CONTROL UNIT, LASER PROCESSING SYSTEM AND LASER PROCESSING PROCESS

Abstract

A control device (3) includes a trained model holding unit (26) for storing trained models, which stores a plurality of trained models that are the result of learning the relationship between features indicating the processing state by a processing machine and the processing state by the processing machine; a processing state evaluation unit (23) that evaluates the processing state by the processing machine by inputting the features into each of the plurality of trained models or into a combination of the plurality of trained models; and a monitoring model determination unit (24) that determines the trained model or the combination of the plurality of trained models as the monitoring model to be used for monitoring the processing state by the processing machine.based on the result of the evaluation of the machining state by observing a workpiece being machined by the machine tool and the result of the evaluation by the machining state evaluation unit (23), and a machining control unit (28) which controls the machine tool based on the result of the monitoring of the machining state by the machine tool using the monitoring model.

Description

Area

[0001] The present invention relates to a control unit that controls a laser processing machine, a laser processing system and a laser processing method. background

[0002] A control unit that operates a laser processing machine reads a preset processing condition and controls the laser processing machine accordingly. The control unit contains a multitude of processing conditions preset, related to workpiece materials, workpiece thicknesses, processing gas types, and the like. From this multitude of processing conditions, a suitable one for the task at hand is selected.

[0003] A processing condition that enables good processing by the laser processing machine may differ from the preset processing condition due to individual variations between workpieces from different manufacturers, variations between workpiece batches, or similar factors. Therefore, the control unit can monitor the processing status of the laser processing machine and adjust the processing parameters used based on the monitoring results. Appropriate adjustment of the processing parameters enables the laser processing machine to perform successful processing.

[0004] Patent document 1 describes a numerical control device that detects light emitted from the vicinity of a machining point on a workpiece and evaluates the quality of the workpiece's machining condition based on the detected light intensity distribution. The numerical control device described in patent document 1 generates a trained model by learning the relationship between the detected light intensity distribution and the quality of the machining condition, and evaluates the quality of the workpiece's machining condition using this trained model.

[0005] Patent document 1: Japanese patent application disclosure JP 2022-118 774 A Summary of the invention Problem to be solved by the invention

[0006] The state of the light emitted from the vicinity of the workpiece's processing location, when the processing condition is good, can vary due to the condition of the laser processing machine, variations in the workpiece's components, or similar factors. While the state of the light can vary as described, the conventional technique described in Patent Document 1 uses a previously generated trained model to evaluate the quality of the processing condition and therefore sometimes fails to accurately assess it. Consequently, the conventional technique described in Patent Document 1, which sometimes fails to accurately assess the quality of the processing condition, has the problem that a good processing operation by the laser processing machine may sometimes be prevented.

[0007] The present invention was made taking into account the foregoing. An objective of the present invention is to provide a control unit that enables a laser processing machine to continue processing effectively. Means to solve the problem

[0008] To solve the problems described above and achieve the objective, a control unit according to the present invention includes: a trained model holding unit for storing a plurality of trained models, which are the results of learning a relationship between features indicating a processing state of a processing machine and the processing state of the processing machine; a processing state evaluation unit for evaluating the processing state by the processing machine by inputting the features into each of the plurality of trained models or by inputting the features into a combination of the plurality of trained models;A monitoring model determination unit for determining the trained model or the combination of a multitude of trained models as the monitoring model, which is used to monitor the processing state by the processing machine, based on a result of the processing state evaluation by observing a workpiece processed by the processing machine and a result of the evaluation by the processing state evaluation unit; and a processing control unit for controlling the processing machine based on a result of the processing state monitoring by the processing machine using the monitoring model. Effects of the invention

[0009] The control unit according to the present invention has the advantage that a laser processing machine can continue good processing. Brief description of the drawings Fig. Figure 1 is a diagram illustrating an exemplary configuration of a laser processing system according to a first embodiment. Fig. Figure 2 is a diagram illustrating an exemplary configuration of a control unit included in the laser processing system according to the first embodiment. Fig. Figure 3 is a diagram illustrating an exemplary configuration of a trained model in the first embodiment. Fig. Figure 4 is a flowchart that represents a first example of a processing operation performed by the laser processing system according to the first embodiment. Fig. Figure 5 is a flowchart that represents a second example of the processing operation carried out by the laser processing system according to the first embodiment. Fig. Figure 6 is a flowchart that represents a third example of the processing operation carried out by the laser processing system according to the first embodiment. Fig. Figure 7 is a diagram illustrating an exemplary configuration of a monitoring model determination unit included in a control unit according to a second embodiment. Fig. Figure 8 is a diagram illustrating an exemplary configuration of a control unit included in a laser processing system according to a third embodiment. Fig. Figure 9 is a diagram showing an exemplary configuration of a monitoring model determination unit included in the control unit according to the third embodiment. Fig. Figure 10 is a diagram showing an exemplary configuration of a control circuit according to the first to third embodiments. Fig. Figure 11 is a diagram showing an exemplary configuration of a dedicated hardware circuit according to the first to third embodiments. Description of the embodiments

[0010] In the following, a control unit, a laser processing system and a laser processing method are described in detail according to embodiments with reference to the drawings. First embodiment.

[0011] Fig. Figure 1 is a diagram illustrating an exemplary configuration of a laser processing system 1 according to a first embodiment. The laser processing system 1 includes a laser processing machine 2 and a control unit 3. The laser processing machine 2 is a processing machine that processes a workpiece 10 with laser light. The laser processing machine 2 locally melts the workpiece 10 by irradiating it with laser light in order to process the workpiece 10. The laser processing machine 2 performs laser processing operations such as cutting, welding, additive manufacturing, or heat treatment. In the diagram shown in Figure 1, the laser processing machine 2 is used to process the workpiece 10. Fig. In the example shown, workpiece 10 is a metal plate. The laser processing machine 2 performs laser processing to cut workpiece 10. The control unit 3 controls the laser processing machine 2.

[0012] The laser processing machine 2 includes a laser oscillator 4 as a light source that emits laser light, a processing head 6, a cable 5 as a transmission line for the laser light from the laser oscillator 4 to the processing head 6, and a table on which the workpiece 10 is placed. The table is not shown. The cable 5 contains, for example, an optical fiber. The outgoing end of the cable 5 is inserted into the processing head 6.

[0013] Fig. Figure 1 schematically shows an internal configuration of the processing head 6. Inside the processing head 6 are a collimating optical system 11, an optical imaging system 12, and a protective glass 13. The laser light diverging from the outgoing end of the cable 5 falls onto the collimating optical system 11. The collimating optical system 11 emits parallel light. The optical imaging system 12 forms an image of the outgoing end of the cable 5. The protective glass 13 is a flat glass plate. The protective glass 13 is an optical component for protecting the internal configuration of the processing head 6. Fig. Figure 1 shows the collimating optical system 11 as a single lens and the optical imaging system 12 as a single lens. At least one of the collimating optical system 11 and the optical imaging system 12 can consist of a plurality of lenses.

[0014] The processing head 6 includes a processing nozzle 7 through which the laser light to be applied to the workpiece 10 and the processing gas to be blown onto the workpiece 10 pass. The laser light, which has passed through the collimating optical system 11, the optical imaging system 12, and the protective glass 13, passes through the processing nozzle 7. The processing gas is supplied from a gas supply source on the outside of the processing head 6 to the inside of the processing head 6. The laser processing machine 2 blows a jet of the processing gas from the processing head 6 through the processing nozzle 7 onto the workpiece 10. The gas supply source is not shown.

[0015] The laser processing machine 2 includes a processing head drive unit 14, which drives the processing head 6. The laser processing machine 2 moves the laser light and the workpiece 10 relative to each other by moving the processing head 6 with the processing head drive unit 14 relative to the table. The laser processing machine 2 moves the laser light and the workpiece 10 relative to each other by controlling the impact position of the laser light on the workpiece 10. The laser processing machine 2 can move the laser light and the workpiece 10 relative to each other by moving the table relative to the processing head 6 without moving the processing head 6 itself.

