Estimation model generating device and machining state estimation device

Abstract

This estimation model generation apparatus generates an estimation model for estimating a processing state of laser processing, the estimation model generation apparatus comprising: an information acquisition unit whereby first thermal radiation, first visible light, first reflected light, and first laser light are observed from a workpiece during laser processing by a first device, and first waveform data including a first waveform and a second waveform with respect to at least two from among the first thermal radiation, first visible light, first reflected light, and first laser light are acquired; an estimation model generation unit for performing machine learning using teacher data in which the first waveform data are used as an explanatory variable and a processing state is associated as an object variable, and generating a first estimation model for estimating the state of processing by the first device; and a storage unit for storing the first estimation model.

Description

[Technical field]

[0001] The present disclosure relates to a method for generating an estimation model of a laser machining state and a machining state estimation device. [Background technology]

[0002] When determining the processing state in laser processing, a method is known in which visible light, reflected light, heat radiation, or the like emitted from a workpiece during laser processing is detected.

[0003] For example, the welding condition judgment device described in Patent Document 1 detects the intensity of plasma light and reflected light emitted from the workpiece during laser welding, and judges the condition of the workpiece using the extreme value of a feature value extracted based on the intensity of the detected light in a specified section. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2000-153379 A Summary of the Invention

[0005] An estimation model generating device according to one aspect of the present disclosure includes: An apparatus for generating an estimation model for estimating a processing state of laser processing, comprising: During laser processing by the first device, first thermal radiation, first visible light, first reflected light, and first laser light are observed from the workpiece; an information acquiring unit that acquires first waveform data including a first waveform and a second waveform for at least two of the first thermal radiation, the first visible light, the first reflected light, and the first laser light; an estimation model generating unit that performs machine learning using teacher data in which the first waveform data is associated with the machining state as an explanatory variable and the machining state is associated with the objective variable, and generates a first estimation model that estimates the machining state by the first device; A storage unit that stores the first estimation model; Equipped with.

[0006] An estimation model generating device according to another aspect of the present disclosure, An apparatus for generating an estimation model for estimating a processing state of laser processing by a first apparatus and a second apparatus, During laser processing by the first device, a third thermal radiation, a third visible light, a third reflected light, and a third laser light are observed from the workpiece; During laser processing by the second device, a fourth thermal radiation, a fourth visible light, a fourth reflected light, and a fourth laser light are observed from the workpiece; an information acquiring unit that acquires third waveform data including a fifth waveform and a sixth waveform for at least two of the third thermal radiation, the third visible light, the third reflected light, and the third laser light, and fourth waveform data including a seventh waveform and an eighth waveform for at least two of the fourth thermal radiation, the fourth visible light, the fourth reflected light, and the fourth laser light; an information conversion unit that generates second converted data by multiplying the fifth waveform by a third coefficient calculated based on the relationship between the fifth waveform and the seventh waveform, and by multiplying the sixth waveform by a fourth coefficient calculated based on the relationship between the sixth waveform and the eighth waveform; an estimation model generating unit that generates a second estimation model that estimates a machining state by the first device and a third estimation model that estimates a machining state by the second device; a storage unit that stores the second estimation model and the third estimation model; Equipped with The estimation model generation unit A second estimation model is generated by performing machine learning using teacher data in which the third waveform data is used as an explanatory variable and the machining state is used as a target variable; A third estimation model is generated by performing machine learning using training data in which the second converted data is associated with the processing state as a response variable, with the second converted data being an explanatory variable. [Brief description of the drawings]

[0007] [Figure 1A] FIG. 1 is a block diagram showing an estimation model generating device according to a first embodiment; [Figure 1B]FIG. 1 is a block diagram showing a machining state estimation device according to a first embodiment. [Figure 1C] Block diagram showing a first device [Figure 1D] Block diagram showing a second device [Figure 2A] Graph showing first waveform data acquired by an information acquisition unit [Figure 2B] Graph showing second waveform data acquired by the information acquisition unit [Figure 3A] Normalized graph of the first waveform data [Figure 3B] Normalized graph of first transformation data [Figure 4A] Graph showing the relationship between the average signal intensity and the defocus amount in the first waveform data and the first converted data [Figure 4B] 1 is a graph showing a relationship between a correct value and an estimated value when the first converted data and the second waveform data are used as input data when estimating a machining state based on a first estimation model in a machining state estimating device. [Figure 5A] FIG. 13 is a block diagram showing an estimation model generating device according to a second embodiment. [Figure 5B] FIG. 13 is a block diagram showing a machining state estimation device according to a second embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0008] (Background to this disclosure) When monitoring the processing state during laser processing, the thermal radiation, reflected light, visible light, and laser light emitted from the laser processing portion may be detected by sensors, as in the welding state determination device described in Patent Document 1.

