JP7752020B2 - Welding system and welding monitor device - Google Patents

Description

The present invention relates to a welding system and a welding monitoring device.

In conventional welding, when determining appropriate processing conditions that meet the required welding specifications, appropriate processing conditions have been searched for by automatically determining the processing conditions to be used and conducting as few trials as possible in processing experiments as needed (see Patent Document 1), or by repeatedly conducting search experiments by trying all possible combinations of numerous elements, such as the materials that make up the weld and the welding time.

Furthermore, after welding is completed, non-destructive testing using ultrasonic waves (see Patent Document 2) and destructive testing, in which the weld is actually destroyed to check its condition, have been carried out to check whether the weld was welded correctly and whether the required weld strength has been achieved.

Japanese Patent Application Laid-Open No. 2012-236267 Japanese Patent Application Laid-Open No. 2009-156701

However, the device disclosed in Patent Document 1 requires multiple machining experiments, and exploratory experiments that try all possible combinations rely heavily on the intuition and experience of the operator, resulting in the problem that it takes a long time to find the appropriate machining conditions.

Furthermore, in order to confirm the weld strength of welds through non-destructive testing using the device disclosed in Patent Document 2, the device requires components such as sensors and probes, making it difficult to minimize the device configuration. Furthermore, there is the problem that non-destructive testing is unavoidable in order to confirm whether the weld strength of welds welded under certain processing conditions is comparable between those that have been subjected to destructive testing and those that have not.

One aspect of the present invention is a welding system and welding monitoring device that can easily search for appropriate welding processing conditions in a short amount of time with a simple configuration, and that can non-destructively estimate the welding results of a welded portion welded under appropriate processing conditions.

A welding system according to one aspect of the present invention includes a welding device that welds workpieces; a learning device that inputs, as training data, processing condition information including processing conditions for welding previously performed by the welding device and welding result information related to the welding results obtained after welding based on the processing conditions, and creates and outputs a welding result estimation model based on the training data; and an estimation device that inputs, as estimation data, processing condition information including the processing conditions set for the welding device to the welding result estimation model created by the learning device, and estimates and outputs the welding results of welding performed by the welding device based on the processing conditions.

In a welding system according to one aspect of the present invention, a welding result estimation model is created based on training data consisting of processing condition information for welding previously performed by a welding device and welding result information obtained after welding based on this information. Furthermore, by inputting processing condition information set in the welding device as estimation data into the welding result estimation model, the welding results of welding performed by the welding device are estimated and output based on the processing conditions, allowing for non-destructive estimation of the welding results of welds welded under appropriate processing conditions. Furthermore, since welding results can be estimated by inputting processing condition information set in the welding device into the welding result estimation model, a simple configuration can be used to reduce the amount of processing condition search work and the number of searches, making it possible to easily search for appropriate welding processing conditions in a short amount of time.

One aspect of the present invention makes it possible to easily search for appropriate welding processing conditions in a short amount of time using a simple configuration, and to non-destructively estimate the welding results of a welded portion welded under appropriate processing conditions.

FIG. 1 is an explanatory diagram showing the basic configuration of a welding system according to a first embodiment of the present invention. FIG. 2 is an explanatory diagram showing the basic hardware configuration of a learning device and/or an estimation device of a welding system. FIG. 3 is an explanatory diagram showing the basic configuration of a welding system according to a second embodiment of the present invention. FIG. 4 is an explanatory diagram showing the basic configuration of a welding system according to a third embodiment of the present invention. FIG. 5 is a graph showing an example of time-series data. FIG. 6 is an explanatory diagram showing the basic configuration of a welding system according to a fourth embodiment of the present invention. FIG. 7 is an explanatory diagram showing the basic configuration of a welding system according to a fifth embodiment of the present invention.

The following describes in detail a welding system and welding monitor device according to an embodiment of the present invention, with reference to the accompanying drawings. However, the following embodiments do not limit the invention according to each claim, and not all combinations of features described in the embodiments are necessarily essential to the solution of the invention. Furthermore, in the following embodiments, the arrangement, scale, dimensions, etc. of each component may be shown exaggerated or minimized, and the description of some components may be omitted.

[First embodiment]
Fig. 1 is an explanatory diagram showing the basic configuration of a welding system according to a first embodiment of the present invention. Fig. 2 is an explanatory diagram showing the basic hardware configuration of a learning device and/or an estimation device of the welding system.

As shown in FIG. 1, the welding system 100 according to the first embodiment is a system capable of estimating and outputting welding results. The welding system 100 includes a welding device (laser welder 40) that welds workpieces; a learning device 10 that receives as training data processing condition information 2, including processing conditions for welding previously performed by the welding device (laser welder 40), and welding result information 1 related to the welding results obtained after welding based on the processing conditions; and an estimation device 30 that receives as estimation data processing condition information 4, including processing conditions set for the welding device (laser welder 40), into the welding result estimation model 3 generated by the learning device 10; and an estimation device 30 that estimates and outputs the welding results of welding performed by the welding device (laser welder 40) based on the processing conditions. The estimation device 30 is included in a welding result determination device 20 that determines the welding results based on the estimated welding results.

In this welding system 100, for example, the teacher data for the objective variable input to the learning device 10 may be welding result information 1 related to the welding strength F (N/ ^mm2 ) of the laser welded portion, and the teacher data for the explanatory variable may be processing condition information 2 including various processing conditions (laser power (output value), laser irradiation time, etc.). Note that the objective variable and explanatory variables used in the welding system 100 are not limited to these examples.

In the welding system 100 of the first embodiment, the welding device is a laser welder 40, and the processing condition information 2, 4 includes at least one processing condition: the material of the workpiece, the thickness of the workpiece, the laser power, the laser irradiation time, the time from the start of laser irradiation to reaching peak power, the time from the peak power to the end of laser irradiation, the fiber diameter, the lens focal length, the focal position, and the laser diameter at the irradiation point. The spot diameter calculated from the combination of the lens, fiber, and focal position is used, for example, for machine learning in the welding result learning unit 11.

The laser welder 40 is a device that irradiates a workpiece with laser light under processing conditions specified, for example, by external input, to perform welding. The laser welder 40 is configured to be able to acquire or send and receive various conditions for irradiating the laser light via various types of communication (network communication or communication via direct connection). The welding results include weld strength, spot diameter, weld quality, etc.

