JPH05261545A - Control method for welding robot - Google Patents

Abstract

PURPOSE:To perform automatic profile welding by the robot on complicated three-dimensional works by decreasing the welding speed and a welding current and increasing weaving frequency and the locus correction quantity when an actual welding starting point is deviated from a taught starting point. CONSTITUTION:Since the welding starting point P1 taught correctly on the work position is deviated above slantingly the works, there is no fillet weld part and the actual welding starting point comes on a metal plate. When the weaving welding profile is started on the point P1', it approaches in the resultant vector direction in the left direction and in the lower direction, namely, at an angle of zetamax of the fillet weld part so as to eliminate the current difference and maintain the welding current constant with a welding torch in the weaving direction on the metal plate as a reference and normal welding is started from a point P1''. The welding speed and a welding current value are extremely decreased, the weaving frequency is increased, the locus correction quantity is increased and zeta is increased. Accordingly, welding of the robot can be automated for works with inferior accuracy even though the work fitting position is deviated to some extent.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control method in welding line tracing control of an articulated arc welding robot.

[0002]

2. Description of the Related Art Conventionally, a consumable electrode type arc welding machine has been used as shown in FIG.
When the fillet welding as shown in (a) and the V groove welding as shown in FIG. 1 (b) are performed, a dolly equipped with a welding torch 1 that makes a weaving motion is run along the groove line. However, when the carriage does not travel properly along the groove line of the work, the welded portion 2 is unevenly welded. In order to eliminate this problem, when the center of the weaving at the tip of the welding torch is deviated from the welding line, the welding current or voltage at both ends of the weaving is different, so that the welding torch 1 is different from the welding line. , An actuator that moves horizontally in the weaving direction is provided, and the actuator is controlled so that the difference between the detection values at the both ends of the weaving is 0 so that a weld bead that is not displaced from side to side can be obtained. There has been proposed a welding line automatic follow-up copying control device that controls an actuator in a vertical direction (consumable electrode direction) so that it is always constant. This copying method is based on known cylindrical coordinate robots (cylindrical coordinates).
robot), polar coordinate robot (polar coo)
rdinates robot, Cartesian coordinate robot
bot), articulated robot (articulated)
Considering the case of executing the robot, the arc welding robot has a configuration as shown in FIG. 2, for example. This Figure 2
Shown in FIG. 1 is an example of applying the above method to an articulated robot, which is equipped with a weaving device 3, a horizontal drive actuator 4, and a vertical drive actuator 5 on the wrist of the robot. Since a total of three drive sources of the actuator and the weaving device are mounted on the wrist of the robot, its weight and size become a problem. That is, if the work tool becomes heavy, the load on the robot wrist becomes large, which is not preferable from the durability of the wrist, and if the work tool becomes large, it does not go into a narrow place, so the welding place is restricted and versatility is high. There is a drawback that spoils. In addition, the proposed welding line automatic tracking copying control device has the following technology required for a welding robot:
No means is disclosed. In other words, if the welding torch is not set correctly at the welding start point of the work piece due to wrist angle (welding torch posture) control, block switching method (bending angle detection method) of the work piece consisting of multiple blocks, or work misalignment A method for approaching a special welding start point and a method for driving a three-dimensional welding torch between trajectory correction signals at both ends of the weaving which are intermittently given are not shown in this prior art, and a general-purpose three-dimensional work It has the drawback that it cannot be used for welding copying.

[0003]

The present invention is similar to the conventional case where the weaving device 3, the horizontal drive actuator 4 and the vertical drive actuator 5 are not provided on the wrist of the robot as in the conventional case. An object of the present invention is to obtain a control method capable of providing an articulated welding robot that operates.

[0004]

SUMMARY OF THE INVENTION The present invention has been made to achieve the above object, and a robot for performing three-axis operation of a weaving axis, a horizontal axis actuator, and a vertical axis actuator or an operation of the horizontal and vertical axis actuators. The basic three axes or the basic three elements for performing the three-dimensional movement of the robot are substituted, and the weaving center is controlled so as to travel on a previously taught locus or a locus obtained by three-dimensionally parallel shifting the locus, and weaving. Welding current or welding voltage is detected every time it comes to the edge, compared with that at the weaving edge of the previous time, and if there is a difference, it is the direction to eliminate that difference, that is, to the traveling direction of the weaving center (actual welding line). Left and right (weaving direction) and the welding current or welding voltage at the previous weaving end and this weaving end Comparing the mean value with the preset value, if there is a difference, the direction that eliminates the difference, that is, the direction of the composite vector defined by the upper and lower sides of the weaving center (direction orthogonal to the weaving direction) The weaving center is the robot's basic 3 axes or basic 3
By controlling the elements to shift (correction control), the positional deviation between the weaving center and the actual welding line is corrected, and proper weaving welding is performed by making the welding current or welding voltage constant and at the same time complicated 3 Provided is a control method for an articulated welding robot that enables automatic profile welding of a three-dimensional workpiece by a robot.