[0009] For example, a method is known in which an estimation model is generated by performing machine learning using training data in which such sensing data is used as explanatory variables and the processing state is used as a target variable, and the estimation model is used to determine the processing state from the sensing data during processing.

[0010] Since laser processing is a non-contact processing, the reproducibility of processing phenomena is high. In other words, if the processing state is similar, the intensities of thermal radiation, reflected light, visible light, and laser light will be similar. However, due to variations in light transmission efficiency or amplifier gain, etc., the sensing data acquired by the sensor will have machine differences for each processing device. For this reason, if an estimation model generated using sensing data from a specific processing device is used in another processing device, there is a problem that it is difficult to accurately estimate the processing state. In other words, in the welding state determination device described in Patent Document 1, there is a problem that light of different intensities is detected even in the same processing state due to machine differences in the sensor that detects the light intensity.

[0011] In addition, for devices used in production lines, it is difficult to change parameters for learning, i.e., to obtain sensing data under conditions other than the standard processing conditions. For this reason, for example, sensing data for learning may be obtained using experimental devices, and an estimation model may be created. For the reasons described above, it is difficult to use the estimation model created in this way to estimate the processing state of devices used in production lines.

[0012] Therefore, the inventors have considered a method for enabling an estimation model generated using sensing data from a specific processing device to be used in other processing devices, and have arrived at the following invention. The present disclosure provides an estimation model generating device and a machining state estimating device that enable an estimation model generated in a specific processing device to be used in other processing devices.

[0013] An estimation model generating device according to one aspect of the present disclosure includes: An apparatus for generating an estimation model for estimating a processing state of laser processing, comprising: During laser processing by the first device, first thermal radiation, first visible light, first reflected light, and first laser light are observed from the workpiece; an information acquiring unit that acquires first waveform data including a first waveform and a second waveform for at least two of the first thermal radiation, the first visible light, the first reflected light, and the first laser light; an estimation model generating unit that performs machine learning using teacher data in which the first waveform data is associated with the machining state as an explanatory variable and the machining state is associated with the objective variable, and generates a first estimation model that estimates the machining state by the first device; A storage unit that stores the first estimation model; Equipped with.

[0014] A machining state estimating device according to one aspect of the present disclosure includes: An apparatus for estimating a processing state of laser processing by a second apparatus different from the first apparatus, During laser processing by the second device, second thermal radiation, second visible light, second reflected light, and second laser light are observed from the workpiece; an information acquiring unit that acquires second waveform data including a third waveform and a fourth waveform for at least two of the second thermal radiation, the second visible light, the second reflected light, and the second laser light; an information conversion unit that multiplies the third waveform by a first coefficient calculated based on a relationship between the third waveform and the first waveform, and multiplies the fourth waveform by a second coefficient calculated based on a relationship between the fourth waveform and the second waveform, to generate first converted data; A storage unit that stores a first estimation model generated by the estimation model generation device; an estimation unit that estimates the machining state from the first conversion data based on the first estimation model; Equipped with.

[0015] According to this configuration, it is possible to provide an estimation model generating device and a machining state estimating device that can use an estimation model generated by a predetermined machining device in other machining devices. The estimation model generating device generates a first estimation model using first waveform data in a first device. By multiplying the waveform data in a second device by a predetermined coefficient, the waveform data in the second device is converted to match the waveform data in the first device, so that the same first estimation model can be used in different devices.

[0016] The first coefficient and the second coefficient may be determined such that first conversion data is generated while maintaining a ratio of average values ​​of the first waveform and the second waveform.

[0017] According to this configuration, it is possible to provide an estimation model generating device and a machining state estimating device with higher estimation accuracy.

