Welding support device, welding support method, and program

Abstract

To provide a welding support device that can stabilize a welding quality and suppress occurrence of defect, in a welding work that is executed by a welder.SOLUTION: A welding support device is provided with: an obtaining unit that obtains welding data constituted of a plurality of parameters showing a welding state; a determining unit that calculates an abnormality degree of the welding data, on the basis of learning model constructed by learning normal data constituted of welding data collected in a period of time during which the welding state is normal and the welding data obtained by the obtaining unit; an extracting unit that extracts, as an optimal condition range, portions of the normal data in the learning model, on the basis of the welding data; a setting unit that sets a recommended range of a value of at least one parameter included in the welding data, on the basis of the optimal condition range; and an output unit that outputs support information including the recommended range.SELECTED DRAWING: Figure 1

Description

Technical Field

[0001] The present disclosure relates to a welding support device, a welding support method, and a program.

Background Art

[0002] As a technique for improving welding quality, for example, in Patent Document 1, physical quantities related to arc welding such as the appearance of a welding bead and arc welding conditions such as welding speed and protrusion length are machine-learned, and based on the physical quantities obtained from imaging data, a technique for adjusting arc welding conditions to perform automatic welding is described.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, since the technique described in Patent Document 1 targets an automatic welding robot, it is difficult to apply it to welding work performed by a welder. Therefore, there is a need for a method of assisting the welding work performed by a welder to stabilize welding quality and suppress the occurrence of defects.

[0005] The present disclosure has been made in view of such problems, and provides a welding support device, a welding support method, and a program capable of stabilizing welding quality and suppressing the occurrence of defects in welding work performed by a welder.

Means for Solving the Problems

[0006] According to one aspect of the present disclosure, a welding support device includes: an acquisition unit that acquires welding data consisting of a plurality of parameters representing the welding state; a learning model constructed by learning normal data consisting of welding data collected during a period in which the welding state is normal; a determination unit that calculates the degree of abnormality of the welding data based on the welding data acquired by the acquisition unit; an extraction unit that extracts a portion of the normal data of the learning model as an optimal condition range based on the welding data; a setting unit that sets a recommended range for the value of at least one parameter included in the welding data based on the optimal condition range; and an output unit that outputs support information including the recommended range.

[0007] According to one aspect of the present disclosure, a welding support method includes the steps of: acquiring welding data consisting of a plurality of parameters representing a welding state; calculating the degree of abnormality of the welding data based on a learning model constructed by learning normal data consisting of welding data collected during a period in which the welding state is normal, and the acquired welding data; extracting a portion of the normal data of the learning model as an optimal condition range based on the welding data; setting a recommended range for the value of at least one parameter included in the welding data based on the optimal condition range; and outputting support information including the recommended range.

[0008] According to one aspect of the present disclosure, the program causes a welding support device to perform the following steps: acquire welding data consisting of a plurality of parameters representing a welding state; calculate the degree of abnormality of the welding data based on a learning model constructed by learning normal data consisting of welding data collected during a period in which the welding state is normal and the acquired welding data; extract a portion of the normal data of the learning model as an optimal condition range based on the welding data; set a recommended range for the value of at least one parameter included in the welding data based on the optimal condition range; and output support information including the recommended range. [Effects of the Invention]

[0009] According to the welding support device, welding support method, and program described herein, welding quality can be stabilized and defects can be suppressed in welding work performed by a welder. [Brief explanation of the drawing]

[0010] [Figure 1] This block diagram shows the functional configuration of a welding support device according to one embodiment of the present disclosure. [Figure 2] This is a first flowchart showing an example of processing by a welding support device according to one embodiment of the present disclosure. [Figure 3] This is a first figure illustrating the function of a welding support device according to one embodiment of the present disclosure. [Figure 4] This is a second figure illustrating the function of a welding support device according to one embodiment of the present disclosure. [Figure 5] This is a third figure illustrating the function of a welding support device according to one embodiment of the present disclosure. [Figure 6] This is a second flowchart showing an example of processing by a welding support device according to one embodiment of the present disclosure. [Modes for carrying out the invention]

[0011] Hereinafter, a welding support system 1 and a welding support device 10 according to one embodiment of this disclosure will be described with reference to Figures 1 to 6.