[0005]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The control method will be described below with reference to the drawings. FIG. 3 shows an example of a multi-joint type teaching / playback robot for carrying out the present invention, and 11 is a consumable electrode welding torch. This welding robot has two arms 1 that bend on a base 12.
The welding torch 11 attached to the tip of the arm 13 is provided with a revolving body 15 including the armatures 3 and 14, and the swinging angle and the twisting angle can be changed by a drive motor. And 14 and the revolving structure 15 are controlled. As is well known, when a teaching playback robot teaches two points with a teach box (not shown), the welding torch 11 moves on a straight line connecting the two points by linear interpolation. Therefore,
In the case of tracing a work welding line R having a curve as shown in FIG. 4, proper welding cannot be performed without giving a large number of teaching points to a robot having no trajectory correcting function. Therefore, in the present invention, roughly, P ₁ ,
Only by giving 3 teaching points of P ₂ and P ₃ , the work welding line R
It is intended to provide a control method for tracking. This practical value simplifies the teaching work before the start of welding, facilitates the handling of the robot, and enables proper welding to be performed even when the accuracy of the work is poor.

First, the principle will be described. In order for the tip of the welding torch 11 of the robot to move along the locus of the workpiece on the welding line R, the weaving motion for detecting the deviation amount is continued, and the weaving center of the welding torch is always held on the welding line. You can do this. Then, in order for the weaving center to follow the welding line R, the weaving center is moved in a direction corresponding to a trajectory correction signal at both ends of the weaving provided by a known sensor, and then,

[0007]

[Outer 1]

These three movements, that is, weaving movement, correction of the locus of the weaving center, and movement of the weaving center on the locus of the teaching locus three-dimensionally parallel-shifted are executed by the robot basic three axes or basic three elements, and weaving is performed. When weaving to the left and right with respect to the weld line by the same distance from the center, when the trajectory correction signals at both ends are given, the weaving center is moved in that direction,
After the movement is completed, the weaving center moves on the new trajectory that is three-dimensionally parallel-shifted, and when it is not given, the weaving center moves on the original trajectory and repeats the operation with the next trajectory correction signal to reach P ₂ ″. ..

[0009]

[Outside 2]

If the tracking operation is performed, the tracking control of the work welding line R is executed. To detect the inflection point P ₂ ″ for switching the parallel shift direction, the bending angle detection monitoring point P ₂₀ is set on the complementary angle side of ∠P ₁ P ₂ P ₃ .
The distance m between the tip of the welding torch and the weaving center is constantly calculated, and P ₂ ″ can be obtained by detecting a point that has become larger than the minimum value by a certain fluctuation range. Note that P ₂₀ is ∠P ₁ P ₂ P It is desirable to set on the bisector of _{3. If} tracking control is performed in this way, compared to the conventional method,
Appropriate weaving welding is performed with extremely rough teaching, which is extremely large in contributing to the driving operation of the welding robot. Next, a weaving motion by the robot's three basic axes or three basic elements will be described with reference to FIG. Prior to robot operation, it is first necessary to teach the conditions necessary for copy welding, such as the weaving direction, amplitude and frequency. Q between the teaching points P ₁ and P _2
The weaving direction and amplitude are set by arbitrarily teaching the three points ₁ , Q ₂ , and Q ₃ . Since the weaving surface with respect to the welding line cannot be determined only by the two points Q ₁ and Q ₂ , the point Q ₃ is set.

[0011]

[Outside 3]

[0012]

[Equation 1]

Since P ₁ , P ₂ , Q ₁ , Q ₂ , and Q ₃ are the taught points, and their coordinate values are known, α, β,
γ, α ', β', γ ', λ, μ, ν or x _W ,
y _W and z _W are easily obtained by the arithmetic device. Although the weaving center is H, the intersection W is tentatively referred to as the weaving center in the present description for the sake of easy understanding. This weaving center W is P in FIG.
Starting from ₁ , the point P ₂ 'is reached while performing the copying operation. Welding torch tip attached to the wrist of the robot

[0014]

[Outside 4]

The reference clock here is a signal generally called a servo clock, which is an output signal of the frequency dividing / rising differentiation circuit 37 shown in FIG. Although the details will be described later in the description regarding FIG. 18, the BRM (Binary R) in FIG.
The output of the aerated multi-layer), that is, the position servo command as a pulse train signal to the position servo of each axis forming the robot is the clock φ and the BRM shown in FIG.
A clock having a cycle in which payout is completed in a time determined by the number of bits. The cycle is called the reference clock cycle. Within this reference clock cycle, the next position servo command data for each axis, that is, the buffer 4 shown in FIG.
The input data to 1 is calculated by the calculator 31. The operation clock period is usually set to several ms, and the movement amount of each axis of the robot is incrementally calculated for each reference clock period to perform a fine-textured weaving motion. In carrying out the weaving exercise,
Initially, this K is set to K = 1, and K = K every reference clock cycle.
K is incremented to a value of +1. Here, if the reference clock period is C ₀ , the weaving frequency is h, and half of the weaving amplitude is j,

[0016]

[Equation 2]

[0017]

[Outside 5]

[0018]

[Equation 3]

It is calculated by The left and right direction cosines are α, β and γ. FIG. 6 shows eight left, right, up, and down correction vectors.