[0018] An estimation model generating device according to another aspect of the present disclosure, An apparatus for generating an estimation model for estimating a processing state of laser processing by a first apparatus and a second apparatus, During laser processing by the first device, a third thermal radiation, a third visible light, a third reflected light, and a third laser light are observed from the workpiece; During laser processing by the second device, a fourth thermal radiation, a fourth visible light, a fourth reflected light, and a fourth laser light are observed from the workpiece; an information acquiring unit that acquires third waveform data including a fifth waveform and a sixth waveform for at least two of the third thermal radiation, the third visible light, the third reflected light, and the third laser light, and fourth waveform data including a seventh waveform and an eighth waveform for at least two of the fourth thermal radiation, the fourth visible light, the fourth reflected light, and the fourth laser light; an information conversion unit that multiplies the fifth waveform by a third coefficient calculated based on a relationship between the fifth waveform and the seventh waveform, and multiplies the sixth waveform by a fourth coefficient calculated based on a relationship between the sixth waveform and the eighth waveform, to generate second converted data; an estimation model generating unit that generates a second estimation model that estimates the machining state by the first device and a third estimation model that estimates the machining state by the second device; a storage unit that stores the second estimation model and the third estimation model; Equipped with The estimation model generation unit generating the second estimation model by performing machine learning using teacher data in which the third waveform data is used as an explanatory variable and the machining state is used as a response variable; The third estimation model is generated by performing machine learning using teacher data in which the second converted data is associated with the machining state as an explanatory variable.

[0019] According to this configuration, the waveform data in the first device can be converted to match the waveform data in the second device. Then, a third estimation model for the second device can be generated using the waveform data of the first device.

[0020] The third coefficient and the fourth coefficient may be determined so that second conversion data is generated while maintaining a ratio of the average values ​​of the seventh waveform and the eighth waveform.

[0021] According to this configuration, it is possible to provide an estimation model generating device and a machining state estimating device with higher estimation accuracy.

[0022] A machining state estimating device according to another aspect of the present disclosure includes: An apparatus for estimating a processing state of laser processing, comprising: A storage unit that stores a third estimation model generated by the estimation model generating device according to claim 4 or 5; an information acquiring unit that acquires the fourth waveform data; an estimation unit that estimates the machining state from the fourth waveform data based on the third estimation model; Equipped with.

[0023] According to this configuration, an estimation model can be generated for each processing device.

[0024] Hereinafter, the embodiments of the present disclosure will be described in detail with reference to the drawings as appropriate. However, more detailed explanations than necessary may be omitted. For example, detailed explanations of already well-known matters or duplicate explanations of substantially identical configurations may be omitted. This is to avoid the following explanation becoming unnecessarily redundant and to facilitate understanding by those skilled in the art. Note that the inventor provides the accompanying drawings and the following explanation so that those skilled in the art can fully understand the present disclosure, and does not intend for them to limit the subject matter described in the claims.

[0025] (Embodiment 1) [Overall configuration] FIG. 1A is a block diagram showing an estimation model generating device 100 according to the first embodiment. FIG. 1B is a block diagram showing a machining state estimating device 200 according to the first embodiment. FIG. 1C is a block diagram showing a first device 300. FIG. 1D is a block diagram showing a second device 400. Each of them may be installed in the same factory or in two or more sites. The estimation model generating device 100 and the machining state estimating device 200 may be integrated.

[0026] 1A to 1D, an estimation model generating device 100 and a processing state estimating device 200 according to the present embodiment will be described. The estimation model generating device 100, the processing state estimating device 200, and the devices 300 and 400 are connected to each other in a wired or wireless manner so as to be able to communicate with each other. The communication may be performed using a public line such as the Internet and / or a dedicated line.

[0027] The estimation model generating device 100 shown in Fig. 1A is a device that generates a first estimation model for estimating a machining state by using sensing data in a first device 300 shown in Fig. 1C. The estimation model generating device 100 can be constructed by using a computer system such as a PC or a workstation. The estimation model generating device 100 includes an information acquiring unit 11, an estimation model generating unit 12, and a storage unit 13. The internal configuration of the estimation model generating device 100 will be described later.

[0028] The machining state estimation device 200 shown in FIG. 1B is a device that estimates a machining state in a first device 300 or a second device 400 different from the first device 300 based on a first estimation model generated by the estimation model generation device 100 in FIG. 1A. The machining state estimation device 200 can be configured with, for example, a microcomputer, a CPU, an MPU, a GPU, a DSP, an FPGA, or an ASIC. The function of the machining state estimation device 200 may be configured with only hardware, or may be realized by combining hardware and software. The machining state estimation device 200 includes an information acquisition unit 21, an information conversion unit 22, an estimation unit 23, and a storage unit 24. The internal configuration of the machining state estimation device 200 will be described later.

