KR20200007960A - Arc end adjusting device, welding system, arc end adjusting method and computer program - Google Patents

Abstract

In the arc end adjustment apparatus 5 which adjusts the arc end procedure in the welding process repeated, WHEREIN: The acquisition part which acquires the welding data which shows the welding state which concerns on a next welding process, and the welding data acquired by the said acquisition part. Based on this, there is a procedure adjustment unit for adjusting the arc end procedure so that the cycle time of the welding process is shortened.

Description

Arc end adjusting device, welding system, arc end adjusting method and computer program

The present invention relates to an arc end adjusting device, a welding system, an arc end adjusting method and a computer program for adjusting an arc end procedure in a welding process.

As one of the welding methods, there is a gas shield arc welding method of a consumption electrode type. The gas shield arc welding method is a method of generating an arc between a welding wire supplied to a welded portion of a base material and a base material, and welding the base material by the heat of the arc. In order to prevent oxidation of a base material, welding is performed, spraying shield gas around the weld part.

In the continuous production, in the case of arc welding using a welding robot, after the welding torch is moved to the welding start position, the arc start is started to start the wire slow down to supply the welding current at the timing at which the welding wire contacts the base metal. Generate an arc by the procedure. Further, at the end of welding, an anti-stick treatment is performed to supply a small current to the welding wire so that the welding wire rises for a predetermined time (for example, Patent Document 1). The reason for performing the anti stick treatment is as follows. Even after the stop signal is input to the wire feeder, the feed motor feeds the welding wire by the inertia force. Therefore, when the welding wire enters a molten pool and the molten pool cools, the tip of the welding wire sticks to the weld metal. In order to prevent this fixation, it is necessary to input a stop signal to the wire feeder and then melt the welding wire by energizing a current smaller than the welding current value to prevent the welding wire from entering the melting pool. have.

In addition, Patent Literature 2 automatically describes welding conditions by machine learning using image data obtained by imaging a welded portion, appearance data of beads obtained by processing the image data, sputtering generation amount data, and the like. A technique for setting is disclosed.

In addition, Patent Literature 3 discloses a technique for performing a good or bad determination of a welding result based on welding monitor data such as a welding current, a welding current, a wire feeding speed, and the like measured during a welding step.

[Patent Document 1] Japanese Patent Application Laid-Open No. 2011-200867 [Patent Document 2] Japanese Patent Application Publication No. 2017-30014 [Patent Document 3] Japanese Patent Application Laid-Open No. 2017-39160

By the way, since anti-stick process affects cycle time, it is good that the process time is short. However, if the time required for the anti-stick treatment is shortened, defects such as contact tip or base metal welding may occur, or the welding wire protruding from the contact tip may have an inappropriate length. Therefore, there was a problem that the time required for the arc start treatment in the next welding process is long, or the arc start fails, and the cycle time is rather long. In addition, when the time required for anti stick treatment is shortened, the weldability may be adversely affected.

In order to ensure the welding quality, it was necessary for the worker to repeatedly perform the arc end treatment by hand and set it to a condition which is considered to be optimal from the results of many trials.

An object of the present invention is to automatically adjust the time required for arc end treatment such as anti-stick treatment, de-welding treatment, and so on, and to reduce the cycle time of the welding process. A method and a computer program are provided.

In the arc end adjusting device according to the present aspect, in the arc end adjusting device for adjusting the arc end procedure in a repeated welding process, the welding data indicating the result of the anti-stick treatment and the welding state associated with the next welding process is acquired. And an acquisition adjusting section and a procedure adjusting section for adjusting the arc end procedure so that the cycle time of the welding process is shortened based on the welding data acquired by the obtaining section.

According to this aspect, an acquisition part acquires welding data, and a procedure adjustment part adjusts an arc end procedure based on the acquired welding data. Welding data is information indicating the result of the anti-stick treatment and the state of welding in the next welding process following the anti-stick treatment, and whether or not it is possible to shorten the cycle time by shortening the arc end procedure or to extend the cycle time. Includes information to help you decide whether or not you should. By using such welding data, the procedure adjustment unit can adjust the arc end procedure so that the cycle time of the welding process is shortened without deteriorating the welding result, and can shorten the cycle time of the welding process.

In addition, the arc end adjustment apparatus which concerns on this aspect performs the arc end process in one welding power supply, More specifically, the arc end process which concerns on one welding wire. The next welding step is not a welding step performed on the downstream side of the production line, but a welding step performed next by the one welding power source. Even when a plurality of welding power sources are provided in the production line, the arc end adjusting device adjusts the arc end procedure for each welding power source, that is, for each welding wire. Needless to say, an arc end adjusting device may be provided in each welding power supply, and one arc end adjusting device may be configured to adjust the arc end procedures in a plurality of welding power sources, respectively.

The arc end adjustment apparatus which concerns on this aspect is equipped with the quality determination part which determines the quality of the welding result of an anti stick process and the next welding process based on the welding data acquired by the said acquisition part, The said procedure adjustment part is the said The change in the arc end procedure is determined so that the cycle time is shortened when the pass determination unit judges positive, and when the pass determination unit judges negative, the cycle time is extended.

According to this aspect, when the result of anti-stick processing and the welding result in the next welding process are favorable, since an arc end adjustment apparatus may shorten an arc end procedure, there exists a possibility that it may shorten the cycle time of a welding process, Shorten the cycle time of the welding process. When the result of the anti stick treatment and the welding result in the next welding process are poor, the arc end adjusting device extends the cycle time of the welding process. By such an adjustment process, the cycle time of a welding process can be shortened so that a welding result may not be deteriorated extremely.

In the arc end adjusting device according to the present aspect, the procedural adjustment part shortens the cycle time of the welding process, and when the anti-stick treatment and the welding result of the next welding process change from a good state to a bad state, the cycle time is shortened. The adjustment is confirmed in the previous arc end procedure, and the determined arc end procedure is stored in the storage.

According to this aspect, the cycle time of a welding process can be shortened, and a memory | storage part memorizes the arc end procedure of which cycle time is the shortest. The shortest arc end procedure is not necessarily the logical arc time shortest arc end procedure. The shortest arc end procedure was performed by shortening the cycle time of the welding process and performing the arc end procedure before the cycle time reduction when the anti-stick treatment or the welding result of the next welding process changed from a good state to a bad state. it means.

The cycle time can then be immediately shortened using the arc end procedure stored in the memory.

In the arc end adjusting device according to the present aspect, in the arc end adjustment device, when the welding data indicating a welding state related to the next welding process is input, the quality of the welding result related to the welding process when the welding data is obtained is determined. A neural network is trained to train the neural network to output data representing

According to this aspect, the acceptance determination neural network is, for example, a deep neural network that has been learned, and can appropriately determine whether the welding result is good or bad. The type of neural network is not particularly limited. It may be appropriately selected according to the characteristics of welding data such as a convolutional neural network (CNN), a recurrent neural network (RNN), and a long short-term memory (LSTM).

In the arc end adjusting device according to the present aspect, the procedure adjusting unit learns a neural network to output data indicating a change in the arc end procedure, which can shorten a cycle time of the welding process when the welding data is input. Equipped with a neural network.

According to this aspect, the procedure coordination neural network is, for example, a trained deep neural network, which can appropriately adjust the arc end procedure. The type of neural network is not particularly limited. It may be appropriately selected according to the characteristics of the welding data such as CNN, RNN, LSTM, and the like.

In the arc end adjusting device according to this aspect, the procedural adjusting neural network outputs data indicative of the amount of change in the arc end procedure.

According to this aspect, the procedure adjustment neural network can output the amount of change in the adjustable arc end procedure, rather than whether to shorten the cycle time of the welding process. For example, the procedure adjustment neural network may output a large amount of change if the welding result is very stable, and a small amount of change if the welding result is good but unstable. Therefore, the cycle time of a welding process can be shortened more quickly.

The arc end adjustment device which concerns on this aspect is obtained after adjusting the said arc end procedure, and the quality determination part which judges the quality of the welding result of an anti stick process and the next welding process based on the welding data acquired by the said acquisition part. On the basis of the determination result of the acceptance determination unit, a learning processing unit for learning the procedural adjustment neural network is provided.

According to this aspect, a procedure adjustment neural network learns using the data which shows the welding result at the time of adjusting the arc end procedure. Therefore, the cycle time of a welding process can be shortened more effectively so that a welding result may not deteriorate.

