JP2023061287A - Welding system and welding monitor device - Google Patents

Abstract

To easily search for a processing condition suitable for welding with a simple structure in a short time and estimate a welded result of a welded part by welding under the suitable processing condition in a non-destructive manner.SOLUTION: A welding system according to one aspect comprises: a welding device that welds a work-piece; a learning device that inputs, as teacher data, processing condition information including a processing condition for welding previously executed by the welding device and welded-result information relevant to a welded result obtained after welding based on the processing condition, and generates and outputs a welded-result estimation model, on the basis of the teacher data; and an estimating device that inputs, as data for estimation, processing condition information including a processing condition set on the welding device in the welded-result estimation model generated by the learning device, and estimates and outputs a welded result of welding performed by the welding device on the basis of the processing condition.SELECTED DRAWING: Figure 1

Description

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

Conventionally, in welding, when determining the appropriate processing conditions corresponding to the required specifications of welding, it is necessary to automatically determine what processing conditions should be used in as few trials as possible. , searching for appropriate processing conditions (see Patent Document 1), and repeating search experiments by testing all possible combinations of many factors such as materials constituting welding and welding time. It was

In addition, non-destructive inspection using ultrasonic waves (see Patent Document 2) and actual destruction of the welded portion are performed to check whether the welded portion is correctly welded after welding is completed and whether the required weld strength is obtained. A destructive test was conducted to confirm the state of

JP 2012-236267 A JP 2009-156701 A

However, the apparatus disclosed in Patent Document 1 requires a plurality of processing experiments, and the search experiment for a method of trying all possible combinations relies on the intuition and experience of the operator in many ways. There is a problem that it takes a long time to find the appropriate processing conditions.

In addition, in order to check the welding strength of the welded portion by non-destructive inspection using the device disclosed in Patent Document 2, the device requires components such as sensors and probes, so the device configuration should be minimized. Furthermore, in order to confirm whether the weld strength of welds that have been welded under certain processing conditions is the same between those that have undergone destructive inspection and those that have not undergone destructive inspection, non-destructive inspection should be avoided. I have a problem that I can't pass.

One aspect of the present invention is to enable a search for suitable processing conditions for welding to be performed easily in a short time with a simple configuration, and to non-destructively estimate the welding result of a welded portion by welding under suitable processing conditions. A welding system and welding monitoring device capable of

A welding system according to an aspect of the present invention includes a welding device that welds a workpiece, processing condition information including processing conditions for welding performed in advance by the welding device, and welding obtained after welding based on the processing conditions. a learning device for inputting welding result information related to the welding results as training data, creating and outputting a welding result estimation model based on the teaching data; an estimating device for inputting processing condition information including processing conditions set in a welding device as estimation data, and estimating and outputting a welding result of welding performed by the welding device based on the processing conditions.

According to the welding system according to one aspect of the present invention, the welding result estimation model is generated based on the teaching data of the welding processing condition information previously performed by the welding device and the welding result information obtained after welding based thereon. is created. In addition, by inputting the processing condition information set in the welding device into the welding result estimation model as data for estimation, the welding results of the welding performed by the welding device are estimated and output based on the processing conditions. The welding result of the weld zone by welding below can be estimated non-destructively. In addition, since the welding result can be estimated by inputting the processing condition information set in the welding device to the welding result estimation model, the search work and the number of searches for the processing condition can be reduced with a simple configuration, and the appropriate processing condition for welding can be obtained. can be easily searched for in a short time.

ADVANTAGE OF THE INVENTION According to one aspect of the present invention, it is possible to easily search for suitable processing conditions for welding in a short time with a simple configuration, and to non-destructively estimate the welding result of a welded portion by welding under suitable processing conditions. .

FIG. 1 is an explanatory diagram showing the basic configuration of the welding system according to the first embodiment of the present invention. FIG. 2 is an explanatory diagram showing the basic hardware configuration of the learning device and/or estimation device of the welding system. FIG. 3 is an explanatory diagram showing the basic configuration of the welding system according to the 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 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 invention. FIG. 7 is an explanatory diagram showing the basic configuration of the welding system according to the fifth embodiment of the present invention.

Hereinafter, a welding system and a welding monitor device according to embodiments of the present invention will be described in detail with reference to the accompanying drawings. However, the following embodiments do not limit the invention according to each claim, and all combinations of features described in the embodiments are essential to the solution of the invention. Not exclusively. Further, in the following embodiments, there are cases where the arrangement, scale, dimensions, etc. of each component are exaggerated or dwarfed, and descriptions of some components are omitted.

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

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

In this welding system 100, for example, the training data of the objective variable input to the learning device 10 includes welding result information 1 related to the welding strength F (N/mm ² ) of the welded portion of the laser welding. Examples of variable teaching data include processing condition information 2 including various processing conditions (laser power (output value), laser irradiation time, etc.). Note that the objective variable and the explanatory variable 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 the laser welder 40, and the processing condition information 2 and 4 includes the material of the workpiece, the thickness of the workpiece, the laser power, the laser irradiation time, It includes at least one processing condition from the time from the start of laser irradiation to reaching the peak power, the time from the peak power to the end of laser irradiation, fiber diameter, lens focal length, focal position, and laser diameter at the irradiation point. Note that the spot diameter calculated by the combination of the lens, fiber, and focal position is used for machine learning in the welding result learning unit 11, for example.