[0029] The first device 300 and the second device 400 shown in Fig. 1C and Fig. 1D are laser processing devices that perform welding or cutting using laser light. By irradiating the metal plate to be processed with laser light, the metal plate can be welded or cut. The first device 300 includes a laser welding unit 31 and a first sensor 32. The second device 400 includes a laser welding unit 41 and a second sensor 42.

[0030] The first device 300 is, for example, a laser processing device for generating a first estimation model, such as an experimental device, etc. The second device 400 is, for example, a device used in a production line.

[0031] In the first device 300 and the second device 400, when the workpiece is irradiated with laser light, thermal radiation, visible light, and reflected light are generated. The thermal radiation, visible light, and reflected light generated during laser processing in the first device 300 are detected by the first sensor 32 as first thermal radiation, first visible light, and first reflected light. Furthermore, in the first device 300, the laser light irradiated during processing is detected by the first sensor 32 as a first laser light. Similarly, during laser processing in the second device 400, the second sensor 42 detects second thermal radiation, second visible light, second reflected light, and second laser light. The first sensor 32 and the second sensor 42 are, for example, photodetectors. Note that the first sensor 32 and the second sensor 42 may include different sensors for thermal radiation, visible light, reflected light, and laser light, respectively.

[0032] Thermal radiation occurs when the temperature of the workpiece irradiated with the laser increases when the workpiece is irradiated with the laser. Visible light is plasma light generated when the workpiece is irradiated with the laser and the melted workpiece absorbs the laser light. Reflected light is light that is reflected from the laser light irradiated on the workpiece.

[0033] <Estimation model generation device> The estimation model generating device 100 is a device that generates a first estimation model for estimating a processing state of laser processing in the first device 300. As described above, the estimation model generating device 100 includes the information acquiring unit 11, the estimation model generating unit 12, and the storage unit 13.

[0034] The information acquiring unit 11 acquires first waveform data including a first waveform and a second waveform for at least two of the first thermal radiation, the first visible light, the first reflected light, and the first laser light from the workpiece detected during laser processing by the first device 300. The first waveform data is acquired based on a detection value detected by the sensor 32 of the first device 300.

[0035] FIG. 2A is a graph showing the first waveform data acquired by the information acquisition unit 11. As shown in FIG. 2A, the first waveform data is data including the waveforms of the first thermal radiation, the first visible light, the first reflected light, and the first laser light during laser processing. In the example of FIG. 2A, the first waveform data includes four waveforms of the thermal radiation, the visible light, the reflected light, and the laser light. For example, among these four waveforms, the waveform of the laser light can be the first waveform, and the waveform of the visible light can be the second waveform. The first waveform data includes at least two waveforms, the first waveform and the second waveform. Note that the first waveform and the second waveform are not limited to the waveform of the laser light and the waveform of the visible light, and may be any waveform that can be detected by the sensor 32.

[0036] The number of waveforms included in the first waveform data may be at least two. The waveform of the laser light is detected by the sensor 32 from the laser light outputted from the laser welding portion 31, and therefore the output is relatively stable. Therefore, it is preferable that the first waveform data includes the waveform of the first laser light. Moreover, the number of waveforms included is not limited to two, and may be three or more.

[0037] By including two or more waveforms in the first waveform data, an estimation model with higher estimation accuracy can be generated.

[0038] The estimation model generating unit 12 performs machine learning using teacher data in which the first waveform data is associated with the processing state as an explanatory variable, and generates a first estimation model for estimating the processing state by the first device 300. The processing state indicates a state during processing, such as a shift in the focal position of the laser light, a shift in the irradiation position of the laser light, or the presence or absence of a hole.

[0039] In the first device 300, laser processing is performed under standard processing conditions used in the production line and processing conditions other than the standard processing conditions, and the first waveform data can be acquired based on the respective sensing data by the information acquisition unit 11. In this way, by acquiring data under various processing conditions, a first estimation model with higher estimation accuracy can be generated.

[0040] The storage unit 13 stores the first estimation model.

[0041] <Machining state estimation device> The processing state estimating device 200 is a device that estimates a processing state of the laser processing in the second device 400. As described above, the processing state estimating device 200 includes the information acquiring unit 21, the information converting unit 22, the estimating unit 23, and the storage unit 24.