[0012] (Overall configuration of the welding support system) Figure 1 is a block diagram showing the functional configuration of a welding support device according to one embodiment of the present disclosure. As shown in Figure 1, the welding support system 1 according to this embodiment comprises a welding support device 10, a welding device 20, a data logger 30, and a display device 40.

[0013] The welding device 20 includes devices such as switches and levers for adjusting various parameters that affect the welding state, and an operation panel provided with instruments. The welder performs the welding operation while adjusting each parameter on the operation panel of the welding device. These parameters are, for example, welding current, welding voltage, electrode protrusion length, etc. The welding state is, for example, the degree of abnormality estimated based on welding data.

[0014] The data logger 30 acquires welding data consisting of a plurality of parameters measured by sensors (not shown) provided in each part such as the welding device 20 and near the welding site.

[0015] The welding support device 10 monitors the welding state based on the welding data acquired by the data logger 30, displays the current welding state and the recommended range of the values of each parameter on the display device 40, and supports the welder's welding operation.

[0016] (Functional Configuration of Welding Support Device) The welding support device 10 includes a processor 11, a memory 12, a storage 13, and a communication interface 14.

[0017] The processor 11 functions as an acquisition unit 110, a determination unit 111, an extraction unit 112, a setting unit 113, an output unit 114, and a learning unit 115 by operating according to a predetermined program.

[0018] The acquisition unit 110 acquires welding data from the data logger 30.

[0019] The determination unit 111 calculates the degree of abnormality of the welding data based on the learning model M constructed by learning the normal data consisting of the welding data collected during the period when the welding state is normal, and the welding data acquired by the acquisition unit 110.

[0020] The extraction unit 112 extracts a part of the normal data of the learning model M as the optimal condition range based on the welding data acquired by the acquisition unit 110.

[0021] The setting unit 113 sets a recommended range for the value of at least one parameter included in the welding data, based on the optimal condition range extracted by the extraction unit 112.

[0022] The output unit 114 outputs support information, including the recommended range set by the setting unit 113. The support information is displayed on the display device 40.

[0023] The learning unit 115 learns and updates the learning model M based on the welding data collected from the data logger 30.

[0024] Memory 12 has a memory area necessary for the operation of the processor 11.

[0025] Storage 13 is a so-called auxiliary storage device, such as an HDD (Hard Disk Drive) or SSD (Solid State Drive). Welding data collected from the data logger 30 and the trained learning model M are recorded in storage 13.

[0026] The communication interface 14 is an interface for sending and receiving various types of information (signals) with external devices (such as the data logger 30 and the display device 40).

[0027] (Regarding welding support processing using welding support equipment) Figure 2 is a first flowchart showing an example of processing by a welding support device according to one embodiment of the present disclosure. Here, with reference to Figure 2, we will explain the flow of welding support processing by the welding support device 10.

[0028] First, the acquisition unit 110 acquires welding data X from the data logger 30 (step S10).

[0029] The determination unit 111 calculates the degree of abnormality of the welding data X based on the acquired welding data X and the trained learning model M, and determines whether the welding condition is normal or abnormal (step S11).

[0030] Figure 3 is a first diagram illustrating the function of a welding support device according to one embodiment of the present disclosure. The learning model M shown in Figure 3 is an evaluation model that performs unsupervised learning on welding data (normal data P) collected during past periods when the welding state was normal, and defines the normal range (normal space) of the welding data. For example, if the normal data P consists of two parameters, the normal data P is represented as a point in a two-dimensional space, as in the example in Figure 3. In reality, the normal data P consists of a larger number of parameters, n, and the learning model M is an evaluation model that defines the normal space in an n-dimensional space. It is assumed that the learning model M has already been constructed by the learning unit 115 before the welding support processing is performed.

[0031] The determination unit 111 selects the k1th (e.g., the 10th) closest normal data Pk1 from the normal data P included in the learning model M, and calculates the distance D between this k1th normal data Pk1 and the welding data X. The distance D represents the degree of abnormality of the welding data X. The distance D may be, for example, the Euclidean distance obtained by summing the differences between each parameter of the welding data X and the normal data Pk1, or it may be the Mahalanobis distance, Manhattan distance, etc. The determination unit 111 determines that the welding state is abnormal (indicating a defect) if the distance D exceeds a predetermined threshold. On the other hand, the determination unit 111 determines that the welding state is normal if the distance D is less than the threshold.

[0032] Furthermore, the extraction unit 112 extracts the optimal condition range based on the learning model M and welding data X (step S12).