In this way, the welding system 100 of the first embodiment is used, for example, as a system for estimating the weld strength included in the welding results of laser welding performed by the laser welder 40. The welding result information 1 may also include the spot diameter, and may be an index that evaluates the metal structure based on the appearance or cross section as the welding quality, or an index that evaluates the corrosion resistance. Therefore, the welding system 100 can also be used as a system for estimating the spot diameter and weld quality included in the welding results of laser welding.

The learning device 10 includes a welding result learning unit 11. The welding result learning unit 11 performs machine learning using, for example, a predetermined machine learning algorithm based on training data, including processing condition information 2 input to the learning device 10 and welding result information 1 related to the welding results of the welded portion obtained, for example, by destructive testing after welding based on these processing conditions, to create a welding result estimation model 3. Note that destructive testing may refer to any of the destructive tests listed on the website of the Welding Information Center of the Japan Welding Engineering Society, such as JIS Z 2241 Tensile Test Method for Metallic Materials, or other listed JIS standards applicable to welding tests and inspections. The experimental and evaluation methods are also those specified in the JIS standards. The trained welding result estimation model 3 is composed of, for example, software run by a computer. In this example, a program that outputs a welding result estimation value 8, including weld strength, when processing condition information 4 is input as estimation data.

Learning methods include, but are not limited to, algorithms such as regression, classification, clustering, discrimination, interpolation, feature extraction, and time series modeling. The welding result estimation model 3 trained by the welding result learning unit 11 includes various parameters, formulas, and/or algorithms for prediction and estimation that characterize the trained model. The learning device 10 then outputs the welding result estimation model 3 to the welding result judgment device 20, which includes the estimation device 30.

Meanwhile, the estimation device 30 is included in the welding result determination device 20 and includes a welding result estimation unit 31. The welding result determination device 20 includes, together with the estimation device 30, a welding result estimation model 3 output from the learning device 10. The welding result estimation unit 31 inputs the processing condition information 4 set for the laser welder 40 input to the estimation device 30 as estimation data into the welding result estimation model 3 included in the welding result determination device 20, estimates the welding result of welding performed by the laser welder 40 based on the processing conditions, and outputs a welding result estimation value 8.

The welding result determination device 20 can determine, for example, the weld strength and/or spot diameter (hereinafter referred to as "weld strength, etc.") of the weld based on the welding result estimation value 8 output from the welding result estimation unit 31 of the estimation device 30. Note that, with regard to weld strength, the welding result estimation value 8 refers to information that can numerically express, for example, the tensile strength (unit: N) and strength distribution range of a certain weld point (weld). With regard to spot diameter, the welding result estimation value 8 refers to information that can numerically express, for example, the size of the laser light irradiation area on the surface of a certain weld point (weld), or the size of the melted portion of the weld.

Note that, when performing machine learning and estimation in the learning device 10 and the estimation device 30, algorithms such as regression analysis (RA), principal component analysis (PCA), singular value decomposition (SVD), linear discriminant analysis (LDA), independent component analysis (ICA), Gaussian process latent variable model (GPLVM), logistic regression (LR), and support vector machine (SVM) may be used. Various algorithms can be used, including Vector Machine (SVM), Discriminant Analysis (DA), Random Forest (RF), Ranking Support Vector Machine (RSVM), Gradient Boosting (GB), Naive Bayes (NB), and K-Nearest Neighbor Algorithm (K-NN).

In this way, in the welding system 100 of the first embodiment, the welding result learning unit 11 of the learning device 10 performs, for example, machine learning based on the processing condition information 2 and the welding result information 1 to create a welding result estimation model 3. Then, the welding result estimation unit 31 of the estimation device 30 inputs the processing condition information 4 set in the laser welder 40 into the welding result estimation model 3, thereby estimating the welding results (weld strength, spot diameter) of the welding performed by the laser welder 40 and outputting a welding result estimation value 8.

The welding system 100 can obtain a welding result estimate 8 simply by inputting the processing condition information 4 set in the laser welder 40 into the trained welding result estimation model 3. This eliminates the cumbersome task of actually irradiating a laser beam using multiple pieces of processing condition information 4 and measuring the results, and reduces the number of processing condition information combinations required to search for processing conditions that will achieve the desired weld strength. This makes it possible to easily search for appropriate processing conditions for laser welding in a short amount of time with a simple configuration, and to non-destructively estimate the welding results of a welded part created by laser welding under appropriate processing conditions.

In addition, each piece of data, welding result information (teaching data for objective variables) 1 and processing condition information (teaching data for explanatory variables) 2 in welding system 100, may each be composed of one or more variables or parameters, and may be stored, for example, in a stationary or portable storage device or storage medium (not shown). Furthermore, each piece of data may be transmitted and received via an information and communication medium such as the Internet, and may be raw data acquired by a measuring device such as a sensor (not shown). Furthermore, welding result estimation model 3 may be input and output between learning device 10 and estimation device 30 via the above-mentioned storage medium or information and communication medium.

As shown in FIG. 2, the learning device 10 and/or estimation device 30 of the welding system 100 includes, as basic hardware components, a CPU (central processing unit) 201, a RAM (random access memory) 202, a ROM (read-only memory) 203, a HDD (hard disk drive) 204 and/or an SSD (solid state drive) 205, and a memory card 206.

The learning device 10 and/or the estimation device 30 also include, for example, an input I/F (interface) 207, an output I/F (interface) 208, and a communication I/F (interface) 209. The components 201 to 209 are interconnected by a bus 200.

The CPU 201 controls the learning device 10 and/or the estimation device 30 by executing various programs stored in the RAM 202, ROM 203, HDD 204, SSD 205, etc. In the learning device 10, the CPU 201 executes the learning program to realize the functions of each part of the learning device 10, including the function of the welding result learning unit 11.

In addition, the CPU 201 executes an estimation program in the estimation device 30, thereby realizing the functions of each part of the estimation device 30, including the function of the welding result estimation unit 31. The CPUs 201 of the learning device 10 and the estimation device 30 can be configured to cooperate to control the entire welding system 100.

RAM 202 can be used as a working area for the CPU 201's arithmetic processing. ROM 203 stores the various programs described above in a readable manner. HDD 204 and SSD 205 store the various data described above in a readable and writable manner. Memory card 206 stores these various data in a readable and writable manner and serves as a storage medium that is detachable from each device 10, 30. HDD 204, SSD 205, and memory card 206 realize the functions of the storage devices or storage media described above.

