KR20110032851A - Method of recognizing shape of welding zone, method of generating welding path adopting the same, and automatic welding robot adopting the same methods - Google Patents

Abstract

PURPOSE: A method for recognizing welding part shape, weaving welding path generating method applied with the same, and automatic welding robot applied with the both above are provided to weld using an automatic welding robot by forming a welding path in which the shape of welding area is simplified. CONSTITUTION: An automatic welding robot comprises a main body, a sensor part(150), a welding part(160), a conveying part and a control part. The sensor part is arranged towards the welding part of the main body. The welding part is connected to the main body and includes a welding rod(155) to weld the welding area. The conveying part is connected to the main body and transfers the sensor part and welding part. The control part controls the sensor part, welding part and conveying part.

Description

Method of recognizing shape of welding zone, method of generating welding path adopting the same, and automatic welding robot adopting the same methods}

The present invention relates to a welding robot that performs butt welding of a longi and a longage in a double hull, and more particularly, to a welding site shape for automating butt welding of a T bar shape. The present invention relates to a recognition method, a weaving welding path generation method using the same, and an automatic welding robot using the same.

In general, welding works such as hull blocks or steel structures are carried out in a narrow working space and a harsh working environment in which harmful gases are generated. Therefore, various automation devices are devised to protect workers in the production site and to improve welding quality and productivity. It is applied to the production site.

The shipyard docking process consists of three stages: setting, mounting and welding. Setting is the process of moving a block to the dock and adjusting the block position to minimize the step with the block previously placed in the dock. At the time of setting, it is fixed by welding (hereafter attaching) to the part where a block and abutment contact. After setting, the first stage welding is performed by minimizing the step difference between the longages of the parts in contact with each other, so that the longage and the longage are fixed. After the mounting operation, the mounting process is completed by performing butt welding to weld the portions where the ronze abuts each other and the portions where the blocks abut each other.

1 is a photograph of a block being mounted at the shipyard. The red line in the picture and the inside of the longage are the parts that need butt welding when mounted.

After setting, the operator must pass through a narrow access hole as shown in FIG. 2 to enter the inside of the double hull. The interior of the double hull through the access passages is blocked by steel plates in all directions except the access passages. Also, as shown in FIG. 1, the spacing between side-by-side longines and longines is narrow. Workers are standing or sitting between side-by-side and longines to weld. As described above, the mounting is performed in a poor environment such as a closed space, an uncomfortable posture, and so far, a welding robot that performs welding in the double hull during the mounting process has not been developed. The reason is that it is difficult to develop a robot that is small and light enough to pass through the access passage in FIG. 2 and that can perform welding along an irregular weld improvement surface.

In particular, the present invention has been derived in the course of studying a robot and a method for automatically welding a T-shaped welding portion of a portion where a long cross section having a T-shaped contact with each other. The first layer welding used in connection with welding a T-bar is a term used to mean welding a welding part in one layer to fix a part to be welded. In addition, welding the T bar is also called butt welding or weaving welding. Butt welding is an indication of how the workpieces are laid out and means welding the objects that are in horizontal contact with each other, such as the T-bars. Weaving welding refers to a welding method, which means welding while moving a torch in a zigzag shape. That is, when mounting on the T-bar, we select the method of weaving welding (welding while moving the torch in a zigzag) to perform butt welding (welding objects that are in contact with each other horizontally), and superlayer welding (one-pass welding only) To do. When welding the T-bar welded on the first floor, butt welding is performed because the arrangement of the workpiece is the same, weaving welding is performed in the same way.

Figure 3a is a photograph showing the state of the portion to perform the butt welding between the longage and the inside of the double hull when the setting work to remove the step between the block and the block during the mounting operation. In addition, Figure 3b is a view showing the configuration of the portion to perform the butt welding between the long and long.

The longines 10 and 20 of the type shown in FIGS. 3A and 3B are referred to as long bars as the cross-sectional shape is similar to the capital letter 'T' when viewed in the longitudinal direction of the longines. The wide iron plate on the upper part of the long paper is called the long side (11, 21), and the lower part of the long side is called the long side web (12, 22). The two parts 30 and 40 facing the two longages 10 and 20 are the parts to be welded.

Each of the facing long side and long side webs is welded once to perform a "mount" operation to fix the weld prior to main welding. The mounting work consists of removing the step between the two longines that are in contact with each other, welding the first floor, removing slag, and recovering the equipment installed for step removal.

The step of removing the step between the longines and longines that are in contact with each other includes (i) the paralleling of the front side of the long side using a chain block, and (ii) a cylindrical fit-up long jig on the front side of the long side. jig) to remove the upper and lower steps on the front of the longines, and (iii) to remove the steps on the longines web.

If the gap between the long paper and the long paper after removing the step between the long paper is too narrow, it is difficult to perform the first layer welding. In such cases, use a cutter to widen the gap. Due to this operation, irregular gaps are usually formed.

After removing the step between the longages and in some cases extending the gap, a backing is attached on the opposite side to the welding side for welding. The backing material closes the gap between the long paper and the long paper, so that the weld beads (beads) are well accumulated. Attaching the backing material is also performed when welding the front side of the longage. However, when welding the front of the long side, as well as attaching the backing material, it also performs the work of padding the iron plate. This serves as a support to prevent the welds from flowing down to the floor when welding the front of the long piece.

After attaching the backing material, first layer welding is performed on the longage web. When a person directly welds the long web and the front of the long paper, the worker has no choice but to weld in an uncomfortable position in a narrow space, and continues to work in an uncomfortable position until the welding of the T bar is completed.

Finishing the first floor welds for the long side front and long side webs respectively, the operator removes the slag accumulated on the weld bead. Finally, the installation is completed by recovering the equipment installed to remove the step between the long and long.

4 is a photograph showing the front side of the longage of the T-bar after the first floor welding is completed, Figure 5 is a photograph showing the shape of the longage web after the first floor welding is completed. In the case of the first floor welded long paper, the weld beads were relatively evenly stacked, while in the long web, the weld beads were very unevenly stacked. Also, in both cases, the shape of the improvement is not a straight line but a rugged curve. The left side improvement of the front side of the long side of Figure 4 is rugged compared to the improvement on the right side, such a shape is created while expanding the T bar spacing.