[0033] Figure 4 is a second diagram illustrating the function of a welding support device according to one embodiment of the present disclosure. The learning model M shown in Figure 4 is the same as the learning model M in Figure 3. The optimal condition range R is a part of the normal space composed of normal data P, and defines a range in which welding quality is more stable and defects are less likely to occur. Specifically, the extraction unit 112 extracts the range containing the k2th to k3rd (for example, the 9000th to 10000th) normal data P from the current welding data X as the optimal condition range R. In other words, each time welding data X is acquired, the extraction unit 112 extracts a different optimal condition range R corresponding to the acquired welding data X.

[0034] The extraction unit 112 extracts normal data Pk2 to Pk3, which are further from the welding data X than the normal data Pk1 used to calculate the abnormality, as the optimal condition range R. In other words, the values ​​of k2 and k3 are set to be greater than the values ​​of k1. As a result, the extraction unit 112 can extract as the optimal condition range R a range within the normal space composed of normal data P where the density of normal data P is higher (welding quality is stable) than the area around normal data Pk1 used to calculate the abnormality. Furthermore, the values ​​of k1, k2, and k3 may be arbitrarily changed considering the simulation results or the welding quality when welding work is actually performed using the welding support device 10.

[0035] Next, the setting unit 113 sets a recommended range for the parameter values ​​included in the welding data X based on the optimal condition range R extracted by the extraction unit 112 (step S13). Specifically, the setting unit 113 sets the range from the minimum value to the maximum value of each parameter included in the normal data Pk2 to Pk3 within the optimal condition range R as the recommended range for the value of each parameter.

[0036] Figure 5 is a third diagram illustrating the function of a welding support device according to one embodiment of the present disclosure. As shown in the example in Figure 5, the welding data X includes parameters such as welding current [A], welding voltage [V], and overhang length [mm], which can be adjusted by the welder via the control panel of the welding apparatus 20. Furthermore, if the minimum welding current value included in the normal data Pk2 to Pk3 within the optimal condition range R is 575 [A] and the maximum value is 634 [A], the setting unit 113 sets the recommended range for the welding current value to 575 to 634 [A]. The setting unit 113 similarly sets recommended ranges for other parameters.

[0037] Next, the output unit 114 generates support information 41 (Figure 5) including the recommended range for each parameter set by the setting unit 113 and outputs (displays) it to the display device 40 (step S14).

[0038] As shown in Figure 5, the support information 41 includes the recommended range 412 for each parameter, the measured value 413, and the judgment result 414. The output unit 114 sets the judgment result 414 to "OK (no problem)" if the measured value 413 for each parameter is within the recommended range 412. The output unit 114 sets the judgment result 414 to "NG (problem)" if the measured value 413 exceeds the recommended range 412. The output unit 114 may also add an improvement suggestion display 415 that highlights this parameter with a frame, background color, or text color to encourage improvement if the judgment result 414 is "NG". Furthermore, the output unit 114 may also include the abnormality level 411 calculated by the judgment unit 111 and the welding condition judgment result in the support information 41.

[0039] The welder refers to the support information 41 displayed on the display device 40 and operates the welding apparatus 20 to change or maintain welding conditions so that the measured values ​​413 of each parameter fall within the recommended range 412. Furthermore, if the measured values ​​413 of each parameter exceed the recommended range 412, the welder promptly changes the welding conditions to bring them back within the recommended range 412. This stabilizes welding quality and suppresses the occurrence of defects, regardless of the welder's skill level.

[0040] (Regarding the process of updating the learning model using welding support equipment) Figure 6 is a second flowchart showing an example of processing by a welding support device according to one embodiment of the present disclosure. As shown in Figure 6, the learning unit 115 may update (retrain) the learning model M after constructing it, based on the welding data collected while performing welding support processing. Here, the flow of the learning model M update process by the welding support device 10 will be explained with reference to Figure 6.

[0041] The learning unit 115 determines whether to use the welding data X acquired in step S10 of Figure 2 and the anomaly score calculated in step S11 for learning. Specifically, the learning unit 115 determines whether the anomaly score of the welding data X is less than a predetermined threshold (step S20).

[0042] If the abnormality level of the welding data X is above the threshold (step S20; NO), the welding condition is not normal, and therefore this welding data X cannot be used for learning. For this reason, the learning unit 115 terminates processing.

