can all be photographed, so that when information on the tool (24) and the tool holder (25) is changed, this can be known in real time, which has the advantage of being possible.

[0095] The image acquired by the camera (C) can be processed through image processing using machine vision to acquire tool information and tool holder information (see Fig. 4). At this time, either a tool edge search technique or a tool material prediction technique using color analysis can be selected as the image processing technique.

[0096]

[0097] FIG. 9 is an example of a method for a prediction device according to another embodiment of the present disclosure to acquire tool information and tool holder information.

[0098] Referring to FIG. 9, a method for a prediction device according to another embodiment of the present disclosure to acquire tool information and tool holder information is described.

[0099] The camera (C') of the prediction device (1) according to another embodiment is arranged to photograph the tool (24) and tool holder (25) stored in the ATC (23).

[0100] That is, since the camera (C') is positioned in a space separated from the working position, contaminants, etc. generated by the processing operation can be prevented from spreading to the camera (C'). In the present disclosure, the ATC (23) is illustrated as being positioned adjacent to the workpiece (22), but this is merely an illustration for understanding, and in reality, the ATC (23) is generally positioned at a location separated by a certain distance or more from the workpiece (22). Therefore, by positioning the camera (C') adjacent to the ATC (23) to photograph the tool (24) and tool holder (25) stored in the ATC (23), there is an advantage in that the tool (24) and tool holder (25) can be photographed while protecting the camera (C') from contaminants.

[0101] The image acquired by the camera (C') can be processed through image processing using machine vision to acquire tool information and tool holder information (see Fig. 4). At this time, either a tool edge search technique or a tool material prediction technique using color analysis can be selected as the image processing technique.

[0102]

[0103] FIG. 10 is an example of another method by which a prediction device according to one embodiment of the present disclosure acquires tool information and tool holder information.

[0104] Referring to FIG. 10, a prediction device (1) according to one embodiment of the present disclosure can obtain tool information and tool holder information by collaborating with a tool and tool holder DB (3).

[0105] The tool and tool holder DB (3) can be configured to store information provided from a tool manufacturer or tool holder manufacturer.

[0106] When a marker (not shown) attached to a tool (24) and a tool holder (25) is recognized by a marker reader (not shown), the type and information of the corresponding tool (24) and tool holder (25) are confirmed from the tool and tool holder DB (3). The confirmed information is input into the tool information input unit (100) and the tool holder information input unit (102), respectively, and can be transmitted to the learning module (12).

[0107] Here, the marker can be of any type, such as RFID, QR code, or barcode.

[0108] Alternatively, tool and tool holder information can be obtained through collaboration between the tool and tool holder DB (3) and the camera (C, C'). For example, tool information and tool holder information not stored in the tool and tool holder DB (3) can be confirmed by the camera (C, C'). This has the effect of obtaining tool information and tool holder information with higher reliability and accuracy.

[0109]

[0110] The above description is merely an example of the technical idea of ​​the present embodiment, and those skilled in the art will appreciate that various modifications and variations can be made without departing from the essential characteristics of the present embodiment. Therefore, the present embodiments are not intended to limit the technical idea of ​​the present embodiment, but rather to explain it, and the scope of the technical idea of ​​the present embodiment is not limited by these embodiments. The scope of protection of the present embodiment should be interpreted by the claims below, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of rights of the present embodiment.

Claims

1. A method for predicting the dynamic characteristics of a tool included in a machine tool, which is equipped with a tool for processing a workpiece, a tool holder for supporting the tool, a spindle for transmitting power to the tool and the tool holder, and an ATC for storing the tool and the tool holder. (a) a step in which, when the camera (C, C') photographs the tool and the tool holder, an image processor acquires tool information and tool holder information using the photographed image; (b) A step in which information including the tool information and the tool holder information is input into the learning information generation module (10), and the FRF (frequency response function) of the tool is generated by the learning information generation module (10); (c) A step in which the learning module (12) learns a prediction model using the information input to the learning information generation module (10) and the FRF generated by the learning information generation module (10); (d) a step of outputting FRF when tool information and tool holder information are input into the above prediction model; including, The above tool information includes tool length, tool diameter, number of tool teeth, tooth length and tool properties. The above tool holder information includes tool holder length, tool holder diameter and tool holder properties. How to predict.

2. In paragraph 1, When the tool and the tool holder are in the working position, the camera (C) photographs the tool and the tool holder. How to predict.

3. In paragraph 1, When the tool and the tool holder are stored in the ATC, the camera (C') photographs the tool and the tool holder. How to predict.

4. In any one of paragraphs 1 to 3, The images captured by the above cameras (C, C') are image processed using a tool edge search technique or a tool material prediction technique using color analysis. How to predict.

5. In paragraph 1, The learning information generation module (10) receives tool information and tool holder information from the tool and tool holder DB (3) in which the type of tool, tool information matching the type of tool, type of tool holder, and tool holder information matching the type of tool holder are stored. How to predict.

6. In paragraph 1, The tool information acquired in the above step (a) includes the overall length of the tool, fluted length, in-hold length, cutting diameter, fluted diameter, number of tool teeth, length of teeth, modulus of elasticity, Poisson's ratio, structural damping ratio, and volumetric density. How to predict.

7. In paragraph 6, The tool holder information acquired in the above step (a) includes an overhang length of the tool holder, a tapered length, a clamping length, an overhang diameter, a larger diameter of taper, a smaller diameter of taper, a diameter of adjustable screw, a modulus of elasticity, a Poisson's ratio, a structural damping ratio, and a volumetric density. How to predict.

Images