For example, a sensor 212 is connected to the input I/F 207 to acquire detected information. The sensor 212 includes various sensors such as a temperature sensor, an optical sensor, an acoustic sensor, and an image sensor. A touch panel 211 that functions as an operation or input unit for the learning device 10 and/or the estimation device 30 is connected to the input I/F 207, and receives information associated with operation input from the user of the welding system 100. Various input devices, such as a keyboard and a mouse (including a trackball mouse), not shown, may also be connected to the input I/F 207.

The output I/F 208 is connected to, for example, a display 210 as a display device, and outputs various information to be displayed on the monitor of the learning device 10 and/or the estimation device 30. The touch panel 211 may be provided on the display 210. Furthermore, the learning device 10 and/or the estimation device 30 may be indirectly or directly connected via the communication I/F 209 to a server device, external device, etc. connected to a network such as the Internet (not shown).

Second Embodiment
Fig. 3 is an explanatory diagram showing the basic configuration of a welding system according to a second embodiment of the present invention. In the following description, including Fig. 3, the same components as those in the first embodiment and its modified examples are denoted by the same reference numerals, and therefore, redundant description will be omitted below.

As shown in FIG. 3, the welding system 100A according to the second embodiment includes a welding monitoring device 50 that outputs monitoring information 5 and 6 obtained by monitoring (status observation) the progress of welding by a welding device (laser welder 40) using a sensor 212. The learning device 10 calculates feature information as training data for auxiliary variables from the training data of the monitoring information 5 output from the welding monitoring device 50, creates and outputs a welding result estimation model 3 based on the training data to which the auxiliary variables have been added, and the estimation device 30 calculates feature information as estimation data for auxiliary variables from the estimation data of the monitoring information 6 output from the welding monitoring device 50, inputs the estimation data to which the auxiliary variables have been added, and estimates and outputs the welding result. The welding monitoring device 50 monitors the welded portion of the workpiece using a sensor 212, and is capable of acquiring and outputting output information from the sensor 212 as monitoring information 5 and 6.

In other words, the welding system 100A of the second embodiment is a system applied to laser welding that uses monitoring information 5, 6 from the welding monitor device 50 as auxiliary variables (teaching data and estimation data) that are correlated with the processing condition information 2, 4. Examples of auxiliary variables that are correlated with the explanatory variables (processing condition information) include physical quantities (welding temperature, sound, light, color, etc.) observed during or after laser welding. However, the auxiliary variables used in the welding system 100A are not limited to those exemplified above.

To use auxiliary variables, specifically, the learning device 10 of the welding system 100A includes a first auxiliary variable calculation unit 12 that calculates feature information as training data for auxiliary variables from the training data of the monitoring information 5, and a welding result learning unit 11 that creates and outputs a welding result estimation model 3, for example by performing machine learning, based on the training data of the welding result information 1, the training data of the processing condition information 2, and the training data (feature information) for the auxiliary variables calculated by the first auxiliary variable calculation unit 12.

The estimation device 30 includes a second auxiliary variable calculation unit 32 that calculates feature information from the estimation data of the monitoring information 6 as estimation data for the auxiliary variables, and a welding result estimation unit 31 that inputs the estimation data (feature information) for the auxiliary variables calculated by the second auxiliary variable calculation unit 32 and the estimation data for the processing condition information 4 into a welding result estimation model 3, and estimates and outputs the welding result.

The monitoring information 5, 6 output from the welding monitoring device 50 may include, for example, time-series data on physical quantities (welding temperature, sound, light, color, etc.) observed during laser welding. The feature information calculated based on the monitoring information 5 is used as training data for auxiliary variables provided when the learning device 10 learns the welding result estimation model 3, and the feature information calculated based on the monitoring information 6 is used as estimation data for auxiliary variables provided when the estimation device 30 estimates the welding result estimated value 8.

In the welding system 100A of the second embodiment, the welding device is a laser welder 40. Furthermore, the processing condition information 2 and 4 includes at least one processing condition: the material of the workpiece, the thickness of the workpiece, the laser power, the laser irradiation time, the time from the start of laser irradiation to reaching peak power, the time from the peak power to the end of laser irradiation, the fiber diameter, the lens focal length, the focal position, and the laser diameter at the irradiation point. The spot diameter calculated from the combination of the lens, fiber, and focal position is used for machine learning in the welding result learning unit 11. Furthermore, the monitoring information 5 and 6 includes time-series fluctuation data of the laser power and time-series fluctuation data of the radiated near-infrared light intensity at the irradiation point.

In the welding system 100A of the second embodiment, the first auxiliary variable calculation unit 12 of the learning device 10 calculates feature information from the training data of the monitoring information 5 as training data for the auxiliary variables, and the welding result learning unit 11 uses data including this feature information and the training data of the welding result information 1 and processing condition information 2 to learn and create the welding result estimation model 3. The welding result estimation model 3 created by the learning device 10 is then provided to the estimation device 30 of the welding result judgment device 20.

In the estimation device 30 provided with the welding result estimation model 3, the second auxiliary variable calculation unit 32 calculates feature information as estimation data for auxiliary variables from the estimation data of the monitoring information 6 using, for example, the same algorithm as that used by the first auxiliary variable calculation unit 12. Data including this feature information and the estimation data of the processing condition information 4 is input to the welding result estimation model 3, and the welding result estimation unit 31 estimates the welding result and outputs the welding result estimate value 8.

The first auxiliary variable calculation unit 12 may detect multiple predetermined features from the training data of the monitoring information 5, which is time-series data, and calculate the values of each detected feature as feature amount information. The second auxiliary variable calculation unit 32 may detect multiple predetermined features from the estimation data of the monitoring information 6, which is time-series data, and calculate the values of each detected feature as feature amount information. Calculation of this feature amount information will be described later.

As a result, the welding system 100A of the second embodiment can achieve the same effects as the first embodiment. Furthermore, by using the monitoring information 5 and 6 as auxiliary variables, it is possible to estimate physical phenomena caused by welding as events that occurred during welding, including fluctuations in the welding conditions. This enables more precise estimation and improves the accuracy of the estimation.