[0042] The information acquisition unit 21 acquires second waveform data including a third waveform and a fourth waveform for at least two of the second thermal radiation, the second visible light, the second reflected light, and the second laser light detected during laser processing by the second device.

[0043] The second waveform data includes the same type of waveform as that included in the first waveform data. FIG. 2B is a graph showing the second waveform data acquired by the information acquisition unit 21. The second waveform data includes the same waveform data as the first waveform data shown in FIG. 2A, that is, four waveforms of thermal radiation, visible light, reflected light, and laser light. For example, among these four waveforms, the waveform of laser light is the third waveform, and the waveform of visible light is the fourth waveform. That is, the first waveform of the first waveform data and the third waveform of the second waveform data indicate the same laser light, and the second waveform of the first waveform data and the fourth waveform of the second waveform data indicate the same visible light.

[0044] Generally, laser processing is non-contact processing and therefore has high reproducibility. Therefore, under the same processing conditions, the first device 300 and the second device 400 have similar trends in the waveforms of the above-mentioned thermal radiation, visible light, reflected light, and laser light. However, as shown in Figures 2A and 2B, due to variations in light transmission efficiency or amplifier gain, the detected values ​​of the above-mentioned thermal radiation, visible light, reflected light, and laser light in the first device 300 and the second device 400 differ from each other.

[0045] Therefore, when estimating the processing state from the processing data acquired during processing by the second device 400 based on the first estimation model generated by performing machine learning using the first waveform data, the estimation result may not be correct.

[0046] Therefore, in this embodiment, the machining state estimating device 200 generates first converted data by converting the second waveform data acquired based on the detection value from the second device 400 to match the first waveform data by a predetermined coefficient. The first converted data is data converted to match the first waveform data based on the detection value from the first device 300.

[0047] By converting the second waveform data into the first converted data, it is possible to estimate the processing state of the laser processing by the second device 400 based on the first estimation model.

[0048] Information conversion unit 22 multiplies each waveform of the second waveform data based on the detection value in second device 400 by a predetermined coefficient to generate first conversion data. The predetermined coefficient is calculated based on the relationship between waveforms of the same type. Specifically, the first coefficient is calculated based on the relationship between the third waveform (second laser light) and the first waveform (first laser light). Also, the second coefficient is calculated based on the relationship between the fourth waveform (second visible light) and the second waveform (first visible light). When the first waveform data and the second waveform data include three or more waveforms, coefficients are calculated between waveforms of the same type, and the waveform included in the second waveform data is multiplied by the calculated coefficient.

[0049] Fig. 3A is a graph in which the first waveform data has been normalized. Fig. 3B is a graph in which the first conversion data has been normalized. The first waveform data shown in Fig. 3A is obtained by normalizing the graph in Fig. 2A so that the intensity of the laser light is 1V. In other words, the graph in Fig. 3A is a graph in which the signal intensity of each waveform in the graph in Fig. 2A has been normalized by multiplying it by 0.34.

[0050] Similarly, the first converted data shown in Fig. 3B is obtained by multiplying the graph in Fig. 2B by a coefficient and then normalizing the laser light intensity to 1 V. That is, the graph in Fig. 3B is a graph obtained by multiplying the laser light intensity of the graph in Fig. 2A by 0.15 (first coefficient), the reflected light intensity by 0.15, the visible light intensity by 0.10 (second coefficient), and the thermal radiation intensity by 0.13. Each coefficient can be calculated, for example, from the average value of the signal intensity of each waveform of the first waveform data from 1 ms to 4 ms and the average value of the signal intensity of each waveform of the second waveform data from 1 ms to 4 ms.

[0051] That is, the first coefficient is calculated so that the average value of the signal intensity of the second laser light (third waveform) of the second waveform data from 1 ms to 4 ms is converted into the average value of the signal intensity of the first laser light (first waveform) of the first waveform data from 1 ms to 4 ms. For example, the first coefficient is the ratio of the average value of the signal intensity of the first laser light (first waveform) to the average value of the signal intensity of the second laser light (third waveform). Similarly, the second coefficient is calculated so that the average value of the signal intensity of the second visible light (fourth waveform) of the second waveform data is converted into the average value of the signal intensity of the first visible light (second waveform) of the first waveform data. For example, the second coefficient is the ratio of the average value of the signal intensity of the first visible light (second waveform) to the signal intensity of the second visible light (fourth waveform). By calculating the first coefficient and the second coefficient in this way, the second waveform data can be converted into the first conversion data while maintaining the ratio of the average values ​​of the signal intensity of each waveform in the first waveform data.