[0016] A motion mechanism 15, which moves the collimating optical system 11, and a motion mechanism 16, which moves the optical imaging system 12, are provided within the processing head 6. The motion mechanism 16 moves the optical imaging system 12 in the direction of the optical axis to change the image acquisition position. By changing the image acquisition position, the laser processing machine 2 changes the positional relationship between the image acquisition position and the workpiece 10 without changing the positional relationship between the processing nozzle 7 and the workpiece 10. The motion mechanism 15 adjusts the position of the collimating optical system 11 by moving the collimating optical system 11 in the direction of the optical axis while the optical imaging system 12 is moved.Alternatively, the movement mechanism 15 can adjust the divergence angle of the laser light incident on the optical imaging system 12 by moving the collimating optical system 11 in the direction of the optical axis. The optical axis in the above description is the direction of the optical axes of both the collimating optical system 11 and the optical imaging system 12. The optical axes of the collimating optical system 11 and the optical axis of the optical imaging system 12 coincide. These optical axes coincide with the central axis of the laser light incident on the workpiece 10.

[0017] Within the processing head 6, an optical system other than the collimating optical system 11 and the optical imaging system 12 may be provided. A zoom optical system may be provided within the processing head 6, which changes the size of a formed image by moving it along the optical axis. The laser processing machine 2 moves at least one of a plurality of optical systems provided within the processing head 6 along the optical axis. The image formation position may be a position on the beam waist or a position offset from the beam waist. Instead of the imaging optical system 12, a condensing optical system may be provided within the processing head 6. In the following description, the optical systems provided within the processing head 6 are sometimes collectively referred to as a processing optical system.

[0018] An example of the laser oscillator 4 is a fiber laser oscillator. The laser oscillator 4 can be a direct diode laser, a carbon dioxide laser, a copper vapor laser, or any ion laser. Alternatively, the laser oscillator 4 can be a solid-state laser incorporating a yttrium aluminum garnet (YAG) crystal or the like as the excited medium. The laser processing machine 2 can include a wavelength conversion unit that performs wavelength conversion of the laser light emitted by the laser oscillator 4.

[0019] The control unit 3 controls the laser oscillator 4, the processing head drive unit 14, the movement mechanism 15 and the movement mechanism 16 by outputting a control signal to the laser oscillator 4, the processing head drive unit 14, the movement mechanism 15 and the movement mechanism 16.

[0020] The laser processing system 1 includes a sound sensor 8 and a light sensor 9. The sound sensor 8 and the light sensor 9 each detect the processing status of the laser processing machine 2. The sound sensor 8 detects noise generated near the processing point and outputs a signal indicating the volume of the detected noise to the control unit 3. The light sensor 9 detects light emitted from the vicinity of the processing point and outputs a signal indicating the intensity of the detected light to the control unit 3. The processing point is the position on the workpiece 10 where the processing is performed and is the position where the laser light is incident.

[0021] In the following description, the data acquired by sound sensor 8 and light sensor 9, respectively, are referred to as sensor data. The sensor data acquired by sound sensor 8 indicates the volume of the sound generated during processing. The sensor data acquired by light sensor 9 indicates the intensity of the light emitted during processing. Both the sensor data acquired by sound sensor 8 and the sensor data acquired by light sensor 9 indicate the processing status of the laser processing machine 2.

[0022] The laser processing system 1 may include a sensor other than the sound sensor 8 and the light sensor 9. For example, the laser processing system 1 may include a vibration sensor that detects vibrations generated near the processing point, or a camera that captures an image of the area surrounding the processing point. The sensor data acquired by the vibration sensor indicates the intensity of the vibrations generated during processing. The control unit 3 processes an image captured by the camera to acquire data indicating the dimensions of a processed area, the melting state of the workpiece 10, or similar information. The sensor data acquired by the camera is obtained by processing a captured image of the area surrounding the processing point.The laser processing system 1 includes at least one of the following sensors: the sound sensor 8, the light sensor 9, the vibration sensor, and the camera. The laser processing system 1 can include two or more of the following components: the sound sensor 8, the light sensor 9, the vibration sensor, and the camera. The sensors included in the laser processing system 1 are not limited to those described in the first embodiment, but can be any sensors capable of detecting the processing status by the laser processing machine 2.

[0023] In the Fig. In the example shown, the light sensor 9 is arranged outside the processing head 6. The light sensor 9 can also be arranged inside the processing head 6. The light sensor 9 has a short reaction time. Thus, the laser processing system 1 equipped with the light sensor 9 can detect the processing status at high speed. The sound sensor 8 can detect noise over a large area and therefore has the advantage that its installation position can be easily determined.

[0024] Next, a configuration of control unit 3 will be described. Fig. Figure 2 is a diagram showing an exemplary configuration of the control unit 3 included in the laser processing system 1 according to the first embodiment. The control unit 3 comprises a processing monitoring unit 20, which monitors the processing status by the laser processing machine 2, and a processing control unit 28, which controls the laser processing machine 2.

[0025] During continuous processing, the laser processing machine 2 may experience processing errors due to heat accumulation in the areas comprising the processing head 6, heat accumulation in the workpiece 10, or similar factors. The processing monitoring unit 20 monitors processing errors by evaluating the processing status of the laser processing machine 2.

[0026] In the Fig. In the laser processing system 1 shown in Figure 1, the control unit 3 is a device located outside the laser processing machine 2. The laser processing machine 2 and the control unit 3 communicate with each other and transmit and receive information between the laser processing machine 2 and the control unit 3. The control unit 3 can be a device integrated within the laser processing machine 2. Alternatively, the processing control unit 28 of the control unit 3 can be implemented by a device integrated within the laser processing machine 2, and the processing monitoring unit 20 of the control unit 3 can be implemented by a device located outside the laser processing machine 2.In the case where the processing monitoring unit 20 is implemented by the device outside the laser processing machine 2, the device which implements the processing monitoring unit 20 and the device which implements the processing control unit 28 are in communication connection with each other.

[0027] The processing monitoring unit 20 includes a data acquisition unit 21, a feature calculation unit 22, a processing status evaluation unit 23, a monitoring model determination unit 24, a processing parameter correction unit 25, a trained model holding unit 26 and a monitoring model holding unit 27.

[0028] The data acquisition unit 21 acquires the sensor data output by the sound sensor 8 and the light sensor 9. That is, the data acquisition unit 21 acquires the sensor data resulting from the detection of the light generated during processing by the laser processing machine 2, and the sensor data resulting from the detection of the sound generated during processing by the laser processing machine 2. Since the sensor data is continuously input from each of the sound sensors 8 and light sensors 9, the data acquisition unit 21 acquires the sensor data as time-series data. The data acquisition unit 21 outputs the acquired sensor data to the feature calculation unit 22.

[0029] The feature calculation unit 22 calculates features from the sensor data acquired by the data acquisition unit 21. In the first embodiment, the features indicate the processing state of the laser processing machine 2. The feature calculation unit 22 outputs the calculated features to the processing state evaluation unit 23.

[0030] The features are values ​​obtained by processing sensor data and include, for example, statistics such as means or standard deviations. Features can also be values ​​obtained through processing such as frequency analysis, filter banks, or wavelet transformations. The feature calculation unit 22 can calculate combinations of a multitude of values ​​as features.

[0031] It should be noted that the features calculated by the feature calculation unit 22 are not limited to those shown here. The features can be any features that can be calculated using any typical time series data analysis method. Furthermore, the feature calculation unit 22 can directly output the sensor data values ​​as features. The feature calculation unit 22 can store the position of a feature vector, which specifies features at the start of processing, in the feature space and output the magnitude of the change in the feature vector's position after processing has begun as features. In the case where the data acquisition unit 21 acquires sensor data from a multitude of sensors, the feature calculation unit 22 can calculate features that reflect the output of each sensor, or it can calculate features that reflect combinations of the outputs from the multitude of sensors.

[0032] The trained model holding unit 26 stores a plurality of trained models. Each of the plurality of trained models is the result of learning the relationship between the features indicating the processing state of the laser processing machine 2 and the processing state of the laser processing machine 2. Each of the plurality of trained models stored in the trained model holding unit 26 is the result of learning the relationship between the features and the processing state in a corresponding case from a plurality of cases that differ from each other in at least one of the following points: thickness of the plate-shaped workpiece 10 and type of processing gas during processing by the laser processing machine 2.Thus, the control unit 3 stores the trained models in the trained model holding unit 26 for trained models for the multitude of cases that differ from each other in the thickness of the workpiece 10 or the type of processing gas.

[0033] The processing state evaluation unit 23 reads the multitude of trained models from the trained model holding unit 26. The processing state evaluation unit 23 inputs the features calculated by the feature calculation unit 22 into each of the multitude of trained models to evaluate the processing state by the laser processing machine 2. The processing state evaluation unit 23 assesses the processing state by determining its quality. That is, the processing state evaluation unit 23 determines whether the processing state by the laser processing machine 2 is good or bad. The processing state evaluation unit 23 outputs the result of the processing state assessment to the monitoring model determination unit 24.