[Third embodiment]
FIG. 4 is an explanatory diagram showing the basic configuration of a welding system according to a third embodiment of the present invention.
As shown in FIG. 4 , in a welding system 100B according to the third embodiment, a learning device 10 includes a first latent variable calculation unit 13 that calculates teacher data (not shown) of a latent variable that can abstractly express the correlation between the teacher data of the machining condition information 2 from the teacher data of the auxiliary variables (feature amount information) by a technique such as dimension reduction or dimension compression, based on the teacher data of the machining condition information 2 and the teacher data of the auxiliary variables (feature amount information), and outputs latent variable calculation information 7, and a welding result learning unit 11 performs, for example, machine learning based on the teacher data of the machining condition information 2, the teacher data of the welding result information 1, and the teacher data of the latent variables, to create and output a welding result estimation model 3.

The estimation device 30 includes a second latent variable calculation unit 33 that calculates estimation data for latent variables (not shown) that can abstractly express the correlation between the estimation data for the processing condition information 4 from the estimation data for the auxiliary variables (feature information) using techniques such as dimension reduction or dimension compression, based on the estimation data for the processing condition information 4, the estimation data for the auxiliary variables (feature information), and the latent variable calculation information 7. The welding result estimation unit 31 inputs the estimation data for the processing condition information 4 and the estimation data for the latent variables into the welding result estimation model 3, and estimates and outputs the welding result.

In other words, the welding system 100B of the third embodiment is a system applied to laser welding that uses latent variables (teaching data and estimation data) that abstractly represent the correlation between auxiliary variables (feature information) calculated based on monitoring information 5 and 6 and processing condition information 2 and 4 using techniques such as dimensionality reduction or dimensionality compression.

The first latent variable calculation unit 13 outputs the latent variable calculation information 7 determined when calculating the teacher data for the latent variables to the estimation device 30 of the welding result judgment device 20 so that it can be used. The first latent variable calculation unit 13 may calculate one-dimensional or two-dimensional information obtained by dimensionally reducing or compressing the teacher data for the machining condition information 2 and the teacher data (feature information) for the auxiliary variables as the teacher data for the latent information. The second latent variable calculation unit 33 may calculate one-dimensional or two-dimensional information obtained by dimensionally reducing or compressing the estimation data for the machining condition information 4 and the estimation data (feature information) for the auxiliary variables as the estimation data for the latent information.

The training data and estimation data for latent variables are obtained by, for example, reducing or compressing the dimensionality of the auxiliary variables, or the auxiliary variables and explanatory variables, respectively, so as to reduce the correlation between the auxiliary variables and the explanatory variables relative to the auxiliary variables. In this embodiment, latent variables, like explanatory variables, characterize the objective variable and are variables that are not directly observed but are estimated from observed variables (auxiliary variables). In other words, latent variables refer to variables that characterize observed variables and have been dimensionally reduced or compressed from at least the auxiliary variables. The explanatory variables, objective variables, latent variables, and latent variable calculation information 7 used in welding system 100B may each be composed of one or more variables or parameters. In this specification, "prediction/estimation of objective variables" means "prediction of objective variables" or "estimation of objective variables." "Prediction of objective variables" refers to assuming objective variables that are expected to be realized in the future, and "estimation of objective variables" refers to inferring objective variables that are currently realized but cannot be directly observed. Note that "correlation" is not limited to a linear relationship between multiple variables, but refers to a relationship between multiple variables in which a change in one variable causes a change in the other variables, whether linear or nonlinear.

The latent variable calculation information 7 refers to various information determined by the first latent variable calculation unit 13 when calculating the teacher data for the latent variables and used when calculating additional latent variables, and includes, for example, various parameters, various formulas, and/or algorithms. For example, if the latent variable calculation information 7 includes an algorithm, the second latent variable calculation unit 33 can calculate the estimation data for the latent variables using the same algorithm as the algorithm used to calculate the teacher data for the latent variables by the first latent variable calculation unit 13 of the learning device 10. In this way, by calculating the latent variables using the same algorithm during estimation as during learning, estimation using the same welding result estimation model 3 becomes possible.

Furthermore, the latent variable calculation information 7 passed from the learning device 10 to the estimation device 30 can be selected as appropriate for calculating latent variables in each device 10, 30. For example, when the learning device 10 calculates the teacher data for latent variables using principal component analysis, the latent variable parameters may be principal component vectors and fixed values. In addition, latent variable parameters may be selected that exclude principal component vectors whose teacher data contribution rate for a specific auxiliary variable (monitoring information 5) is higher than a predetermined value. This allows, for example, latent variables that have a smaller correlation with explanatory variables than auxiliary variables to be used in the teacher data and estimation data together with the explanatory variables, thereby avoiding deterioration in estimation accuracy due to multicollinearity.

The first auxiliary variable calculation unit 12 and the second auxiliary variable calculation unit 32 detect multiple features that have been automatically extracted in advance, for example, by computer processing, based on selection conditions previously confirmed through experiments, etc., from the training data of monitoring information 5 and the estimation data of monitoring information 6, which are, for example, time-series data, and calculate the value of each detected feature as feature amount information. The multiple features detected at this time are expected to be, for example, the following:

FIG. 5 is a graph showing an example of time-series data.
Examples of the time-series data included in the monitoring information 5 and 6 include time-series fluctuation data D1 of laser power and time-series fluctuation data D2 of the intensity of radiated near-infrared light at the irradiation point, as shown in Fig. 5. In Fig. 5, the horizontal axis represents time (seconds) and the vertical axis represents temperature (×100°C), and multiple features from which feature amount information in the data can be calculated are shown.

As one of the multiple feature values in Figure 5, "rise_period" is a value that indicates the time from when the laser is turned on to the inflection point (predicted (70%)). This "rise_period" value is calculated as the number of measurement points (time x 5 μs) from the time the laser is turned on until it reaches 70% of the measured value at the time the laser rise is complete.

"inflection1_period" is a value that indicates the time from when the laser is turned on until melting begins. The value of "inflection1_period" is calculated using the number of measurement points (time x 5 μs) from the time the laser is turned on to inflection point 1 when the laser rise is complete. "rise_value" is a value that indicates the inflection point (predicted (70%)). The value of this "rise_value" is calculated as 70% of the value measured at the time the laser rise is complete. "inflection1_value" is a value that indicates the temperature at which melting begins (melting point). The value of this "inflection1_value" is calculated as the measurement value at inflection point 1.