[0020]

[Outside 6]

[0021]

[Equation 4]

It becomes The traveling direction of the weaving center without modification of W is parallel to P ₁ P ₂ , and its direction cosine is (λ, μ, ν). FIG. 7 is a diagram showing that the weaving center W is corrected by moving by γ in the direction of (λ, μ, ν). The x, y, z components δx, δy, δz of the combined actual correction vector δ of W are obtained by the following equation.

[0023]

[Equation 5]

In order to realize the specified welding speed V, the movement amount for each reference clock in the parallel movement mode is

[0025]

[Equation 6]

[0026]

[Outside 7]

[0027]

[Equation 7]

[0028]

[Outside 8] In the correction mode in which the vector δ in FIG. 7 is followed, the r direction is controlled at the specified speed V, so W moves by δ at the time of r / V. Therefore, the movement amounts X _n1 , y _n1 , and Z _n1 in the δ direction per one reference clock are

[0029]

[Equation 8]

[0030]

[Outside 9] When δ _x1 , δ _y1 , and δ _z1 are at the end of the correction mode, δ x,
The difference is corrected in the final round of the correction mode so as to match δy and δz.

[0031]

[Equation 9]

Is calculated for each reference clock, the minimum value is stored, the minimum value is updated as W approaches P ₂ , and W is P
_{After the 2} "point, D becomes larger than the minimum value. P _{1 of} W
The average distance D decreases as P ₂ ″ progresses, but D does not necessarily decrease in a microscopic sense depending on how the correction vector is output in the correction mode. The minimum value has a small fluctuation range, and the point beyond that is the end point of the block.
This bend angle recognition method is effective for detecting a gentle bend angle as shown in FIG. 4, but cannot be applied to a bend angle such as a steep right-angled corner. Therefore, it can be used together with the turning angle recognition method described in FIGS. Next, the control of the robot wrist axis will be described. The wrist posture must be controlled so that the torch angle and the advancing angle along the welding line are within a certain variation range determined by welding. Teaching points P ₁ , P ₂ , P
The wrist postures at the _three points are naturally taught correctly. In the case of welding copying to P ₁ P ₂ , the values at the points P ₁ and P _{2 of} the wrist swing axis B and the twist axis T are B ₁ , T ₁ , B ₂ , T ₂
As a result, the increments ΔB and ΔT per reference clock C ₀ are obtained by the following equation.

[0033]

[Equation 10]

B and T at the end point P ₂ ″ of the block are B ₂ ,
In general, it is not equal to T ₂ , but in FIG. 4, the deviation of the work is only straddled and the distance between the P ₂ and P ₂ ″ points in the actual target work is small, so this difference is Therefore, the welding torch posture is always maintained during welding, because the X, Y, and Z values in equation (1) are the values at the tip of the welding torch, which is the control point. 1)
In order to obtain the formula, regardless of whether the robot has an orthogonal shape, an articulated shape, a cylindrical shape, or a polar coordinate shape, the drive axes of P ₁ , P ₂ , Q ₁ , Q ₂ , and Q ₃ stored in the memory Conversion from coordinate data to Cartesian coordinates is required. Also, equation (17),
The position (X _n , Y _n , Z _n , B _n , _Bn , _where the welding torch should be, which is calculated every moment (every reference clock) by the equation (20),
In order to realize T _n ) as a robot, the solutions X _n , Y _n , and Z _n in Equation (17) must be inversely converted into drive axis data of the robot basic three axes. Robot wrist axis data is B
_n and T _n . Basic 3 of the embodiment robot shown in FIG.
The axes, that is, the three-element swing body 15 arms 13 and 14 are controlled by rotation angles φ, ψ, and θ, respectively, and the wrist swing axis and the twist axis are controlled by rotation angles B and T, respectively. The control point P is the tip of the welding torch 11. Distance A from the center of rotation of the wrist swing axis,
Since the point P, which is a distance d from the center of rotation of the twist axis, is a taught control point, its orthogonal coordinate value is

[0035]

[Equation 11]

The equations (21) and (22) are defined according to each robot form. FIG. 8 shows a sensor circuit section for generating a trajectory correction signal for moving a predetermined amount q upward, downward, leftward or rightward from the welding voltage or welding current at both ends of the weaving. Q ₁ explained in FIG.
When Q ₂ and Q ₃ are given by teaching, foot H of the perpendicular line
The coordinates Motomari, it _{x h,} y h, when a _{z h,} the current value _x k of the _{weaving, y k, Z k ((} 6) type) current Fuhaba j 'is determined from.

[0037]

[Equation 12]

When j'is j, it is the weaving end, and at this time, a measurement start command is given to the sensor circuit section. A "left" signal is generated when the weaving center of the welding torch is offset from the actual weld line to the right with respect to the weld line direction. If left, a "right" signal is generated. An "up" signal is generated when the average value of both ends of the weaving is larger than a preset value, and a "down" signal is generated when the average value is smaller than the preset value. This up / down / left / right signal is 8
There are street combinations (see Figure 6). Fig. 9 shows the up, down, left, and right directions. The left and right are weaving directions,
The upper and lower sides are the welding torch direction, that is, the direction in which the wire electrode 11a emerges. FIG. 10 is a diagram for explaining automatic continuation of the weaving teaching pattern. In the figure, P ₁ , P ₂ , and P ₃ are teaching points, Q ₁ and Q, as in FIG.
₂ ', _{Q 3} means exactly the same thing as FIG.