[0043] On the other hand, if the abnormality level of the welding data X is below the threshold (step S20; YES), the welding condition is normal, so the learning unit 115 adds this welding data X as new normal data P (step S21). The normal data P is recorded and stored in the storage 13.

[0044] Next, the learning unit 115 determines whether it is time to update the learning model M (step S22). For example, the learning unit 115 determines that it is time to update if the newly added normal data P exceeds a predetermined amount, or if a predetermined time has elapsed since the last update of the learning model M (step S22; YES). In this case, the learning unit 115 retrains and updates the learning model M based on both the previously accumulated normal data P and the newly added normal data P (step S23). The learning unit 115 records the updated learning model M in the storage 13 and terminates the process.

[0045] Furthermore, if it is not time to update (step S22; NO), the learning unit 115 terminates the process. Each time new welding data X is acquired, the learning unit 115 executes the series of processes shown in Figure 6 to add new normal data P and update the learning model M.

[0046] (Effect, Action) As described above, the welding support device 10 according to this embodiment includes an acquisition unit 110 that acquires welding data X, an extraction unit 112 that extracts an optimal condition range R corresponding to the welding data X from normal data P included in a learning model M used for calculating the degree of abnormality, a setting unit 113 that sets a recommended range 412 of parameter values ​​included in the welding data based on the optimal condition range R, and an output unit 114 that outputs support information 41 including the recommended range 412.

[0047] In this way, the welding support device 10 can suggest appropriate values ​​for each parameter to the welder to stabilize the welding quality. As a result, the welding support device 10 can stabilize the welding quality and suppress the occurrence of defects regardless of the welder's skill level. Furthermore, by using the same learning model M for calculating the degree of abnormality and extracting the optimal condition range R, the welding support device 10 can shorten the processing time in the judgment unit 111 and the extraction unit 112, and provide rapid feedback of the recommended range to the welder.

[0048] Furthermore, the determination unit 111 selects the k1th normal data Pk1 from the normal data P included in the learning model M, and calculates the abnormality score as the distance between the selected normal data Pk1 and the welding data X. The extraction unit 112 extracts the k2nd to k3rd normal data Pk2 to Pk3, which are further from the welding data X than normal data Pk1, as the optimal condition range R.

[0049] In this way, the welding support device 10 can suppress the calculation of an abnormality score that would be judged as excessively abnormal by using normal data Pk1 that is relatively close to the welding data X when calculating the abnormality score. On the other hand, when extracting the optimal condition range R, the welding support device 10 can appropriately extract a range of normal data P where the density of normal data P is high (representing a more stable welding state). As a result, the welding support device 10 can appropriately set the recommended values ​​for each parameter based on the optimal condition range R, thereby further stabilizing the welding quality.

[0050] Furthermore, the setting unit 113 sets the range from the minimum value to the maximum value of each parameter in the optimal condition range R as the recommended range.

[0051] Conventional technologies required highly skilled welders to visually inspect the welding area and determine the appropriate parameter adjustments themselves. In contrast, the welding support device 10 according to this embodiment clearly proposes a recommended range consisting of the minimum and maximum values ​​of each parameter, enabling welders to adjust parameters appropriately regardless of their skill level. This suppresses variations in welding quality caused by different welders.

[0052] Furthermore, the output unit 114 outputs support information 41, including a suggested improvement display 415 for the parameter, if the parameter value is not within the recommended range.

[0053] In this way, the welding support device 10 allows the welder to easily and quickly recognize whether or not parameter improvements are necessary. This makes it possible to prevent the occurrence of defects.

[0054] Furthermore, the learning unit 115 updates the learning model M by adding the welding data X acquired when the abnormality level is below a predetermined value (i.e., the welding condition is normal) to the normal data P.

[0055] In this way, the welding support device 10 can learn the characteristics of the welding device 20 and the effects of aging deterioration by accumulating a large amount of normal data P while performing welding work and updating the learning model M, making it possible to propose parameters suitable for the welding device 20.

[0056] In the above-described embodiment, the processes of various operations of the welding support device 10 are stored in program form on a computer-readable recording medium, and the various operations are performed by a computer reading and executing this program. The computer-readable recording medium refers to magnetic disks, magneto-optical disks, CD-ROMs, DVD-ROMs, semiconductor memory, etc. Alternatively, this computer program may be distributed to a computer via a communication line, and the computer that receives the distribution may execute the program.