The arc end adjustment apparatus according to the present aspect is characterized in that the learning processing unit is configured such that the cycle time is shortened when the affirmative determination unit judges positive, and the cycle time is extended when the affirmative determination unit judges negative. Train the coordination neural network.

According to this aspect, the procedure adjustment neural network can be trained in the direction which shortens the cycle time of a welding process. By the said learning, the cycle time of a welding process can be shortened.

In the arc end adjustment apparatus according to the present aspect, the learning processing unit learns the procedure adjustment neural network so that the cycle time is maintained when the welding result is in an intermediate state of good and bad.

According to this aspect, when the welding result is an intermediate state between good and bad, the procedure adjustment neural network can be trained so that the cycle time of the welding process is maintained. The intermediate state is a state in which the welding result is relatively good, but when the cycle time is further shortened, the welding result may deteriorate. By this learning, the cycle time of a welding process can be shortened and a welding result can be stabilized in a favorable state.

In the arc end adjusting device according to the present aspect, in the arc end adjustment device, when the welding data indicating a welding state related to the next welding process is input, the quality of the welding result related to the welding process when the welding data is obtained is determined. It is equipped with a positive decision neural network that has trained the neural network to output data indicative of.

According to this aspect, the acceptance determination neural network is, for example, a deep neural network that has been learned, and can appropriately determine whether the welding result is good or bad. By using the result of the acceptance decision of the acceptance determination neural network, the procedure adjustment neural network can be learned more effectively.

In the arc end coordination apparatus according to the present aspect, the procedural coordination neural network includes a network configuration that is substantially the same as all or part of the positive determination neural network.

In accordance with this aspect, the procedural coordinating neural network comprises a neuronal configuration that is substantially the same as all or part of the positive decision neural network. For example, part of the procedure coordination neural network has an intermediate layer and weighting factor that is equal to or substantially the same as all or part of the positive decision neural network. Since the determination of the acceptance of the welding result and the adjustment of the arc end procedure have some common features, the acceptance determination neural network can be useful for the procedure adjustment neural network. In other words, the initial value of the weighting coefficient of the procedure-adjusted neural network can be set to a more appropriate value. Therefore, even if the training data for learning the arc end procedure is insufficient, if the training data of the welding data and the data indicating the results of the welding results can be sufficiently prepared, the initial value of the weighting coefficient of the procedure adjustment neural network is appropriately set, The training neural network can be trained more efficiently. It goes without saying that the network structures of the procedure coordination neural network and the positive judgment neural network may be configured in the same way.

The arc end adjustment apparatus which concerns on this aspect is equipped with the state data acquisition part which acquires the state data containing the image data obtained by image | photographing the welding torch, the welding wire, and the base material in multiple time views in an arc end process, The said procedure adjustment part is the said An evaluation unit that calculates an evaluation value for the action in the state indicated by the state data based on the state data acquired by the state data acquisition unit and the action data indicating the action associated with the arc end procedure; It has an action selector that selects the action with the highest evaluation value.

According to this aspect, the action associated with the optimal arc end procedure using the reinforcement learning evaluation unit based on the state data including image data obtained by imaging the welding torch, the welding wire, and the base metal at multiple time points in the arc end processing. Select.

The arc end adjustment apparatus which concerns on this aspect is a quality determination part which determines the quality of the anti-stick process and the next welding process based on the welding data acquired by the said acquisition part, and the said quality plate obtained after adjusting the said arc end procedure. A repair calculation unit that calculates a repair for the arc end procedure based on the determination result of the government and the time from the end of welding until the start of welding in the next welding process, and the state data acquisition unit And a reinforcement learning unit that learns the evaluation unit based on the state data acquired in step 2, the action data indicating the action associated with the arc end procedure, and the reward calculated by the reward calculation unit.

According to this aspect, it is possible to reinforce learning of the arc end procedure in which the cycle time of a welding process is shortened.

The arc end adjustment device according to the present aspect includes, in the state where the state data indicates, when the evaluation unit inputs state data acquired by the state data acquisition unit and action data indicating an action related to the arc end procedure. It is equipped with an evaluation neural network that outputs an evaluation value for the action.

According to this aspect, it is possible to deepen and learn the arc end procedure in which the cycle time of a welding process is shortened.

In the arc end adjusting device according to the present aspect, the welding data indicating the result of the anti-stick treatment includes at least one of data indicating current during welding release processing, an image of the tip of the welding wire, and a temperature of the tip of the welding wire. And the welding data indicating the welding state associated with the next welding process includes the welding current and welding voltage detected during the welding process, the feeding speed of the welding wire, the short-circuit situation, and the sound collected during the welding process. Data representing at least one of a welding sound and an image of a welded portion photographed after completion of welding.

According to this aspect, the quality of anti-stick processing can be determined based on at least one of data indicating the current during the de-welding process, the image of the tip of the welding wire, and the temperature of the tip of the welding wire. Moreover, at least one of the welding current and welding voltage detected during the welding process in the next welding process, the feeding speed of a welding wire, a short circuit situation, the welding sound picked up during the welding process, and the image of the welding site image | photographed after completion | finish of welding is shown. Using the data, the arc end procedure can be adjusted.

In the arc end adjusting device according to the present aspect, the arc end procedure includes a welding voltage and a welding current in an anti-stick treatment, a time of the anti-stick treatment, a retract time of a welding wire, and a time of de-welding treatment. And at least one of a retry number.

According to this aspect, the arc end treatment or the like is performed by adjusting the welding voltage and the welding current in the anti-stick treatment, the time of the anti-stick treatment, the retract time of the welding wire, and the time and the number of retries of the de-welding treatment. The cycle time of the welding process can be shortened.

The arc end adjusting device according to the present aspect includes a receiving unit that receives an adjustment strength for adjusting the arc end procedure by the procedure adjusting unit, and the procedure adjusting unit adjusts the arc end procedure at the adjustment strength received by the receiving unit. Is carried out.

According to this aspect, a user can arbitrarily set the degree of automatic adjustment of the arc end procedure implemented by the said procedure adjustment part.

The welding system which concerns on this aspect is equipped with any one of the arc end adjustment apparatus mentioned above, the welding robot which hold | maintains a welding torch, and the welding power supply which supplies a welding current to the said welding torch.

According to this aspect, the welding system provided with a welding robot and a welding power supply can shorten the cycle time of a welding process. Incidentally, the arc end adjusting device may be installed inside the welding robot and the welding power supply, and may be installed inside the control device for controlling the operation of the welding robot and the welding power supply. You can equip it outside separately. The arc end adjusting device may be a server, and the control device or the welding power supply may communicate with the server to shorten the cycle time of the welding step.

In the arc end adjustment method according to this aspect, in an arc end adjustment method for adjusting an arc end procedure in a repeatedly performed welding process, welding data indicating a result of an anti-stick treatment and a welding state associated with the next welding process is acquired. Based on the obtained welding data, the arc end procedure is adjusted so that the cycle time of the welding process is shortened.

According to this aspect, the cycle time of a welding process can be shortened. The arc end adjustment method may be an aspect which the welding power supply, the control apparatus, etc. which comprise a welding system automatically perform, and a worker may connect an arc end adjustment apparatus to a welding system, and may perform an arc end adjustment method. Do.

The computer program according to the present aspect is a computer program for adjusting the arc end procedure in a repeatedly performed welding process, wherein the computer indicates to the computer the result of the anti-stick treatment and the welding state associated with the next welding process. A program for acquiring welding data and executing a process of adjusting the arc end procedure so that the cycle time of the welding process is shortened based on the acquired welding data.

According to this aspect, a computer can function as the said arc end adjustment apparatus.

According to the present invention, it is possible to automatically adjust the time required for the arc end treatment such as the anti-stick treatment and the de-welding treatment to shorten the cycle time of the welding step.

1 is a schematic diagram showing an arc welding system according to a first embodiment.
2 is a block diagram showing an arc end adjusting device according to the first embodiment.
3A is a conceptual diagram illustrating an arc end procedure.
3B is a conceptual diagram illustrating an arc end procedure.
4 is a functional block diagram showing an arc end adjusting device according to the first embodiment.
FIG. 5 is a flowchart illustrating an arc end adjustment method according to the first embodiment. FIG.
FIG. 6 is a functional block diagram showing an arc end adjusting device according to Embodiment 2. FIG.
7 is a conceptual diagram illustrating a network configuration of a procedure coordination unit.
8 is a functional block diagram showing an arc end adjusting device according to the third embodiment.
9 is a functional block diagram showing an arc end adjusting device according to the fourth embodiment.
10 is a schematic diagram illustrating an arc welding system according to a fifth embodiment.
It is a schematic diagram which shows an arc end adjustment screen.