[0052] It should be noted that data normalization is not necessarily performed on the first waveform data and the first converted data, and the first converted data may be generated by multiplying each waveform of the second waveform data by a predetermined coefficient.

[0053] As shown in FIGS. 3A and 3B, by multiplying each waveform of the second waveform data by a predetermined coefficient to obtain the first conversion data, it is possible to obtain a waveform that exhibits the same tendency as the first waveform data.

[0054] The storage unit 24 stores the first estimation model generated by the estimation model generating device 100.

[0055] The estimation unit 23 estimates the machining state from the first converted data based on the first estimation model. When the first converted data is input, the estimation unit 23 estimates the machining state based on the first estimation model and outputs the estimation result. The first converted data maintains the ratio of the average values ​​of the respective waveforms of the first waveform data. Therefore, when the first converted data is input to the estimation unit 23, the machining state can be estimated with high accuracy using the first estimation model.

[0056] <Verification> 4A is a graph showing the relationship between the average signal strength and the defocus amount in the first waveform data and the first converted data. The change in the average signal strength of the first waveform data and the first converted data when the defocus amount is changed was examined.

[0057] In the verification, the position where the focal position of the laser light is near the surface of the workpiece was set as defocus 0 mm, and the first waveform data and the first converted data were obtained when the focal position of the laser light was moved away from the surface of the workpiece. In Fig. 4A, the average signal intensity indicates the average value of the signal intensity, for example, between 1 ms and 4 ms in the first waveform data after normalization as shown in Fig. 3A and the first converted data after normalization as shown in Fig. 3B.

[0058] In the graph of Fig. 4A, the first laser light, the first reflected light, the first visible light, and the first thermal radiation indicate the average signal intensity of the first waveform data based on the detection values ​​in the first device 300. The first waveform data is normalized so that the average signal intensity of the first laser light is 1V. The second laser light, the second reflected light, the second visible light, and the second thermal radiation indicate the average signal intensity of the first converted data generated by multiplying the second waveform data based on the detection values ​​in the second device 400 by the respective predetermined coefficients. The first converted data is also normalized so that the average signal intensity of the second laser light is 1V.

[0059] As shown in the graph of Fig. 4A, the change in average signal intensity when the defocus amount is increased shows approximately the same transition in the first waveform data and the first converted data. Therefore, the processing state estimated from the first waveform data based on the first estimation model and the processing state estimated from the first converted data based on the first estimation model are approximately the same results.

[0060] In this embodiment, the information conversion section 22 of the machining state estimation device 200 converts the second waveform data based on the detection value in the second device 400 into first converted data. The estimation section 23 can estimate the machining state based on the first estimation model using the first converted data as an input. That is, the first estimation model generated using the first waveform data based on the detection value in the first device 300 can be used in the second device 400.

[0061] FIG. 4B is a graph showing the relationship between the correct value and the estimated value when the first conversion data and the second waveform data are input data in the machining state estimation device 200 when estimating the machining state based on the first estimation model. As shown in the graph of FIG. 4B, the estimated values ​​obtained by inputting the second waveform data to the estimation unit 23 without converting it in the information conversion unit 22 are distributed in the range of about -2.2 mm to -1.6 mm. This is a large gap compared to the correct values ​​of -0.6 mm to 0.3 mm. On the other hand, the estimated values ​​obtained by converting the second waveform data to the first conversion data in the information conversion unit 22 and inputting the first conversion data to the estimation unit 23 have an error from the correct value within ±0.15 mm. In this embodiment, it is sufficient to distinguish a defocus of about 0.5 mm, so it can be seen that a sufficiently practical estimation result is obtained.

[0062] [effect] According to the above-described embodiment, it is possible to provide an estimation model generating device and a machining state estimating device that enable an estimation model generated by a predetermined machining device to be used by other machining devices.

[0063] In a production line, it is difficult to use conditions other than the standard processing conditions used in normal production. In the above-described embodiment, sensing data suitable for various processing conditions is acquired by a first device 300 separate from the device used in the production line, and an estimation model is generated using the first waveform data based on the sensing data. Therefore, an estimation model with higher estimation accuracy can be generated.