The laser welder 40 is a device that performs welding by irradiating a workpiece with a laser beam under processing conditions specified by an external input, for example. The laser welder 40 is configured to be able to acquire or transmit/receive various conditions under which the laser beam is irradiated through various types of communication (network communication or communication by direct connection). Welding results also include weld strength, spot diameter, weld quality, and the like.

Thus, the welding system 100 of the first embodiment is used as a system for estimating the welding strength included in the welding result of laser welding by the laser welder 40, for example. The welding result information 1 may include the spot diameter, may be an index for evaluating the metal structure from the appearance or cross section as the welding quality, or may be an index for evaluating the corrosion resistance. good too. Therefore, the welding system 100 can also be used as a system for estimating the spot diameter and welding quality included in the welding results of laser welding.

The learning device 10 includes a welding result learning section 11 . The welding result learning unit 11 combines the processing condition information 2 input to the learning device 10 and the welding result information 1 related to the welding result of the weld obtained by, for example, a destructive inspection after welding based on this processing condition. A welding result estimation model 3 is created by performing machine learning using, for example, a predetermined machine learning algorithm based on the teacher data. In addition, destructive inspection, etc., can be used for welding tests and inspections listed as destructive tests listed on the website of the Japan Welding Association Welding Information Center, for example, JIS Z 2241 Metal Material Tensile Test Method and other posted Any of the JIS standards may be selected, and the experimental method and evaluation method are also based on those described in the JIS standard. The learned welding result estimation model 3 is composed of, for example, computer-operated software. In this example, when processing condition information 4 is input as estimation data, welding result estimation values 8 including welding strength, etc. are output. program.

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

On the other hand, estimating device 30 is included in welding result determination device 20 and includes welding result estimating section 31 . The welding result determination device 20 includes the welding result estimation model 3 output from the learning device 10 together with the estimation device 30 . The welding result estimation unit 31 inputs the processing condition information 4 set in the laser welding machine 40 input to the estimation device 30 into the welding result estimation model 3 provided in the welding result determination device 20 as estimation data. , the welding result of the welding performed by the laser welding machine 40 is estimated based on the processing conditions, and the welding result estimated value 8 is output.

Based on the welding result estimation value 8 output from the welding result estimation unit 31 of the estimation device 30, the welding result determination device 20 determines, for example, the welding strength and/or the spot diameter (hereinafter referred to as "welding strength etc."). ) can be determined. Note that the welding result estimated value 8 is information that can numerically express, for example, the tensile strength (unit: N) of a certain welding point (welded part) and the distribution range of the strength. In addition, the welding result estimated value 8 is information that can numerically express, for example, the size of the irradiation range of the laser beam on the surface of a certain welding point (weld portion) or the size of the melted portion of the welded portion with respect to the spot diameter. That's what I mean.

In machine learning and estimation in the learning device 10 and the estimation device 30, algorithms include Regression Analysis (RA), Principal Component Analysis (PCA), and Singular Value Decomposition (SVD). , Linear Discriminant Analysis (LDA), Independent Component Analysis (ICA), Gaussian Process Latent Variable Model (GPLVM), Logistic Regression on: LR), support vector machine (SUPPORT VECTOR MACHINE: SVM), identification analysis (Discriminant Analysis: Da), Random Forest (RANDOM FOREST: RF), Ranking SVM (RANKING SUPPORT VECT OR Machine: RSVM), gradient boosting (GRADIENT BOOSTING: GB), Naive Bayes ( Various algorithms such as Naive Bayes (NB) and K-Nearest Neighbor Algorithm (K-NN) can be used.

As described above, in the welding system 100 of the first embodiment, the welding result estimation model 3 is created by performing machine learning, for example, based on the processing condition information 2 and the welding result information 1 in the welding result learning unit 11 of the learning device 10. do. Then, by inputting the processing condition information 4 set in the laser welder 40 to the welding result estimation model 3 by the welding result estimation unit 31 of the estimation device 30, the welding result (welding strength , spot diameter) and output the welding result estimated value 8 .

The welding system 100 can obtain the welding result estimated value 8 simply by inputting the processing condition information 4 set in the laser welder 40 to the learned welding result estimation model 3, so that a plurality of pieces of processing condition information 4 can be obtained. It is possible to eliminate the complicated work of actually irradiating a laser beam using the and measuring the results, and reduce the combination of processing condition information used for searching for processing conditions that achieve the desired welding strength. It becomes possible. Therefore, it is possible to easily search for suitable processing conditions for laser welding in a short time with a simple configuration, and to non-destructively estimate the welding result of the welded portion by laser welding under suitable processing conditions. can.

Each data of the welding result information (teaching data of the objective variable) 1 and the processing condition information (teaching data of the explanatory variable) 2 in the welding system 100 can be composed of one or more variables or parameters. For example, it can be stored in a stationary or portable storage device or storage medium (not shown). Further, each data can be transmitted and received via an information communication medium such as the Internet, and may be raw data obtained by a measuring device such as a sensor (not shown). Also, the welding result estimation model 3 may be input/output between the learning device 10 and the estimation device 30 via the above storage medium or information communication medium.

As shown in FIG. 2, the learning device 10 and/or the estimating device 30 of the welding system 100 has a basic hardware configuration such as a CPU (Central Processing Unit) 201, a RAM (Random Access Memory) 202, and a ROM. (Read Only Memory) 203 , HDD (Hard Disk Drive) 204 and/or SSD (Solid State Drive) 205 , and memory card 206 .