EMBODIMENT OF THE INVENTION Hereinafter, this invention is explained in full detail based on drawing which shows embodiment. In addition, you may combine arbitrarily at least one part of embodiment described below.

Embodiment 1

1 is a schematic diagram showing an arc welding system according to a first embodiment. The arc welding system according to the present embodiment is a gas shielded arc welding machine of a consuming electrode type, which is a welding robot 1, a welding power supply 2, a control device 3, an imaging device 4, and an arc end adjustment device 5. ) The arc end adjustment device 5 is provided in the control device 3. In addition, although the unit of the arc end adjustment apparatus 5 is demonstrated to be included in the control apparatus 3 for the convenience of drawing and description, the control apparatus 3 and the arc end adjustment apparatus 5 may be a unitary structure. In appearance, the hardware and software of the control device 3 may realize the function of the arc end adjustment device 5. In addition, the arc end adjustment apparatus 5 may be provided in the welding power supply 2, and may be installed in another apparatus. In addition, you may comprise so that the function of the arc end adjustment apparatus 5 may be distributed-processed by a some apparatus and the server.

The welding robot 1 performs arc welding of the base material A automatically. The welding robot 1 is equipped with the base part fixed to the suitable location of a floor surface. A plurality of arms are rotatably connected to the base portion via a shaft portion, and a welding torch 11 is held at the tip portion of the arm. Moreover, the wire supply apparatus 12 is provided in the suitable location of an arm. A motor is provided in the connection part of each arm, and each arm rotates about an axis part by the rotational drive force of a motor. The rotation of the motor is controlled by the control device 3. The control device 3 can move the welding torch 11 up, down, front, back, left, and right with respect to the base material A by rotating each arm. Moreover, the encoder which outputs the signal which shows the rotation position of an arm to the control apparatus 3 is provided in the connection part of each arm, The control apparatus 3 is based on the signal output from the encoder, and the welding torch ( 11) recognize the position.

The welding torch 11 is made of a conductive material such as a copper alloy and guides the welding wire W to the base material A to be welded, and at the same time, a cylindrical contact for supplying a welding current necessary for generating an arc. Have a tip. The welding current is supplied from the welding power supply 2. The welding wire W is supplied to the welding torch 11 by the wire supply apparatus 12 from the wire supply source which is not shown in figure. The welding wire W is a solid wire, for example, and functions as a consuming electrode.

The contact tip contacts the welding wire W through which the inside is inserted, and supplies a welding current to the welding wire W. FIG. In addition, the welding torch 11 has a hollow cylindrical shape for restoring the contact tip, and has a nozzle for injecting shield gas from the opening of the tip to the base material A. The shield gas is for preventing oxidation of the base material A and the welding wire W melted by the arc. The shield gas is, for example, a mixed gas of carbon dioxide gas, carbon dioxide gas and argon gas, and inert gas such as argon. Shield gas is supplied from the welding power supply 2.

The welding power supply 2 includes a power supply unit 21, a wire supply control unit 22, a shield gas supply unit 23, and a detection unit 24. The power supply part 21 is connected to the contact tip of the welding torch 11 and the base material A via a feed cable, and supplies a welding current. The wire supply control unit 22 controls the supply speed of the welding wire W by the wire supply device 12. The shield gas supply part 23 supplies the shield gas to the welding torch 11. The detection part 24 is a current detection part which detects the welding current which flows through an arc during a welding process, and the electric current in anti-stick process and de-welding process, and the voltage detection part which detects the voltage applied to the welding torch 11 and the base material A. FIG. It includes. The power supply unit 21 includes a power supply circuit, a signal processing circuit, and the like which output a PWM controlled DC current based on the welding current and the welding voltage detected by the detection unit 24. Moreover, the welding power supply 2 outputs the welding monitor data which shows the welding state in an welding process, the state of an anti stick process, and a welding release process to the control apparatus 3. The welding monitor data is, for example, welding current data or welding voltage data indicating a welding current or welding voltage detected during the welding process. In addition, the welding monitor data includes data of voltage or current detected during the anti-stick process and the de-welding process. In addition, as welding monitor data, the supply speed data which shows the supply speed of the welding wire W, the short circuit condition data which shows a short circuit situation, and the welding sound data obtained by collecting a micro not shown may be output to the control apparatus 3. .

In addition, the said welding monitor data is an example of the welding data which shows the result of an anti stick process, and the welding state related to the next welding process.

The imaging device 4 image | photographed the tip part or welding site | part of the welding wire W during an anti-stick process and a welding release process, and image | photographed and imaged the welding site | part of the base material A after a welding process. The data is output to the control device 3. Moreover, the imaging device 4 may be an infrared camera which detects the temperature of the front-end | tip part of the welding wire W. FIG.

In addition, the said image data is an example of the welding data which shows the result of an anti stick process, and the welding state related to the next welding process.

The control device 3 controls the operation of the welding robot 1, and simultaneously controls welding conditions such as a welding current, a welding voltage, a supply speed of the welding wire W, and a supply amount of shield gas to the welding power supply 2. Output to the welding power supply 2 to control the operation of the welding power supply (2). The control apparatus 3 memorize | stores the various welding conditions about the material of the base material A, the kind of improvement, etc. Moreover, the control apparatus 3 outputs an arc end procedure, and performs an arc end process. The welding conditions stored in the control device 3 are not necessarily optimal, and for the arc end procedure, the arc end adjusting device 5 such that the cycle time of the welding process is minimized within a range in which the welding result does not deteriorate. Is being adjusted by.

In addition, although only one set of the welding power supply 2 and the welding torch 11 is shown in FIG. 1, when the several welding power supply 2 is installed in a production line, one arc end adjustment apparatus 5 is carried out. The arc end adjustment device 5 may be separately installed in each of the plurality of welding power sources 2, and may be configured to adjust the arc end procedure of the corresponding power source for each of the plurality of welding power sources 2. It may be configured to adjust the arc end procedure of the.

2 is a block diagram showing the arc end adjusting device 5 according to the first embodiment. The arc end adjustment apparatus 5 is equipped with the control part 50 which controls the operation | movement of each component part of the said arc end adjustment apparatus 5. The control unit 50 is connected to an input unit 50a, an output unit 50b, and a storage unit 50c.

The storage unit 50c is a nonvolatile memory such as an EEPROM (Electrically Erasable Programmable ROM), a flash memory, or the like. The storage unit 50c stores the computer program 50d for optimizing the arc end processing and minimizing the cycle time of the welding process within a range in which the welding result does not deteriorate.

The control unit 50 is a computer having a processor such as a central processing unit (CPU), a graphics processing unit (GPU) or a multi-core CPU, a read only memory (ROM), a random access memory (RAM), an input / output interface, and the like. The input unit 50a, the output unit 50b, and the storage unit 50c are connected to each other. The control part 50 performs the arc end adjustment method which shortens the cycle time of the said welding process in continuous production by which the welding part is repeatedly performed in a production line by executing the computer program 50d memorize | stored in the memory | storage part 50c. The computer functions as the arc end adjustment device 5. In addition, the welding process performed repeatedly means the welding which one welding power supply 2 or the welding torch 11 provided in the production line repeatedly performs.

The input unit 50a is connected to the welding power supply 2 and the imaging device 4. The welding monitor data output from the welding power supply 2 and the image data output from the imaging device 4 are input to the input part 50a. The welding monitor data is, for example, time series data indicating a welding current and a welding voltage during a welding process, a current and a voltage during an anti-stick treatment and a welding release treatment, a feeding speed of the welding wire W, a short circuit situation, a welding sound, and the like. . Image data is data which showed the external appearance of the front-end | tip part of the welding wire W, and the external appearance of the bead after welding. The image data may be data indicating the temperature of the tip portion of the welding wire W captured by the infrared camera.

The output part 50b is connected to the welding robot 1 and the welding power supply 2. The controller 50 controls the welding process and the arc end procedure, and also outputs control data for changing the arc end procedure to the welding robot 1 and the welding power source 2. The control data for changing the arc end procedure may be indicative of a change in the arc end procedure, or may indicate an arc end procedure after the procedure change.