[0039]

[Outside 10]

[0040]

[Equation 13]

[0041]

[Outside 11]

Weaving is automatically continued in an appropriate direction while returning. Even if there is an air-cut block that does not weave in the middle, the above calculation is continued in the calculation, so the weaving point Q ₁ ,
Teaching of Q ₂ and Q ₃ is unnecessary. In other words,
Teaching of Q ₁ , Q ₂ and Q ₃ can be done only once. Within the same block between the teaching points P ₁ and P ₂ , there is a restriction on the welding contour between the trajectory of teaching and the actual work welding line deviation angle ξ (see FIGS. 11 and 12). That is, weaving frequency h (Hz), welding concentration V (mm
/ Min), the locus correction amount q (mm) shown in FIG. 6, or √2q (mm) determines the allowable maximum value ζ _max of ζ.

[0043]

[Equation 14]

If it is less than ζ _max , welding copying can be performed in the same block, and it is not necessary to consider it as a bending angle. However, here, ζ _max / 2 is set as an allowable angle within the same block. Now h = 2Hz, q = 0.7mm, V
= 300 mm / min, ζ _max ≈ 30 ° and ζ
It can be said that the use of ζ _max / 2 is appropriate because there is generally no workpiece to be welded with _max / 2 ≈15 ° and a work variation of 15 ° even in view of welding automation. FIG. 13 shows the behavior of profile welding at a steep right-angled corner portion in fillet welding.

[0045]

[Outside 12]

[0046]

[Equation 15]

The average value of ξ and η is calculated each time before and after the welding copying correction mode, and is sequentially stored in the FIFO.
When the average value in the IFO approaches ζ _max with ζ _max / 2 or more, the point P ₂ ″ in FIG. 13 is recognized as a turning angle. As described with reference to FIG. 10, the weaving direction is switched to P ₂ P. ₃ the direction perpendicular to the .P 2 _"was used as a starting point P _{2 P} 2" profiling welding of the next block starts the _{'displacement} vector by P ₃ P ₃ point obtained by shifting _the' as calculated in a straight line to the end point.

[0048]

[Outside 13]

The average value of ξ and η after the recognition of the turning angle is also ζ _max
Therefore, an interlock is necessary to prevent the recognition of the corner. In this case P ₂ '→ P ₂ "→
But it will be an over-turn so that P 3 _"', this over-turn is not necessarily bad. In other words, originally welding of sharp corners difficult, in order to eliminate an undercut, by the robot without mimic In the case of teaching with an acute angle, there are many examples of bending outside the welding line as shown in Fig. 15. Fig. 15 is a diagram showing the teaching position and the welding torch posture in fillet welding as seen from above and from the side. In fillet welding that uses welding current, the aim position of the welding torch is usually shifted to the flange side by t in order to prevent undercut on the web side, and since the amount of meat increases especially at the corners, the web side Since the undercut of the welding torch is likely to occur, the aim position of the welding torch is made larger than t at the corner as shown in the figure. It can be said that in 識法 there is validity.
Further, the offset amount of t shown in FIG. 15 can naturally be obtained by copying the main welding by biasing the left and right signal comparison circuit (not shown) of the sensor circuit unit and offsetting the zero point.
16 and 17 are explanatory views of the welding start point search function when the actual welding start point, which often occurs in mass production, deviates significantly from the point originally taught as the welding start point. In FIG. 16, the welding start point P _{1 that} is correctly taught at the work position indicated by the dotted line is
Since the work is the actual welding starting point is no fillet for displaced obliquely upward from a dotted line to the solid line position is on a flat plate, it has been described above flat when starting the copying weaving welding at the point P _'1 in FIG. 9 Welding torch 11 in the weaving direction (left-right direction)

[0050]

[Outside 14]