As such, the welding portion of the T bar to perform the weaving welding is not constant in shape, so it is necessary to recognize the shape before performing each welding. If the welding path can be automatically generated based on the recognized shape, the welding robot can perform the welding of the T-bar, which requires much labor of the operator when the human is directly performing the welding. Therefore, there is a great need to develop a method of recognizing the shape of the welded portion of the T-bar welded to the first floor and generating a welding path therefrom and an automatic welding robot applying the method.

The present invention has been made to solve various problems including the above problems, and an object of the present invention is to recognize a shape of a welding part for enabling automatic welding by a welding robot, and a welding path of a welding robot to which the method is applied. It is to provide an automatic welding robot capable of performing a welding operation with good welding quality by following the generation method and the welding path generated by this welding path generation method.

An object of the present invention as described above, as a method for recognizing the shape of the welded site having the shape of a groove extending in one direction by welding two sheets of first layer,

The direction of traversing the weld using a one-dimensional laser sensor in order to divide the weld into a plurality of sections in the extended longitudinal direction of the groove and to recognize the shape of the welded section where the two plates are welded to each layer for each section. A scanning step (a) of obtaining a profile of cross sections of the plurality of welded portions by performing a process of measuring a distance between the one-dimensional laser sensor and the welded portion while moving to;

A filtering step (b) of removing noise data corresponding to a point where a slope changes rapidly in each profile curve data obtained in the scanning step (a); And

It is achieved by providing a shape recognition method of a welded site including a reference point generation step (c) for determining a reference point that is a reference of the cross-sectional shape of the welded site from the data filtered in the filtering step (b).

In addition, the object of the present invention as described above, as a method of automatic welding path generation of the welded portion having the form of a groove extending in one direction by welding two sheets of first layer,

The direction of traversing the weld using a one-dimensional laser sensor in order to divide the weld into a plurality of sections in the extended longitudinal direction of the groove and to recognize the shape of the welded section where the two plates are welded to each layer for each section. A scanning step (a) of obtaining a profile of cross sections of the plurality of welded portions by performing a process of measuring a distance between the one-dimensional laser sensor and the welded portion while moving to;

A filtering step (b) of removing noise data corresponding to a point where a slope changes rapidly in each profile curve data obtained in the scanning step (a);

A reference point generation step (c) of determining a reference point which is a reference of the cross-sectional shape of the welded portion from the data filtered in the filtering step (b);

It is achieved by providing a welding path generation method comprising a welding path generation step (d) for generating a weaving welding path using the reference point generated in the reference point generation step (c).

In addition, the object of the present invention as described above, the main body;

A sensor unit disposed from the main body toward the welding part;

A welding unit arranged to be connected to the main body, the welding unit having a welding rod to perform welding to a welding site;

A transfer unit connected to the main body to transfer one or more of the sensor unit and the welding unit; And

A control unit for controlling the sensor unit, the welding unit, and the transfer unit,

The sensor unit may emit a laser beam at the welding site and receive the reflected light of the reflected laser beam to measure the distance between the sensor unit and a specific point of the welding site, and the sensor unit may be connected to the welding site by the transfer unit. The distance is measured while moving in the direction transverse to the longitudinal direction of the welded portion to obtain the profile of the cross section of the welded portion, and then moved again in the longitudinal direction of the welded portion at a position spaced apart from the profile of the previous welded portion. Measuring the cross-sectional profile data of a plurality of welds in such a way as to obtain a cross-sectional profile of the welded area,

The control unit generates a reference point constituting the profile from the plurality of cross-sectional profile data measured by the sensor unit, generates a welding path from the reference point, and the transfer unit transfers the welding unit to follow the generated welding path. It is achieved by providing an automatic welding robot, characterized in that it is controlled to enable automatic welding.

In addition, the object of the present invention as described above, the main body;

A sensor unit disposed from the main body toward the welding part;

A welding unit arranged to be connected to the main body, the welding unit having a welding rod to perform welding to a welding site;

A transfer unit connected to the main body to transfer one or more of the sensor unit and the welding unit; And

A control unit for controlling the sensor unit, the welding unit, and the transfer unit,

On the basis of the welding path information generated by the above-described welding path generation method, the conveying unit automatically welds the welding portion while the welding rod is transported according to the weaving angle of the weaving axis of the electrode and the longitudinal feeding interval of the welding portion. By providing a welding robot.

According to the present invention, the shape of the welded portion can be easily recognized at the level necessary for the welding operation. In addition, by applying the shape recognition method to generate a welding path it is possible to weld by the automatic welding robot. According to the present invention, in consideration of the shape error of the welded portion, the welding operation can be performed with good quality.

In particular, the present invention has the advantage of using the multi-dimensional laser sensor even when scanning with a one-dimensional laser sensor, the price, size and weight of the one-dimensional laser sensor is less than the multi-dimensional laser sensor Has a very important meaning in welding work for assembling double hull blocks in shipyards. First of all, the passage through which the worker moves for welding in a double hull structure is very narrow. In addition, the larger the ship, the farther the worker has to carry the equipment for the welding operation, and must go through several ladders. In such an environment, it is very burdensome to carry a piece of heavy equipment in such an environment and prefer to use a piece of light equipment. Therefore, it can be said that the significance of developing a welding robot capable of recognizing a three-dimensional shape using the one-dimensional laser sensor of the present invention is very large.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the present invention will be described in more detail.

To write a weld path generation algorithm, we need to know the location and size of the T-bar spacing. To this end, in the present invention, by using a one-dimensional laser sensor to scan the work object (scan) to obtain information about the position and size of the T-bar spacing, using this to generate a welding path.

6 is a view schematically showing a method for generating a welding path according to the present invention.

As shown in FIG. 6, the welding path generating method according to the present invention includes a scanning step of measuring a distance between a workpiece and a laser sensor using a 1D laser sensor, and filtering to remove noise included in the scanned data. Step, generating a reference point (abbreviated as "Ref" in the drawings below) to obtain the shape of the T-bar interval from the filtered data, and a weld path generation step of generating a weaving weld path using the generated reference point do.

When creating the weld path, the change in the width of the T-bar spacing and the change in the position of the T-bar spacing must be taken into account. In order to recognize the position and shape of the T-bar spacing (hereinafter, referred to as a profile), the depth of the T-bar spacing is measured using a one-dimensional laser sensor.