3A and 3B are conceptual diagrams illustrating an arc end procedure. In particular, FIG. 3A shows an anti-stick treatment, and FIG. 3B shows a procedure of de-welding treatment.

As shown in FIG. 3A, the control device 3 that has completed the welding process supplies a smaller current Ia and voltage Va at the time of welding to supply the required anti-stick treatment time ta to provide a welding wire. The distal end of (W) is melted and raised to form a solidification hole of appropriate dimensions. Moreover, you may perform the process of pulling up the welding wire W for the predetermined retracting time.

By optimizing the anti-stick treatment, the length of the welding wire W protruding from the tip of the welding torch 11 can be the length that can start the next welding process without clogging, and also the shape and size of the coagulation sphere. The weldability of the next welding process is also good.

In addition, since the welding wire W may be welded to the base material A at the end of welding, as shown in FIG. 3B, the control apparatus 3 has an electric current related to the required welding release. (Ir) is supplied to the welding wire W during the required welding release processing time tr to thereby perform a process of melting the welding location. In addition, since welding may not be able to be canceled by one melting process, when welding is not eliminated, welding removal processing is performed by the required number of retries Nr. The presence or absence of welding can be determined by detecting the electric current which flows into the welding wire W. FIG.

4 is a functional block diagram showing the arc end adjusting device 5 according to the first embodiment. The arc end adjustment apparatus 5 is a function block as a welding monitor data acquisition part 51a, the image data acquisition part 51b, the 1st acceptance determination part 52a, and the 2nd acceptance determination part 52b. And an anti-stick pass / fail determination unit 52c, a walk-in-and-stop comprehensive determination unit 54, a procedure adjustment unit 55, an arc end control unit 56, and a shortest procedure storage unit 57.

The welding monitor data acquisition unit 51a acquires the welding monitor data output from the welding power supply 2, and outputs the acquired welding monitor data to the first transfer determination unit 52a and the anti-stick transfer determination unit 52c. .

The image data acquisition unit 51b acquires the image data output from the imaging device 4, and outputs the acquired image data to the second pass-through determination unit 52b and the anti-stick pass-through determination unit 52c.

When the welding monitor data is input, the first quality determining unit 52a outputs data indicating the quality of the welding result related to the welding process when the welding monitor data is obtained. Neural Network The acceptance judgment RNN 53a is, for example, a learned recursive neural network.

The acceptance judgment RNN 53a includes, for example, a first neuron outputting data indicating a probability that a welding result is good, and a second neuron outputting data indicating a probability that the welding result is bad. In this case, the data indicating the affirmativeness are data output from the first and second neurons.

In addition, the pass / fail determination RNN 53a may be equipped with the output layer the neuron which outputs the pass / fail of a welding result by 2 values. In this case, the data indicating the affirmativeness is data of two values output from the neuron.

In addition, the pass / fail determination RNN 53a may include a neuron in the output layer that outputs an analog value indicating the degree of pass / fail of the welding result.

The pass / fail determination RNN 53a uses the welding monitor data (input data) and the data (teacher data) indicating the pass / fail of the welding result corresponding to the weld data as the training data, and the recursive deep neural network before learning. You can learn by giving.

In addition, the structure, such as the number of layers of the intermediate | middle layer of the pass / fail determination RNN 53a, the number of neurons of each layer, is not specifically limited. In addition, the acceptance determination RNN 53a does not necessarily need to be a recursive neural network, and may be constituted by other types of neural networks.

When the image data is input, the second acceptance determination unit 52b outputs the data indicating the acceptance of the welding result related to the welding process when the image data is obtained. The acceptance determination CNN 53b (Convolutional Neural Network) ) The pass / fail determination CNN 53b is a trained overlapping neural network.

The acceptance judgment CNN 53b includes, for example, a third neuron outputting data indicating a probability that a welding result is good, and a fourth neuron outputting data indicating a probability that the welding result is bad. In this case, the data indicating the affirmativeness are data output from the third and fourth neurons.

In addition, the pass / fail determination CNN 53b may be equipped with the output layer the neuron which outputs the pass / fail of a welding result by 2 values. In this case, the data indicating the affirmativeness is data of two values output from the neuron.

In addition, the pass / fail determination CNN 53b may include a neuron in the output layer that outputs an analog value indicating the degree of pass / fail of the welding result.

The pass / fail determination CNN 53b may be trained by giving image data (input data) and data (teacher data) indicating pass / fail of the welding result corresponding to the weld data to the superimposed neural network before learning.

In addition, the structure, such as the number of layers of the intermediate | middle layer of the pass / fail determination CNN 53b, the number of neurons of each layer, is not specifically limited. In addition, the acceptance determination CNN 53b does not necessarily need to be an overlapping neural network, and may be constituted by other types of neural networks.

The anti-stick acceptance determination unit 52c determines whether the anti-stick process is successful by determining whether or not the current flowing at the time of the welding release processing of the arc end processing is equal to or greater than a threshold. The result is output to the positive and negative judgment unit 54.

In addition, the anti-stick acceptance determination unit 52c determines whether the anti-stick process is successful or not based on the image of the distal end portion of the welding wire W imaged at the time of the anti-stick process, and determines the determination result of the acceptance-positive determination unit 54. ) For example, whether or not the size of the coagulation ball formed on the tip of the welding wire W is within a predetermined range and whether the shape of the coagulation ball is suitable for the predetermined pattern is determined whether the anti-stick treatment is successful.

The acceptance comprehensive determination unit 54 is based on the data output from the first acceptance determination unit 52a, the second acceptance determination unit 52b, and the anti-stick acceptance determination unit 52c. Of the welding result is determined, and the determination result is output to the procedure adjustment unit 55.

For example, the acceptance comprehensive determination unit 54 includes data output from the first and second neurons of the acceptance determination RNN 53a, data output from the third and fourth neurons of the acceptance determination CNN 53b, and the like. The determination result of the anti-stick pass / failure determination unit 52c is comprehensively determined. Specifically, by comparing the sum of the values of the data output from the first and third neurons with the sum of the values of the data output from the second and fourth neurons, it may be determined whether the next welding result is good or bad. In addition, you may weight-add and compare the value of the data output from each neuron. And when the welding result in the next welding process is favorable and the result of anti-stick process is also favorable, the both parts comprehensive determination part 54 will judge it as a positive whole collectively.

In addition, in the case of a configuration in which two-value data is output from the acceptance determination RNN 53a and the acceptance determination CNN 53b, the acceptance comprehensive determination unit 54 includes the first acceptance determination unit 52a and the second acceptance determination unit ( When both of 52b) output the data which shows that it is favorable, the welding result of the next welding process is judged to be positive, and one of the 1st acceptance determination part 52a and the 2nd acceptance determination part 52b is In the case of outputting data indicating that the defect is defective, the welding result of the next welding process is determined as defective. In addition, the method of comprehensive determination is an example, and when one of the 1st acceptance determination part 52a and the 2nd acceptance decision part 52b is outputting the data which shows that it is favorable, the welding result of the next welding process is positive. You may comprise so that it may be determined.

The procedure adjustment unit 55 adjusts the arc end procedure so that the cycle time of the welding process is shortened when the determination result of the integrated quality determination unit 54 is positive, and the cycle time of the welding process is extended when the determination result is bad. Then, the adjustment result is output to the arc end control part 56. The adjustment result is, for example, data indicating the increase or decrease of various parameters of the arc end procedure, that is, the current, voltage and time in the anti-stick process, the retract time, the current, time, retry count, etc. in the de-welding process. The procedure adjustment unit 55 outputs data to the arc end control unit 56 indicating increasing or decreasing at least one of various parameters of the arc end procedure.

The procedure adjustment part 55 may change the value of a some parameter in one adjustment process, and may change the value of one parameter. In addition, when adjusting the arc end procedure by repeating step S11-step S19 mentioned later until a cycle time becomes the shortest, you may comprise so that a different parameter may be adjusted in each repeated adjustment process. It's okay. For example, in the first adjustment, the anti-stick treatment time may be adjusted, and in the second adjustment, the welding release treatment time may be adjusted.

In addition, the variable may be configured to have a correlation with the amount of increase and decrease of each parameter. Moreover, you may comprise so that a change amount may be limited in the range of a predetermined ratio from a standard parameter value.