"inflection3_value" is a value that indicates the temperature at which solidification begins (melting point). This "inflection3_value" value is calculated as the measurement value at inflection point 3, where solidification begins after the laser is stopped. "laseroff_inf3_period" is a value that indicates the time from laser off to the start of solidification. This "laseroff_inf3_period" value is calculated using the number of measurement points from laser off to inflection point 3 (time x 5 μs). "inflection3_5ms_value" is a value that indicates the first 5 ms of solidification (speed of cooling (hardness of solidified metal) x (related to molten volume)). This "inflection3_5ms_value" value is calculated as the measurement value 5 ms after inflection point 3.

"inflection2_value" is a value that indicates the temperature at which the temperature begins to drop. This "inflection2_value" value is calculated as the measurement value at inflection point 2, where the measurement value begins to drop after the laser is turned off. "inf2_inf3_period" is a value that indicates the time from when the temperature begins to drop to when solidification begins. This "inf2_inf3_period" value is calculated using the number of measurement points from inflection point 2 to inflection point 3 (time x 5 μs). "laseroff_inf2_period" is a value that indicates the time from when the laser is turned off to when the temperature begins to drop. This "laseroff_inf2_period" value is calculated using the number of measurement points from when the laser is turned off to when the temperature begins to drop.

"trend_start_value" refers to the first value (minimum) of the regression line interval. This "trend_start_value" value is calculated as the value at inflection point 1 of the linear regression equation of the measured values for the period from inflection point 1 to inflection point 2 (hereinafter referred to as the "specific period"). "trend_finish_value" refers to the last value (maximum) of the regression line interval. This "trend_finish_value" value is calculated as the value at inflection point 2 of the linear regression equation of the measured values for the specific period. "trend_period" refers to the time from the start of melting to the start of the temperature drop (time calculated). This "trend_period" value is calculated at the time (μs) of the specific period. "trend_slope" refers to the rate of temperature increase during melting. This "trend_slope" value is calculated as the slope of the linear regression equation of the measured values for the specific period.

Other item elements, not shown in Figure 5, include "rise_value_sum," "laserontime_value_sum," "count," "mean," "std," "min," "25%," "50%," "75%," "max," and "dence1-5." "rise_value_sum" is a value that indicates the total temperature rise. The value of this "rise_value_sum" is calculated by integrating the measured values from the time the laser is turned on to "rise_period." "laserontime_value_sum" is a value that indicates the total temperature during laser output. The value of this "laserontime_value_sum" is calculated by integrating the measured values from the time the laser is turned on to the time the laser is turned off.

"Count" is a value that indicates the time from the start of melting until the temperature begins to drop. This "count" value is calculated using the number of measurement points in a specific period (time x 5 μs). "Mean" is a value that indicates the degree of deviation from a linear trend (waviness). This "mean" value is calculated as the average value of the measured values excluding the linear trend (linear regression) for a specific period. "Std" is a value that, like "mean," indicates the degree of deviation from a linear trend (waviness). This "std" value is calculated as the standard deviation of the measured values for a specific period.

"min" is a value that indicates the minimum temperature during melting. This "min" value is calculated as the minimum value of the measured values over a specific period. "25%" means the 25th/100th measured value in ascending order. This "25%" value is calculated as the 25th percentile of the measured values over a specific period. "50%" means the 50th/100th measured value in descending order. This "50%" value is calculated as the 50th percentile of the measured values over a specific period. "75%" means the 75th/100th measured value in descending order. This "75%" value is calculated as the 75th percentile of the measured values over a specific period. "max" is a value that indicates the maximum temperature during melting. This "max" value is calculated as the maximum value of the measured values over a specific period.

"dence1" is a value that indicates the (rough) degree of temperature fluctuation (jaggedness) based on the time-series fluctuation data D2. This "dence1" value is calculated using a spectral density of 195-585 Hz (the sum of the spectral densities of three specified frequencies). "dence2" is a value that indicates the (slightly coarser) degree of temperature fluctuation (jaggedness) based on the time-series fluctuation data D2. This "dence2" value is calculated using a spectral density of 781-1171 Hz (the sum of the spectral densities of three specified frequencies).

"dence3" is a value that indicates the (medium) degree of temperature fluctuation (jaggedness) based on the time-series fluctuation data D2. This "dence3" value is calculated using a spectral density of 1367-1757 Hz (the sum of the spectral densities of three specified frequencies). "dence4" is a value that indicates the (slightly finer) degree of temperature fluctuation (jaggedness) based on the time-series fluctuation data D2. This "dence4" value is calculated using a spectral density of 1953-2343 Hz (the sum of the spectral densities of three specified frequencies). "dence5" is a value that indicates the (finer) degree of temperature fluctuation (jaggedness) based on the time-series fluctuation data D2. This "dence5" value is calculated using a spectral density of 2539-2929 Hz (the sum of the spectral densities of three specified frequencies).

The first auxiliary variable calculation unit 12 and the second auxiliary variable calculation unit 32 detect, for example, automatically extracted features from among the multiple features described above and calculate their values as feature amount information, but this is not limited to this. For example, the first auxiliary variable calculation unit 12 may detect multiple automatically extracted features based on the magnitude of their contribution to the estimation of the objective variable. The contribution level is weighted as the contribution level to the estimation of various physical phenomena related to welding quality, such as spatter and porosity, for multiple features included in the time-series data of the training data in the monitoring information 5, using, for example, a predetermined machine learning algorithm used when the welding result learning unit 11 created the welding result estimation model 3. The first auxiliary variable calculation unit 12 may detect multiple automatically extracted features based on this contribution level and calculate their values as feature amount information.

The welding system 100B of this third embodiment makes it possible to more precisely and accurately estimate welding results, including fluctuations in the welding conditions, by using the training data and estimation data for auxiliary variables calculated by the first and second auxiliary variable calculation units 12 and 32 based on the monitoring information 5 and 6, as well as the training data and estimation data for latent variables calculated by the first and second latent variable calculation units 13 and 33. It also makes it possible to more effectively avoid deterioration in estimation accuracy due to the correlation between the monitoring information 5 and 6 and the processing condition information 2 and 4.