With reference to the above, the current difference is eliminated, and in order to make the welding current constant, the combined vector direction in the left direction and the downward direction, that is, toward the fillet, approaches at an angle of ζ _{max, and} becomes normal from the point P ₁ ″. Welding is started in order to minimize the length m ₁ and to prevent the unnecessary welding ped from P ₁ '→ P ₁ ″ to adversely affect the work piece.
During P ₁ ′ → P ₁ ″, welding copying is performed under different weaving conditions and welding speeds from the actual welding line (after P ₁ ″). That is, the welding speed V and the welding current value are extremely decreased, the weaving frequency h is increased, the locus correction amount q is increased, and ζ _{max in the} equations (26) and (27) is increased. To recognize the P ₁ ″ point, the method described with reference to FIGS. 13 and 14 may be applied mutatis mutandis. FIG. 17 is a diagram when the work is reversed, and is processed in exactly the same way as in FIG. Is a main CPU (microprocessor) for supervising the operation of the entire robot and a welding profile controller composed of a computing unit and a BRM with a two-stage buffer in one embodiment of the present invention.
FIG. 9 is a block diagram showing a connection between the welding current command changeover switch and the sensor circuit unit described in FIG. 8. In the case of performing welding copying from the entire robot operation sequence in response to a command request from a computing unit, the main CPU 20
Is the start point P ₁ and end point P ₂ of the block to be copied specified by the teach box (not shown), the end point P ₃ of the block following the block to be copied, the weaving pattern defining points Q ₁ , Q ₂ , Q ₃ and the distance monitoring point P. _The number of pulses from the origin of the ₂₀ robot basic 3 axes and the robot wrist axis, the welding speed V, the weaving frequency h, and the trajectory correction amount q are read from the memory (not shown), and the registers 21 to 3 are read.
It has a function of setting to 0 and outputting a welding copying start macro command. The arithmetic unit 31 includes a sequence controller 32, a micro program memory 33, a pipeline register 34, a multiplexer 35, and a RALU (Re
gister and Arithmetic Log
ical unit) 36, register 21 to register 3
0, frequency division of clock φ, rising differential circuit 37, and reference clock / address generator 38. The sequence controller 32 is an address controller that controls the execution sequence of microinstructions stored in the microprogram memory 33.
Various addressing and stack control are performed by a control instruction from the pipeline register 34. More specifically, the pipeline at the time of the address jump being executed, the address selection specified by the macro command, the address selection specified by the reference clock address generator 38, and the conditional jump according to the test condition including RALU status. This is a part for processing a jump address given from the register 34, a jump address given from the pipeline register 34 in the case of an unconditional jump, a stack control at the time of a micro subroutine call, and the like.

There are three types of input information for addressing: a macro command from the CPU, an output of the reference clock address generator 38, and an output of the pipeline register 34. The microprogram, that is, the control instruction of the pipeline register 34 determines which of the three, or the current controller increments the current address by the sequence controller 32. There are the following four macro commands for welding profile control. It is an arc start block immediately after air cut,
Moreover, the turning angle with the next block is recognized by the distance monitoring method. It is an arc start block immediately after air cut,
Moreover, the bending angle with the next block is recognized by the (ξ + η) / 2 angle calculation method. The arc start has already been issued in the previous block and the welding is continuously performed. The bend angle with the next block is recognized by the distance monitoring method. The arc start has already been issued in the previous block, and the welding angle is recognized continuously with the next block in the (ξ + η) / 2 angle calculation method. From the viewpoint of hardware, the output of the macro command and the reference clock address generator 38 is given in a form indicating the start address of each processing microprogram. When these are not given to the sequence controller, the jump address, the subroutine call address and the current address increment from the output of the pipeline register 34 are given. The micro program memory 33 is the central part of the arithmetic unit 31, and all arithmetic processing is executed according to the instructions of this micro program. The pipeline register 34 outputs to the RALU 36 the operation microinstruction currently to be executed by the buffer register of the microprogram memory 33, and outputs the control instruction for the next microaddress determination to the sequence controller 32 and the multiplexer 35.
To the sequence controller 32, and when a bend angle is detected in the block in which welding copying is being executed, the main command request is to request the coordinate data of the next block. Output to the CPU 20.

This pipeline register 34 is for forming two signal paths, allowing them to proceed in parallel at the same time, shortening the micro cycle time, and speeding up the operation. That is, one path is a control system path connected to the pipeline register 34 → sequence controller 32 → microprogram memory 33, and another path is a pipeline system 34 → RALU 36 operation path. To operate in parallel during the same clock cycle, pipeline register 34
Is prepared. At the rise of the clock CP, the next instruction of the microprogram already prepared in the control system path appears at the input of the pipeline register 34, so that a high-speed operation equivalent to zero memory fetch time becomes possible. The multiplexer 35 gives the RALU status and the eight kinds of test conditions of the locus correction signal from the sensor unit shown in FIG. 6 to the sequence controller 32 according to the control command of the pipeline register 34, and makes a conditional jump to each processing program. It is meant to be executed. The RALU 36 is composed of a logic / arithmetic operation unit and a programmable register, and executes operation instructions defined by a microprogram. The number of increment pulses of each drive axis of the robot for each reference clock, which is the calculation result, is stored in a predetermined register in the RALU 36. Register 21 has start point P ₁ , register 22 has end point P ₂ , register 23 has P ₃ point, register 24 has Q ₁ point,
The register 25 is a Q ₂ point, the register 26 is a Q ₃ point, and the register 27 is a P ₂₀ point. The basic 3 axes of the robot and the number of pulses from the origin of the robot wrist axis are stored in the register. The register 28 is a register that stores the profiling welding speed V, the register 29 stores the weaving frequency h, and the register 30 stores the trajectory correction amount q. The buffer 41 is a register for storing the number of incremental pulses to be paid out at the next reference clock of each drive axis of the robot, and the buffer 42 is a register for storing the number of incremental pulses currently being paid out. BRM is buffer 4
The number of pulses stored in 2 within the reference clock period,
Distribute evenly and uniformly as a pulse synchronized with clock φ,
A payout completion signal for buffer transfer is output for each reference clock.