FIG. 7 is a view showing an example of a scan method for profile recognition of a welded portion and a profile of a welded cross section obtained through one one-dimensional scan.

Since the profile changes its position and shape according to the height, the profile should be scanned for the height of the entire T bar. Therefore, as shown in FIG. 7A, the scan is performed on the entire height of the T bar. The scan moves along the x- axis and repeats the upward movement along the z- axis at equal intervals dz .

In (b) of FIG. 7, the scanning process may be confirmed. For example, when scanning the lowest profile, move the sensor from left to right along the x axis. At the same time, measure the distance between the sensor and the T bar at equal intervals dx using the sensor. Through this, the shape of the profile can be obtained.

8 is a view showing an example of a profile obtained as a result of performing a one-dimensional scan in the x-axis direction on the actual specimen. In FIG. 8, an enlarged portion of the red ellipse in the upper profile is the lower profile. As shown in FIG. 8, the profile scan result includes an error (incorrect scan point) and noise (problems where the profile is not smooth and splashes). These errors and noises included in the scanned profile must be removed in advance to find the weld reference point.

9 is a diagram showing a method of removing errors and noise in a profile measured using a laser sensor.

First, how to remove the error will be described. The error included when using the one-dimensional laser sensor is significantly smaller than the non-error points as shown in the above profile of FIG. 9. By using these characteristics, errors that are noticeably different in size are eliminated.

First, the average value yavg of the y values of the scanned points is obtained. If necessary, give a little margin to use as a criterion for determining whether an error is an error or not. That is, points below the straight line ' y = y _avg -margin' can be regarded as an error and removed. Of course, the margin can be set to 0, and the user can set an appropriate value considering the distance between the actual laser sensor and the welding part.

After the error is removed, the location of the data in error is empty. In order to apply a Gaussian filter in a later step, it is necessary to pre-populate an appropriate value in this empty position. This is because the Gaussian filter is used to smoothly change the curve of the profile by calculating the average value of the point at the current position by multiplying the weights of the surrounding points. Where there was an error, linearly interpolate the non-error points around it and enter the value. That is, when the empty space is originally referred to as the i- th scanned point, the i-th y value y _i is determined by the following equation and filled in the empty space.

[Equation 1]

Where ( i - k ) is not an error and is the index of the point closest to i among points that are smaller than i, and ( i + t ) is not an error and is the closest point to i among points that are larger than i. . The red dots in FIG. 9 show the dots filled by this process.

Differential coefficients are used to generate reference points that represent the shape of the profile. However, if there is noise as shown in Fig. 9, the derivative coefficient changes rapidly, and this method cannot be used. Therefore, noise cancellation is necessary. A Gaussian filter is used for this.

The Gaussian filter modifies values of existing data using weights (hereinafter, referred to as masks) determined by a Gaussian distribution. Below is the equation for Gaussian distribution.

[Equation 2]

Where μ is the average and σ is the standard derivative.

FIG. 10 is a graph illustrating a Gaussian distribution when σ = 1.5 and μ = 0, and FIG. 11 is a graph illustrating a method of extracting mask values from a Gaussian distribution.

The process of applying the Gaussian filter is as follows.

(1) From the Gaussian distribution, l (odd) values are extracted at equal intervals around x = 0. In FIG. 11, (a) of the left side shows that five values were extracted with equal intervals 1 as the gaussian distribution of FIG. 10.

(2) Divide each extracted value by the sum of the extracted values. The set of values obtained through this process is called mask. The sum of the components of the mask is always one. The values shown in (b) of the right side of FIG. 11 are examples of masks.

(3) Filtering is completed by multiplying and adding data to the mask.

If the i th value is y _i , and m _k is the k th component of the mask and l _h = ( l −1) / 2, the value Yi obtained by filtering y _i is as follows.

&Quot; (3) "

When Gaussian filtering is performed on the data in the form of FIG. 12A using the mask of FIG. 11B, the result is as shown in FIG. 12B. For example, the fourth value y ₃ = 8.2 (index starts from 0) in FIG. 12A is calculated from the following equation.

&Quot; (4) "

y _3, the product had the largest value of 0.2921, the values of y ₃ is surrounded by the more distant from the quality y ₃ and further multiplied by a small value. That is, the largest weight is given to the current value, and the further weight is added as the distance increases.

FIG. 13 is a graph showing results of applying a Gaussian filter with l = 29 for the noisy profile of FIG. 9 and different standard deviations. The larger the standard deviation value, the softer the filtered profile, but at the same time the width becomes wider, reducing the value as a whole and different from the original shape. Therefore, it is necessary to select an appropriate standard deviation value to remove the noise of the noisy profile while minimizing the shape change.

Hereinafter, a method of obtaining a welding reference point from the profile recognized as described above will be described.

FIG. 14 is a diagram showing the shape of an ideal profile of a welded site to perform butt welding. A reference point becomes a point which shows the shape of a profile in the butt welding of T bar. Analyzing the shape of the ideal profile of FIG. 14 before obtaining this reference point, it is noted that four green points (hereinafter referred to as reference points) of FIG. 14 are needed to define the profile of the butt weld portion of the T-bar of interest in the present invention. Able to know. As shown, the reference point is a point where the slope of the straight line changes drastically, and the reference point [0] and the reference point [3] are points at which the slope is rapidly increased, and the reference point [1] and reference point [2] are points at which the slope is rapidly decreased. The slope of the straight line between the reference point [0] and the reference point [1] has a maximum value in the overall profile, and the slope of the straight line between the reference point [2] and the reference point [3] has a minimum value.

FIG. 15 is a diagram illustrating a process of generating a temporary reference point for a filtered profile [0] using the above characteristics. Here, the profile [n] means the nth profile. In FIG. 15 (a), the point p having the maximum first derivative of the profile and the point q having the minimum are found. Based on the center of p and q, divide the left and right sides to find the maximum and minimum values of the second derivative. On the left, the maximum point of the second derivative is the temporary reference point [0], and the minimum point is the temporary reference point [1]. On the right, the minimum point of the second derivative is the temporary reference point [2], and the maximum point is the temporary reference point [3].

FIG. 16 shows the results of generating temporary reference points of the profile [0] filtered in the above manner. However, the provisional reference points of FIG. 16 differ from their reference points in the shape of the ideal profile of FIG. 14. The provisional reference point [0], the provisional reference point [3], the provisional reference point [1] and the provisional reference point [2] of FIG. 16 are not parallel to each other, and in the case of the provisional reference point [3], they are located above the actual profile. Therefore, it is necessary to correct the current reference points.