[0064] Also, the first conversion data is generated from the second waveform data based on the sensing data of the second device 400 used in the production line. The estimation unit 23 of the machining state estimation device 200 receives the first conversion data as an input, and can output a more accurate estimation result by using the first estimation model to predict the machining state even when there is a machine difference in the sensors between the devices. That is, the first estimation model based on the sensing data of the first device 300 can be used to estimate the machining states of other devices including the second device 400.

[0065] (Embodiment 2) A second embodiment will be described with reference to Figures 5A and 5B. In the second embodiment, the same or equivalent configurations as those in the first embodiment will be denoted by the same reference numerals. In the second embodiment, descriptions that overlap with those in the first embodiment will be omitted.

[0066] Fig. 5A is a block diagram showing an estimation model generating device 110 according to the embodiment 2. Fig. 5B is a block diagram showing a machining state estimating device 210 according to the embodiment 2. The embodiment 2 is different from the embodiment 1 in that the estimation model generating device 110 includes an information conversion unit 17, and the machining state estimating device 210 does not include an information conversion unit.

[0067] In the first embodiment, the first estimation model based on the sensing data of the first device 300 generated by the estimation model generating device 100 can also be used to estimate the processing state of the second device 400. On the other hand, in the second embodiment, the estimation model generating device 110 generates a second estimation model for estimating the processing state of the first device 300 and a third estimation model for estimating the processing state of the second device 400.

[0068] The estimation model generating device 110 includes an information acquiring unit 16 , an information converting unit 17 , an estimation model generating unit 18 , and a storage unit 19 .

[0069] The information acquiring unit 16 acquires the third waveform data and the fourth waveform data. The third waveform data includes a fifth waveform and a sixth waveform for at least two of the third thermal radiation, the third visible light, the third reflected light, and the third laser light, which are observed during the laser processing by the first device 300 (see FIG. 1C). That is, the third waveform data corresponds to the first waveform data of the first embodiment. The fourth waveform data includes a seventh waveform and an eighth waveform for at least two of the fourth thermal radiation, the fourth visible light, the fourth reflected light, and the fourth laser light, which are observed during the laser processing by the second device 400 (see FIG. 1D). That is, the fourth waveform data corresponds to the second waveform data of the first embodiment.

[0070] In this embodiment, the information conversion unit 17 of the estimation model generating device 110 converts the third waveform data for the first device 300 into fourth waveform data for the second device 400, and generates second converted data.

[0071] For example, if the third laser light is the fifth waveform and the third visible light is the sixth waveform, the fourth laser light is the seventh waveform and the fourth visible light is the eighth waveform. In this embodiment, the information conversion unit 17 multiplies the fifth waveform by the third coefficient and the sixth waveform by the fourth coefficient to convert the third waveform data into the second conversion data. Just as the first coefficient in the first embodiment is calculated based on the relationship between the first waveform and the third waveform, the third coefficient in the second embodiment is calculated based on the relationship between the fifth waveform and the seventh waveform. Similarly, the fourth coefficient in the second embodiment is calculated based on the relationship between the sixth waveform and the eighth waveform. The third coefficient and the fourth coefficient are determined so that the second conversion data is generated while maintaining the ratio of the average values ​​of the seventh waveform and the eighth waveform.

[0072] The estimation model generating unit 18 generates a second estimation model by performing machine learning using teacher data in which the third waveform data is an explanatory variable and the processing state is associated with the objective variable. The second estimation model is an estimation model for estimating the processing state for the first device 300. Furthermore, the estimation model generating unit 18 generates a third estimation model by performing machine learning using teacher data in which the second conversion data is an explanatory variable and the processing state is associated with the objective variable. The third estimation model is an estimation model for estimating the processing state for the second device 400.

[0073] The machining state estimation device 210 includes an information acquisition unit 26, an estimation unit 27, and a storage unit 28. The storage unit 28 stores the second estimation model and the third estimation model generated by the estimation model generation device 110. The information acquisition unit 26 acquires the third waveform data and the fourth waveform data. The estimation unit 28 estimates the machining state of the workpiece at the first device 300 from the third waveform data based on the second estimation model. The estimation unit 28 further estimates the machining state of the workpiece at the second device 400 from the fourth waveform data based on the third estimation model.

[0074] In the first embodiment, an example has been described in which the machining state estimation device 200 estimates the machining state by using the first estimation model for the first device 300 by converting the second waveform data for the second device 400 during estimation. On the other hand, in the present embodiment, the estimation model generating device 110 generates the second estimation model and the third estimation model for the first device 300 and the second device 400, respectively.

