1 BACKGROUND
Precision welding of complex 3D profiles is usually performed by skilled manual
welders. The automation of the Gas Tungsten Arc Welding (GTAW) process
for high performance alloys is difficult on account of various factors, principally the
need to provide inert gas shielding. More generic problems are related to the
complexity of a multi-parameter process, that involves many interrelated
phenomena, such as heat transfer, fluid dynamics, plasma effects, electrical power
supply, etc. An automatic system must be capable of emulating the ability of the skilled welder.
The system architecture has to be such that its functional specification encompasses intelligence/decision making, adaptation to cope with uncertainties, application of knowledge,
rules and procedures. In addition to the "intelligent" attributes listed above, the system must have the physical attributes to perform a dexterous task under close control in a difficult
environment. The torch has to be manipulated along a trajectory, where the
weld speed is a critical process parameter. The size of the welding tool,
supply cables, sensors and gas fixtures complicate trajectory planning due to increased chance of collisions. Care must be taken to ensure that the molten weld pool, electrode and trailing areas of
solidifying metal are protected by an inert gas. The system is not merely controlling a
robotic manipulator but managing and controlling a process. There are many subsystems and auxiliary processes that have to be integrated as they all influence the final quality of the weld
Individual areas of welding automation have received considerable interest and research effort over the last thirty years. Bringing them together in a working practical system raises many other issues that have yet to be resolved. For example equipment that has proven successful in a
research laboratory may not be applicable in an industrial solution due to cost, robustness and being intrusive to the process. Each subsystem cannot be viewed in isolation. The work described in this report represents the first stages in the commissioning of a demonstrator and
research facility that is capable of intelligent 3D welding. This system will evolve and increase in functionality. The fundamental attributes of this system are:
• in-process sensingwelders. The automation of the Gas Tungsten Arc Welding (GTAW) process
for high performance alloys is difficult on account of various factors, principally the
need to provide inert gas shielding. More generic problems are related to the
complexity of a multi-parameter process, that involves many interrelated
phenomena, such as heat transfer, fluid dynamics, plasma effects, electrical power
supply, etc. An automatic system must be capable of emulating the ability of the skilled welder.
The system architecture has to be such that its functional specification encompasses intelligence/decision making, adaptation to cope with uncertainties, application of knowledge,
rules and procedures. In addition to the "intelligent" attributes listed above, the system must have the physical attributes to perform a dexterous task under close control in a difficult
environment. The torch has to be manipulated along a trajectory, where the
weld speed is a critical process parameter. The size of the welding tool,
supply cables, sensors and gas fixtures complicate trajectory planning due to increased chance of collisions. Care must be taken to ensure that the molten weld pool, electrode and trailing areas of
solidifying metal are protected by an inert gas. The system is not merely controlling a
robotic manipulator but managing and controlling a process. There are many subsystems and auxiliary processes that have to be integrated as they all influence the final quality of the weld
Individual areas of welding automation have received considerable interest and research effort over the last thirty years. Bringing them together in a working practical system raises many other issues that have yet to be resolved. For example equipment that has proven successful in a
research laboratory may not be applicable in an industrial solution due to cost, robustness and being intrusive to the process. Each subsystem cannot be viewed in isolation. The work described in this report represents the first stages in the commissioning of a demonstrator and
research facility that is capable of intelligent 3D welding. This system will evolve and increase in functionality. The fundamental attributes of this system are:
• adaptation
• sufficient degrees of freedom for 3D
welding
• interfacing with off-the-shelf welding
equipment
• trajectory generation
• open architecture control
• fixturing for inert gas weld protection
2 OBJECTIVE
The overall objective of this work is the building of a welding demonstrator. This can be divided into the following task objectives:
Hardware To upgrade the Whyetech Cartesian robot so that it is capable of performing industrially relevant welding tasks, i.e. add two rotary axes to allow orientation of the weld toolin 3D space Trajectory Generation To develop algorithms for following 3D weld paths, two distinct types of problem are identified:
• welding of large box structures
• precision welding of small aero-engine
parts with double curvature surfaces Sensing/Process Monitoring To report on techniques for laser seam sensing on titanium parts, to identify problems and possible solutions, and to assess the performance of three commercial sensors. To study low cost methods for monitoring of welding process parameters using standard vision systems. The devices must be robust, affordable, and nonintrusive. Control Architecture To implement GiCNC control architecture, interfaces must be modular and accepted by industry standard equipment, such as welding power supplies.