FIG. 17 is a diagram illustrating a method of generating a final reference point by modifying provisional reference points. As illustrated in FIG. 17, the following method is used to generate the final reference point by modifying the coordinates of the temporary reference points. First, the average value y ₀ of the temporary reference point [0] and the temporary reference point [3] is obtained, and the average value y ₁ of the temporary reference point [1] and the temporary reference point [2] is obtained. Then, a straight line 0 (line 0 in Fig. 17) passing through the temporary reference point [0] and a point p is obtained, and a straight line 1 (line 1 in Fig. 17) passing through the temporary reference point [3] and the temporary reference point [2] is obtained. Next, find the intersection of y = y ₀ , y = y ₁ , line 0 and line 1 as the new reference point. That is, the intersection of y = y ₀ and the straight line 0 is the reference point [0], the intersection of y = y ₁ and the straight line 0 is the reference point [1], the intersection of y = y ₁ and the straight line 1 is the reference point [2], The intersection of y = y ₀ and the straight line 1 is the reference point [3]. The final reference points are shown as green points in FIG. 17.

When the straight line 0 and the straight line 1 are obtained, the point p and the point q are used instead of the temporary reference point [1] and the temporary reference point [2] because of the shape of the profile. As shown in FIG. 5, the profile of the T-bar welded to the first layer has a relative reference point [1] and the reference point [2] because uneven welding beads are accumulated near the reference point [1] and the reference point [2]. It is likely to be inaccurate. Therefore, points p and q are used in place of the temporary reference point [1] and the temporary reference point [2]. Since point p and point q are the maximum and minimum first derivatives respectively, it can be determined that they will be located in the improvement plane. Therefore, it is possible to express the profile more precisely by using the temporary reference point [0] and the temporary reference point [3] indicating the point where the improvement starts and ends, and the points p and point q included in the improvement.

Hereinafter, a method of generating a welding path for performing one pass welding using the reference points of the respective cross-sectional profiles obtained above will be described.

First, the weaving axis (θ axis) is determined. (theta) axis is an axis which becomes the center of rotation of a torch at the time of weaving welding. The welding robot moves in the z- axis direction and uses the motor of the θ-axis to weav the torch. Here, a method of defining the x- and y- coordinates of the θ-axis on the same xy plane as the plane on which the cross-sectional profile is placed will be described in order to determine the relative position with respect to the cross-sectional profile of the θ-axis.

FIG. 18 is a diagram illustrating a method of determining left and right welding reference points based on previously determined reference points. Using reference points [1] and reference points [2] representing the shape of the profile, the welding reference points WL and WR representing the reference points of the welding are determined and the WC is calculated therefrom. WL is the left welding reference point, WR is the right welding reference point, and WC is the bisection of the line connecting WL and WR. As shown in Fig. 18, in each profile, WL is an intersection of a straight line connecting the reference point [0] and the reference point [1] and a straight line separated in the -y direction by an offset from the reference point [1]. WR is an intersection of a straight line (horizontal line) separated from the reference point [2] in the -y direction by an offset and a straight line connecting the reference point [2] and the reference point [3]. Here, the offset is the distance that the wire of the torch is separated from the reference point [1] and the reference point [2] of the profile, and an offset of about 2 mm is suitable for good welding quality. The offset relative to the reference point does not necessarily mean that the end of the wire does not touch the superlayer weld bead. When the first layer weld beads are formed in the form of droplets, the torch may pass through the weld beads, or may not come into contact with the first layer beads. However, experiments with non-uniform superficial weld beads confirm that this effect does not need to be taken into account.

19 is a diagram showing the manner of determining the position of the weaving axis of the welding torch. After WL and WR for each profile for several profiles, WL and WR and WC for the average profile are obtained. The red reference points in FIG. 19 are average values of the reference points of the entire profile. WL and WR are obtained from the mean profile, and their center is defined as WC. The θ axis is on the waterline along the straight line between WL and WR at the WC of the mean profile. The θ axis is located along the waterline at WC, away from WC by the length of the wire and the distance between the torch and weaving axis. The length between the torch and the weaving axis is a value determined by which part of the torch is held by the mechanism of the welding robot. In this way, the position of the weaving axis (θ axis) of the torch can be determined.

The torch moves in the z direction and weaves between WL and WR of each profile. When generating the welding path, it is assumed that the weaving speed and the moving direction in the z direction are constant. When weaving, the torch's trajectory draws an arc, so when viewed from the xz plane, the torch's movement path during weaving welding is not a straight line. However, here we are not actually moving the torch, but looking for the target point where the tip of the torch is located. Therefore, the above error can be ignored and the target points can be assumed to exist on a straight line with a constant slope.

20 is a diagram illustrating a method of determining a target point at which the end of the torch is to be positioned. The target point starts from the left weld reference point WL of the filtered profile [0] as in FIG. If the straight line through WL of the filtered profile [0] is z = ax + b , the next target point is the straight line between the WR of the filtered profile [0] and the WR of the filtered profile [1] and the straight line z = ax is the intersection of + b . The next target point is the intersection of the straight line passing through the current target point and connecting the straight line z = -ax + c with the slope- a and the WLs of the filtered profile [0] and the filtered profile [1]. That is, when the WL of the filtered profile [0] when said 0th target point, the odd-numbered target points, a straight line with a slope of a 'and a cross point of the "connecting straight line of the welding reference point WR', even-numbered target point is - the intersection of "a straight line with a slope of a 'and a" straight line connecting the reference point welding WL'. The torch performs weaving welding by reciprocating between even and odd target points.

21 is a diagram illustrating the need to consider the weaving motion of the torch when determining the position of the final torch tip.

As shown in FIG. 21, since the torch draws an arc during weaving welding, there is a gap between the torch tip and the target points no matter how the position of the torch tip is determined. Therefore, the target points determined in FIG. 20 may not be the position of the torch tip. The position of the torch tip is set to the position of the torch when the torch rotates exactly as shown in FIG. 21 and the torch points exactly at the target point. Since the position of the θ axis, that is, the rotation axis of the torch, is determined, the position of the end of the torch is determined by the weaving angle θ and the z axis direction position z around the rotation axis.