[0075] In this embodiment, the machining state estimating device 210 can estimate the machining state by inputting the fourth waveform data based on the detection value in the second device 400 to the estimating section 27 without any particular conversion.

[0076] [effect] According to the above-described embodiment, the machining state estimating device 210 does not perform conversion processing of the waveform data, so that the time required for estimation in the estimating section 27 can be shortened. [Industrial Applicability]

[0077] The estimation model generating device and machining state estimating device according to the present disclosure are widely applicable to prediction of the machining state in a machining device that performs laser machining. [Explanation of symbols]

[0078] 100, 110 Estimation model generating device 11, 16 Information acquisition section 17 Information conversion section 12, 18 Estimation model generation section 13, 19 Memory section 16 Information acquisition department 200, 210 Machining state estimation device 21, 26 Information acquisition section 22 Information conversion unit 23, 27 Estimation part 24, 28 Storage section

Claims

1. An apparatus for estimating a processing state of laser processing in a second apparatus different from a first apparatus based on a first estimation model for estimating a processing state of laser processing, comprising: During laser processing by the first device, a first thermal radiation, a first visible light, a first reflected light, and a first laser light are observed from the workpiece; During laser processing by the second device, second thermal radiation, second visible light, second reflected light, and second laser light are observed from the workpiece; the first estimation model is a model that estimates the machining state by the first apparatus by performing machine learning using teacher data in which first waveform data including a first waveform and a second waveform for at least two of the first thermal radiation, the first visible light, the first reflected light, and the first laser light is used as an explanatory variable and the machining state is associated with the first waveform data as an objective variable, an information acquiring unit that acquires second waveform data including a third waveform and a fourth waveform for at least two of the second thermal radiation, the second visible light, the second reflected light, and the second laser light; an information conversion unit that multiplies the third waveform by a first coefficient calculated based on a relationship between the third waveform and the first waveform, and multiplies the fourth waveform by a second coefficient calculated based on a relationship between the fourth waveform and the second waveform, to generate first converted data; A storage unit that stores the first estimation model; an estimation unit that estimates the machining state of the second device from the first conversion data based on the first estimation model; Equipped with Machining state estimation device.

2. the first coefficient and the second coefficient are determined so that first conversion data is generated while maintaining a ratio of average values ​​of the first waveform and the second waveform. The machining state estimating device according to claim 1 .

3. An apparatus for generating an estimation model for estimating a processing state of laser processing by a first apparatus and a second apparatus, During laser processing by the first device, third thermal radiation, third visible light, third reflected light, and third laser light are observed from the workpiece; During laser processing by the second device, a fourth thermal radiation, a fourth visible light, a fourth reflected light, and a fourth laser light are observed from the workpiece; an information acquiring unit that acquires third waveform data including a fifth waveform and a sixth waveform for at least two of the third thermal radiation, the third visible light, the third reflected light, and the third laser light, and fourth waveform data including a seventh waveform and an eighth waveform for at least two of the fourth thermal radiation, the fourth visible light, the fourth reflected light, and the fourth laser light; an information conversion unit that multiplies the fifth waveform by a third coefficient calculated based on a relationship between the fifth waveform and the seventh waveform, and multiplies the sixth waveform by a fourth coefficient calculated based on a relationship between the sixth waveform and the eighth waveform, to generate second converted data; an estimation model generating unit that generates a second estimation model that estimates the machining state by the first device and a third estimation model that estimates the machining state by the second device; a storage unit that stores the second estimation model and the third estimation model; Equipped with The estimation model generation unit generating the second estimation model by performing machine learning using teacher data in which the third waveform data is used as an explanatory variable and the machining state is used as a response variable; generating the third estimation model by performing machine learning using teacher data in which the second converted data is associated with the machining state as an explanatory variable and the machining state is associated with the second converted data as an objective variable; Estimation model generator.

4. the third coefficient and the fourth coefficient are determined so that second conversion data is generated while maintaining a ratio of the average values ​​of the seventh waveform and the eighth waveform. The estimation model generating device according to claim 3 .

5. An apparatus for estimating a processing state of laser processing, comprising: a storage unit that stores a third estimation model generated by the estimation model generating device according to claim 3 or 4; an information acquiring unit for acquiring the fourth waveform data; an estimation unit that estimates the machining state from the fourth waveform data based on the third estimation model; Equipped with Machining state estimation device.

Images