[Welding system operation]
In welding system 100B, welding result information 1 and processing condition information 2 stored in, for example, an external storage device or storage medium, and monitoring information 5 output from welding monitor device 50 are input to learning device 10 as, for example, teacher data for the objective variable, teacher data for the explanatory variable, and teacher data for the auxiliary variable, respectively, and stored in a memory unit not shown.

The learning device 10 calculates feature information (teaching data for auxiliary variables) in the first auxiliary variable calculation unit 12 from the input data, including the teaching data of the monitoring information 5 from the welding monitor device 50. Furthermore, based on the feature information calculated in the first auxiliary variable calculation unit 12 and the teaching data of the machining condition information 2, the first latent variable calculation unit 13 calculates teaching data for latent variables that are abstractly expressed (e.g., with low correlation) by dimensionally reducing or compressing the correlation between the feature information and the teaching data of the machining condition information 2, and outputs latent variable calculation information 7.

The learning device 10 also performs, for example, machine learning based on the training data for the welding result information 1, the training data for the processing condition information 2, and the training data for the latent variables calculated by the first latent variable calculation unit 13 in the welding result learning unit 11 to create a welding result estimation model 3 that estimates the welding result, including the welding strength, etc. The learning device 10 stores the latent variable calculation information 7 used to calculate the training data for the latent variables and the created welding result estimation model 3 in a memory unit within the device, and outputs them so that they can be used by the estimation device 30 of the welding result judgment device 20.

The learning device 10 may output the latent variable calculation information 7 and the welding result estimation model 3 to an external storage device (not shown), for example, via a connected network. In this case, the external storage device temporarily or permanently stores the input latent variable calculation information 7 and the welding result estimation model 3. The various data may be stored not in the above-mentioned external storage device or storage medium, but in a storage device (not shown) that constitutes part of the welding system 100B, for example.

Meanwhile, in the laser welding machine 40, laser welding is performed based on the set processing condition information 4. The welding monitoring device 50 acquires monitoring information 6, which is status observation information during laser welding of the estimation target (e.g., the welded portion of the workpiece) using a sensor 212 attached to the laser welding machine 40. The processing condition information 4 set in the laser welding machine 40 and the monitoring information 6 from the welding monitoring device 50 are input into the estimation device 30 of the welding result judgment device 20 so that they can be used as estimation data for explanatory variables and estimation data for auxiliary variables, respectively.

The estimation device 30 stores various data input together with the latent variable calculation information 7 and the welding result estimation model 3 in a memory unit (not shown), and calculates feature information (estimation data for auxiliary variables) in the second auxiliary variable calculation unit 32 from data including estimation data for the monitoring information 6 from the welding monitoring device 50. Furthermore, based on the feature information calculated in the second auxiliary variable calculation unit 32, the estimation data for the machining condition information 4, and the latent variable calculation information 7, the second latent variable calculation unit 33 calculates estimation data for latent variables that are abstractly expressed (e.g., have a low correlation) by dimensionally reducing or compressing the correlation between the feature information and the estimation data for the machining condition information 4.

The estimation device 30 also inputs the estimation data for the latent variables calculated by the second latent variable calculation unit 33 in the welding result estimation unit 31 and the estimation data for the processing condition information 4 into the welding result estimation model 3, estimates the welding result including the welding strength, etc., and outputs the welding result estimation value 8. The welding result estimation value 8 is stored in the memory unit and is output from the estimation device 30 in an appropriate usable form such as display or printing, and can be used by the welding result determination device 20.

When calculating feature information, calculating latent variables, and creating the welding result estimation model 3, the learning device 10 may add the estimation data for the processing condition information 4 set in the laser welding machine 40 and the estimation data for the monitoring information 6 from the welding monitor device 50 as training data for explanatory variables and auxiliary variables to the training data for the processing condition information 2 and the training data for the monitoring information 5, or may reference and use this as training data for explanatory variables and auxiliary variables. In this way, more data can be used, contributing to improved estimation accuracy.

In the welding system 100B of the third embodiment, feature information and latent variables are calculated based on monitoring information 5 as training data for auxiliary variables of welding result information 1, and a welding result estimation model 3 is created using this information. Furthermore, feature information and latent variables are calculated based on monitoring information 6 as estimation data for auxiliary variables using latent variable calculation information 7, and this information and processing condition information 4, which is estimation data for explanatory variables, are input into the welding result estimation model 3, and a welding result estimation value 8 including weld strength, etc. is output.

Therefore, with the welding system 100B of the third embodiment, the welding results of the estimation target (the welded portion of the workpiece) can be estimated non-destructively from data related to the laser welding processing conditions and data monitored during welding, making it possible to estimate the weld strength contained in the welding results and the spot diameter resulting from the combination of lens, fiber, and focal position.

Generally, with laser welding, even when laser welding is performed under the same processing conditions, weld strength varies depending on factors such as the material of the workpiece and variations in the welding environment, so there is a problem in that it is impossible to measure weld strength without actually destroying the product. Another problem is that it is impossible to know the various processing conditions required to achieve the desired welding results without conducting experiments in which actual laser light is irradiated.

However, the welding system 100B of the third embodiment solves these problems by adopting a configuration that can improve estimation accuracy by using feature information from monitoring information as auxiliary variables and avoid deterioration of estimation accuracy due to correlation, as well as a configuration that can estimate physical phenomena caused by welding, including fluctuations in the welding situation, as events that occurred during welding. At the same time, it is possible to achieve the same effects as the first and second embodiments, but also to enable more precise estimation, improving estimation accuracy and avoiding deterioration of accuracy due to correlation.

Note that while the welding system 100B of the third embodiment is intended to estimate welding results, including the weld strength of laser welding, it is not limited to this. For example, it is also possible to estimate welding results, including the quality of welding, based on various physical phenomena related to welding quality, such as spatter and porosity. In this case, the objective variable in the learning device 10 can be set to the welding quality of laser welding.

Data related to welding quality includes at least one of data indicating good welding and data indicating poor welding, and the processing condition information 2 and monitoring information 5 at that time are used as training data for explanatory variables and auxiliary variables to perform machine learning in the learning device 10 to create a welding result estimation model 3. By using this welding result estimation model 3 and making an estimation based on the estimation data of the processing condition information 4 and monitoring information 6 in the estimation device 30, it is possible to estimate the welding result, including whether the welding is good or bad, even during laser welding.