When the x coordinate of the target points is T _x , the y coordinate of the target points is T _y , the z coordinate of the target points is T _z , and the x coordinate of the θ axis is W _x , and the y coordinate of the θ axis is W _y . Weaving angle θ and height z of the end of the torch are as follows.

[Equation 5]

&Quot; (6) "

22 shows the position of the torch tip of the filtered profile [0] through the filtered profile [3], that is, the weaving weld path generation result. The torch performs zigzag weaving along the welding path.

FIG. 23 is a perspective view schematically illustrating a configuration of an automatic welding robot to which the welding site shape recognition method and the welding path generation method of the present invention are applied, and FIG. 24 shows a side of a long web using the automatic welding robot shown in FIG. 23. A drawing showing the operation of welding a welded portion of is shown.

As shown in FIGS. 23 and 24, the automatic welding robot 100 according to the present invention includes a main body, a first guide member 140, a sliding block, and a support unit.

The body is basically a member having a welding rod 155. The main body is a welding unit 160 in which the electrode 155 is rotatably disposed around the axis of rotation perpendicular to the longitudinal direction of the electrode 155 in the embodiment shown in FIGS. The coupling unit so as to be movable in the horizontal direction along the first guide member 140 and the coupling unit 170 slidably coupled to the welding unit to move the welding unit in the front-rear direction with respect to the welded portion. It is connected to the 170, and includes a sensor unit 150 that can recognize the shape of the welding portion.

In the hull manufacture, the welding site is large, the welding operation can not be completed by only one pass of the welding rod 155 as the welding rod 155 passes. In other words, weld beads should be stacked in several layers in the depth direction, and in many cases, the work should be divided into 23 or more times even when a single layer of beads is stacked. The electrode 155 should be able to perform welding while weaving (weaving) to the left and right with respect to the welding site. Therefore, the electrode 155 is rotatably connected to the rotating shaft to enable the weaving.

The sensor unit 150 may include a laser sensor capable of recognizing a shape of a welded portion, and a cover 159 that covers a portion in which the laser sensor is embedded so as to protect the beam radiation portion of the laser sensor during a welding operation. ) May be provided.

When connecting large hull blocks to each other, there may be a large shape error or a dimension error that can be visually identified. Even in this case, in order to perform welding without recognizing the shape of the corresponding part automatically, it is necessary to recognize the shape of the welded part and analyze the shape to set the welding path and perform the welding. To this end, in the present invention, the sensor unit is provided with a laser sensor so as to recognize the shape of the welded portion while scanning the welded portion with the laser sensor.

In addition, sparks may break out during the welding operation. Therefore, it is necessary to protect the laser sensor, which is a precise part, so that it is desirable to protect the laser sensor with a cover when the laser sensor does not perform the task of recognizing the welded portion.

The first guide member 140 is slidably coupled to the main body so that the main body moves in a left and right direction with respect to the welded part while facing the welded part.

The second guide member extends in a direction orthogonal to the extending direction of the first guide member 140. The first guide member 140, the sensor unit moving along the first guide member 140, the coupling unit, and the welding unit all move along the second guide member. The second guide member may include two guide rails 111 and 112 disposed in parallel with each other, and at least one fixed to each of the guide rails 111 and 112 to maintain a distance between the guide rails 111 and 112. It may include a crossbar.

Guide rails (111, 112) constituting the second guide member is formed on one side of the rail for guiding the movement of the sliding blocks (131, 132), the other side of the support unit (121, 122, 123, 124) Rails may be formed to guide the movement of the vehicle. That is, as shown in the drawing, the guide rails 111 and 112 are members having a predetermined width and thickness, and have rails for guiding the movement of the sliding blocks 131 and 132 on one surface in the thickness direction, and the opposite surface. A rail may be provided to guide the movement of the sliding blocks 131 and 132.

The sliding blocks 131 and 132 are fixed to the first guide member 140 and slidably coupled along the second guide member. When the second guide member has two guide rails 111 and 112 as shown in the drawing, the sliding blocks 131 and 132 are also movable one by one along the respective guide rails 111 and 112. The sliding blocks 131 and 132 are disposed. Even when two sliding blocks 131 and 132 are installed, the two sliding blocks 131 and 132 are integrally connected and move along the guide rails 111 and 112 using a configuration such as a motor 134 and a roller 133. Can be. Of course, the movements can be automatically realized by all the components moving along the first and second guide members described above, and known conveying means using the mobility motor in the front and rear direction of the welding unit.

When two sliding blocks 131 and 132 are installed, the first guide portion may be arranged to connect two sliding blocks 131 and 132.

The support units 121, 122, 123, and 124 may move along the second guide member on the second guide member, may be fixed at a specific position of the second guide member, and the second guide member may be fixed to the second guide member. Fix it to the welding area. The support units 121, 122, 123, and 124 are provided with magnets 121a, 122a, 123a, and 124a at their ends, respectively, to weld the metal material, which is a magnetic material, to the second guide using magnetic force on the metal material. The member is fixed and consequently the automatic welding robot 100 of the present invention can be fixed.

Two support units 121, 122, 123, and 124 may be installed on the guide rails 111 and 112 constituting the second guide member, respectively. The support units 121, 122, 123, and 124 are disposed at the ends in a state where they are spaced apart along the guide rails 111 and 112 as shown in FIG. Using the magnets to perform the function of fixing the automatic welding robot 100 of the present invention to the longe web.

The automatic welding robot 100 having the above configuration is disposed adjacent to the welding portion at the front surface of the rosy as shown in FIG. 24, and then welds the welding portion of the rosy web while moving in the vertical direction. In addition, since the welding portion is large along with the movement in the vertical direction, the welding is performed while moving in the left and right directions, and the welding operation may be performed in the depth direction of the welding portion, and thus the welding operation may be performed in several layers. Of course, as mentioned above, the welding rod 155 may be weaved, and thus, four degrees of freedom of the welding rod 155 may be realized as a whole. Four degrees of freedom are as shown by the arrows in FIG.

The mechanical part of the automatic welding robot of the present invention includes a step motor corresponding to a one-dimensional laser sensor and four axes ( x , y , z , θ) as a sensor unit, and a driver for driving the motors ( driver). In addition, there are two proximity sensors on each axis, and serve as limit sensors in the directions of left and right ( x ), front and rear ( y ), up and down ( z ), and clockwise and counterclockwise (θ). In addition, a torch (welding rod) is installed under the θ axis for welding.