Further, the learning device 10 and/or the estimation device 30 are provided with an input I/F (interface) 207, an output I/F (interface) 208, and a communication I/F (interface) 209, for example. Each component 201 to 209 is interconnected by a bus 200, respectively.

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, and the like. The CPU 201 executes a learning program in the learning device 10 to realize the function of each section of the learning device 10 including the function of the welding result learning section 11 .

Further, CPU 201 executes an estimation program in estimating device 30 to realize the function of each section of estimating device 30 including the function of welding result estimating section 31 . Note that the CPUs 201 of the learning device 10 and the estimating device 30 can cooperate to control the entire welding system 100 .

The RAM 202 can be used as a work area for arithmetic processing of the CPU 201 . The ROM 203 stores at least the various programs described above in a readable manner. The HDD 204 and SSD 205 store the various data described above in a readable and writable manner. The memory card 206 stores these various data in a readable and writable manner, and constitutes a removable storage medium for each of the devices 10 and 30 . The HDD 204, SSD 205, and memory card 206 implement the functions of the storage device or storage medium described above.

For example, a sensor 212 is connected to the input I/F 207 to acquire detection information. The sensor 212 includes various sensors such as a temperature sensor, an optical sensor, an acoustic sensor, and an image sensor. Input I/F 207 is connected to touch panel 211 functioning as an operating unit or input unit of learning device 10 and/or estimating device 30 , and receives information accompanying operation input from the user of welding system 100 . Various input devices such as a keyboard and a mouse (including a trackball mouse) (not shown) can also be connected to the input I/F 207 .

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

[Second embodiment]
FIG. 3 is an explanatory diagram showing the basic configuration of the welding system according to the second embodiment of the present invention. In the following description including FIG. 3, the same reference numerals are given to the same constituent elements as those of the first embodiment and its modification, so redundant description will be omitted below.

As shown in FIG. 3, the welding system 100A according to the second embodiment includes monitoring information 5 and 6 obtained by monitoring (state observation) the progress of welding by the welding device (laser welder 40) with a sensor 212. , and the learning device 10 calculates feature amount information as teacher data of auxiliary variables from the teacher data of the monitoring information 5 output from the welding monitor device 50, and adds the teacher data with the auxiliary variables. Based on the data, the welding result estimation model 3 is created and output, the estimation device 30 calculates the feature amount information from the estimation data of the monitoring information 6 output from the welding monitor device 50 as data for estimation of auxiliary variables, By inputting the data for estimation including the auxiliary variables, the welding result is estimated and output. Note that the welding monitor device 50 is a device capable of monitoring the welded portion of the workpiece with the sensor 212, acquiring the output information from the sensor 212 as the monitoring information 5, 6, and outputting it.

That is, the welding system 100A of the second embodiment uses the monitoring information 5 and 6 from the welding monitor device 50 as auxiliary variables (teaching data and estimation data for) having a correlation with the processing condition information 2 and 4. It is a system applied to welding. The auxiliary variables correlated with explanatory variables (processing condition information) include, for example, physical quantities (welding temperature, sound, light, color, etc.) observed during or after laser welding. However, the auxiliary variables used in welding system 100A are not limited to those illustrated.

Specifically, since the auxiliary variables are used, the learning device 10 of the welding system 100A includes a first auxiliary variable calculator 12 that calculates feature amount information from the teacher data of the monitoring information 5 as teacher data of the auxiliary variables, and a welding result Based on the teacher data of the information 1, the teacher data of the processing condition information 2, and the teacher data (feature amount information) of the auxiliary variables calculated by the first auxiliary variable calculator 12, for example, machine learning is performed to perform welding. and a welding result learning unit 11 that creates and outputs the result estimation model 3 .

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

The monitoring information 5 and 6 output from the welding monitor device 50 includes, for example, information including time-series data of physical quantities (welding temperature, sound, light, color, etc.) observed during laser welding. The feature amount information calculated based on the monitoring information 5 is used as teacher data for auxiliary variables given when the welding result estimation model 3 is learned in the learning device 10, and the feature amount information calculated based on the monitoring information 6 is , are used as data for estimating auxiliary variables given when estimating the welding result estimated value 8 in the estimating device 30 .

In the welding system 100A of the second embodiment, the welding device is a laser welder 40. As shown in FIG. The processing condition information 2 and 4 includes 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 until reaching the peak power, and the time from the peak power to the end of laser irradiation. It includes at least one processing condition of time to, fiber diameter, lens focal length, focal position, and laser diameter at the irradiation point. Note that the spot diameter calculated by combining the lens, fiber, and focal position is used for machine learning in the welding result learning unit 11 . The monitoring information 5 and 6 includes time-series variation data of laser power and time-series variation data of radiated near-infrared light intensity at the irradiation point.

In the welding system 100A of the second embodiment, the first auxiliary variable calculator 12 of the learning device 10 calculates the feature amount information from the teacher data of the monitoring information 5 as the teacher data of the auxiliary variable. The welding result estimation model 3 is learned and created by the welding result learning unit 11 using the data including the information 1 and the teaching data of the processing condition information 2 . Then, the welding result estimation model 3 created by the learning device 10 is provided to the estimation device 30 of the welding result determination device 20 .

In the estimating device 30 to which the welding result estimation model 3 is given, the second auxiliary variable calculating unit 32 uses the same algorithm as the first auxiliary variable calculating unit 12, for example, from the data for estimation of the monitoring information 6 to The feature amount information is calculated as data for estimating the auxiliary variables, and the data including this feature amount information and the data for estimating the machining condition information 4 is input to the welding result estimation model 3, and the welding result estimation unit 31 estimates the welding result. is estimated and the welding result estimated value 8 is output.