[0034] The result of the evaluation by the processing state evaluation unit 23 is entered into the monitoring model determination unit 24. In addition to the result of the evaluation by the processing state evaluation unit 23, the result of the evaluation by observation of the workpiece 10 being processed by the laser processing machine 2 is also entered into the monitoring model determination unit 24. The monitoring model determination unit 24 determines a trained model as the monitoring model based on the result of the processing state evaluation by observation of the workpiece 10 being processed by the laser processing machine 2 and the result of the evaluation by the processing state evaluation unit 23. In the example described here, a user evaluates the processing state by observing the workpiece 10 being processed by the laser processing machine 2.The monitoring model determination unit 24 determines as a monitoring model a trained model whose evaluation result output by the processing state evaluation unit 23 matches the user evaluation result, from the multitude of trained models that are stored by the trained model holding unit 26.

[0035] A trained model to be designated as a monitoring model can be selected by the user or by the monitoring model selection unit 24. If a trained model to be designated as a monitoring model is selected by the user, the user checks the evaluation result output by the processing status evaluation unit 23 for each of the many trained models and selects a trained model whose evaluation result output by the processing status evaluation unit 23 matches the user's evaluation result. Alternatively, the user selects a trained model whose evaluation result output by the processing status evaluation unit 23 is closest to the user's evaluation result.

[0036] The user enters information specifying the selected trained model into the monitoring model determination unit 24, for example by operating an input device provided on the control unit 3. Fig. Figure 2 shows the input device not shown. The monitoring model determination unit 24 determines the trained model selected by the user as the monitoring model. The laser processing system 1 can be equipped with a display device that shows the result of the evaluation by the processing state evaluation unit 23. The user can select a trained model by checking the evaluation result displayed on the display device. Fig. 2 The display device is not shown.

[0037] In the event that a model to be trained and designated as a monitoring model is selected by the monitoring model selection unit 24, the user inputs the result of the processing state evaluation by the laser processing machine 2 into the monitoring model selection unit 24. For example, the user inputs the processing state evaluation result into the monitoring model selection unit 24 by actuating the input device associated with the control unit 3. The monitoring model selection unit 24 then selects a trained model from the multitude of trained models whose evaluation result, output by the processing state evaluation unit 23, matches the user's evaluation result.Alternatively, the monitoring model determination unit 24 selects a trained model from the multitude of trained models whose evaluation result, output by the processing status evaluation unit 23, most closely matches the user's evaluation result. The monitoring model determination unit 24 then designates the trained model selected by the monitoring model determination unit 24 as the monitoring model.

[0038] The monitoring model determination unit 24 stores the determined monitoring model in the monitoring model holding unit 27. The monitoring model holding unit 27 stores the monitoring model. The control unit 3 can obtain the monitoring model, which can perform the same evaluation as the user evaluation, by determining the monitoring model through comparison of the user evaluation result with the evaluation result output by the processing state evaluation unit 23.

[0039] The processing monitoring unit 20 monitors the processing state of the laser processing machine 2 using the monitoring model determined by the monitoring model determination unit 24. While the processing monitoring unit 20 monitors the processing state of the laser processing machine 2, the data acquisition unit 21 acquires the sensor data output by the sound sensor 8 and the light sensor 9. The feature calculation unit 22 calculates features from the sensor data acquired by the data acquisition unit 21.

[0040] While the processing monitoring unit 20 monitors the processing state of the laser processing machine 2, the processing state evaluation unit 23 reads the monitoring model from the monitoring model holding unit 27. The processing state evaluation unit 23 inputs the features calculated by the feature calculation unit 22 into the monitoring model in order to evaluate the processing state of the laser processing machine 2. The processing state evaluation unit 23 outputs the result of the processing state evaluation to the processing parameter correction unit 25.

[0041] The processing parameter correction unit 25 calculates the correction values ​​of the processing parameters based on the result of the evaluation by the processing status evaluation unit 23. In the first embodiment, the processing parameters are parameters used during processing by the laser processing machine 2. The processing parameter correction unit 25 outputs information about the calculated correction values ​​to the processing control unit 28.

[0042] The processing control unit 28 adjusts the processing parameters according to the processing conditions. Upon receiving information about the correction amounts, the processing control unit 28 corrects the processing parameters based on these amounts. The processing control unit 28 controls the laser oscillator 4, the processing head drive unit 14, the motion mechanism 15, and the motion mechanism 16 according to the corrected processing parameters.

[0043] In the description above, the processing condition evaluation unit 23 determines whether the processing condition achieved by the laser processing machine 2 is good or bad. In this case, the processing condition evaluation unit 23 outputs information indicating whether the processing condition is good or bad. Whether the processing condition is good or bad is represented by two values. The processing condition evaluation unit 23 outputs a value indicating the result of the determination. It should be noted that the evaluation by the processing condition evaluation unit 23 is not limited to determining whether the processing condition is good or bad. The processing condition evaluation unit 23 can evaluate the processing condition by calculating an evaluation value that indicates the degree of quality of the processing condition. An example of an evaluation value is a value in the range of 0% to 100%.The better the processing condition, for example, the higher the rating. The rating can be a value in the range of 0 to 10, or similar. The processing condition evaluation unit 23 outputs the calculated rating.

[0044] The processing state evaluation unit 23 can define a multitude of elements for the processing state and output an evaluation result for each element. For example, if the processing state evaluation unit 23 determines whether the processing state is good or bad, it determines for each element whether the processing state is good or bad. The processing state evaluation unit 23 outputs the determination result for each element.

[0045] Examples of elements influencing the machining condition include the presence of slag, the roughness of a cut surface, and the like. Slag is molten material that adheres to the workpiece 10 during cutting. The slag adheres near the underside of the cut surface of the workpiece 10. It can be said that the machining condition is better the smaller the slag. Roughness of the cut surface is a periodic unevenness that forms on the top surface of the cut. When roughness is present on the cut surface, the depth of the grooves formed on the cut surface is greater than when no roughness is present. It can be said that the machining condition is better the lower the roughness of the cut surface.

[0046] If oxygen is used as the processing gas in laser processing, an oxide layer forms on the cut surface. The degree to which this oxide layer flakes off can be considered a characteristic of the processing condition. In other words, the less the oxide layer flakes off, the better the processing condition.

[0047] The processing condition criteria are not limited to those mentioned above. Processing condition criteria can include elements such as discoloration of the workpiece 10 or the presence or absence of vibrations on the surface of the workpiece 10. The processing condition evaluation unit 23 can output evaluation results for one or more of the elements described in the first embodiment. The processing condition evaluation unit 23 can modify the elements to be evaluated according to the processing conditions or the like. For example, the processing condition evaluation unit 23 can modify the elements to be evaluated according to a combination of laser power, processing speed, and workpiece thickness 10. The workpiece thickness 10 is the thickness measured along the central axis of the laser light incident on the workpiece 10.

[0048] Alternatively, the processing condition evaluation unit 23 can change the elements to be evaluated according to the type of processing gas. For example, if the processing gas used in laser processing is oxygen, the processing condition evaluation unit 23 includes the detachment of the oxide layer in the elements to be evaluated. If, on the other hand, the processing gas used in laser processing is nitrogen, no oxide film forms on the cut surface, so the processing condition evaluation unit 23 omits the detachment of the oxide film from the elements to be evaluated. In the case of the laser processing machine 2, which performs welding operations, the processing condition elements can include an element such as the occurrence of spatter.In the case of laser processing machine 2, which performs additive processing, the elements for the processing state can include an element such as the height of an object or excessive melting of an object.

[0049] The processing status evaluation unit 23 can determine the processing status for two or more elements and output the result of summarizing the determinations for each element as a processing status determination result. The processing status evaluation unit 23 can output a determination result indicating that the processing status is poor, for example, if the processing status evaluation unit 23 determines that the processing status is poor for a preset number or more of the elements. Alternatively, if the processing status evaluation unit 23 determines that the processing status is poor, it can analyze the quality for each element.

[0050] As described above, the laser processing system 1 can include a display device that shows the result of the evaluation by the processing condition evaluation unit 23. The display device can be located on the laser processing machine 2 or on a device outside the laser processing machine 2. For example, the display device shows information indicating the quality of the processing condition. The display device can also show the result of the evaluation by the processing condition evaluation unit 23 only if the processing condition evaluation unit 23 determines that the processing condition is poor. In this case, the display device shows information indicating that the processing condition is poor, but does not show any information indicating that the processing condition is good.