The welding degree shape recognition method and the welding path generation method described above are applied to the four degree of freedom automatic welding robot illustrated in FIGS. 23 and 24.

25 is a flow chart of the welding operation of the automatic welding robot according to the present invention.

As shown in FIG. 25, the welding operation of the automatic welding robot according to the present invention includes four stages, Stage A to Stage D. Stage A installs the welding robot, placing the welding robot adjacent to the T-bar spacing of the T-bar, connecting the torch, and supplying power. Determine the position of the stopper of the two limit sensors on the z axis. The lower limit sensor is used to set the zero point on the z axis, and the upper limit sensor is used to set the work area to where the work will be done. Initialize each axis with limit sensors. Stage B scans the profiles of n sections using a one-dimensional laser sensor. A reference point is created based on the data obtained from the scan, and the reference points are used to calculate the area when scanning the next layer. After repeating the above operation and receiving a signal from the upper limit sensor set in Stage A, the welding robot stops scanning. Use a base point generated by the Stage B to perform welding while in Stage C generates a welding path between the n-th profile and the (n + 1) th profile and controlling movement of the four degrees of freedom automatic welding robot so as to follow the weld path. Welding to the profile of the last layer removes slag and withdraws the welding robot from Stage D.

Of stages A to D, portions directly related to the shape recognition method of the welded portion of the present invention are Stage B, and portions directly related to the weld path generation method of the present invention are Stage B and Stage C.

FIG. 26 is a flowchart illustrating a scanning method using a laser sensor as a flowchart illustrating Stage B of FIG. 25 in detail, and FIG. 27 is a diagram illustrating a scanning method of a welded portion.

As shown in FIG. 26, in the shape recognition method of the welded portion of the present invention, a scan is performed from 0 to 1000 along the x axis at an initial position. Where 0 and 1000 are indexes divided by the total scan interval by 1000. Scan at the nth layer (indicating the scanned position, the result of scanning the nth layer is the nth profile), store the data obtained by the scan, filter it, and generate a reference point.

When the reference point generation of the profile [ n ], which is the nth profile, is successful, the scan range may be determined using the reference point of the profile [ n ] when scanning the n + 1th layer. This is intended to speed up computation by reducing the scan range and reduce the load on the main CPU. When the scan range of n + 1st layer is expressed as [SINCE] and [UNTIL], as shown in FIG. 27, [SINCE] is a certain interval (ADDGAP) from the reference point [0] of profile [ n ]. It is located on the left side, and [UNTIL] is located on the right side of the reference point [3] of the profile [ n ]. If the reference point generation of the nth profile fails, the scan range when the nth layer is scanned is used when scanning the n + 1st layer. Since the straightness of the T bar is small within 1 mm, if the ADDGAP is made large enough, there will always be a gap of the T bar within the scan range.

FIG. 28 is a flowchart illustrating a method of generating a welding path from filtering of profile data to generating a reference point. The main CPU receives data from the sub CPU, stores the data, and executes an algorithm.

The main CPU eliminates data errors, applies a Gaussian filter, and derivatives the data to find the maximum and minimum of the first derivative. However, as explained earlier, the total scannable range is from 0 to 1000, and the actual scan range is reduced from [SINCE] to [UNTIL]. Therefore, indexes before [SINCE] and after [UNTIL] do not contain distance values, but other arbitrary values. This may result in the maximum or minimum in [SINCE] or [UNTIL], when considering the maximum and minimum of the first derivative. Therefore, to prevent this, when calculating the maximum and minimum of the first derivative, do not include both ends of the data. That is, considering [SINCE + alpha] to [UNTIL-alpha], the maximum and minimum of first derivative are calculated.

After obtaining the first-order derivatives, the first-order derivatives are differentiated once more to find the second-order derivatives. A smooth second derivative curve is obtained by applying a Gaussian filter to the second derivative. In order to reduce the load, the AVR calculates 1000 times the value of the second derivative before reducing the minimum and the maximum of the second derivative. Thereafter, as described above, the left part and the right part are divided into maximum and minimum values. Then, the maximum and minimum of each part is set as a temporary reference point, and the reference point is obtained using the reference point.

In the following description of the present invention, the embodiments illustrated in the drawings have been described with reference to the present invention. I will understand the point. Therefore, the true technical protection scope of the present invention will be defined by the technical spirit of the appended claims.

1 is a photograph of a block being mounted at the shipyard.

2 is a photograph showing a narrow access hole for the operator to enter the double hull after setting.

Figure 3a is a photograph showing the state of the part to perform the butt welding between the long and the inside of the double hull when the setting work to remove the step of the block and the block during the mounting operation.

Fig. 3b schematically shows the shape of a T-shaped weld site in contact with the longage;

Figure 4 is a photograph showing the state of the front side of the T bar after the first floor welding is completed.

Fig. 5 is a photograph showing the shape of the longage web after the first floor welding is completed.

Figure 6 is a schematic view showing a welding path generation method according to the present invention.

7 is a view showing an example of a scan method for profile recognition of a welded portion and a profile of a welded section cross section obtained through one one-dimensional scan.

8 shows an example of a profile obtained as a result of performing a one-dimensional scan in the x-axis direction on an actual specimen.

9 illustrates a method of removing errors and noise from a profile measured using a laser sensor.

10 is a graph showing a Gaussian distribution when the standard polarization σ is 1.5 and μ = 0.

11 is a graph illustrating a method of extracting mask values from a Gaussian distribution.

12 shows changes in profile before and after applying a Gaussian filter.

FIG. 13 is a graph showing results of applying a Gaussian filter with l = 29 for the noisy profile of FIG. 9 and having different standard deviations. FIG.

14 shows the shape of an ideal profile of a welded site to perform butt welds.

15 is a diagram illustrating a process of generating a temporary reference point for a filtered profile [0].

16 shows the results of generating temporary reference points of the filtered profile [0].

17 is a view showing a method of generating a final reference point by modifying temporary reference points.

18 is a view illustrating a method of determining left and right welding reference points based on previously determined reference points.

19 shows a way of determining the position of the weaving axis of a welding torch.

20 illustrates a method of determining a target point at which an end of a torch is to be positioned.

21 illustrates the need to take into account the weaving motion of the torch when determining the position of the final torch tip.