In addition, when the procedure adjustment unit 55 determines that the cycle time of the welding process is shortened, and the welding result is changed from a good state to a bad state, the procedure end unit 55 performs the arc end procedure before the cycle time shortening. Remember to.

In addition, when the determination result is poor, an example of adjusting the arc end procedure so that the cycle time of the welding process is extended has been described above. However, the parameter may be adjusted so that the determination result is good while maintaining the cycle time. .

The arc end control part 56 controls an arc end by outputting control data for changing an arc end procedure to the welding power supply 2 based on the adjustment result of the procedure adjustment part 55. FIG. However, when the shortest procedure storage unit 57 stores the arc end procedure in which the cycle time of the welding process is the shortest, the arc end control unit 56 stores the arc end procedure stored in the shortest procedure storage unit 57. To control the arc end.

Next, the processing procedure of the control part 50 which concerns on adjustment of an arc end procedure is demonstrated.

5 is a flowchart illustrating an arc end adjusting method according to the first embodiment. The controller 50 repeatedly executes the following processing for each welding step in continuous production, for example. The control part 50 determines whether the memory | storage part 50c memorize | stores the arc end procedure of the shortest cycle time (step S11). If it is determined that the shortest arc end procedure is stored (step S11: YES), the control section 50 arcs the end based on the shortest arc end procedure stored by the storage section 50c (step S11). (S12)). For example, the controller 50 controls the arc end by outputting control data indicating the shortest arc end procedure to the welding power supply 2. Of course, the controller 50 may control the arc end by outputting control data indicating the amount of change for making the arc end procedure with the shortest cycle time to the welding power supply 2.

When the storage unit 50c determines that the arc end procedure of the cycle time shortest is not stored (step S11: NO), the control unit 50 obtains welding monitor data (step S13), Image data is acquired (step S14). Then, the control unit 50 determines whether the anti-stick processing and the welding result are successful based on the acquired welding monitor data and image data (step S15). For example, the control part 50 determines the quality of a welding result using the quality determination RNN 53a and the quality determination CNN 53b of completion of learning. Both parts of the anti-stick treatment and the de-welding process are determined by the current at the time of the de-welding process.

Next, when it determines with a welding result favorable (step S15: YES), the control part 50 shortens the cycle time of a welding process (step S16). When it is determined that the welding result is defective (step S15: NO), as a result of shortening the welding process last time, it is determined whether the welding result has changed from a good state to a defective state (step S17). . Needless to say, the change to the defective state is not limited to the configuration of determining only one welding result, but also includes a configuration of judging using two or more welding results. For example, in 10 times, when a welding result is in a bad state more than a fixed ratio, you may determine that it changed to the bad state. When it is determined that the welding result is not changed from the good state to the bad state (step S17: NO), the control part 50 extends the cycle time of a welding process (step S18). The control part 50 which completed the process of step S16 or step S18 controls an arc end based on the arc end procedure after adjustment (step S19). Specifically, the controller 50 controls the arc end by outputting control data indicating the arc end procedure after the adjustment process to the welding power supply 2. Of course, the controller 50 may control the arc end by outputting control data indicating the amount of change in the arc end procedure to the welding power supply 2.

As a result of shortening the cycle time of the welding step, when it is determined that the welding result has changed from a good state to a bad state (step S17: YES), the controller 50 determines the arc end procedure before shortening the cycle time of the welding step. (Step S20), the arc end procedure before the cycle time reduction is stored in the storage unit 50c as the shortest arc end procedure (step S21), and the process returns to step S12. .

According to the above-described arc end adjusting device 5, the welding system, the arc end adjusting method and the computer program 50d, the arc end processing can be optimized without effectively deteriorating the welding result, thereby effectively reducing the cycle time of the welding process. Can be.

Moreover, since it is the structure which stores the shortened arc end procedure in the memory | storage part 50c, the arc end adjustment apparatus 5 can control welding by shortening the cycle time of a welding process quickly after that.

In addition, although the arc end adjustment apparatus 5 demonstrated the example in which the arc end adjustment apparatus 5 was equipped with the learning completion determination RNN 53a and the acceptance determination CNN 53b in this Embodiment 1, the 1st acceptance determination part 52a and the 2nd assignment part Various parameters defining the neural network of the determination unit 52b may be configured to be downloaded from an external server and updated. The parameter is information including, for example, the number of layers in the middle layer, the number of neurons in each layer, the weighting coefficient of each neuron, the type of activation function, and the like. Furthermore, the arc end adjustment apparatus 5 memorize | stores the flag which shows whether it is allowed to reflect the downloaded various parameters to the 1st acceptance determination part 52a and the 2nd acceptance decision part 52b, and a flag In the case where the permission is indicated, the neural network of the acceptance determination RNN 53a and the acceptance determination CNN 53b may be updated using the downloaded parameter.

In addition, when the welding system provided with the arc end adjustment apparatus 5 is installed in a factory, the arc end adjustment apparatus 5 of each welding system may replace the said parameter as needed.

In addition, you may comprise the arc end adjustment apparatus 5 as a cloud server. The welding power supply 2 or the control device 3 may request the server to adjust the arc end procedure, receive the adjustment amount of the arc end procedure sent from the server as required, and adjust the arc end procedure. .

In addition, the arc end adjustment device 5 may be provided in the welding power supply 2. In addition, you may implement the arc end adjustment apparatus 5 as a dedicated apparatus for arc end procedure adjustment. The operator can connect the dedicated device to the welding system and automatically adjust the arc end procedure.

Moreover, although the 1st acceptance determination part 52a and the 2nd acceptance determination part 52b demonstrated the example provided with the acceptance determination RNN 53a and the acceptance determination CNN 53b, both or one of each determination part was neural. It is also possible to configure so as to determine the quality of the welding result without using a network. For example, the first pass / fail determination unit 52a may determine pass / fail by simple determination processing of comparing the welding current value with a predetermined threshold value. Further, the second pass / failure determination unit 52b may determine pass / fail by simple determination processing of extracting a predetermined feature amount from the image data, and comparing the presence or absence of the feature amount, the number of feature amounts, and the like with a threshold value. . In addition, it is not necessary to provide both the first pass / fail determination unit 52a and the second pass / fail determination unit 52b, and any one may be provided. In this case, the integrated status determination unit 54 is unnecessary.

Embodiment 2

In the arc end adjustment device 205, the welding system, the arc end adjustment method and the computer program 50d according to the second embodiment, the procedure adjustment part 55 and the shortest procedure storage part 57 of the first embodiment are connected to a deep neural network. Since the point comprised is different from Embodiment 1, the said difference is mainly demonstrated below. Since the other structure and effect are the same as that of embodiment, the same code | symbol is attached | subjected to the corresponding location, and detailed description is abbreviate | omitted.

6 is a functional block diagram showing an arc end adjusting device 205 according to the second embodiment. The arc end adjustment apparatus 205 which concerns on Embodiment 2 is the welding monitor data acquisition part 51a, the image data acquisition part 51b, the 1st good-quality determination part 52a, and the 2nd good-quality board similarly to Embodiment 1. The government unit 52b, the father and son comprehensive determination unit 254, the procedure adjustment unit 255, the learning processing unit 259, and the arc end control unit 56 are provided.

The welding monitor data acquisition unit 51a acquires welding monitor data output from the welding power supply 2, and acquires the acquired welding monitor data from the first acceptance determination unit 52a, the anti-stick acceptance determination unit 52c, and the procedure adjustment unit. Output as (255).

The image data acquisition unit 51b acquires the image data output from the imaging device 4, and acquires the acquired image data from the second transfer determination unit 52b, the anti-stick transfer determination unit 52c, and the procedure adjustment unit 255. Will output

The integrated / failure determination unit 254 of the second embodiment outputs the data indicating the welding result and the probability of the anti-stick processing to be good and the data indicating the welding result and the probability of the anti-stick processing to the learning processing unit 259. . For example, the value of the data output from the first neuron of the acceptance determination RNN 53a, the value of the data output from the third neuron of the acceptance determination CNN 53b, and the output from the anti-stick acceptance determination unit 52c. The probability of having a good welding result and an anti-stick treatment may be calculated using the obtained data value. Similarly, the value of the data output from the second neuron of the acceptance determination RNN 53a, the data output from the fourth neuron of the acceptance determination CNN 53b, and the value of the data output from the anti-stick acceptance determination unit 52c. It is possible to calculate the probability that the welding result is bad by using.