[0057] The above program may be intended to implement some of the functions described above. Furthermore, it may be a so-called differential file (differential program) that can implement the above functions in combination with a program already recorded in the computer system.

[0058] As described above, several embodiments relating to this disclosure have been explained, but all of these embodiments are presented as examples and are not intended to limit the scope of the invention. These embodiments can be carried out in various other forms, and various omissions, substitutions, and modifications can be made without departing from the spirit of the invention. These embodiments and their variations are included in the scope and spirit of the invention, as well as in the claims and their equivalents.

[0059] <Note> The welding support apparatus, welding support method, and program described in the above-described embodiment can be understood, for example, as follows.

[0060] (1) According to a first aspect of the present disclosure, the welding support device 10 includes an acquisition unit 110 that acquires welding data X consisting of a plurality of parameters representing the welding state, a learning model M constructed by learning normal data P consisting of welding data collected during a period in which the welding state is normal, and a determination unit 111 that calculates the degree of abnormality of the welding data X based on the welding data X acquired by the acquisition unit 110, an extraction unit 112 that extracts a portion of the normal data P of the learning model M as an optimal condition range R based on the welding data X, a setting unit 113 that sets a recommended range for the value of at least one parameter included in the welding data X based on the optimal condition range R, and an output unit 114 that outputs support information including the recommended range.

[0061] In this way, the welding support device 10 can suggest appropriate values ​​for each parameter to the welder to stabilize the welding quality. As a result, the welding support device 10 can stabilize the welding quality and suppress the occurrence of defects regardless of the welder's skill level. Furthermore, by using the same learning model M for calculating the degree of abnormality and extracting the optimal condition range R, the welding support device 10 can shorten the processing time in the judgment unit 111 and the extraction unit 112, and provide rapid feedback of the recommended range to the welder.

[0062] (2) According to a second aspect of the present disclosure, in the welding support device 10 according to the first aspect, the determination unit 111 selects the k1th normal data Pk1 from the normal data P of the learning model M that is closest to the welding data X, calculates the distance between the selected normal data Pk1 and the welding data X as the degree of abnormality of the welding data X, and the extraction unit 112 extracts the k2nd to k3rd normal data Pk3, which are further from the welding data X than the normal data Pk1 selected by the determination unit 111, as the optimal condition range R.

[0063] In this way, the welding support device 10 can suppress the calculation of an abnormality score that would be judged as excessively abnormal by using normal data Pk1 that is relatively close to the welding data X when calculating the abnormality score. On the other hand, when extracting the optimal condition range R, the welding support device 10 can appropriately extract a range of normal data P where the density of normal data P is high (representing a more stable welding state). As a result, the welding support device 10 can appropriately set the recommended values ​​for each parameter based on the optimal condition range R, thereby further stabilizing the welding quality.

[0064] (3) According to a third aspect of the present disclosure, in the welding support device 10 according to the first or second aspect, the setting unit 113 sets the range from the minimum value to the maximum value of the parameter in the optimal condition range as the recommended range.

[0065] In this way, the welding support device 10 clearly proposes a recommended range consisting of the minimum and maximum values ​​of each parameter, enabling welders to adjust parameters appropriately regardless of their skill level. This helps to suppress variations in welding quality caused by different welders.

[0066] (4) According to a fourth aspect of the present disclosure, in a welding support device 10 according to any one of the first to third aspects, the output unit 114 outputs support information including suggestions for improving the parameter when the parameter value is not within the recommended range.

[0067] In this way, the welding support device 10 allows the welder to easily and quickly recognize whether or not parameter improvements are necessary. This makes it possible to prevent the occurrence of defects.

[0068] (5) According to a fifth aspect of the present disclosure, the welding support device 10 according to any one of the first to fourth aspects further comprises a learning unit 115 that updates the learning model M by adding welding data X to normal data P when the degree of abnormality is less than a predetermined threshold.

[0069] In this way, the welding support device 10 can learn the characteristics of the welding device 20 and the effects of aging deterioration by accumulating a large amount of normal data P while performing welding work and updating the learning model M, making it possible to propose parameters suitable for the welding device 20.