The frequency dividing / rising differentiation circuit is a circuit for generating a reference clock divided by the number of BRM bits from the clock φ and differentiating its rising, and is for synchronizing the BRM payout completion signal and the reference clock. .. The welding current command to the welding machine and the sensor circuit section 50 is sent to the main CPU 2
Although it is given from 0, when the welding start point is detected, the welding current command changeover switch 51 is switched by the output of the calculator, and a fixed value is selected, and when the welding start point is detected, it is switched to the main CPU command value side by the output of the calculator. Change FIG. 19 shows FIG.
The basic 3 axes (φ axis, θ axis, of the articulated robot controlled by the command pulse issued from the BRM described in 8
Position servo of two axes (B axis, T axis) and wrist axis.
The difference between each axis command pulse given to make the welding profile while the tip of the welding torch makes a weaving motion and the feedback pulse from each pulse generator 616, 626, 636, 646, 656 is the deviation counter 611, 62.
1, 631, 641, 651 outputs analog speed commands via D / A converters 612, 622, 632, 642, 652 and servo amplifiers 613, 623, 63.
3, 643 and 653, respectively. The servo amplifier uses the speed command and tachogenerator 615, 625, 63.
The outputs (detection speeds) of 5, 645 and 655 are compared with each other, and each drive motor is controlled so that the difference is eliminated.
With this position servo system, the tip of the welding torch attached to the robot follows the command pulse to perform the desired welding copying operation. Next, the welding copying control operation by the control circuit shown in FIG. 18 will be described. The calculator 31 first executes a wait routine. When a macro command is given from the main CPU 20, the sequence controller 32 controls the address so that the arithmetic unit 31 executes a wait routine while being given a control instruction for selecting the start address of the service program from the pipeline register 34. is doing. Since the wait routine includes an instruction to reset the reference clock frequency dividing circuit 37, the reference clock is not generated.

[0055]

[Outside 15]

When the main CPU 20 outputs one of the four macro commands, the head address of the corresponding service microprogram is selected. These four service programs set a welding start point detection flag (not shown) to 1 when the block includes the welding start point detection and reset it to 0 when the block does not include the welding start point detection. If the method is a method, a bend angle detection flag (not shown) is set to 1, and if the method is a (ξ + η) / 2 calculation method, the bend angle detection flag is reset to 0.
It is a program to enter #. Further, since the four types of programs include an instruction to release the reset signal of the reference clock frequency dividing circuit 37, the frequency dividing circuit 37 starts counting the clock φ. After that, the sequence controller 32 operates so as to perform addressing necessary for arithmetic processing until the next output from the reference clock address generator 38 comes. Next, let us explain the welding profile processing program starting from N #. Since welding copying is started from the block at the time of arc start, the welding start point flag is set to 1 at first. Therefore, the welding speed V, the weaving frequency h, and the locus correction amount q are set to the register 28 to the register 30 until the welding copying operation reaches the regular welding start point.
Use the value stored as a fixed value in the calculator instead of using the value set in. Further, the welding current command changeover switch 51 is switched by the output of the calculator so that the welding current command given to the welding machine and the sensor circuit section 50 becomes a fixed value.

[0057]

[Outside 16]

Next, the equations (4) and (5) are solved using h stored as a fixed value in the arithmetic unit. Further, the equation (11) is solved with the fixed value V. Further, the equations (2) and (3) are solved, and then the equation (7) is solved. Equation (19) is solved from the wrist axis angles of P ₁ and P ₂ . It is assumed that the head address where the processing programs thereafter are stored is M #. At the first reference clock for starting the welding contour, the trajectory correction signal is not generated because the welding torch is still located on the center of the weaving. Therefore, at this point in time, expressions (8) to (10) are
Equation (13) and equation (14) are not involved, and arithmetic unit 31 is in the parallel movement mode. N for the first reference clock
Set = 1 and solve equation (6). Since K '= 0, K = 1 from the equation (16). Expression (12), (15)
X _n , Y _n , and Z _n of the equation (17) are obtained from the equation and the equation (7). _Bn and _Tn are calculated from the equation (20).

[0059]

[Outside 17]

When the turning angle detection flag is 1, (1
The formula 8) is calculated, the value is set in a predetermined register in the RALU, and the calculator 31 waits. This wait is the main CP
Unlike the wait routine that waits for a macro command from U20, the next reference clock address generator output M
This routine waits for #, and the frequency divider circuit 37 is not reset. Since the clock φ and the number of BRM bits are set so that the reference clock cycle becomes longer than the time until the above calculation is completed, the problem that the calculation time is not in time does not occur. When the arithmetic unit 31 is executing the wait routine waiting for M #, the reference clock is generated and at the same time the payout completion signal is output from the BRM.
.DELTA..phi., .DELTA..theta., .DELTA..phi., .DELTA.B, and .DELTA.T are loaded from the predetermined registers in the RALU 36 into the respective buffers 41, and the contents of the buffer 41 are loaded into the buffer 42. At the first reference clock, the buffer 41 remains cleared, so 0 is entered in the buffer 42 and BR
No pulse payout by M is performed. This reference clock causes the reference clock address generator 38 to operate, and the microprogram starts to execute from M # N =
Then, the equation (6) is obtained by setting the value to 2. Since j'in the equation (23) at the time of the last reference clock is not j, the trajectory correction signal is not generated and the arithmetic unit continues the parallel movement mode. Therefore, K '= 0, and K = 2 from the equation (16). Therefore, from equations (12) and (15), X _n in equation (17),
Y _n and Z _n are calculated. Further, B _n and T _n in the equation (20) are obtained, and the differences ΔB and ΔT from the previous time are set in a predetermined register in the RALU. Substituting these X _n , Y _n , Z _n and B _n , T _n into the equation (22) to obtain ψ _n , θ _n , and φ _n , and calculating the differences Δψ, Δθ, and Δφ from the previous time in predetermined registers in the RALU. Set to. Also, the equation (18) is calculated, and the value is RALU.
Confirm that the value is smaller than the previous value stored in and replace the value.