[0051] In the description above, the processing state evaluation unit 23 assesses the processing state of the laser processing machine 2 by inputting the features calculated by the feature calculation unit 22 into each of the multitude of trained models. The processing state evaluation unit 23 can also assess the processing state of the laser processing machine 2 by inputting information other than the features calculated by the feature calculation unit 22 into each of the multitude of trained models. Examples of information input into each of the multitude of trained models include the values ​​of the processing parameters at the time of assessment, the temperature of the optical processing system at the time of assessment, any changes in the temperature of the optical processing system, the thickness of the workpiece 10, or the material of the workpiece 10.In this case, each of the many trained models is the result of learning the relationships between the information, the features, and the processing state. The processing state evaluation unit 23 inputs the information and the features into the monitoring model to evaluate the processing state by the laser processing machine 2.

[0052] The processing parameter correction unit 25 calculates the correction values ​​of the processing parameters based on the evaluation result of the processing condition evaluation unit 23. For example, if an evaluation result indicating that the processing condition is poor due to the laser processing machine 2 is entered into the processing parameter correction unit 25, the processing parameter correction unit 25 calculates the correction values ​​of the processing parameters. The processing control unit 28 corrects the processing parameters using the correction amounts calculated by the processing parameter correction unit 25. The processing monitoring unit 20 repeats the calculation of the correction amounts by the processing parameter correction unit 25 and the correction of the processing parameters by the processing control unit 28 until a determination result is obtained indicating that the processing condition is good.

[0053] The machining parameter correction unit 25 can calculate the correction amounts based on the machining parameter values ​​set in the machining control unit 28 and the result of the determination by the machining status evaluation unit 23. In this case, the machining parameter correction unit 25 retrieves the machining parameter values ​​set in the machining control unit 28 from the machining control unit 28.

[0054] Based on the evaluation results from the processing status evaluation unit 23, the processing parameter correction unit 25 can determine the presence or absence of an indication of a processing error and calculate the correction amounts if an indication of a processing error is detected. The processing monitoring unit 20 can repeat the calculation of the correction amounts by the processing parameter correction unit 25 and the correction of the processing parameters by the processing control unit 28 until no further indication of a processing error is detected.

[0055] An example of processing parameters corrected on the basis of the correction amounts calculated by the processing parameter correction unit 25 is a parameter that includes the laser power, the beam quality of the laser light, the pressure of the processing gas, the processing speed, the focal length of the optical imaging system 12, the diameter of the image produced by the optical imaging system 12, the pulse frequency of the laser oscillator 4, the pulse duty cycle of the laser oscillator 4, the magnification of the optical imaging system 12, the nozzle diameter, the distance between the workpiece 10 and the processing nozzle 7, the mode type of the laser light, or the like. The nozzle diameter is the diameter of an opening through which the laser light passes in the processing nozzle 7.A machining parameter corrected on the basis of the correction value calculated by the machining parameter correction unit 25 can be a parameter that specifies the positional relationship between the center position of the opening through which the laser light passes in the machining nozzle 7 and the center axis of the laser light.

[0056] In the event that the determination result for each of the several elements is entered into the processing parameter correction unit 25, the processing parameter correction unit 25 can determine the processing parameters to be corrected and the correction amounts of the processing parameters to be corrected based on a combination of the determination results for the individual elements.

[0057] Here, the information indicating that the processing condition is good is "1", and the information indicating that the processing condition is poor is "0". The processing condition evaluation unit 23 outputs determination results for the elements slag occurrence, cut surface roughness, and detachment of the oxide layer formed on the cut surface. If "1" for slag occurrence, "0" for cut surface roughness, and "0" for detachment of the oxide layer formed on the cut surface are entered into the processing parameter correction unit 25, the processing parameter correction unit 25 determines the processing parameters to be corrected and the correction amounts based on a combination of the entered values ​​"1, 0, 0". For example, the processing parameter correction unit 25 determines laser power and processing gas pressure as processing parameters to be corrected based on this combination.Furthermore, the processing parameter correction unit 25 determines the correction amount for the laser power and the processing gas pressure based on their combination. For example, the processing parameter correction unit 25 calculates the correction amount to increase the laser power and the correction amount to decrease the processing gas pressure.

[0058] In the processing parameter correction unit 25, the relationships between the processing parameters to be corrected and the correction amounts are predefined for each of a multitude of combinations of the determination results for the individual elements. The processing parameter correction unit 25 can determine the processing parameters to be corrected and the correction amounts based on the predefined relationships and a combination of the determination results for the individual elements.

[0059] In the description above, the processing parameter correction unit 25 calculates the correction amounts based on the result of the evaluation by the processing state evaluation unit 23, but it can also calculate the correction amounts based on the characteristics. The processing parameter correction unit 25 can calculate the correction amounts by inputting the characteristics calculated by the characteristic calculation unit 22 into the monitoring model. Each of the multitude of trained models stored by the trained model retention unit 26 can be any trained model that outputs at least either the result of the processing state evaluation or the correction amounts when the characteristics are input. If the trained model outputs the correction amounts, the correction amounts are calculated after the processing state has been evaluated using the trained model.

[0060] In the description above, the processing monitoring unit 20 repeats the calculation of the correction amounts by the processing parameter correction unit 25 and the correction of the processing parameters by the processing control unit 28 until a determination result is obtained indicating that the processing condition is satisfactory. The laser processing system 1 can stop the laser processing by the laser processing machine 2 if a situation persists in which a determination result indicating that the processing condition is satisfactory cannot be obtained.

[0061] As described above, the monitoring model determination unit 24 determines the monitoring model based on the result of the user's assessment of the machining condition by the laser machining machine 2 and the result of the assessment by the machining condition evaluation unit 23. For example, the user can determine the quality of the machining by visually inspecting the cut surface of the workpiece 10. The result of the user's assessment of the machining condition can be the result of measuring the roughness of the cut surface with a measuring device such as a 3D measuring instrument or a roughness measuring instrument. The monitoring model determination unit 24 can receive information indicating the result of the user's assessment of the machining condition.For example, by operating the input device available to the control unit 3, the user enters the result of the evaluation of the processing status into the monitoring model determination unit 24.

[0062] In the description above, the user evaluation result, derived from the assessment of the processing status by observing the workpiece 10 being processed by the laser processing machine 2, is entered into the monitoring model determination unit 24. The evaluation result entered into the monitoring model determination unit 24 is not limited to the user evaluation result. That is, the result of the processing status assessment by observing the workpiece 10 is not limited to the user evaluation result. The monitoring model determination unit 24 can receive input of the processing quality determination result from a determination device that determines the processing status by observing the workpiece 10. For example, the determination device automatically determines the processing status based on an image captured by a camera.Thus, either the result of the user's assessment of the machining state or the result of the assessment of the machining state by the monitoring device can be entered into the monitoring model determination unit 24. The monitoring model determination unit 24 can select a trained model from the multitude of trained models whose assessment result, output by the machining state evaluation unit 23, most closely approximates the result of the machining state assessment by observing the workpiece 10. The monitoring device is not shown.

[0063] In the event that the result of the user or the determining entity's evaluation is entered into the monitoring model determination unit 24, and the processing state evaluation unit 23 determines whether the processing state is good or bad, the result of the user or determining entity's determination of whether the processing state is good or bad is entered into the monitoring model determination unit 24. If the result of the determination by the processing state evaluation unit 23 matches the result of the determination by the user or the determining entity, the monitoring model determination unit 24 designates as the monitoring model the trained model that output the result of the determination by the processing state evaluation unit 23.

[0064] In the event that the result of the evaluation by the user or the determining entity is entered into the monitoring model determination unit 24, and the processing state evaluation unit 23 outputs the result of the evaluation for each of the multiple elements, the result of the determination of the processing state for each of the multiple elements by the user or the determining entity is entered into the monitoring model determination unit 24. If the result of the determination for each element by the processing state evaluation unit 23 matches the result of the determination for each element by the user or the determining entity, the monitoring model determination unit 24 designates as the monitoring model the trained model that output the result of the determination by the processing state evaluation unit 23.

[0065] If the result of the evaluation by the user or the determining entity is entered into the monitoring model determination unit 24, and the processing state evaluation unit 23 outputs the evaluation value for the processing state, an evaluation value determined by the user or the determining entity for the processing state is entered into the monitoring model determination unit 24. If the evaluation value output by the processing state evaluation unit 23 is within a valid range that includes the evaluation value determined by the user or the determining entity, the monitoring model determination unit 24 designates as the monitoring model the trained model that output the evaluation value from the processing state evaluation unit 23.The permissible range is a range based on the evaluation value determined by the user or the determining institution and is a preset numerical range.