The first auxiliary variable calculation unit 12 detects a plurality of predetermined features from the teacher data of the monitoring information 5, which is time-series data, and calculates the value of each detected feature as feature amount information. You may do so. In addition, the second auxiliary variable calculation unit 32 detects a plurality of predetermined features from the data for estimation of the monitoring information 6, which is time-series data, and calculates the value of each detected feature as feature amount information. You may make it Calculation of this feature amount information will be described later.

Thus, according to the welding system 100A of the second embodiment, it is possible to achieve the same effects as those of the first embodiment. In addition, by using the monitoring information 5 and 6 as auxiliary variables, it is possible to estimate the physical phenomenon caused by welding as an event that occurred during welding, including fluctuations in the welding state. This enables more precise estimation and improves the accuracy of estimation.

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

Based on the data for estimation of the processing condition information 4, the data for estimation of the auxiliary variables (feature information), and the latent variable calculation information 7, the estimation device 30 extracts from the data for estimation of the auxiliary variables (feature information) A second latent variable calculation unit that calculates latent variable estimation data (not shown) that can abstractly represent the correlation between the estimation data of the processing condition information 4 and the dimension reduction or dimension compression. 33, the welding result estimation unit 31 inputs the estimation data of the processing condition information 4 and the estimation data of the latent variables to the welding result estimation model 3, estimates and outputs the welding result.

That is, the welding system 100B of the third embodiment performs dimension reduction or dimension compression on the correlation between the auxiliary variables (feature information) calculated based on the monitoring information 5 and 6 and the processing condition information 2 and 4. This system is applied to laser welding using latent variables (teaching data and estimation data) expressed abstractly by a method.

The first latent variable calculator 13 outputs the latent variable calculation information 7 determined at the time of calculating the latent variable teacher data to the estimation device 30 of the welding result determination device 20 so that it can be used. The first latent variable calculation unit 13 uses one-dimensional or two-dimensional information obtained by dimension reduction or dimension compression of the teacher data of the processing condition information 2 and the teacher data (feature amount information) of the auxiliary variables as latent information teacher data. You may make it calculate. The second latent variable calculation unit 33 converts one-dimensional or two-dimensional information obtained by dimension reduction or dimension compression of the estimation data of the processing condition information 4 and the estimation data of the auxiliary variables (feature amount information) into latent information. may be calculated as data for estimation of .

The training data and estimation data of the latent variables are such that the auxiliary variables or the auxiliary variables and the explanatory variables are dimensionally reduced or dimensionally compressed as described above, respectively, so that the correlation between the explanatory variables is reduced more than the auxiliary variables. It is a thing. In the present embodiment, the latent variable is a variable that characterizes the objective variable in the same way as the explanatory variable, and is a variable that is estimated from observed variables (auxiliary variables) but not directly observed. In other words, a latent variable means a variable that characterizes the observed variable and that is dimensionality-reduced or dimensionality-compressed from at least the auxiliary variables. The explanatory variables, the objective variables, the latent variables, and the latent variable calculation information 7 used in the welding system 100B can each be composed of one or more variables or parameters. In this specification, "prediction/estimation of the objective variable" is used to mean "prediction of the objective variable" or "estimation of the objective variable". "Objective variable prediction" means to assume an objective variable that is expected to be realized in the future (future), and "objective variable estimation" is an objective variable that is currently realized but cannot be directly observed. means to estimate "Correlation" is not limited to a linear relationship between multiple variables, but whether it is linear or non-linear, between multiple variables, if one variable changes, the other variables will change. means relationship.

The latent variable calculation information 7 means various types of information that are determined in the first latent variable calculation unit 13 when calculating teacher data for latent variables and used when calculating additional latent variables. or algorithms are included. For example, when the latent variable calculation information 7 includes an algorithm, the second latent variable calculation unit 33 uses the same algorithm as the one used to calculate latent variable teacher data in the first latent variable calculation unit 13 of the learning device 10. Algorithms may be used to calculate data for latent variable estimation. In this way, even during estimation, estimation using the same welding result estimation model 3 is possible by calculating latent variables using the same algorithm as during learning.

In addition, the latent variable calculation information 7 passed from the learning device 10 side to the estimation device 30 side can be selected by each of the devices 10 and 30 so as to be suitable for calculating the latent variables. For example, when calculation of latent variable teacher data in the learning device 10 is performed by principal component analysis, the latent variable parameters may be principal component vectors and fixed values, and in addition to these, specific auxiliary variables (monitoring information 5) The latent variable parameters may be selected by excluding the principal component vectors whose teacher data contribution rate is higher than a predetermined value. As a result, it is possible to use, for example, a latent variable that has a smaller correlation with an explanatory variable than an auxiliary variable as the teacher data and the estimation data together with the explanatory variable, thereby avoiding deterioration in estimation accuracy due to multicollinearity. becomes.

The first auxiliary variable calculation unit 12 and the second auxiliary variable calculation unit 32 calculate, for example, selection conditions confirmed in advance by experiments, etc., from the teacher data of the monitoring information 5 and the estimation data of the monitoring information 6, which are composed of time-series data, for example. A plurality of features automatically extracted in advance by computer processing etc. are detected based on the above, and the value of each detected feature is calculated as feature amount information. something is assumed.