Thereafter, the arithmetic unit waits and waits for the next reference clock address generator output M #. At the next reference clock, the contents of the previous buffer 1 are stored in the buffer 42,
The buffer 41 has the current Δφ, Δθ, Δφ, ΔT, and ΔB.
Are loaded, and the command pulse is issued to each axis position servo by BRM. This operation is repeated until j'in Expression (23) becomes j. At the time of j ′ = j, x _n ,
y _n and z _n are stored, a measurement start command is given to the sensor circuit unit 50, and the calculator 31 enters the locus correction mode at the next reference clock. That is, the vertical and horizontal locus correction signals from the sensor circuit unit are received as test conditions in M #, and α, β, γ, λ, μ, ν already calculated and q stored as a fixed value in the calculator. With a fixed value γ (8)
Equations (10) and (13) are solved. The equation (14) is solved with K ′ = 1, and the weaving centers x _n , y _n , and z _n obtained by solving the equation (15) with K being the previous value are x ′.
_n, y _'n, z' stored as _n. N = K + K '
Toki (6) is solved. Solution of equation (17) X _n , Y _n , Z
_{Find n} . B _n and T _n are calculated, and Δφ, Δθ, Δφ and Δ
Converted to T, ΔB, set in a predetermined register in RALU, and position servo controlled by BRM. Also (18)
Calculate the expression and update its value. Next, the computing unit enters the parallel movement mode again, and at each reference clock, ψ _n , θ _n ,
φ _n , T _n , B _n are calculated and the difference from the previous reference clock Δ
Position servo control is performed using ψ, Δθ, Δφ, ΔT, and ΔB. Equation (18) is also calculated each time and the minimum value is saved.

[0062]

[Outside 18]

Further, the x _n , y _n , and z _n obtained this time are replaced with the previous values and stored. The measurement start command is sent to the sensor circuit unit 50.
To the locus correction mode at the next reference clock. The equations (8), (10), and (13) are solved according to the trajectory correction signal from the sensor circuit unit 50, and K ′
= 2 and the equation (14) is solved, and the weaving centers x _n , y _n , and z _n obtained by solving the equation (15) with K being the value at the time of the previous reference clock are x ′ _n , y ′ _n. , Z ′ _n . The _{x 'n, y' n,} z 'n and the previous memorized _{x' n, y 'n,} z' n (x 'n-1, y' n-1,
direction cosine (a ', of a straight line connecting z'n _-1 )
b ', c' _{and) x 'n, y' n} , z 'n x' n-1,
y determined from _{'n-1, z'} coordinate value of _n-1, already determined that the direction cosines (λ, μ, ν) and (λ ', μ', ν ') by (29) calculates a η of Formula To do. The value of η is stored in the FIFO memory. In addition, x _n ', y _n ',
Z _n 'is replaced with the previous value and stored. Solve equation (6) with N = K + K '. The solutions X _n , Y _n , and Z _n of the equation (17) are obtained. B _n , T _n are calculated and Δφ, Δθ, Δφ, ΔT, Δ
Convert it to B and set it in the prescribed register in RALU, then BR
Position servo control by M. Also, the equation (18) is calculated and the value is updated. This ends the correction mode and repeats the parallel movement mode ⇒ correction mode ⇒ parallel movement mode. The average value of (ξ + η) / 2 stored in the FIFO memory before and after this correction mode is ζ _{max when it} is 1/2 or more of ζ _max calculated by the equation (26) or (27) (27). Whether or not it is approaching is calculated, and when it is a value close to ζ _max , it is recognized that the welding start point is being detected, V, h, q are fixed values as they are, and the welding current command changeover switch 51 is also set to the fixed value side. Continue the welding copying as it is. Then, when the average value of (ξ + η) / 2 becomes ζ _max / 2 or less, it is determined that the welding start point has been reached, and the welding current command changeover switch 51 selects the welding current command value from the main CPU 20. Control output, V, h,
Change the value of q to the value of Register 28 to Register 30,
Weaving per reference clock by solving equations (4) and (5)

[0064]

[Outside 19]

Thereafter, when the bending angle detection flag is 1, the minimum value of the equation (18) is updated, and when the bending angle detection flag is 0, ξ and η are sequentially stored in the FIFO and the above welding copying operation is continued. .. When the bend angle detection flag is 1, the end point of the block is set when the minimum value of the equation (18) exceeds a certain fluctuation range. When the bend angle detection flag is 0, the end point of the block is set when the average value of (ξ + η) / 2 in the FIFO memory is ζ _max / 2 or more and approaches ζ _max . After the detection of the turning angle, the calculator 31 determines the weaving direction in the next block, and sets the weaving pattern so that the weaving amplitude becomes equal to the weaving amplitude in the previous block.