Fig. 22 shows the position of the torch tip of the filtered profile [0] through the filtered profile [3], i.e., the result of weaving weld path generation.

Figure 23 is a perspective view showing the configuration of the automatic welding robot according to the present invention.

FIG. 24 is a view showing an operation of welding a welding part on the front side of the longage using the automatic welding robot shown in FIG. 23.

25 is a flowchart of a welding operation of an automatic welding robot.

FIG. 26 is a flowchart specifically illustrating a stage B of FIG. 25, illustrating a scanning method using a laser sensor. FIG.

27 is a view for explaining a scanning method of a welded portion together with the movement of the electrode.

28 is a flowchart illustrating a method of generating a welding path from filtering of profile data to generating a reference point.

<Explanation of symbols for the main parts of the drawings>

10, 20: Longines 11, 21: Longines Web

12, 22: long side front 30: horizontal welding area

40: up and down direction welding part 100: automatic welding robot

111, 112: guide rails 113, 114: cross bars

121, 122, 123, 124: support unit 121a, 122a, 123a, 124a: magnet

131, 132: sliding block 133: roller

134: motor 140: first guide member

150: sensor unit 155: welding rod

156: axis of rotation 159: cover

160: welding unit 170: coupling unit

Claims (17)

As a shape recognition method of a welded portion having a shape of a groove extending in one direction by welding two sheets of first layer, The direction of traversing the weld using a one-dimensional laser sensor in order to divide the weld into a plurality of sections in the extended longitudinal direction of the groove and to recognize the shape of the welded section where the two plates are welded to each layer for each section. A scanning step (a) of obtaining a profile of cross sections of the plurality of welded portions by performing a process of measuring a distance between the one-dimensional laser sensor and the welded portion while moving to; A filtering step (b) of removing noise data corresponding to a point where a slope changes rapidly in each profile curve data obtained in the scanning step (a); And And a reference point generating step (c) of determining a reference point as a reference of the cross-sectional shape of the welded portion from the data filtered in the filtering step (b). The method of claim 1, The method of recognizing a welded portion of the welded part is further performed before the filtering step (b) of removing an error among the measured values of the one-dimensional laser sensor measured in the scanning step (a). The method of claim 1, The filtering step (b), (1) From the Gaussian distribution, extract l (odd) values at equal intervals around x = 0, (2) a mask is formed by dividing each of the extracted values by the sum of the extracted values, (3) a shape recognition method of a welded portion, characterized in that filtering is completed by applying a mask to data according to the following equation. [Equation]

Where y _i is the value of the i th data, m _k is the k th component of the mask, l _h = ( l -1) / 2, and Yi is the value obtained by filtering y _i ) The method of claim 1, The reference point generation step (c), (i) Find the point p with the largest first derivative and the point q with the smallest differential coefficient in the curve of the profile, and find the maximum and minimum values of the second derivative by dividing them to the left and right, based on the center of p and q. Set the point where the second derivative is the maximum as the temporary reference point [0], the minimum point as the temporary reference point [1], and on the right side the minimum point of the second derivative as the temporary reference point [2] and set the maximum point. A provisional reference point setting step defined by the provisional reference point [3], (ii) Obtain the average value y ₀ of the temporary reference point [0] and the temporary reference point [3], the average value y ₁ of the temporary reference point [ ₁ ] and the reference point [2], and the straight line 1 passing through the temporary reference point [0] and point p. After finding the straight line 2 passing the temporary reference point [3] and the temporary reference point [2], the intersection of y = y ₀ and the straight line 0 is the reference point [0], and the intersection of y = y ₁ and the straight line 0 is the reference point [ 1], a final reference point generation step of determining final reference points by determining an intersection point of y = y ₁ and a straight line 1 as a reference point [2] and an intersection point of y = y ₀ and a straight line 1 as a reference point [3]. The shape recognition method of the welding part characterized by the above-mentioned. As a method of automatic welding path generation of a welded area in which two plates are welded to each other and have a groove extending in one direction, The direction of traversing the weld using a one-dimensional laser sensor in order to divide the weld into a plurality of sections in the extended longitudinal direction of the groove and to recognize the shape of the welded section where the two plates are welded to each layer for each section. A scanning step (a) of obtaining a profile of cross sections of the plurality of welded portions by performing a process of measuring a distance between the one-dimensional laser sensor and the welded portion while moving to; A filtering step (b) of removing noise data corresponding to a point where a slope changes rapidly in each profile curve data obtained in the scanning step (a); A reference point generation step (c) of determining a reference point which is a reference of the cross-sectional shape of the welded portion from the data filtered in the filtering step (b); And a welding path generation step (d) of generating a weaving welding path using the reference point generated in the reference point generation step (c). The method of claim 5, And removing the error among the measured values of the one-dimensional laser sensor measured in the scanning step (a) before the filtering step (b). The method of claim 5, The filtering step (b), (1) From the Gaussian distribution, extract l (odd) values at equal intervals around x = 0, (2) a mask is formed by dividing each of the extracted values by the sum of the extracted values, (3) Welding path generation method characterized in that the filtering is completed by applying a mask to the data according to the following equation. [Equation]

Where y _i is the value of the i th data, m _k is the k th component of the mask, l _h = ( l -1) / 2, and Yi is the value obtained by filtering y _i ) The method of claim 5, The reference point generation step (c), (i) Find the point p with the largest first derivative and the point q with the smallest differential coefficient in the curve of the profile, and find the maximum and minimum values of the second derivative by dividing them to the left and right, based on the center of p and q. Set the point where the second derivative is the maximum as the temporary reference point [0], the minimum point as the temporary reference point [1], and on the right side the minimum point of the second derivative as the temporary reference point [2] and set the maximum point. A provisional reference point setting step defined by the provisional reference point [3], (ii) Obtain the average value y ₀ of the temporary reference point [0] and the temporary reference point [3], the average value y ₁ of the temporary reference point [ ₁ ] and the reference point [2], and the straight line 1 passing through the temporary reference point [0] and point p. After finding the straight line 2 passing the temporary reference point [3] and the temporary reference point [2], the intersection of y = y ₀ and the straight line 0 is the reference point [0], and the intersection of y = y ₁ and the straight line 0 is the reference point [ 1], a final reference point generation step of determining final reference points by determining an intersection point of y = y ₁ and a straight line 1 as a reference point [2] and an intersection point of y = y ₀ and a straight line 1 as a reference point [3]. Characterized in that the welding path generation method. The method of claim 5, The welding path generation step (d), (i) Using a reference point [1] and a reference point [2] representing the shape of the profile, the torch's wire is spaced along the hypotenuse of the profile by an offset that is the distance apart from the reference point [1] and the reference point [2], respectively. Set the left and right weld reference points WR and WR for each profile, calculate WL and WR for each profile, and then calculate the WL, WR, and WC between the average weld reference points, and the average profile. Determination of the y-axis direction coordinates of the θ axis to determine the position of the θ axis where the wire length and the distance between the torch and the weaving axis are moved away from the WC along the waterline along the straight line connecting the WL and the WR in the WC. ; And and (ii) a target point determining step of determining an angle θ to weave based on the position of the θ axis and a distance z in the z-axis direction to be raised by the following equation in one unidirectional movement of the torch. How to produce. [Equation]