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 variation data D1 of laser power and time-series variation data D2 of radiant near-infrared light intensity at the irradiation point, as shown in FIG. . In FIG. 5 , the horizontal axis represents time (seconds) and the vertical axis represents temperature (×100° C.), showing a plurality of features from which feature amount information in data can be calculated.

As the multiple feature values in FIG. 5, “rise_period” is a value that means the time from laser-on to the inflection point (prediction (70%)). The value of this "rise_period" is calculated by the number of measurement points (time in 5 μs) from the time when the laser is turned on until it reaches 70% of the measured value at the time when the laser rise is completed.

Also, "inflection1_period" is a value that means the time from laser-on to the start of melting. The value of this "inflection1_period" is calculated by the number of measurement points (time is x 5 µs) from the time when the laser is turned on to the point of inflection 1 when the laser rise is completed. "rise_value" is a value that means an inflection point (prediction (70%)). The value of this "rise_value" is calculated as 70% of the measured value at the laser rise completion time. "inflection1_value" is a value that means the temperature at which melting starts (melting point). The value of this “inflection1_value” is calculated as the measured value at the inflection point 1.

"inflection3_value" is a value that means the temperature at which solidification starts (melting point). The value of this "inflection3_value" is calculated as a measured value at inflection point 3 where coagulation starts after the laser stops. "laseroff_inf3_period" is a value that means the time from laser off to the start of coagulation. The value of this "laseroff_inf3_period" is calculated by the number of measurement points from laser-off to inflection point 3 (time is x5 μs). "inflection3_5ms_value" is a value that means the first 5ms of solidification (cooling speed (hardness of solidified metal) x (relative to molten volume)). The value of this "inflection3_5ms_value" is calculated as a measured value after 5ms have elapsed from the inflection point 3.

Also, "inflection2_value" is a value that means the temperature at which the temperature begins to drop. The value of this "inflection2_value" is calculated as the measured value at the inflection point 2 where the measured value starts decreasing after the laser is turned off. "inf2_inf3_period" is a value that means the time from the start of temperature drop to the start of solidification. The value of this "inf2_inf3_period" is calculated by the number of measurement points from the inflection point 2 to the inflection point 3 (time is 5 μs). "laseroff_inf2_period" is a value that means the time from when the laser is turned off until the temperature starts to drop. The value of this "laseroff_inf2_period" is calculated by the number of measurement points from laser-off to inflection point 2 (time is x5 μs).

Also, "trend_start_value" means the first value (minimum) of the regression line section. The value of this "trend_start_value" is the time of inflection point 1 of Calculated as a value at time. "trend_finish_value" means the last value (maximum) of the regression line section. The value of this "trend_finish_value" is calculated as the value at the time of the inflection point 2 of the linear regression equation of the measured values in the specific period. "trend_period" is a value that means the time (calculated time) from the start of melting to the start of temperature drop. The value of this "trend_period" is calculated with the time (μs) of the specific period. "trend_slope" is a value that signifies the rate of temperature rise during melting. The value of this "trend_slope" is calculated from the slope of the linear regression equation of the measured values in the specific period.

In addition, although illustration in FIG. 5 is omitted, as a plurality of item elements, "rise_value_sum", "laserontime_value_sum", "count", "mean", "std", "min", "25%", "50 %", "75%", "max", "dence 1 to 5", and the like. "rise_value_sum" is a value that means the total rise temperature. The value of this "rise_value_sum" is calculated by integrating the measured values in the period from the laser-on time to "rise_period". "laserontime_value_sum" is a value that means the total amount of temperature in the laser output. The value of this “laserontime_value_sum” is calculated by integrating the measured values during the period from laser-on to laser-off.

Also, "count" is a value that means the time from the start of melting until the temperature starts to drop. The value of this "count" is calculated by the number of measurement points in a specific period (time is multiplied by 5 μs). "Mean" is a value that means the degree of deviation from the linear trend (the degree of undulation). This "mean" value is calculated as the average value of the measured values excluding the linear trend (linear regression) for a specific period. Also, "std" is a value that means the degree of deviation (waviness) from the linear trend, like "mean". This "std" value is calculated as the standard deviation of the measurements over a specified period of time.

Also, "min" is a value that means the minimum temperature during melting. This "min" value is calculated as the minimum value of the measurements over a specified period. "25%" means 25/100 number measurements in ascending order. This "25%" value is calculated as the 25% tile of the measurements for the specified time period. "50%" means the 50/100th measured value in ascending order. This "50%" value is calculated as the 50% tile of the measurements for the specified time period. "75%" means 75/100 measurements in ascending order. This "75%" value is calculated as the 75% tile of the measurements for the specified time period. "max" is a value denoting the maximum temperature during melting. This "max" value is calculated as the maximum measured value for a specific period.

Further, "dence1" is a value that means the degree of (coarse) fluctuation (jaggedness) of the temperature based on the time-series fluctuation data D2. The value of "dence1" is calculated from the spectral density of 195-585 Hz (the sum of the spectral densities of three predetermined frequencies). "dence2" is a value that means the degree of (slightly rough) fluctuation (jaggedness) of the temperature based on the time-series fluctuation data D2. The value of this "dence2" is calculated from the spectral density of 781-1171 Hz (the sum of spectral densities of three predetermined frequencies).