[0066]

[Outside 20]

After that, in response to a command request from the arithmetic unit 31, the main CPU 20 sets the number of pulses from the origin of the robot 5 axis at the end point of the next block in the register 22, and the distance between the end point of the next block and the next block. The number of pulses from the origin of the robot 5 axis at the monitoring point
3. After setting in the register 27, the macro command is given to the arithmetic unit 31. This macro command does not include welding start point detection.

[0068]

[Outside 21]

[Advantages of the Invention] As described above, the welding copying system according to the present embodiment enables automation of welding of a work having poor accuracy and has the following advantages. (1) Nothing other than the welding torch is attached to the robot wrist, so welding can be performed in any narrow place where the welding torch can enter. (2) Even if the work accuracy is poor and the actual work welding point deviates greatly from the taught point, the welding start point can be detected without an expensive shape recognition device such as a vision system. (3) Even without a shape recognition device, it is possible to smoothly perform proper welding at any bending angle of any shape while following welding. Also, proper welding is possible even if the bending angle varies widely depending on the work. (4) Work welding lines having a welding profile angle equal to or less than the welding profile angle determined by the set values of the weaving frequency, the welding speed, and the locus correction amount may be included in the same block, so that the initial teaching work is easy and the operability is good. (5) Since the weaving pattern automatic continuation function is provided, it is only necessary to teach the three points that define the weaving pattern only in the first welding block, so the teaching workability is good. (6) Since the sensing information is obtained from the arc phenomenon (welding current, welding voltage) itself, rather than attaching a special sensor to the wrist of the robot, it is usually a problem with other sensors regardless of whether it is a non-contact sensor or a contact sensor. There are no drawbacks such as "the sensor gets in the way of welding", "there is a blind spot", and "there is a weak point in reliability under bad environment due to welding heat, spatter, fume, etc.". Therefore, according to the present invention, it is possible to automate welding by a robot even if the work has poor accuracy, the work mounting position is slightly displaced, or the welding line is wavy.
It can be said that it greatly contributes to the automation of welding.

[Brief description of drawings]

FIG. 1 is an explanatory diagram of weaving welding.

FIG. 2 is a perspective view of an arc welding robot to which a conventional weaving welding method is applied.

FIG. 3 is a perspective view of an arc welding robot for carrying out the present invention.

FIG. 4 is an explanatory diagram of a copying control method according to the present invention.

FIG. 5 is an explanatory diagram of teaching points for setting a weaving motion.

FIG. 6 is a correction vector diagram.

FIG. 7 is a modified vector diagram.

FIG. 8 is an explanatory diagram of input / output signals of the sensor circuit unit.

FIG. 9 is a view showing a control direction of a welding torch.

FIG. 10 is a diagram for explaining automatic continuation of a weaving teaching pattern.

FIG. 11 is a diagram for explaining a deviation angle between a teaching locus and an actual welding line.

FIG. 12 is a diagram for explaining a deviation angle between a teaching locus and an actual welding line.

FIG. 13 is an explanatory diagram showing the behavior of profile welding at a steep right-angled corner portion in fillet welding.

14 is a partially enlarged explanatory view of FIG.

FIG. 15 is a diagram showing teaching points and welding torch postures in fillet welding.

FIG. 16 is an operation explanatory diagram when the actual welding start point is different from the teaching point.

FIG. 17 is an operation explanatory diagram when the actual welding start point is different from the teaching point.

FIG. 18 is a block diagram of a control circuit in the example.

FIG. 19 is a block diagram of a servo control unit in the example.

[Explanation of symbols]

 11 Welding torch 12 Base 13 Arm 14 Arm 15 Revolving structure

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shigemi Nobayashi 2346 Fujita, Yawatanishi-ku, Kitakyushu, Fukuoka Prefecture Yasukawa Electric Co., Ltd.

Claims (1)

[Claims] 1. A multi-joint welding robot that weaves a welding torch tip and moves along a teaching line, wherein a welding current or a welding voltage at both ends of the weaving is detected, and the object to be welded is detected by a change in the detected value. In the articulated welding robot, which detects the actual deviation of the welding line and corrects the movement locus of the welding torch tip along the welding line by the locus correction signal calculated from the detected value, Robot basic 3 axes or basic 3
By controlling the elements, the weaving operation of the tip of the welding torch and the correction operation of the locus of the weaving center are performed, and when the actual welding start point deviates from the welding start point already taught, the actual welding line A method for controlling an articulated welding robot, characterized in that the welding speed and the welding current until reaching the limit are extremely reduced from the normal values and the weaving frequency and the trajectory correction amount are made larger than the normal values.