Furthermore, in the welding system 100B, the learning device 10 and the estimation device 30 can be centrally installed at the same site as the laser welder 40, or separately in a remote location. This enables a wide variety of operations, even for global deployment, regarding the estimation of laser welding strength, etc.

[Fourth embodiment]
6 is an explanatory diagram showing the basic configuration of a welding system according to a fourth embodiment of the present invention. As shown in FIG. 6, welding system 100C according to the fourth embodiment differs from welding system 100B according to the third embodiment in that an estimation device 30 is provided inside welding monitor device 50A.

That is, the welding monitoring device 50A includes a first memory unit (not shown) that stores monitoring information 5 and 6 obtained by using a sensor to monitor the progress of welding by a welding device (laser welder 40) that welds the workpieces; a second memory unit (not shown) that stores processing condition information 2 including the processing conditions of welding previously performed by the welding device (laser welder 40); welding result information 1 related to the welding result obtained after welding based on the processing condition information 2; and monitoring information 5 output from the first memory unit (not shown); and a welding result estimation unit 31 (estimation device 30) that inputs processing condition information 4 including the processing conditions set for the welding device (laser welder 40) and the monitoring information 6 stored in the first memory unit (not shown) into the welding result estimation model 3 stored in the second memory unit (not shown), and estimates and outputs the welding result of welding performed by the welding device (laser welder 40).

Therefore, the welding monitor device 50A is configured to be able to input the latent variable calculation information 7 and welding result estimation model 3 acquired from the learning device 10 into the internal estimation device 30. Then, in the welding monitor device 50A, the internal estimation device 30 inputs the monitoring information 5 and 6 acquired by the sensor 212 and stored in the first memory unit, and the processing condition information 4 set in the laser welder 40, into the welding result estimation model 3 stored in the second memory unit, to estimate the welding result.

This allows the welding monitor device 50A to estimate the welding result, including the weld strength, and output the welding result estimation value 8 by itself. The welding system 100C of the fourth embodiment can achieve the same effects as the third embodiment, and can also realize the functions of the estimation device 30 by itself.

Fifth Embodiment
Fig. 7 is an explanatory diagram showing the basic configuration of a welding system according to a fifth embodiment of the present invention. As shown in Fig. 7, a welding system 100D according to the fifth embodiment is similar to the welding system 100B according to the third embodiment in that it uses monitoring information 5 and 6 from a welding monitor device (resistance welding checker) 51 as auxiliary variables (teaching data and estimation target data) correlated with processing condition information 2 and 4. However, it differs from the welding system 100B according to the third embodiment in that it is applied to resistance welding.

That is, in the welding system 100D of the fifth embodiment, the welding device is a resistance welding machine (resistance welding power supply/head 41). Furthermore, the processing condition information 2 and 4 includes at least one processing condition: welding current, welding pressure, current application time, welding method, workpiece material, workpiece thickness, workpiece surface treatment, workpiece combination conditions, sampling period, displacement detection resolution, and nugget diameter. Furthermore, the monitoring information 5 and 6 includes time-series fluctuation data of the welding current and voltage and time-series fluctuation data of the welding point pressure and displacement.

In the welding system 100D of the fifth embodiment, the first auxiliary variable calculation unit 12 of the learning device 10 calculates feature information (teaching data for auxiliary variables) based on the monitoring information 5, the first latent variable calculation unit 13 calculates teaching data for latent variables based on the feature information and outputs latent variable calculation information 7, the welding result learning unit 11 creates a welding result estimation model 3, and this welding result estimation model 3 and latent variable calculation information 7 are provided to the estimation device 30 of the welding result judgment device 20.

The second auxiliary variable calculation unit 32 of the estimation device 30 of the welding result determination device 20 uses the latent variable calculation information 7 to calculate feature information (data for estimating auxiliary variables) based on the monitoring information 6, the second latent variable calculation unit 33 calculates data for estimating latent variables based on the feature information, and the welding result estimation unit 31 inputs data including this data for estimating latent variables into the welding result estimation model 3 to estimate the welding result. The obtained welding result estimated value 8 can then be used to estimate the welding strength, etc., determine the welding result, or output the welding result estimated value 8 to an external device. In this way, the welding system 100D of the fifth embodiment can also achieve the same effects as the welding system 100B of the third embodiment.

Although several embodiments of the present invention have been described above, these embodiments are presented as examples and are not intended to limit the scope of the invention. These novel embodiments may be embodied in a variety of other forms, and various omissions, substitutions, and modifications may be made without departing from the spirit of the invention. These embodiments and their variations are within the scope and spirit of the invention, and are also included in the scope of the invention and its equivalents as set forth in the claims.

1. Welding result information (training data for objective variables)
2. Processing condition information (teaching data for explanatory variables)
3. Welding result estimation model 4. Processing condition information (data for estimating explanatory variables)
5. Monitoring information (teaching data for auxiliary variables)
6. Monitoring information (data for estimating auxiliary variables)
7 Latent variable calculation information 8 Welding result estimation value 10 Learning device 11 Welding result learning unit 12 First auxiliary variable calculation unit 13 First latent variable calculation unit 20 Welding result determination device 30 Estimation device 31 Welding result estimation unit 32 Second auxiliary variable calculation unit 33 Second latent variable calculation unit 40 Laser welding machine 41 Resistance welding power source/head 50, 50A Welding monitor device 51 Resistance welding checker 100 Welding system

Claims (8)