"dence3" is a value that means the degree of (medium) fluctuation (jaggedness) of the temperature based on the time-series fluctuation data D2. The value of this "dence3" is calculated from the spectral density of 1367-1757 Hz (the sum of spectral densities of three predetermined frequencies). "dence4" is a value that means the degree of (slightly finer) variation (jaggedness) of the temperature based on the time-series variation data D2. The value of this "dence4" is calculated with the spectral density of 1953-2343 Hz (the sum of the spectral densities of three predetermined frequencies). "dence5" is a value that means a (fine) variation component (jaggedness) of the temperature based on the time-series variation data D2. The value of this "dence5" is calculated with spectral densities of 2539-2929 Hz (the sum of spectral densities of three predetermined frequencies).

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

According to the welding system 100B of the third embodiment, along with the teaching data and estimation data of the auxiliary variables calculated by the first and second auxiliary variable calculation units 12 and 32 based on the monitoring information 5 and 6, the first and By using the latent variable teacher data and estimation data calculated by the second latent variable calculators 13 and 33, it is possible to more precisely and accurately estimate the welding result including fluctuations in the welding state. be. Moreover, deterioration of the estimation accuracy due to the correlation between the monitoring information 5, 6 and the processing condition information 2, 4 can be more effectively avoided.

[Welding system operation]
In the welding system 100B, the welding result information 1 and the processing condition information 2 stored in, for example, an external storage device or storage medium, and the monitoring information 5 output from the welding monitor device 50 are, for example, teacher data of objective variables, explanations, and so on. They are input to the learning apparatus 10 and stored in a storage unit (not shown) as teacher data for variables and teacher data for auxiliary variables.

The learning device 10 calculates feature amount information (teaching data of auxiliary variables) in the first auxiliary variable calculating section 12 from the data including teaching data of the monitoring information 5 from the welding monitoring device 50 among these input data. do. Further, based on the feature amount information calculated by the first auxiliary variable calculation unit 12 and the teaching data of the processing condition information 2, the first latent variable calculation unit 13 calculates the teaching data of the processing condition information 2 from the feature amount information. , the correlation between is reduced or compressed to be abstractly represented (for example, the correlation is small), and latent variable calculation information 7 is output.

In addition, the learning device 10 is based on the teacher data of the welding result information 1, the teacher data of the processing condition information 2, and the latent variable teacher data calculated by the first latent variable calculator 13 in the welding result learning unit 11. Then, for example, machine learning is performed to create a welding result estimation model 3 for estimating welding results including welding strength and the like. The learning device 10 stores the latent variable calculation information 7 used for calculating latent variable teacher data and the created welding result estimation model 3 in the storage unit in the device, and the estimation device 30 of the welding result determination device 20. output available to

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) via a connected network or the like. In this case, the external storage device temporarily or permanently stores the inputted latent variable calculation information 7 and welding result estimation model 3 . Various data may be stored in a storage device (not shown) that forms part of the welding system 100B, for example, instead of the external storage device or storage medium as described above.

On the other hand, the laser welding machine 40 performs laser welding based on the set processing condition information 4 . A sensor 212 attached to the laser welder 40 acquires monitoring information 6, which is state observation information during laser welding of an estimation target (for example, a welded portion of a workpiece) by the welding monitor device 50 . The processing condition information 4 set in the laser welder 40 and the monitoring information 6 of the welding monitor device 50 can be used by the estimation device 30 of the welding result determination device 20 as explanatory variable estimation data and auxiliary variable estimation data, respectively. is entered in

The estimating device 30 stores various data input together with the latent variable calculation information 7 and the welding result estimation model 3 in a storage unit (not shown). The second auxiliary variable calculator 32 calculates feature amount information (data for estimating auxiliary variables). Further, based on the feature amount information calculated by the second auxiliary variable calculation unit 32, the estimation data of the processing condition information 4, and the latent variable calculation information 7, the second latent variable calculation unit 33 calculates Estimation data of the latent variable abstractly expressed (for example, with small correlation) is calculated by reducing or compressing the dimension of the correlation with the estimation data of the processing condition information 4 .

The estimating device 30 inputs the data for estimating the latent variables calculated by the second latent variable calculating unit 33 in the welding result estimating unit 31 and the data for estimating the processing condition information 4 to the welding result estimating model 3. Then, the welding result including the welding strength is estimated, and the welding result estimated value 8 is output. The welding result estimated value 8 is stored in the storage unit and output from the estimating device 30 in a form that can be used as appropriate, such as display or printing, and can be used by the welding result determination device 20 .

In addition, when calculating the feature amount information, calculating the latent variables, and creating the welding result estimation model 3, the learning device 10 uses the data for estimation of the processing condition information 4 set in the laser welder 40 and the data from the welding monitor device 50. The data for estimation of the monitoring information 6 is added to the teaching data of the processing condition information 2 and the teaching data of the monitoring information 5 as teaching data of the explanatory variables and auxiliary variables, or is referred to and used as teaching data of the explanatory variables and auxiliary variables. may In this way, more data can be used to contribute to improvement in estimation accuracy.

In the welding system 100B of the third embodiment, after calculating feature amount information and latent variables based on the monitoring information 5 as teaching data of the auxiliary variables of the welding result information 1, the welding result estimation model 3 using these information is created. In addition, after calculating feature amount information and latent variables based on monitoring information 6 as data for estimating auxiliary variables using latent variable calculation information 7, this information and processing condition information as data for estimating explanatory variables are calculated. 4 is input to the welding result estimation model 3, a welding result estimation value 8 including welding strength and the like is output.