[Equation]

Where T _x is the x coordinate of the target points, T _y is the y coordinate of the target points, T _z is the z coordinate of the target points, W _x is the x coordinate of the θ axis, and W _y is the y coordinate of the θ axis) main body; A sensor unit disposed from the main body toward the welding part; A welding unit arranged to be connected to the main body, the welding unit having a welding rod to perform welding to a welding site; A transfer unit connected to the main body to transfer one or more of the sensor unit and the welding unit; And A control unit for controlling the sensor unit, the welding unit, and the transfer unit, The sensor unit may emit a laser beam at the welding site and receive the reflected light of the reflected laser beam to measure the distance between the sensor unit and a specific point of the welding site, and the sensor unit may be connected to the welding site by the transfer unit. The distance is measured while moving in the direction transverse to the longitudinal direction of the welded portion to obtain the profile of the cross section of the welded portion, and then moved again in the longitudinal direction of the welded portion at a position spaced apart from the profile of the previous welded portion. Measuring the cross-sectional profile data of a plurality of welds in such a way as to obtain a cross-sectional profile of the welded area, The control unit generates a reference point constituting the profile from the plurality of cross-sectional profile data measured by the sensor unit, generates a welding path from the reference point, and the transfer unit moves the welding unit to follow the generated welding path. An automatic welding robot, characterized in that the control to transfer to enable automatic welding. The method of claim 10, The sensor unit, A process of dividing a welded portion into a plurality of sections in an extended length direction, and for each section, measuring the distance between the 1D laser sensor and the welded portion while moving in the direction crossing the welded portion using the 1D laser sensor. An automatic welding robot, characterized in that to obtain a profile of a cross section of a plurality of welding sites. The method of claim 10, The control unit, In each profile curve data obtained from the sensor unit, the noise data corresponding to the point where the inclination changes abruptly is removed, and a reference point which is a reference for the cross-sectional shape of the welded part is determined from the data from which the noise is removed. Automatic welding robot, characterized in that for generating a weaving welding path. The method of claim 10, The control unit, And performing a function of eliminating an error among the measured values of the one-dimensional laser sensor in each profile curve data obtained by the sensor unit. The method of claim 10, The control unit, (1) From the Gaussian distribution, extract l (odd) values at equal intervals around x = 0, (2) a mask is formed by dividing each of the extracted values by the sum of the extracted values, (3) Automatic welding robot, characterized in that to filter the profile data by applying a mask to each cross-sectional profile data according to the following equation. [Equation]

Where y _i is the value of the i th data, m _k is the k th component of the mask, l _h = ( l -1) / 2, and Yi is the value obtained by filtering y _i ) The method of claim 10, The control unit, (i) Find the point p with the largest first derivative and the point q with the smallest differential coefficient in the curve of the profile, and find the maximum and minimum values of the second derivative by dividing them to the left and right, based on the center of p and q. Set the point where the second derivative is the maximum as the temporary reference point [0], the minimum point as the temporary reference point [1], and on the right side the minimum point of the second derivative as the temporary reference point [2] and set the maximum point. Set the reference point [3], (ii) Obtain the average value y ₀ of the temporary reference point [0] and the temporary reference point [3], the average value y ₁ of the temporary reference point [ ₁ ] and the reference point [2], and the straight line 1 passing through the temporary reference point [0] and point p. After finding the straight line 2 passing the temporary reference point [3] and the temporary reference point [2], the intersection of y = y ₀ and the straight line 0 is the reference point [0], and the intersection of y = y ₁ and the straight line 0 is the reference point [ 1], the intersection of y = y ₁ and straight line 1 as the reference point [2], and the intersection of y = y ₀ and straight line 1 as the reference point [3] to determine the final reference points to determine the shape of the weld site. Automatic welding robot featured. The method of claim 10, The control unit, (i) Using the reference point [1] and the reference point [2] representing the shape of the profile, the hypotenuse of the profile by the offset of the distance of the torch wire away from the reference point [1] and the reference point [2] in the -y direction, respectively. The spaced apart points are defined as the left welding reference point WL and the right welding reference point WR, and WL and WR for each profile are obtained, and then WC, WR, and WC between the average welding reference points are obtained. The position of the θ axis is determined by moving the wire length and the distance between the torch and the weaving axis away from the WC along the waterline along the straight line connecting the WL and WR in the WC obtained from the average profile. (ii) The automatic welding robot, characterized in the following equation, determines the angle θ to weave and the distance z in the z-axis direction to rise based on the position of the θ axis in one unidirectional movement of the torch. [Equation]

[Equation]

Where T _x is the x coordinate of the target points, T _y is the y coordinate of the target points, T _z is the z coordinate of the target points, W _x is the x coordinate of the θ axis, and W _y is the y coordinate of the θ axis) main body; A sensor unit disposed from the main body toward the welding part; A welding unit arranged to be connected to the main body, the welding unit having a welding rod to perform welding to a welding site; A transfer unit connected to the main body to transfer one or more of the sensor unit and the welding unit; And A control unit for controlling the sensor unit, the welding unit, and the transfer unit, The welding rod according to the welding path information generated by the welding path generation method according to any one of claims 5 to 9, wherein the transfer unit is adapted to the weaving angle of the weaving axis of the welding rod and the longitudinal transfer interval of the welded portion. An automatic welding robot that allows welding to be welded while being transported.

Images