[0066] In the description above, the monitoring model determination unit 24 selects a trained model from the multitude of trained models as the monitoring model. This model's evaluation result, output by the processing state evaluation unit 23, matches the result of the processing state evaluation obtained by observing workpiece 10. If each of the multitude of trained models outputs correction amounts, the monitoring model determination unit 24 can select a trained model to be used as the monitoring model based on the correction amounts output by each of the multitude of trained models.For example, the monitoring model determination unit 24 compares the processing parameters corrected based on the correction amounts output by each trained model with the processing parameters based on the user-customized processing condition. Through this comparison, the monitoring model determination unit 24 can identify a trained model to be used as the monitoring model.

[0067] The trained model holding unit 26, for example, holds a multitude of trained models created by the manufacturer of the laser processing machine 2. In addition to the manufacturer-generated trained models, the multitude of trained models stored in the trained model holding unit 26 can also include a user-generated trained model.

[0068] Here, an example of a trained model in its first embodiment is described. The learning algorithm used by laser processing system 1 can be a well-known algorithm such as supervised learning, unsupervised learning, or reinforcement learning. As an example, a case in which a neural network is used is described.

[0069] The trained model in the first embodiment is the result of learning the relationship between features and the processing state using supervised learning. Here, supervised learning is a technique in which a learning apparatus is given a dataset of inputs and outputs to learn features in the training data in order to derive an output from the inputs. The training data includes inputs and labels, which are the corresponding outputs. The features correspond to the inputs. The processing state contains labeled data and corresponds to the labels. The neural network consists of an input layer made up of a multitude of neurons, a hidden layer (an intermediate layer) made up of a multitude of neurons, and an output layer made up of a multitude of neurons. The intermediate layer can be one of two or more.

[0070] Fig. Figure 3 is a diagram illustrating an exemplary configuration of the trained model in the first embodiment. Fig. Figure 3 illustrates an example configuration of the neural network. The in Fig. The neural network shown in Figure 3 is a three-layer neural network. The input layer contains neurons X1, X2, and X3. The intermediate layer contains neurons Y1 and Y2. The output layer contains neurons Z1, Z2, and Z3. It should be noted that the number of neurons in each layer is arbitrary. A variety of values ​​input to the input layer are multiplied by weights W1, w11, w12, w13, w14, w15, and w16 and fed into the intermediate layer. The same values ​​are multiplied by weights W2, w21, w22, w23, w24, w25, and w26 and output by the output layer. The output from the output layer changes according to the values ​​of weights W1 and W2.The neural network is created by adjusting the weights W1 and W2 so that, after the features are entered into the input layer, the result output by the output layer closely approximates the processing state.

[0071] Next, a processing operation performed by laser processing system 1 is described. Three examples of the processing operation at the time of determining the monitoring model are described here.

[0072] Fig. Figure 4 is a flowchart illustrating a first example of the processing operation performed by the laser processing system 1 according to the first embodiment. In the first example, the laser processing system 1 determines the monitoring model by causing the laser processing machine 2 to perform a trial processing operation. In the first example, the monitoring model determination unit 24 determines the monitoring model based on the result of the user's evaluation of the processing state and the result of the evaluation by the processing state evaluation unit 23.

[0073] In step S1, the laser processing machine 2 starts the trial processing. When the trial processing starts, the acoustic sensor 8 and the light sensor 9 each detect the processing status and output the sensor data. In step S2, the data acquisition unit 21 acquires the sensor data output by the acoustic sensor 8 and the light sensor 9. The data acquisition unit 21 outputs the acquired sensor data to the feature calculation unit 22.

[0074] In step S3, the feature calculation unit 22 calculates features based on the sensor data. The feature calculation unit 22 outputs the calculated features to the processing state evaluation unit 23. In step S4, the processing state evaluation unit 23 evaluates the processing state based on a trained model. The processing state evaluation unit 23 inputs the features into one of the many trained models stored in the trained model retention unit 26, thus obtaining the result of the processing state evaluation. The processing state evaluation unit 23 outputs the obtained evaluation result to the monitoring model determination unit 24.

[0075] In step S5, the monitoring model determination unit 24 determines whether the result of the evaluation of the processing set in step S4 matches the result of the evaluation by the user. In the first in Fig. In the example shown in Figure 4, the result of the user's evaluation is entered into the monitoring model determination unit 24. Monitoring model determination unit 24 compares the result of the processing state evaluation in step S4 with the result of the user's evaluation to determine whether the evaluation result matches the user's evaluation result.

[0076] If the result of the processing state evaluation in step S4 matches the result of the user evaluation (step S5, Yes), the monitoring model determination unit 24 determines the monitoring model in step S6. The monitoring model determination unit 24 designates the trained model used to evaluate the processing state in step S4 as the monitoring model.

[0077] If, however, the result of the processing state evaluation in step S4 does not match the result of the user's evaluation (step S5, No), the processing state evaluation unit 23 selects a trained model to be used for processing state evaluation in step S7. The processing state evaluation unit 23 selects a different trained model from the multitude of models stored in the trained model hold unit 26 than the one used to evaluate the processing set in step S4. The laser processing system 1 then returns to step S4 and performs the processing in steps S4 and S5 using the trained model selected in step S7.

[0078] The laser processing system 1 completes step S6, thus concluding the processing according to the in Fig. The procedure is described in section 4. The laser processing system 1 monitors the processing state while the laser processing machine 2 processes a product, using the monitoring model. When the determining device determines the processing state by observing the workpiece 10, the monitoring model determining unit 24 determines the monitoring model in step S5 based on the result of the processing state evaluation by the determining device and the result of the evaluation by the processing state evaluation unit 23.

[0079] The user can determine whether the processing state evaluation result from the processing state evaluation unit 23 matches the user's evaluation result. The user reviews the evaluation result output by the processing state evaluation unit 23 for each of the many trained models and selects a trained model whose evaluation result from the processing state evaluation unit 23 matches the user's evaluation result. In this case, the monitoring model determination unit 24 designates the user-selected trained model as the monitoring model. In a second example, the monitoring model determination unit 24 determines the monitoring model based on the result of the user's processing state evaluation and the result of the evaluation by the processing state evaluation unit 23.

[0080] Fig. Figure 5 is a flowchart illustrating the second example of the processing operation performed by the laser processing system 1 according to the first embodiment. In the first example described above, the laser processing system 1 receives an evaluation result from each of the many trained models and compares the received evaluation result with the user evaluation result. In the second example, the laser processing system 1 receives evaluation results from the many trained models and compares each received evaluation result with the user evaluation result. In the second example as well, the laser processing system 1 causes the laser processing machine 2 to perform a trial processing operation, thereby determining the monitoring model.

[0081] In step S11, the laser processing machine 2 begins the sample processing. In step S12, the data acquisition unit 21 records the sensor data. In step S13, the feature calculation unit 22 calculates features based on the sensor data. Steps S11 to S13 correspond to those in Fig. 4 steps S1 to S3 are shown.

[0082] In step S14, the processing state evaluation unit 23 assesses the processing state based on each of the multiple trained models. The processing state evaluation unit 23 inputs the features into all multiple trained models stored in the trained model retention unit 26, thus obtaining the result of the assessment by each of the multiple trained models. The processing state evaluation unit 23 outputs the obtained assessment results to the monitoring model determination unit 24.

[0083] In step S15, the monitoring model determination unit 24 selects from the multitude of trained models a trained model whose result in the processing set's evaluation most closely matches the user's evaluation. In the Fig. In the second example shown in Figure 5, the result of the user's evaluation is entered into the monitoring model determination unit 24. Monitoring model determination unit 24 compares the result of the evaluation by each of the many trained models with the result of the user's evaluation and thereby selects a trained model whose evaluation result is closest to the user's evaluation result.

[0084] In step S16, the monitoring model determination unit 24 determines the monitoring model. The monitoring model determination unit 24 designates the trained model selected in step S15 as the monitoring model. The laser processing system 1 terminates step S16, thus completing the processing according to the... Fig. The 5 illustrated methods are used. The laser processing system 1 monitors the processing status while the laser processing machine 2 processes a product, using the monitoring model.