3 METHODOLOGY
3.1 Robot Configuration There is a requirement for a robotic manipulator that can perform welding on 3D structures. As this is needed for developing and testing control algorithms, a commercial robot with a closed architecture is not suitable. The following section details the design considerations for the upgrading of an existing 3-axis manipulator.
3.2 Selection of Axes
The selection of mechanical configuration and number of axes is governed by the type of welds that must be performed and what access restrictions there are in the workspace. A machine that can attain the position and orientation of the weld torch to perform all types of weld on a 3- dimensional structure requires at least 6 degrees of freedom (15,16). (A redundant manipulator—having more than 6 d.o.f.— may be required to navigate around obstructions; i.e. to overcome degeneries in the work volume). Having many degrees of freedom is not always desirable as additional axes introduce extra cost, complexity, weight, difficulty in control, reduced rigidity, increased error and are not always necessary. In general, 3 axes are required to position a tool within a work volume and a 3-axis wrist is needed to provide all possible orientations of the tool. In many cases a two-axis wrist is enough for welding. If required, a missing axis on the robotic manipulator can be compensated for by using a separate device to manipulate the workpiece relative to the tool. This has the advantage of better rigidity than can be obtained in the open kinematic chain of a 6-axis manipulator. Another advantage Another advantage of workpiece manipulation is that the orientation of the weld pool can be maintained so that welding can be performed without regard to gravitational effects (9). When welding small parts, involving changes in orientation, it is easier to manipulate the workpiece than the welding torch, (as well as cables, sensors and the kinematic chain of manipulator links).
The welding of large box structures is a very different problem. The workpiece cannot be moved easily and there are less continuous changes of weld orientation. Referring to the types of welds in figure 2, it can be shown that any of the weld paths can be executed with a 5-axis configuration. (However, it can also be shown that not all of the welds can be performed with the same 5-axis configuration). On figure 2, welds A and G require 3 d.o.f. for position and one rotary axis for orientation of fixtures. Welds B,E and H require an additional rotary axis to
orientate the weld tool at 45 degrees for the Tee-joint. If weld C is possible with a 5-axis manipulator, welds D and F will not be unless the fixture alignment is reconfigured.
As 6-axis co-ordinated motion is never required in any single weld, it was decided that a 5-axis system (figure 3) would provide the best compromise between system capability and cost as well as being, more rigid, less complex and lighter than a six axis manipulator. Positioning is achieved by three orthogonal linear axes X, Y and Z (Cartesian configuration). Orientation of the tool is accomplished using two rotary axes, A and B. When producing horizontal welds, the first rotary axis, A, allows control of the orientation of the tool’s leading and trailing edges to match the
direction of the weld. The second rotary axis, B, allows inclination of the tool to perform different types of weld such as a 45ยบ angle for a Tee-joint.
3.3 Manipulator Kinematics
The kinematics of the manipulator are analysed by assigning coordinate frames to each link and tool (figure 4) and then establishing the Denavit & Hartenburg parameters to define the coordinate transformation matrices for each link (15, 16). The transformation matrix from the tool centre point (TCP) to base frame is given in equation 1. Closer inspection of the first column of the tool transformation matrix shows that the tool’s x-axis has components in X0 and Y0
(governed by joint angle dA) but has no component in Z0; i.e. no vertical component. This means that if the tool is aligned so that its direction of travel points along this axis, the tool cannot pitch. This limitation is due to the manipulator only having 5 degrees of freedom. In practice, this problem can be overcome by changes in the tool configuration. Analysis and simulation demonstrated that the design of the upgraded manipulator satisfied the requirement of system for demonstrating industrially relevant welds. The two extra mechanical links were fabricated and fitted to the Whitch manipulator. Yaskawa Servopack motors and servo controllers were fitted to actuate the rotary links. The original four axis MEI motion control card was upgraded to an 8 axis version. Figure 5 shows details of the last rotary axis and tool, featuring, laser
sensor and shielding fixture.
No comments:
Post a Comment