[0070] (6) According to a sixth aspect of the present disclosure, a welding support method includes the steps of: acquiring welding data X consisting of a plurality of parameters representing a welding state; calculating the degree of abnormality of the welding data X based on a learning model M constructed by learning normal data P consisting of welding data collected during a period in which the welding state is normal, and the acquired welding data X; extracting a portion of the normal data P of the learning model M as an optimal condition range R based on the welding data X; setting a recommended range for the value of at least one parameter included in the welding data X based on the optimal condition range R; and outputting support information including the recommended range.

[0071] (7) According to a seventh aspect of the present disclosure, the program causes the welding support device 10 to perform the following steps: acquire welding data X consisting of a plurality of parameters representing the welding state; calculate the degree of abnormality of the welding data X based on a learning model M constructed by learning normal data P consisting of welding data collected during a period in which the welding state is normal, and the acquired welding data X; extract a portion of the normal data P of the learning model M as an optimal condition range R based on the welding data X; set a recommended range for the value of at least one parameter included in the welding data X based on the optimal condition range R; and output support information including the recommended range. [Explanation of Symbols]

[0072] 1. Welding support system 10 Welding support device 11 processors 110 Acquisition Department 111 Judgment section 112 Extraction part 113 Settings Section Output section of 114 115 Learning Department 12 memory 13 Storage 14. Communication Interface 20 Welding equipment 30 Data Loggers 40 Display device

Claims

1. An acquisition unit that acquires welding data consisting of a plurality of parameters representing the welding state, A learning model constructed by learning normal data consisting of welding data collected during a period when the welding condition is normal, and a determination unit that calculates the degree of abnormality of the welding data based on the welding data acquired by the acquisition unit, An extraction unit that extracts a portion of the normal data of the learning model as an optimal condition range based on the welding data, A setting unit that sets a recommended range for the value of at least one parameter included in the welding data based on the aforementioned optimal condition range, An output unit that outputs support information including the aforementioned recommended range, Equipped with, The determination unit selects the k1-th normal data from the normal data of the learning model that is closest to the welding data, and calculates the distance between the selected normal data and the welding data as the degree of abnormality of the welding data. The extraction unit extracts the k2nd to k3rd normal data points, which are further from the welding data than the normal data selected by the determination unit, as the optimal condition range. Welding support device.

2. The setting unit sets the range from the minimum value to the maximum value of the parameter within the optimal condition range as the recommended range. The welding support device according to claim 1.

3. The output unit outputs the support information, including suggestions for improving the parameter, if the value of the parameter does not fall within the recommended range. The welding support device according to claim 1 or 2.

4. The system further includes a learning unit that updates the learning model by adding the welding data to the normal data when the degree of abnormality is below a predetermined threshold. The welding support device according to claim 1 or 2.

5. A step of acquiring welding data consisting of multiple parameters representing the welding state, A step of calculating the degree of abnormality of the welding data based on a learning model constructed by learning normal data consisting of welding data collected during a period when the welding condition was normal, and the acquired welding data. Based on the welding data, the steps include: extracting a portion of the normal data of the learning model as the optimal condition range; The steps include setting a recommended range for the value of at least one parameter included in the welding data based on the aforementioned optimal condition range, The steps include outputting support information including the aforementioned recommended range, It has, The step of calculating the abnormality of the welding data involves selecting the k1-th normal data from the normal data of the learning model that is closest to the welding data, and calculating the distance between the selected normal data and the welding data as the abnormality of the welding data. The step of extracting the optimal condition range involves extracting the k2nd to k3rd normal data points, which are further from the welding data than the normal data points selected in the step of calculating the abnormality of the welding data, as the optimal condition range. Welding support method.

6. A step of acquiring welding data consisting of multiple parameters representing the welding state, A step of calculating the degree of abnormality of the welding data based on a learning model constructed by learning normal data consisting of welding data collected during a period when the welding condition was normal, and the acquired welding data. Based on the welding data, the steps include: extracting a portion of the normal data of the learning model as the optimal condition range; The steps include setting a recommended range for the value of at least one parameter included in the welding data based on the aforementioned optimal condition range, The steps include outputting support information including the aforementioned recommended range, A program that causes a welding support device to execute, The step of calculating the abnormality of the welding data involves selecting the k1-th normal data from the normal data of the learning model that is closest to the welding data, and calculating the distance between the selected normal data and the welding data as the abnormality of the welding data. The step of extracting the optimal condition range involves extracting the k2nd to k3rd normal data points, which are further from the welding data than the normal data points selected in the step of calculating the abnormality of the welding data, as the optimal condition range. program.

Images