[0085] The user can select a trained model whose machining state assessment result most closely matches the user's assessment. The user reviews the assessment results of each of the many trained models and selects the one whose assessment result most closely matches the user's. In this case, the monitoring model determination unit 24 designates the user-selected trained model as the monitoring model. If the determination device determines the machining state by observing the workpiece 10, the monitoring model determination unit 24, in step S15, selects from the many trained models a trained model whose machining state assessment result most closely matches the determination device's assessment.

[0086] Fig. Figure 6 is a flowchart illustrating a third example of the processing operation performed by the laser processing system 1 according to the first embodiment. In the third example, the laser processing system 1 determines the monitoring model while the laser processing machine 2 performs the product processing. The laser processing system 1 can determine the monitoring model by performing a test processing operation as in the first or second example, or it can determine the monitoring model while the product processing is being performed as in the third example. In the third example, the monitoring model determination unit 24 determines the monitoring model based on the result of the user's evaluation of the processing state and the result of the evaluation by the processing state evaluation unit 23.

[0087] In step S21, the laser processing machine 2 begins processing a product. In step S22, the data acquisition unit 21 records the sensor data. In step S23, the feature calculation unit 22 calculates features based on the sensor data. In step S24, the processing state evaluation unit 23 assesses the processing state based on each of the multiple trained models. In step S25, the monitoring model determination unit 24 selects from the multiple trained models a trained model whose processing state assessment result most closely matches the user's assessment. In step S26, the monitoring model determination unit 24 determines the monitoring model. Steps S22 to S26 correspond to those in Fig. 5 steps shown S12 to S16.

[0088] The laser processing system 1 completes step S26, thus concluding the processing according to the in Fig. The procedure is described in section 6. The laser processing system 1 continues the product processing by the laser processing machine 2 and monitors the processing status using the monitoring model. In the case that processing errors during product processing are acceptable to a certain degree, the laser processing system 1 can determine the monitoring model while the product processing is being carried out, as in the third example. If the determining device determines the processing status by observing the workpiece 10, the monitoring model determination unit 24 selects, in step S25, from a variety of trained models, a trained model whose result in evaluating the processing status is closest to the result of the evaluation by the determining device.

[0089] The monitoring model determination process is performed at a desired time during product processing. For example, laser processing system 1 performs the monitoring model determination process when laser processing machine 2 has completed its product processing operation. Laser processing system 1 can perform the monitoring model determination process after a predetermined number of products have been processed. Laser processing system 1 can perform the monitoring model determination process every time a product is processed. Laser processing system 1 can perform the monitoring model determination process when a processing error occurs.The laser processing system 1 can perform the processing to determine the monitoring model when the operation of the laser processing machine 2 for product processing is started.

[0090] By determining the monitoring model during product processing, the laser processing system 1 can determine a trained model based on a variety of data to be used as a monitoring model. By determining the monitoring model as needed during product processing, the laser processing system 1 can evaluate the processing status using the monitoring model, which reflects the aging of the laser processing machine 2.

[0091] According to the first embodiment, the control unit 3 designates a trained model as a monitoring model to be used for monitoring the processing state by the laser processing machine 2. This model is based on the results of the processing state evaluation by observing the workpiece 10 being processed by the laser processing machine 2 and the results of the evaluation by the processing state evaluation unit 23. The control unit 3 can obtain the monitoring model, which enables an accurate evaluation of the processing state even if the processing state measured by the sensors changes due to the state of the laser processing machine 2, variations in the components of the workpiece 10, or the like. The control unit 3 can accurately evaluate the quality of the processing state and thus enables the laser processing machine 2 to continue the processing successfully.As described above, the control unit 3 has the effect that the laser processing machine 2 can continue processing successfully. Second embodiment.

[0092] In the first embodiment, the monitoring model determination unit 24 selects a trained model from the plurality of trained models as the monitoring model. The evaluation result of this trained model, output by the processing state evaluation unit 23, corresponds to the result of the evaluation by observing the workpiece 10. A second embodiment describes an example in which a combination of a plurality of trained models is selected as the monitoring model. In the second embodiment, the same reference numerals are assigned to the same components as in the first embodiment, and a configuration is described that differs from that of the first embodiment.

[0093] Fig. Figure 7 is a diagram showing an exemplary configuration of a monitoring model determination unit 30, which is included in the control unit 3 according to the second embodiment. The control unit 3 according to the second embodiment includes the in Fig. 7 shown monitoring model determination unit 30 instead of the one in Fig. 2 shown, the monitoring model determination unit 24. For a combination of a multitude of trained models stored by the trained model retention unit 26, the monitoring model determination unit 30 adjusts the weights for the output values ​​of the respective trained models and determines the combination of the multitude of trained models as the monitoring model after the weights for the respective trained models have been adjusted.

[0094] The monitoring model determination unit 30 includes a weight determination unit 31. The weight determination unit 31 adjusts the weights for the respective trained models in order to determine the weights for the respective trained models.

[0095] The monitoring model determination unit 30 reads the multitude of trained models from the trained model retention unit 26. In the Fig. In the example shown in Figure 7, three trained models are stored in the trained model holding unit 26, and the monitoring model determination unit 30 has read the three trained models.

[0096] v1, v2 and v3, which are in Fig. Figure 7 represents the weights for the output values ​​of the respective trained models. The weight determination unit 31 adjusts v1, v2, and v3 accordingly. The in Fig. The seven dashed arrows shown indicate that the weight determination unit 31 adjusts v1, v2, and v3 respectively. Each trained model is, for example, the one in Fig. 3 shown neural networks.

[0097] The processing state evaluation unit 23 evaluates the processing state of the laser processing machine 2 using a combination of the many trained models, after adjusting the weights for each trained model. The monitoring model determination unit 30 adjusts v1, v2, and v3 respectively to reduce the difference between the evaluation result output by the processing state evaluation unit 23 and the evaluation result obtained by observing the workpiece 10. Thus, the monitoring model determination unit 30 adjusts v1, v2, and v3 respectively so that the evaluation result output by the processing state evaluation unit 23 matches the evaluation result obtained by observing the workpiece 10. The evaluation result obtained by observing the workpiece 10 is the result of evaluating the processing state by observing the workpiece 10 being processed by the laser processing machine 2.The result of the evaluation by observation of the workpiece 10 is the result of the evaluation of the machining state by the user by observing the machined workpiece 10 or the result of the evaluation by the determining device that determines the machining state by observing the workpiece 10.

[0098] The monitoring model determination unit 30 determines as the monitoring model the combination of the three trained models whose weights have been adjusted so that the evaluation result output by the machining status evaluation unit 23 matches the result of the evaluation by observing workpiece 10. Alternatively, the monitoring model determination unit 30 determines as the monitoring model the combination of the three trained models whose weights have been adjusted so that the evaluation result output by the machining status evaluation unit 23 most closely approximates the evaluation result by observing workpiece 10. Thus, the monitoring model determination unit 30 determines as the monitoring model the combination of the multitude of trained models after the weights for the respective trained models have been adjusted.The monitoring model determination unit 30 stores the determined monitoring model in the monitoring model holding unit 27. The monitoring model holding unit 27 stores the monitoring model.

[0099] The processing monitoring unit 20 monitors the processing state of the laser processing machine 2 using the monitoring model determined by the monitoring model determination unit 30. When features are input, the monitoring model outputs the sum of the output values ​​of the weighted trained models. The monitoring model can output the weighted mean of the output values ​​of the trained models when features are input. The monitoring model determination unit 30 can determine the weights for the respective trained models by learning the relationships between the weights, the features, and the processing state.

[0100] If each trained model is a neural network, the monitoring model determination unit 30 modifies the weights set in the neural network, i.e., w11 to w16 and w21 to w26, which are in Fig. Figure 3 is not shown. The monitoring model determination unit 30 determines the monitoring model by adjusting the weights for the output values ​​of the respective trained models without changing the weights defined in the neural networks. The processing monitoring unit 20 can use the combination of the multitude of trained models as the monitoring model by adjusting the weights for the output values ​​of the respective trained models with the monitoring model determination unit 30.

[0101] The control unit 3 adjusts the weights for the respective trained models to obtain the monitoring model, which the laser processing machine 2 uses to evaluate the processing status. Any of the many trained models stored in the trained model holding unit 26 can be used as the model on which the monitoring model is based. For example, the model on which the monitoring model is based could be a model set defined for each manufacturer of workpiece 10 or a model set defined for each material of workpiece 10. The control unit 3 can reduce the number of predefined trained models compared to selecting a single trained model to be used as the monitoring model from a large number of trained models. The control unit 3 can determine the monitoring model based on relatively little data.