The procedure adjustment unit 255 includes a procedure adjustment NN (Neural Network) 258 for outputting data indicating an amount of change in the arc end procedure that can reduce the cycle time when welding monitor data and image data are input. The procedure coordination NN 258 is a trained deep neural network.

The procedure adjustment NN 258 is, for example, the current Ia, the voltage Va and the time ta in the anti-stick process, the current Ir in the de-welding process, the time tr and the number of retries. For each of the plurality of adjustment amounts for each adjustment parameter, such as (Nr), the output layer includes a plurality of neurons for outputting data indicating the probability that the adjustment amount is preferable.

In addition, the procedure adjustment NN 258 may be the structure which equips an output layer with the neuron which outputs the data which shows the adjustment amount. In addition, the procedure adjustment NN 258 may be a structure which equips an output layer with the neuron which outputs adjustment amount as 2-value data. Hereinafter, in the second embodiment, the procedure adjustment NN 258 outputs data of probability of indicating that the change amount is appropriate, rather than the two-value data.

In addition, the procedure adjustment part 255 may change the value of a some parameter in one adjustment process, and may change the value of one parameter. In addition, in each adjustment process repeated, you may comprise so that another parameter may be adjusted.

7 is a conceptual diagram illustrating a network configuration of the procedure adjustment unit 255. The procedure adjustment NN 258 of the procedure adjustment unit 255 includes a welding state recognition network unit 258a, an appearance recognition network unit 258b, and a procedure adjustment network unit 258c.

The welding state recognition network unit 258a is a neural network that receives welding monitor data, recognizes the welding state during the welding process, and outputs data corresponding to the state. When the welding monitor data is a welding current, the welding state recognition network unit 258a can recognize a change state of the welding current. The welding state recognition network unit 258a may have a neural network structure similar to that of the first pass / fail determination unit 52a except for the output layer, for example. The output layer has a plurality of neurons, preferably three or more neurons. In addition, as an initial value of the weight coefficient before learning, you may set the weight coefficient of each neuron which comprises the 1st positive determination part 52a. The procedure adjustment unit 255 can be trained more efficiently.

The appearance recognition network part 258b is a neural network which inputs image data, recognizes the state of the welded part after welding, and outputs data according to the state. The appearance recognition network unit 258b may have a neural network structure similar to that of the second pass / fail determination unit 52b except for the output layer, for example. The output layer has a plurality of neurons, preferably three or more neurons. As the initial value of the weight coefficient before learning, the weight coefficient of each neuron constituting the second pass / fail determination unit 52b may be set. The procedure coordination NN 258 can be trained more efficiently.

The procedure coordination network unit 258c receives the data output from the welding state recognition network unit 258a and the appearance recognition network unit 258b, respectively, and outputs data indicating a change amount of the shortened arc end procedure. It is a neural network of completion. The neural network is preferably configured as a deep neural network having a plurality of intermediate layers.

In addition, the neural network configuration of the procedure adjustment unit 255 is an example, and may be configured as one neural network, or a plurality of neural networks may be combined.

The learning processing unit 259 uses the welding monitor data and the image data inputted to the procedure adjusting unit 255 as input data, the welding result when the arc end procedure is changed based on the data, and the data indicating the anti-stick processing quality. As the learning data, it is a processing unit that learns the procedure adjustment NN 258.

Specifically, the learning processing unit 259 is configured such that when the welding result and the anti-stick processing are good, the cycle time is shortened, and the welding result and the anti-stick processing are bad, based on the determination result of the acceptance-specific determination unit 254. When the cycle time is extended and the welding result is in an intermediate state of good and bad, the procedure adjustment NN 258 is trained so that the cycle time is maintained.

A good welding result is a state in which the probability that a welding result is favorable is 50% or more, and the probability that a welding result is bad is less than 50%, and the result of an anti stick process is favorable. A poor welding result is, for example, a state in which the probability that the welding result is good is less than 50%, the probability that the welding result is bad is 50% or more, or a state in which the result of the anti stick treatment is bad. 50% of the threshold is an example, and may be larger than 50%.

The intermediate state of the welding result is, for example, when both of the probability of a good welding result and the probability of a bad welding result are 50% or more, or both are less than 50%, and the result of the anti-stick treatment is good. to be. In addition, when the threshold value is larger than 50%, for example, 60%, the situation in which the probability that the welding result is good or bad is between 40% and 60% is also an intermediate state. In addition, this intermediate state is an example. The intermediate state is a state in which welding results may deteriorate when the abnormal cycle time is shortened.

As described above, by training the procedure adjustment NN 258, the time required for arc end processing such as anti-stick treatment and de-welding treatment is automatically adjusted without deteriorating the welding result, and thus the cycle time of the welding process. Can be minimized.

In addition, in the initial stage of learning of the procedure adjustment NN 258, the arc end procedure may be appropriately changed without maintaining the cycle time in the intermediate state.

In addition, the learning of the procedure adjustment NN 258 may be performed at an appropriate timing, such as when a welding system is installed, when the external environment changes, when the welding conditions are changed, when other procedure replacement is performed, or the like. .

In addition, in the second embodiment, an example of learning the procedure adjustment NN 258 has been described, but the learning completion procedure adjustment NN 258 may be provided so as not to perform further learning.

According to the arc end adjusting device 205, the welding system, the arc end adjusting method, and the computer program 50d according to the above-described embodiment 2, the procedure adjusting unit 255 composed of the deep neural network changes the arc end procedure. Since it is the structure which determines the quantity, it is possible to control and shorten the arc end of a welding process more appropriately, without worsening a welding result.

In addition, the procedure adjusting unit 255 may output the amount of change in the arc end procedure that shortens the cycle time of the welding process. For example, the procedure adjusting unit 255 may output a large change amount when the welding result is very stable, and output a small change amount when the welding result is good but unstable. Therefore, the cycle time of a welding process can be shortened more quickly.

In addition, the procedure adjusting unit 255 can learn using the result of the determination of the result of the welding result, and can adjust the procedure adjusting unit 255 suitable for the environment in which the welding system is installed. Therefore, the cycle time of a welding process can be shortened according to welding conditions and an external environment.

In addition, after the learning processing unit 259 learns the procedure adjusting unit 255 in a direction in which the cycle time of the welding process is shortened, it is possible to output data capable of shortening the cycle time of the welding process. The procedure adjustment neural network can be trained to obtain a stable welding result. Therefore, it is possible to stabilize the welding result in a good state and to shorten the cycle time of the welding process.

In addition, in this Embodiment 2, although the example where the arc end adjustment apparatus 205 provided the learning completion procedure adjustment NN 258 was demonstrated, the various parameters which define the neural network of the procedure adjustment part 255 are downloaded from an external server. It may be configured to update. The parameter is information including, for example, the number of layers in the middle layer, the number of neurons in each layer, the weighting coefficient of each neuron, the type of activation function, and the like. In addition, the arc end adjustment apparatus 205 stores a flag indicating whether to allow the procedure adjustment unit 255 to reflect the downloaded various parameters, and when the flag indicates permission, the procedure is performed using the downloaded parameter. The neural network of the coordination NN 258 may be updated.

In addition, when the welding system provided with the arc end adjustment apparatus 205 is provided in a factory, the arc end adjustment apparatus 205 of each welding system may exchange the said parameter as needed.

In addition, the arc end adjustment device 205 may be configured to upload the various parameters defining the learned procedure adjustment NN 258 to an external server. The other arc end coordination apparatus 205 may update the procedure coordination NN 258 using the parameter uploaded to the server.

In addition, in Embodiment 2, although the procedure adjustment NN 258 and the 1st transfer determination part 52a and the 2nd transfer decision part 52b demonstrated the neural network, the pass determination RNN 53a and pass Either or both of the determination CNN 53b may be configured to determine whether the welding result is good or not without using a neural network.

Embodiment 3

8 is a functional block diagram showing an arc end adjusting device 305 according to the third embodiment. The arc end adjustment device 305, the welding system, the arc end adjustment method, and the computer program 50d according to the third embodiment differ mainly from the second embodiment because the data inputted to the procedure adjustment unit 355 is different from the second embodiment. Explain. Since the other structure and effect are the same as that of embodiment, the same code | symbol is attached | subjected to the corresponding location, and detailed description is abbreviate | omitted.