Therefore, according to the welding system 100B of the third embodiment, the welding result of the estimation target (welded portion of the workpiece) can be non-destructively estimated from the data regarding the laser welding processing conditions and the data monitored during welding. Therefore, it is possible to estimate the welding strength included in the welding result, and to estimate the spot diameter based on the combination of the lens, fiber, and focal position.

Generally, in laser welding, even if laser welding is performed under the same processing conditions, the weld strength varies depending on factors such as the material of the workpiece and variations in the welding environment. There is a problem that the weld strength cannot be measured without Moreover, there is also a problem that various processing conditions for obtaining the desired welding result cannot be known unless an experiment in which the laser beam is actually irradiated is performed.

However, the welding system 100B of the third embodiment has a configuration that can improve the estimation accuracy using the feature amount information from the monitoring information as an auxiliary variable and avoid deterioration of the estimation accuracy due to correlation. Such a problem can be solved by adopting a configuration capable of estimating a physical phenomenon including a change in the welding state as an event that occurred during welding. Along with this, it is possible not only to obtain the same effects as those of the first and second embodiments, but also to enable more precise estimation, improve estimation accuracy, and avoid deterioration in accuracy due to correlation.

The welding system 100B of the third embodiment aims at estimating the welding result including the welding strength of laser welding, but is not limited to this. It is also possible to estimate the welding result including the quality of welding based on the results. In this case, the objective variable in the learning device 10 may be the welding quality of laser welding.

The data on welding quality includes at least one of data indicating good welding and data indicating poor welding. A welding result estimation model 3 is created by learning. By using the welding result estimation model 3 and performing estimation based on the estimation data of the processing condition information 4 and the monitoring information 6 by the estimation device 30, the welding result including the quality of the welding can be estimated even during laser welding. It can be carried out.

Also, in the welding system 100B, the learning device 10 and the estimating device 30 can be separately installed, for example, at the same site as the laser welding machine 40, or at a remote location. For this reason, it is possible to perform a wide variety of operations for estimating the welding strength, etc. of laser welding, assuming global deployment.

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

That is, the welding monitor device 50A is a first storage unit ( not shown), processing condition information 2 including processing conditions for welding performed in advance by a welding device (laser welder 40), and welding result information related to the welding result obtained after welding based on the processing condition information 2. 1 and monitoring information 5 output from a first storage unit (not shown), a second storage unit (not shown) for storing a welding result estimation model 3; Processing condition information 4 including processing conditions set in the welding machine 40) and monitoring information 6 stored in the first storage unit (not shown) are stored in the second storage unit (not shown). and a welding result estimating unit 31 (estimating device 30) that inputs the welding result estimation model 3 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 the welding result estimation model 3 acquired from the learning device 10 to the internal estimation device 30 . In the welding monitor device 50A, the estimating device 30 inside the welding monitor device 50A stores the monitoring information 5 and 6 acquired by the sensor 212 and stored in the first storage unit, and the processing condition information 4 set in the laser welder 40. is input to the welding result estimation model 3 stored in the second storage unit to estimate the welding result.

As a result, the welding result including the welding strength can be estimated by the welding monitor device 50A alone, and the welding result estimated value 8 can be output. According to the welding system 100C of the fourth embodiment, it is possible to achieve the same effects as those of the third embodiment, and to realize the function of the estimation device 30 by the welding monitor device 50A alone.

[Fifth embodiment]
FIG. 7 is an explanatory diagram showing the basic configuration of the welding system according to the fifth embodiment of the present invention. As shown in FIG. 7, a welding system 100D according to the fifth embodiment uses monitoring information 5 and 6 from a welding monitor device (resistance welding checker) 51 as auxiliary variables ( The welding system 100B of the third embodiment is similar to the welding system 100B of the third embodiment in that it is used as teacher data and estimation target data), but is different from the welding system 100B of 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 welder (resistance welding power source/head 41). The processing condition information 2 and 4 includes welding current, welding pressure, energization time, welding method, material of workpiece, plate thickness of workpiece, surface treatment of workpiece, combination condition of workpiece, It includes at least one machining condition of sampling period, displacement detection resolution, and nugget diameter. The monitoring information 5 and 6 includes time-series variation data of welding current/voltage and time-series variation data of welding point pressure/displacement amount.

Also in the welding system 100D of the fifth embodiment, the first auxiliary variable calculator 12 of the learning device 10 calculates feature amount information (teaching data of auxiliary variables) based on the monitoring information 5, and the first latent variable calculator 13 calculates latent variable teacher data based on the feature amount information, outputs latent variable calculation information 7, and creates a welding result estimation model 3 in a welding result learning unit 11. This welding result estimation model 3 and the latent variable The calculation information 7 is given to the estimation device 30 of the welding result determination 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 amount information (data for estimating auxiliary variables) based on the monitoring information 6, and the second latent A variable calculation unit 33 calculates latent variable estimation data based on the feature amount information, and a welding result estimation unit 31 inputs the data including the latent variable estimation data to the welding result estimation model 3 to estimate the welding result. make an estimate. Based on the obtained estimated welding result 8, the welding strength can be estimated, the welding result can be determined, and the estimated welding result 8 can be output to the outside. Thus, 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 can be embodied in various other forms, and various omissions, replacements, and modifications can be made without departing from the scope of the invention. These embodiments and their modifications are included in the scope and gist of the invention, and are included in the scope of the invention described in the claims and equivalents thereof.