[0102] According to the second embodiment, the control unit 3 determines the combination of the multitude of trained models as a monitoring model after the weights for the respective trained models have been adjusted. The control unit 3 can obtain the monitoring model, which enables an accurate assessment of the processing status and allows the laser processing machine 2 to continue processing successfully. As described above, the control unit 3 has the effect of enabling the laser processing machine 2 to continue processing successfully. The laser processing system 1 can determine the monitoring model based on relatively little data and thus improve productivity. Third embodiment.

[0103] The first and second embodiments describe examples in which the multitude of trained models is predetermined. A third embodiment describes an example in which a trained model is added by learning in the laser processing system 1. In the third embodiment, the same reference numerals are assigned to the same components as in the first or second embodiments, and a configuration is described that differs from that of the first or second embodiments.

[0104] Fig. Figure 8 is a diagram showing an exemplary configuration of a control unit 40 included in the laser processing system 1 according to the third embodiment. The laser processing system 1 according to the third embodiment comprises the laser processing machine 2 and the control unit 40. The control unit 40 comprises the processing control unit 28 and a processing monitoring unit 41. The control unit 40 can be a device outside the laser processing machine 2 or a device included in the laser processing machine 2. Alternatively, the processing control unit 28 of the control unit 40 can be implemented by a device included in the laser processing machine 2, and the processing monitoring unit 41 of the control unit 40 can be implemented by a device outside the laser processing machine 2.

[0105] The processing monitoring unit 41 includes the data acquisition unit 21, the feature calculation unit 22, the processing status evaluation unit 23, the processing parameter correction unit 25, the trained model holding unit 26, the monitoring model holding unit 27, a monitoring model determination unit 42 and a learning unit 43.

[0106] Learning Unit 43 captures the features calculated by Feature Calculation Unit 22. Learning Unit 43 captures information indicating the processing state of Laser Processing Machine 2. Learning Unit 43 generates a trained model by learning the relationship between the features and the processing state of Laser Processing Machine 2 and adds the generated trained model to Trained Model Hold Unit 26.

[0107] For a combination of a multitude of trained models stored by the Trained Model Retainer 26, the Monitoring Model Determination Unit 42 adjusts the weights for the output values ​​of the respective trained models and determines the combination of the multitude of trained models as the monitoring model after the weights for the respective trained models have been adjusted. The multitude of trained models stored in the Trained Model Retainer 26 includes the trained model added by the Learning Unit 43.

[0108] Each of the numerous trained models is, for example, the one in Fig. Figure 3 shows a neural network. Learning unit 43 learns the relationship between the features and the processing state through supervised learning and generates a trained model.

[0109] Fig. Figure 9 is a diagram showing an exemplary configuration of the monitoring model determination unit 42, which is included in the control unit 40 according to the third embodiment. The monitoring model determination unit 42 includes a weight determination unit 44. The weight determination unit 44 adjusts weights for the respective trained models in order to determine the weights for the respective trained models.

[0110] The monitoring model determination unit 42 reads the multitude of trained models from the trained model retention unit 26. In the Fig. In the example shown, five trained models are stored in the trained model retention unit 26, and the monitoring model determination unit 42 has read the five trained models. Two of the five trained models are trained models that were added by the learning unit 43.

[0111] v1, v2, v3, v4 and v5, which are in Fig. Figure 9 represents the weights for the output values ​​of the respective trained models. The weight determination unit 44 adjusts v1, v2, v3, v4, and v5 accordingly. The in Fig. The 9 dashed arrows shown indicate that the weight determination unit 44 adjusts each of v1, v2, v3, v4 and v5.

[0112] The processing state evaluation unit 23 assesses the processing state by the laser processing machine 2 using a combination of the many trained models, after the weights for the respective trained models have been adjusted. The monitoring model determination unit 42 adjusts v1, v2, v3, v4, and v5 respectively so that the difference between the evaluation result output by the processing state evaluation unit 23 and the evaluation result obtained by observing the workpiece 10 is reduced. Thus, the monitoring model determination unit 42 adjusts v1, v2, v3, v4, and v5 respectively so that the evaluation result output by the processing state evaluation unit 23 matches the result of the evaluation by observing the workpiece 10.

[0113] The monitoring model determination unit 42 determines as the monitoring model the combination of the five trained models whose weights have been adjusted so that the evaluation result output by the machining status evaluation unit 23 matches the evaluation result obtained by observing workpiece 10. Alternatively, the monitoring model determination unit 42 determines as the monitoring model the combination of the five trained models whose weights have been adjusted so that the evaluation result output by the machining status evaluation unit 23 most closely approximates the evaluation result obtained by observing workpiece 10. Thus, the monitoring model determination unit 42 determines as the monitoring model the combination of the multitude of trained models after the weights for the respective trained models have been adjusted.The monitoring model determination unit 42 stores the determined monitoring model in the monitoring model holding unit 27. The monitoring model holding unit 27 stores the monitoring model.

[0114] The processing monitoring unit 20 monitors the processing state of the laser processing machine 2 using the monitoring model determined by the monitoring model determination unit 42. When features are input, the monitoring model outputs the sum of the output values ​​of the weighted trained models. The monitoring model can output the weighted mean of the output values ​​of the trained models when features are input. The monitoring model determination unit 42 can determine the weights for the respective trained models by learning the relationships between the weights, the features, and the processing state.

[0115] The learning unit 43 generates a trained model using features calculated based on sensor data actually obtained from the laser processing machine 2. By creating the monitoring model based on this trained model, the control unit 40 can obtain a highly accurate monitoring model based on the actual state of the laser processing machine 2.

[0116] Learning Unit 43 creates a new trained model using the features and information indicating the processing state. Alternatively, Learning Unit 43 can perform additional training on a previously created trained model. When a trained model is added by Learning Unit 43 to Trained Model Repository 26, the added trained model can be stored in Trained Model Repository 26. A trained model added to Trained Model Repository 26 can optionally be deleted from Trained Model Repository 26.If the material of the workpiece 10 to be processed by the laser processing machine 2 is changed in the control unit 40, a trained model generated by the learning unit 43 can be added to the trained model holding unit 26, or a trained model generated by the learning unit 43 and added to the trained model holding unit 26 can be deleted from the trained model holding unit 26. When the material of the workpiece 10 is changed, the training models stored in the trained model holding unit 26 are updated so that the control unit 40 can obtain a highly accurate monitoring model corresponding to the material of the workpiece 10.

[0117] If the processing state of the laser processing machine 2 changes, a trained model generated by the learning unit 43 can be added to the trained model holding unit 26 in the control unit 40, or a trained model generated by the learning unit 43 and added to the trained model holding unit 26 can be deleted from the trained model holding unit 26. A change in the processing state is, for example, a change from a good state to a bad state, or the like. When the processing state changes, the training models stored in the trained model holding unit 26 are updated so that the control unit 40 can obtain a highly accurate monitoring model corresponding to the processing state.

[0118] A trained model, which is the result of learning by learning unit 43, can be any trained model that outputs at least either the result of the processing state evaluation or the correction amounts when the features are input. The processing parameter correction unit 25 can calculate the correction amounts by inputting the features calculated by the feature calculation unit 22 into the monitoring model.

[0119] The processing state evaluation unit 23 can evaluate the processing set by the laser processing machine 2 by inputting information other than the features calculated by the feature calculation unit 22 into each of the multitude of trained models. Examples of information input into each of the multitude of trained models include the values ​​of the processing parameters at the time of evaluation, the temperature of the optical processing system at the time of evaluation, any changes in the temperature of the optical processing system, the thickness of the workpiece 10, or the material of the workpiece 10. In this case, the learning unit 43 generates a trained model by learning the relationships between the information, the features, and the processing state.The processing status evaluation unit 23 inputs the information and characteristics into the monitoring model to evaluate the processing status by the laser processing machine 2.

[0120] In the description above, the learning unit 43 is provided within the control unit 40. The learning unit 43 can be implemented by a device outside the control unit 40. A learning device that implements the learning unit 43 can be a device that can be connected to the control unit 40 via a network. The learning device can be a device that resides on a cloud server.

[0121] According to the third embodiment, the control unit 40 determines the combination of the multitude of trained models as a monitoring model after the weights for the respective trained models have been adjusted. The control unit 40 can obtain the monitoring model, which enables an accurate assessment of the processing status and allows the laser processing machine 2 to continue processing successfully. As described above, the control unit 40 has the effect of enabling the laser processing machine 2 to continue processing successfully. The laser processing system 1 can determine the monitoring model based on relatively little data and thus improve productivity. The control unit 40 can obtain a highly accurate monitoring model by generating a trained model using features calculated based on the sensor data actually obtained from the laser processing machine 2.