The arc end adjustment apparatus 305 which concerns on Embodiment 3 further has the welding condition data acquisition part 51c. The welding condition data acquisition part 51c acquires welding condition data. The welding condition data includes, for example, information such as a material of the base material A, an improved shape, a welding current set value, a welding voltage set value, a welding speed set value, and a frequency set value when periodically changing the welding current. Include.

The procedure adjustment unit 355 outputs data indicating the amount of change in the arc end procedure, which can shorten the cycle time of the welding process without deteriorating the welding result, based on the input welding monitor data, image data, and welding condition data. Process coordination of learning completion NN (358). The procedure adjustment NN 358 of learning completion which concerns on Embodiment 3 can further learn by using the same quantity comprehensive determination part 354 and the learning process part 359 similar to Embodiment 2. FIG.

According to the arc end adjusting device 305, the welding system, the arc end adjusting method, and the computer program 50d according to the third embodiment, since the arc end procedure is adjusted by adding welding conditions, the arc end procedure is more effectively performed. It is possible to coordinate the end procedure.

Embodiment 4

9 is a functional block diagram showing an arc end adjusting device 405 according to the fourth embodiment. As for the arc end adjustment apparatus 405, the welding system, the arc end adjustment method, and the computer program 50d which concerns on Embodiment 4, the procedure adjustment part 455 and the shortest procedure memory part 57 of Embodiment 1 are deep reinforcement learning. Since it is different from Embodiment 1 in that it is configured to learn the arc end procedure in the following, the above-described differences will mainly be described. Since the other structure and effect are the same as that of embodiment, the same code | symbol is attached | subjected to the corresponding location, and detailed description is abbreviate | omitted.

The arc end adjustment apparatus 405 which concerns on Embodiment 4 is the welding monitor data acquisition part 51a, the image data acquisition part 51b, the state data acquisition part 51d, the 1st good determination part 52a, and the 2nd. It is equipped with the assignment determination part 52b, the assignment comprehensive decision part 54, the procedure adjustment part 455, and the arc end control part 56. As shown in FIG.

The state data acquisition unit 51d acquires state data indicating the state s of the welding system. The state data includes, for example, image data obtained by imaging the welding torch 11, the welding wire W, and the base material A at a plurality of viewpoints in the arc end processing, for example, moving image data. The welding torch 11, the welding wire W, and the base material A may be picked up by the imaging device 4, or may be picked up using another video pick-up device. The moving image data can recognize the state s of the welding system, that is, the positional relationship between the welding torch 11, the welding wire W and the base material A, and the solidification hole formed at the distal end of the welding wire W as an image. A configuration including image data that can be used is preferable. The image data may be data of an image obtained by capturing the tip end portion of the welding wire W using an infrared camera, that is, an image showing the temperature of the tip end portion of the welding wire W. FIG.

In addition, in this Embodiment 4, although image data is mainly demonstrated as an example of state data, the current data which shows the electric current which flows in the welding wire W, the other temperature data which shows the temperature of the welding wire W, etc. may be sufficient. .

The procedure adjustment unit 455 learns an arc end procedure that minimizes the cycle time of the welding process by in-depth reinforcement learning, and includes an evaluation unit 455a, an action selection unit 455b, a reward calculation unit 455c, and The reinforcement learning part 455d is provided.

The evaluation unit 455a is based on the state data acquired by the state data acquisition unit 51d and the action data indicating the action (a) according to the arc end procedure, to the action (a) in the state indicated by the state data. It is an arithmetic function part which calculates the evaluation value (Q). A state is an image which shows the positional relationship to the said positional relationship of the welding torch 11, the welding wire W, and the base material A, for example. The action (a) according to the arc end procedure is determined by the start of supply of current related to the anti-stick treatment, the current value, the start of supply of current related to the de-welding treatment, the retry of the de-deposition treatment, and the like. When the welding torch 11, the welding wire W, and the base material A are in a specific positional relationship, the evaluation value Q is the cycle time of the welding process when the specific action (a) is taken. Preferably, the shorter the value and the better the welding result, the higher the value.

For example, the evaluation unit 455a receives state data indicating a state s of the welding system acquired by the state data acquisition unit 51d, and action data indicating an action a related to the arc end procedure. In this case, an evaluation NN (Neural Network) 455e outputs an evaluation value Q (s, a) for the action a in the state s.

In addition, the evaluation NN 455e may be equipped with an overlapping neural network at the front end for recognizing the state data showing the state of the welding system as an image.

In addition to the neural network, the evaluation unit 455a may include a table in which state data and behavior data and an evaluation value are related, and may be configured to output the evaluation value using the table.

The action selection unit 455b selects the action a in which the evaluation value Q calculated by the evaluation unit 455a is maximum in a certain state s. The procedure adjustment unit 455 adjusts the arc end procedure based on the action (a) selected by the action selection unit 455b, and the arc end control unit 56 controls the arc end with the adjusted arc end procedure. To control.

The maintenance calculation unit 455c is configured for the arc end procedure based on the determination result output from the combined-integrated determination unit 54 and the time until the arc occurs after the welding torch 11 reaches the welding point. Calculate the reward. The repair is calculated such that the better the welding result, the shorter the time until the arc occurs, the higher the value. The calculation formula for calculating the complement is not particularly limited.

The reinforcement learning part 455d is based on the state data and behavior data input into the evaluation NN 455e, the evaluation value Q output when each data is input, and the reward computed by the reward calculation part 455c. The learning NN 455e is learned. Specifically, the weighting coefficient of the neural network may be learned from the evaluation value Q shown in the following formula (1). Q (s, a)? Q (s, a) +? (R +? MaxQ (s_next, a_next) -Q (s, a))? (1) However, s: state, a: action selected in state (s), α: learning coefficient, r: reward resulting from the action, γ: discount rate, maxQ (s_next, a_next): action that can be taken in the following states: Maximum value of evaluation (Q) for

The learning coefficient α is a positive value of 1 or less, for example, a value of about 0.1. The discount rate γ is a positive value of 1 or less, for example, a value of about 0.9.

By machine learning using said Formula (1), evaluation NN 455e can be trained so that higher evaluation value Q may be provided with respect to behavior (a) from which higher reward is obtained. In addition, when performing reinforcement learning, you may use the (epsilon) -grid method etc. which make it act randomly by a certain probability, and learns the Q value for various behaviors.

According to the arc end adjustment apparatus 405 comprised in this way, the behavior selection part 455b is based on the position s of the welding system, ie, the positional relationship of the welding torch 11, the welding wire W, and the base material A. Accordingly, it is possible to select a more appropriate action (a), that is, start of anti-stick treatment and start of welding release treatment, so that the cycle time of the welding process can be minimized.

In addition, the reinforcement learning of the arc end adjustment apparatus 405 may be performed using data obtained from a plurality of welding apparatuses or welding systems, data such as behavior data, evaluation values, data obtained from a simulation apparatus, and the like.

According to the arc end adjustment apparatus 405, the welding system, the arc end adjustment method, and the computer program 50d which concerns on Embodiment 4, the arc end procedure by which the cycle time of a welding process is shortened can be deeply reinforced.

In addition, although the deep reinforcement learning was demonstrated in the said Embodiment 4, instead of a neural network, you may arrange so that the evaluation value Q corresponding to an action and a state may be arranged, and the arc end procedure may be adjusted.

Embodiment 5

10 is a schematic diagram illustrating an arc welding system according to a fifth embodiment. The welding system according to Embodiment 5 differs from Embodiments 1 to 5 in that the welding system according to Embodiment 5 includes an adjustment method acceptor 506 that accepts an adjustment method of the arc end procedure. The following mainly describes the difference. Since the other structure and effect are the same as that of embodiment, the same code | symbol is attached | subjected to the corresponding location, and detailed description is abbreviate | omitted.

11 is a schematic diagram showing the arc end adjustment screen 507. The adjustment method acceptor 506 displays the arc end adjustment screen 507 on the terminal, for example, and accepts the adjustment method of the arc end procedure. The arc end adjustment screen 507 has, for example, an adjustment method selector 571 that accepts a selection of automatically adjusting or manually adjusting the arc end procedure. The adjustment method selection part 571 is a radio button, for example. By checking the radio button, the user can choose whether to automatically or manually adjust the arc end procedure.