a welding device for welding the workpiece;
a welding monitor device that monitors the progress of welding by the welding device using a sensor and outputs the obtained monitoring information;
a learning device that receives as input, as training data, processing condition information including processing conditions for welding previously performed by the welding device and welding result information related to the welding result obtained after welding based on the processing conditions, and creates and outputs a welding result estimation model based on the training data;
an estimation device that inputs processing condition information including processing conditions set in the welding device as estimation data into the welding result estimation model created by the learning device, and estimates and outputs a welding result of welding performed by the welding device based on the processing conditions ,
The learning device
a first auxiliary variable calculation unit that calculates feature amount information from the training data of the monitoring information as training data of an auxiliary variable;
a first latent variable calculation unit that calculates teacher data of latent variables that can be abstractly expressed by performing dimensional reduction or dimensional compression on the correlation between the teacher data of the auxiliary variables and the teacher data of the machining condition information based on the teacher data of the machining condition information and the teacher data of the auxiliary variables calculated by the first auxiliary variable calculation unit, and outputs latent variable calculation information;
a welding result learning unit that creates and outputs the welding result estimation model based on teacher data of the processing condition information, teacher data of the welding result information, and teacher data of the latent variables,
The estimation device includes:
a second auxiliary variable calculation unit that calculates the feature amount information from the estimation data of the monitoring information as estimation data of an auxiliary variable;
a second latent variable calculation unit that calculates estimation data of a latent variable that can be abstractly expressed by performing dimensional reduction or dimensional compression on the correlation between the estimation data of the auxiliary variable and the estimation data of the machining condition information, based on the estimation data of the machining condition information, the estimation data of the auxiliary variable calculated by the second auxiliary variable calculation unit, and the latent variable calculation information;
a welding result estimation unit that inputs the estimation data of the processing condition information and the estimation data of the latent variables into the welding result estimation model, and estimates and outputs the welding result.
Welding system. a welding device for welding the workpiece;
a welding monitor device that monitors the progress of welding by the welding device using a sensor and outputs the obtained monitoring information;
a learning device that receives as input, as training data, processing condition information including processing conditions for welding previously performed by the welding device and welding result information related to the welding result obtained after welding based on the processing conditions, and creates and outputs a welding result estimation model based on the training data;
an estimation device that inputs processing condition information including processing conditions set in the welding device as estimation data into the welding result estimation model created by the learning device, and estimates and outputs a welding result of welding performed by the welding device based on the processing conditions,
The learning device
a first auxiliary variable calculation unit that calculates feature amount information from the training data of the monitoring information as training data of an auxiliary variable;
a first latent variable calculation unit that calculates one-dimensional or two-dimensional information by dimension-reducing or dimension-compressing the teacher data of the processing condition information and the teacher data of the auxiliary variables calculated by the first auxiliary variable calculation unit as teacher data of latent variables, and outputs latent variable calculation information;
a welding result learning unit that creates and outputs the welding result estimation model based on teacher data of the processing condition information, teacher data of the welding result information, and teacher data of the latent variables,
The estimation device includes:
a second auxiliary variable calculation unit that calculates the feature amount information from the estimation data of the monitoring information as estimation data of an auxiliary variable;
a second latent variable calculation unit that calculates, based on the latent variable calculation information, one-dimensional or two-dimensional information obtained by dimension reduction or dimension compression of the estimation data of the machining condition information and the estimation data of the auxiliary variables calculated by the second auxiliary variable calculation unit, as estimation data of latent variables;
a welding result estimation unit that inputs the estimation data of the processing condition information and the estimation data of the latent variables into the welding result estimation model, and estimates and outputs the welding result. the first auxiliary variable calculation unit detects a plurality of predetermined features from teacher data of the monitoring information, which is time-series data, and calculates values of each of the detected features as the feature amount information;
3. The welding system according to claim 1, wherein the second auxiliary variable calculation unit detects a plurality of predetermined features from the estimation data of the monitoring information, which is time- series data, and calculates values of each of the detected features as the feature amount information. the welding device is a laser welder;
The welding system according to any one of claims 1 to 3, wherein the processing condition information includes at least one processing condition of the material of the workpiece, the thickness of the workpiece, the laser power, the laser irradiation time, the time from the start of laser irradiation to reach peak power, the time from the peak power to the end of laser irradiation, the fiber diameter, the lens focal length, the focal position, and the laser diameter at the irradiation point. The welding system according to claim 4 , wherein the monitoring information includes time-series fluctuation data of laser power and time-series fluctuation data of radiated near-infrared light intensity at the irradiation point. the welding device is a resistance welder;
The welding system according to any one of claims 1 to 3, wherein the processing condition information includes at least one processing condition of a welding current, a welding pressure, a current application time, a welding method, a material of the workpiece, a thickness of the workpiece, a surface treatment of the workpiece, a combination condition of the workpiece, a sampling period , a displacement detection resolution, and a nugget diameter. a first storage unit that stores monitoring information obtained by monitoring the progress of welding by a welding device that welds workpieces using a sensor;
a second storage unit that stores latent variable calculation information and a welding result estimation model created based on teacher data of processing condition information including processing conditions for welding previously performed by the welding device, teacher data of welding result information related to a welding result obtained after welding based on the processing condition information, and teacher data of latent variables calculated so as to be abstractly expressible by dimensionally reducing or dimensionally compressing a correlation between the teacher data of the auxiliary variables and the teacher data of the processing condition information based on teacher data of the auxiliary variables calculated as feature amount information from the teacher data of the processing condition information and the teacher data of the monitoring information output from the first storage unit;
a welding result estimation unit that inputs estimation data of machining condition information including machining conditions set in the welding device, estimation data of auxiliary variables calculated as the feature amount information from the estimation data of the monitoring information stored in the first storage unit, estimation data of latent variables calculated so as to be abstractly expressible by dimensionally reducing or dimensionally compressing a correlation between the estimation data of the auxiliary variables and the estimation data of the machining condition information, based on the estimation data of the auxiliary variables, which includes machining conditions set in the welding device, and the estimation data of the monitoring information stored in the first storage unit, and the latent variable calculation information stored in the second storage unit, and estimates and outputs a welding result of welding performed by the welding device. a first storage unit that stores monitoring information obtained by monitoring the progress of welding by a welding device that welds workpieces using a sensor;
a second storage unit that stores latent variable calculation information and a welding result estimation model created based on teacher data of processing condition information including processing conditions for welding previously performed by the welding device, teacher data of welding result information related to a welding result obtained after welding based on the processing condition information, and teacher data of latent variables calculated as one-dimensional or two-dimensional information obtained by dimension-reducing or dimension-compressing teacher data of auxiliary variables calculated as feature amount information from the teacher data of the processing condition information and the teacher data of the monitoring information output from the first storage unit;
and a welding result estimation unit that inputs estimation data of latent variables calculated as one-dimensional or two-dimensional information obtained by dimension-reducing or dimension-compressing estimation data of machining condition information including machining conditions set in the welding device and estimation data of auxiliary variables calculated as the feature amount information from the estimation data of the monitoring information stored in the first storage unit based on the latent variable calculation information stored in the second storage unit, and the estimation data of the machining condition information to the welding result estimation model stored in the second storage unit, and estimates and outputs a welding result of welding performed by the welding device.