[0122] Next, the hardware for implementing the control unit 3 or 40 according to the first to third embodiments is described. The control unit 3 or 40 is implemented by a processing circuit. The processing circuit can be a circuit in which a processor executes software, or a dedicated circuit.

[0123] If the processing circuit is implemented by software, the processing circuit is, for example, a Fig. 10 control circuits shown 50. Fig. Figure 10 is a diagram illustrating an exemplary configuration of the control circuit 50 according to the first to third embodiments. The control circuit 50 comprises an input unit 51, a processor 52, a memory 53, and an output unit 54. The input unit 51 is an interface circuit that receives data from outside the control circuit 50 and forwards this data to the processor 52. The output unit 54 is an interface circuit that sends data from the processor 52 or from the memory 53 to the outside of the control circuit 50.

[0124] If the processing circuit is in Fig. In the control circuit 50 shown in Figure 10, the control unit 3 or 40 is implemented by software, firmware, or a combination of software and firmware. The software or firmware is written as a program and stored in memory 53. In the processing circuit, the processor 52 reads and executes the programs stored in memory 53, thereby implementing the functions of the control unit 3 or 40. That is, the processing circuit includes memory 53 for storing the programs that lead to the execution of processing in the control unit 3 or 40. It can be said that these programs cause a computer to execute the procedure and method in the control unit 3 or 40.

[0125] The processor 52 is a central processing unit (CPU). The processor 52 can be a central processing unit, a processing unit, a computing unit, a microprocessor, a microcomputer, a processor, or a digital signal processor (DSP). The memory 53 corresponds, for example, to non-volatile or volatile memory such as random access memory (RAM), read-only memory (ROM), flash memory, erasable programmable read-only memory (EPROM), or electrically erasable programmable read-only memory (EEPROM) (registered trademark), or to a magnetic disk, a flexible disk, an optical disk, a compact disc, a mini-disc, a digital versatile disc (DVD), or the like.

[0126] Fig. Figure 10 is an example of hardware where the control unit 3 or 40 is implemented by the general-purpose processor 52 and the memory 53. The control unit 3 or 40 can be implemented by a dedicated hardware circuit. Fig. Figure 11 is a diagram illustrating an exemplary configuration of a dedicated hardware circuit 55 according to the first to third embodiments.

[0127] The dedicated hardware circuit 55 includes the input unit 51, the output unit 54, and the processing circuit 56. The processing circuit 56 can be a single circuit, a combined circuit, a programmed processor, a parallel programmed processor, an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or a circuit combining these. The functions of the control unit 3 or 40 can be implemented by the processing circuit 56 on an individual function basis, or the functions can be implemented jointly by the processing circuit 56. It should be noted that the control unit 3 or 40 can be implemented by a combination of the control circuit 50 and the hardware circuit 55.

[0128] In the case where the learning unit 43 is a device outside the control device 40, a learning device such as this device is implemented by an editing circuit such as the control device 3 or 40. The editing circuit that implements the learning device is the one in Fig. 10 shown control circuit 50 or the one in Fig. 11 Dedicated hardware circuit shown 55.

[0129] The configuration shown in the individual embodiments described above illustrates an example of the subject matter of the present invention. The configuration of each embodiment can be combined with another known technique. The respective configurations of the embodiments can be combined as needed. The configuration of each embodiment can be omitted or modified in any area without departing from the essential nature of the present invention. List of reference symbols 1 laser processing system; 2 laser processing machines; 3.40 Control unit; 4 Laser oscillator; 5 cables; 6. Processing head; 7 processing nozzle; 8 sound sensor; 9 Light sensor; 10 workpieces; 11 collimating optical system; 12 optical imaging system; 13 Protective glass; 14 Machining head drive unit; 15, 16 Mechanism of movement; 20, 41 Processing monitoring unit; 21 Data acquisition unit; 22 Feature calculation unit; 23 Processing status evaluation unit; 24, 30, 42 Monitoring model determination unit; 25 processing parameter correction unit; 26 Trained Model Holding Unit; 27 Monitoring model holding unit; 28 Processing control unit; 31, 44 Unit of weight determination; 43 learning units; 50 control circuit; 51 Input unit; 52 processor; 53 storage locations; 54 output units; 55 Hardware circuitry; 56 processing circuit. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] JP 2022-118 774 A

[0005]

Claims

Control unit comprising: a trained model holding unit to store a variety of trained models, which are the results of learning a relationship between features indicating a processing state of a processing machine and the processing state of the processing machine; a processing state evaluation unit to evaluate the processing state of the processing machine by inputting the features into each of the variety of trained models or into a combination of the variety of trained models;A monitoring model determination unit for determining the trained model or the combination of the multitude of trained models as a monitoring model, which is used to monitor the processing state by the processing machine, based on a result of the processing state evaluation by observing a workpiece processed by the processing machine and a result of the evaluation by the processing state evaluation unit; and a processing control unit for controlling the processing machine based on a result of the processing state monitoring by the processing machine using the monitoring model. Control unit according to claim 1, wherein the monitoring model determination unit determines one of the plurality of trained models stored in the trained model retention unit as the monitoring model. Control unit according to claim 1, wherein the monitoring model determination unit adjusts weights for output values ​​of the respective trained models for the combination of the plurality of trained models stored in the trained model retention unit and determines the combination of the plurality of trained models as the monitoring model after the weights for the respective trained models have been adjusted. Control unit according to one of claims 1 to 3, wherein the processing machine is a laser processing machine for processing the workpiece with laser light and the control unit comprises: a data acquisition unit for acquiring data indicating a result of the acquisition of light generated during processing by the laser processing machine, and data indicating a result of the acquisition of sound generated during processing by the laser processing machine, and a feature calculation unit for calculating the features from the data acquired by the data acquisition unit. Control unit according to one of claims 1 to 4, wherein the processing machine is a laser processing machine for processing the workpiece with laser light and each of the several trained models stored in the trained-model holding unit is a result of learning the relationship between the features and the processing state in a corresponding plurality of cases that differ from each other in at least one of the following points: thickness of the plate-shaped workpiece and type of processing gas during processing by the laser processing machine. Control unit according to one of claims 1 to 5, comprising: a learning unit for generating the trained model by learning the relationship between the features and the processing state by the processing machine and for adding the generated trained model to the trained model holding unit. Control unit according to claim 6, wherein, when a material of the workpiece to be machined by the machine tool is changed, the trained model generated by the learning unit is added to the trained model holding unit or the trained model generated by the learning unit and added to the trained model holding unit is deleted from the trained model holding unit. Control unit according to claim 6, wherein, when the processing state has changed by the processing machine, the trained model generated by the learning unit is added to the trained model holding unit or the trained model generated by the learning unit and added to the trained model holding unit is deleted from the trained model holding unit. Laser processing system comprising: a laser processing machine for processing a workpiece with laser light; and a control unit for controlling the laser processing machine, the control unit comprising: a trained model holding unit for storing a plurality of trained models, which are the results of learning a relationship between features indicating a processing state by the laser processing machine and the processing state by the laser processing machine; a processing state evaluation unit for evaluating the processing state by the laser processing machine by inputting the features into each of the plurality of trained models or by inputting the features into a combination of the plurality of trained models;A monitoring model determination unit for determining the trained model or the combination of the multitude of trained models as a monitoring model to be used for the processing state of the laser processing machine, based on a result of the processing state evaluation by observation of the workpiece processed by the laser processing machine and a result of the evaluation by the processing state evaluation unit; and a processing control unit for controlling the laser processing machine based on a result of the monitoring of the processing state by the laser processing machine using the monitoring model. A laser processing method comprising the following steps: a step of capturing features indicating a processing state by a laser processing machine for processing a workpiece with laser light; a step of evaluating the processing state by the laser processing machine by inputting the captured features into each of a multitude of trained models, which are the results of learning a relationship between the processing state by the laser processing machine and the features, or by inputting the captured features into a combination of the multitude of trained models;a step to determine the trained model or the combination of the multitude of trained models as a monitoring model to be used for the processing state by the laser processing machine, based on a result of the processing state evaluation by observing the workpiece being processed by the laser processing machine and a result of the evaluation in the processing state evaluation step; and a step to control the laser processing machine based on a result of the monitoring of the processing state by the laser processing machine using the monitoring model.

Images