1 Welding result information (teaching data for target 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 (supervised data for auxiliary variables)
6 Monitoring information (data for estimating auxiliary variables)
7 Latent variable calculation information 8 Welding result estimated 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 welder 41 Resistance welding power source/head 50, 50A Welding monitor device 51 Resistance welding checker 100 Welding system

Claims (10)

a welding device for welding workpieces;
Processing condition information including processing conditions for welding performed in advance by the welding device and welding result information related to welding results obtained after welding based on the processing conditions are input as teacher data, and based on these teacher data , a learning device that creates and outputs a welding result estimation model;
Processing condition information including processing conditions set in the welding device is input as estimation data into the welding result estimation model created by the learning device, and welding is performed by the welding device based on the processing conditions. and an estimating device for estimating and outputting a result. A welding monitor device that outputs monitoring information obtained by monitoring the progress of welding by the welding device with a sensor,
The learning device calculates feature amount information as teacher data of auxiliary variables from the teacher data of the monitoring information output from the welding monitor device, and creates the welding result estimation model based on the teacher data to which the auxiliary variables are added. and output
The estimating device calculates feature amount information as data for estimating auxiliary variables from the data for estimating the monitoring information output from the welding monitor device, inputs the data for estimating the auxiliary variables, and inputs the data for estimating the welding. The welding system of claim 1, wherein the results are estimated and output. The learning device
a first auxiliary variable calculation unit that calculates feature amount information as teacher data of auxiliary variables from the teacher data of the monitoring information;
creating and outputting the welding result estimation model based on the teaching data of the welding result information, the teaching data of the processing condition information, and the teaching data of the auxiliary variables calculated by the first auxiliary variable calculating section; a welding result learning unit;
The estimation device is
a second auxiliary variable calculation unit that calculates the feature amount information from the monitoring information estimation data as auxiliary variable estimation data;
Welding result for estimating and outputting the welding result by inputting the data for estimating the auxiliary variables calculated by the second auxiliary variable calculating unit and the data for estimating the machining condition information into the welding result estimating model. and an estimating portion. Based on the training data of the processing condition information and the training data of the auxiliary variables, the learning device reduces the dimension or dimension of the correlation between the training data of the auxiliary variables and the training data of the processing condition information. including a first latent variable calculation unit that calculates latent variable training data that can be compressed and abstractly expressed and outputs latent variable calculation information;
The welding result learning unit creates and outputs the welding result estimation model based on the teaching data of the processing condition information, the teaching data of the welding result information, and the teaching data of the latent variables,
The estimating device converts the data for estimating the machining condition information from the data for estimating the auxiliary variable to the data for estimating the machining condition information based on the data for estimating the machining condition information, the data for estimating the auxiliary variable, and the latent variable calculation information. a second latent variable calculation unit that calculates data for estimating a latent variable that can be represented abstractly by dimension reduction or dimension compression of the correlation between
4. The welding result estimation unit according to claim 3, wherein the welding result estimation unit inputs the data for estimating the processing condition information and the data for estimating the latent variable into the welding result estimation model to estimate and output the welding result. Welding system. The first auxiliary variable calculation unit detects a plurality of predetermined features from the teacher data of the monitoring information, which is time-series data, and calculates the value of each detected feature as the feature amount information,
The second auxiliary variable calculation unit detects a plurality of predetermined features from the data for estimating the monitoring information, which is time-series data, and calculates the value of each detected feature as the feature amount information. Welding system according to claim 3 or 4. The first latent variable calculation unit calculates, as latent information teacher data, one-dimensional or two-dimensional information obtained by dimension reduction or dimension compression of the teacher data of the processing condition information and the teacher data of the auxiliary variables,
The second latent variable calculation unit calculates one-dimensional or two-dimensional information obtained by dimension reduction or dimension compression of the processing condition information estimation data and the auxiliary variable estimation data as latent information estimation data. Welding system according to claim 4. The welding device is a laser welder,
The processing condition information includes 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 the peak power, and the time from the peak power to the end of laser irradiation. , fiber diameter, lens focal length, focal position, and laser diameter of the irradiation point. The welding device is a laser welder,
The processing condition information includes 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 the peak power, and the time from the peak power to the end of laser irradiation. , including at least one processing condition of fiber diameter, lens focal length, focal position, and laser diameter of the irradiation point,
The welding system according to any one of claims 2 to 6, wherein the monitoring information includes time-series variation data of laser power and time-series variation data of radiant near-infrared light intensity at the irradiation point. The welding device is a resistance welder,
The processing condition information includes welding current, welding pressure, energization time, welding method, material of the workpiece, plate thickness of the workpiece, surface treatment of the workpiece, combination conditions of the workpiece, and sampling. The welding system according to any one of claims 1 to 6, including at least one processing condition of period, displacement detection resolution, and nugget diameter. a first storage unit that stores monitoring information obtained by monitoring progress of welding by a welding device that welds workpieces with a sensor;
Processing condition information including processing conditions for welding performed in advance by the welding device, welding result information related to welding results obtained after welding based on the processing condition information, and output from the first storage unit a second storage unit that stores a welding result estimation model created based on the monitoring information;
inputting the processing condition information including the processing conditions set in the welding device and the monitoring information stored in the first storage unit into the welding result estimation model stored in the second storage unit; A welding monitor device, comprising: a welding result estimation unit that estimates and outputs a welding result of welding performed by the welding device.

Images