In addition, the arc end adjustment screen 507 has a priority display unit indicating whether or not the automatic adjustment of the arc end procedure is selected, the priority should be given to the adjustment of the shortest cycle time or the priority given to the welding quality. 572 and a priority adjustment slider 573 for specifying the priority. By sliding the priority adjustment slider 573, the user can specify how much of the welding quality should be adjusted to the arc end procedure, or how short the cycle time should be to adjust the arc end procedure. have.

The adjustment method acceptor 506 that receives the priority related to the welding quality by the priority adjustment slider 573 corresponds to the acceptor that receives the adjustment strength for adjusting the arc end procedure by the procedure adjustment unit 55.

For example, the control device 3 controls the arc end procedure by reducing the adjustment amount of the arc end procedure as the importance of welding quality is higher, and the arc end as the importance of shortening the cycle time is higher. Increase the amount of adjustment in the procedure to perform the arc end treatment.

According to the welding system according to the fifth embodiment, the importance or priority of the shortening of the welding quality or the cycle time can be received, and the arc end procedure can be adjusted in a manner desired by the user.

Embodiment disclosed this time is an illustration in all the points, Comprising: It should be thought that it is not restrictive. It is intended that the scope of the invention be defined by the claims rather than the foregoing, and include all modifications within the meaning and range equivalent to the claims.

1: welding robot
2: welding power
3: control device
4: imaging device
5, 205, 305, 405: arc end adjusting device
11: welding torch
12: wire feeder
21: power supply
22: wire feed control unit
23: shield gas supply
24: detector
50: control unit
50a: input unit
50b: output
50c: memory
50d: computer program
51a: welding monitor data acquisition unit
51b: image data acquisition unit
51c: welding condition data acquisition unit
51d: status data acquisition unit
52a: first acceptance judgment unit
52b: second acceptance judgment unit
52c: anti-stick acceptance determination unit
53a: Acceptance judgment RNN
53b: Acceptance Decision CNN
54, 254, 354: Consolidated comprehensive judgment unit
55, 255, 355, 455: Procedure control unit
56: arc end control unit
57: Shortest Procedure Memory
258: Procedure Coordination NN
258a: welding state recognition network
258b: appearance recognition network unit
258c: Procedure coordination network section
259, 359: learning processing unit
455a: evaluation unit
455b: behavior selector
455c: reward calculation unit
455d: Reinforcement Learning
455e: rating NN
506: Adjustment method receptionist
507: arc end adjustment screen
571: Adjustment method selector
572: priority display unit
573: Priority slider
A: Base material
W: welding wire

Claims (20)

In the arc end adjusting device for adjusting the arc end procedure in a repeated welding process,
An acquisition unit for acquiring welding data indicating a result of the anti-stick treatment and a welding state associated with the next welding step;
A procedure adjustment unit for adjusting the arc end procedure such that the cycle time of the welding process is shortened based on the welding data acquired by the acquisition unit;
End adjustment device to equip. The method of claim 1,
Based on the welding data acquired by the said acquisition part, the acceptance determination part which determines the acceptance of the welding result of an anti-stick process and the next welding process.
Equipped with
The procedure adjustment unit,
The cycle time is shortened when the affirmative determination part judges positive, and when the affirmative judgment part judges negative, the change of the arc end procedure is determined such that the cycle time is extended.
Arc end adjustment device. The method of claim 2,
The procedure adjustment unit,
When the result of shortening the cycle time of the welding process is that the result of the anti-stick treatment and the welding result of the next welding process changes from a good state to a bad state, the arc is confirmed and fixed in the arc end procedure before the cycle time is shortened. Memory end procedure
Arc end adjustment device. The method according to claim 2 or 3,
The acceptance determination unit,
When the welding data indicating the welding state related to the next welding process is input, the affirmative determination neural network in which the neural network has been trained to output data indicating the quality of the welding result related to the welding process when the welding data is obtained.
End adjustment device to equip. The method of claim 1,
The procedure adjustment unit,
A procedure coordination neural network that trains a neural network to output data representing changes in the arc end procedure, which can shorten the cycle time of the welding process when the welding data is input.
End adjustment device to equip. The method of claim 5,
The procedure coordination neural network,
Outputting data indicating the amount of change in the arc end procedure
Arc end adjustment device. The method according to claim 5 or 6,
On the basis of the welding data acquired by the acquisition unit, the acceptance determination unit for determining the acceptance of the welding results of the anti-stick processing and the next welding step,
A learning processing unit that learns the procedural adjustment neural network based on the determination result of the acceptance determination unit obtained after adjusting the arc end procedure.
End adjustment device to equip. The method of claim 7, wherein
The learning processing unit,
Training the procedure-adjusted neural network such that the cycle time is shortened when the affirmative determination unit judges positive, and the cycle time is extended when the affirmative determination unit determines negative.
Arc end adjustment device. The method of claim 8,
The learning processing unit,
If the weld results are in an intermediate state of good and bad, training the procedure control neural network to maintain the cycle time.
Arc end adjustment device. The method according to any one of claims 7 to 9,
The acceptance determination unit,
When the welding data indicating the welding state related to the next welding process is input, the affirmative determination neural network in which the neural network has been trained to output data indicating the quality of the welding result related to the welding process when the welding data is obtained.
End adjustment device to equip. The method of claim 10,
The procedure coordination neural network,
A network configuration that is substantially the same as all or part of the accept decision neural network;
Arc end adjustment device. The method of claim 1,
State data acquisition unit for acquiring state data including image data obtained by photographing a welding torch, a welding wire, and a base metal at multiple time points in the arc end processing.
Equipped with
The procedure adjustment unit,
An evaluation unit for calculating an evaluation value for the action in the state indicated by the state data, based on the state data acquired by the state data acquisition unit, and behavior data indicating an action related to the arc end procedure;
An action selection unit for selecting an action having the maximum evaluation value calculated by the evaluation unit;
End adjustment device to equip. The method of claim 12,
On the basis of the welding data acquired by the acquisition unit, the acceptance determination unit for determining whether the anti-stick treatment and the next welding process,
Compensation for calculating the reward for the arc end procedure based on the determination result of the acceptance judgment portion obtained after adjusting the arc end procedure and the time from the end of welding until the start of welding in the next welding process. A calculation unit,
A reinforcement learning unit that learns the evaluation unit based on the state data acquired by the state data acquisition unit, behavior data indicating an action related to the arc end procedure, and a reward calculated by the reward calculation unit.
End adjustment device to equip. The method according to claim 12 or 13,
The evaluation unit,
An evaluation neural network that outputs an evaluation value for the action in the state indicated by the state data when state data acquired by the state data acquisition unit and action data indicating an action related to the arc end procedure are input;
End adjustment device to equip. The method according to any one of claims 1 to 14,
The welding data indicating the result of the anti-stick treatment,
At least one of data representing a current during the de-welding process, an image of the tip of the welding wire, and a temperature of the tip of the welding wire;
The said welding data which shows the welding state which concerns on the next welding process,
And data representing at least one of a welding current and a welding voltage detected during the welding process, a feeding speed of the welding wire, a short circuit situation, a welding sound collected during the welding process, and an image of the welded portion photographed after the end of the welding.
Arc end adjustment device. The method according to any one of claims 1 to 15,
The arc end procedure,
At least one of a welding voltage and a welding current in an anti-stick treatment, a time of the anti-stick treatment, a retract time of the welding wire, and a time and a retry count of the de-welding treatment.
Arc end adjustment device. The method according to any one of claims 1 to 16,
Receiving unit for accepting the adjustment strength to perform the adjustment of the arc end procedure by the procedure adjustment unit
Equipped with
The procedure adjustment unit,
To adjust the arc end procedure at the adjustment strength received at the
Arc end adjustment device. The arc end adjustment apparatus in any one of Claims 1-17,
A welding robot that holds the welding torch,
A welding power supply for supplying a welding current to the welding torch
Welding system. In the arc end adjustment method of adjusting the arc end procedure in a repeated welding process,
Obtain welding data indicating the result of the anti-stick treatment and the welding state associated with the next welding process,
Based on the acquired welding data, the arc end procedure is adjusted to shorten the cycle time of the welding process.
Arc end adjustment method. In the computer program for adjusting the arc end procedure in the repeated welding process,
To the computer,
Obtain welding data indicating the result of the anti-stick treatment and the welding state associated with the next welding process,
Based on the acquired welding data, the arc end procedure is adjusted to shorten the cycle time of the welding process.
Computer program for executing processing.

Images