Development of laser machining CNC system for the

The development of parallel robot laser machining numerical control system

1 introduction

laser welding, as a new welding process method, has a large number of successful applications in the field of automobile manufacturing, showing the strong vitality and very broad application prospects of laser welding. The use of parallel mechanism can further improve the load/dead weight ratio of the robot and improve the motion characteristics of the robot, Make its structure more dexterous. Take advantage of the advantages of parallel mechanism, such as large stiffness, small kinematic inertia and high precision, to realize high-precision laser welding processing, which is the expansion of parallel robot technology to the field of laser welding.

the traditional NC machining trajectory control concept is based on the Cartesian coordinate system, and the trajectory control of parallel robot is integrated by the spatial motion of several links, If the robot can realize the required movement of the machining end point, the coordinate conversion must be carried out between them. In addition, the diversification of the structure and configuration of the parallel robot makes the closed structure of the traditional NC unable to meet its needs, Therefore, the control system of the parallel robot must be an open structure.

the parallel robot control system developed in this project is a full software open CNC system with Linux 24.20 + real-time kernel rtlinux3.1 as the real-time software platform. The CNC system software is compiled with C language as the programming language. Due to the introduction of real-time multitasking mechanism, the open structure framework is adopted, Therefore, it can realize the high-speed and high-precision control of the parallel robot with special mechanism configuration. The system is suitable for laser welding, cutting and other processing applications that require high precision. The research breakthrough of its key and difficult technologies, It is of great significance to further develop the laser welding parallel robot system and realize its industrialization.

2 components of the system

the parallel robot numerical control system is a real-time multitask software system, which can be divided into two layers: real-time task layer and non real-time task layer. The real-time task layer includes interpolation calculation module, pose forward and inverse solution module, speed processing module, motor control module Soft PLC module, fault diagnosis and processing module and other functional modules that are directly related to NC machining and have high requirements for real-time performance. This layer is handled by RTLinux real-time subsystem; In addition, functional modules that do not require high real-time performance, such as human-computer interaction module, parameter configuration module, code interpretation module, processing simulation module, file management module, etc., are non real-time task layers, which are controlled by the ordinary Linux kernel. The functional structure of CNC system software is shown in Figure 1. The advantage of this layering is that real-time tasks and non real-time tasks run in different process spaces, which is convenient to ensure the real-time performance of the system, The system has a clear hierarchy, enhances openness and flexibility, and can adapt to a variety of practical applications

Figure 1 software structure of CNC system

this CNC system takes linux2.4.20 + rtlinux3.1 as the operating system and PC as the platform. All the above functions are completed by software. The axis control card connected with the external servo only converts the digital quantity of PC and outputs it to the servo system, and sends the external feedback information to PC, No calculation is carried out. The hardware structure of the CNC system is shown in Figure 2.

Figure 2 hardware structure of the CNC system

realization of the forward and inverse displacement modules

in the above functions, the difference from the traditional CNC system is that the system needs to embed a specific forward and inverse displacement model according to the actual mechanism model of the mechanical structure. Therefore, It is very important to establish a correct forward and inverse solution model.

3.1 establish a forward and inverse solution geometric model

the schematic diagram of the robot mechanism is shown in Figure 3. In the mechanism described in this paper, converting the optical axis pose parameters (virtual axis coordinates) to the driving coordinate position (real axis coordinates) is called inverse solution operation

Figure 3 Schematic diagram of robot mechanism

set the rotation angle of Hooke hinge rotating around X axis in L2 branch as α， Rotate the Hooke hinge around the Y axis at an angle of β， The rod lengths of the three branches are L1, L2 and L3 respectively, and the rotation angles of the series joint rotating pair and the pitching rotating pair are respectively γ and θ ， Then the pose coordinate transformation can be obtained:

the pose matrix of the known robot end effector relative to the fixed reference coordinate system o-xyz:

solve this matrix equation α、β 、γ 、θ 、．

according to the solution α、β And L2, ts2o and to1o are solved. The coordinates of S2 and O1 in the fixed reference coordinate system o-xyz are obtained, and the following formula is obtained:

in the fixed platform ob1b3, ob1b3 is a known triangle, and it is easy to get the other two rod lengths:

the forward solution process is exactly opposite to the inverse solution process: the joint variables L1, L2, L3 and L1 of the known robot γ 、θ ， Solve the robot end pose matrix.

according to the joint variables L1, L2 and L3, as well as the coordinates of B1, O and B3 points, the column distance equation can be solved to the rotation angle of L2 branch around X axis and Y axis α and β ． The solution equation is as follows:

according to the obtained α and β ， The known L2 γ 、θ By substituting the in place pose change matrix, the end pose matrix of the robot can be solved to complete the forward kinematics solution.

3.2 forward and inverse solutions to the multiple solution problem

there is a multiple solution problem of the equation in the process of solving the forward and inverse equations of the parallel robot: in the forward solution calculation, a set of real axis coordinates corresponds to two sets of virtual axis coordinates, of which a set of virtual axis coordinates is located above the static platform (plane) and can be omitted; In the inverse solution calculation, a group of virtual axis coordinates corresponds to two groups of real axis coordinates. Because the system has the single axis manual function of the drive axis (real axis), if a group of real axis solutions calculated by the inverse solution are directly omitted, the servo axis tracking error of the system will be out of bounds during the manual automatic switching. The reason for this situation is that the real axis may enter the abandoned real axis solution space when the single axis is manual, while the system does not perform the inverse solution calculation when the real axis is manual, Therefore, no alarm event will occur; When switching to the automatic state, the inverse solution takes effect, but the real axis solution is another group of solutions. Due to the great difference between the theoretical value and the actual value, the following error is out of bounds alarm (as shown in Figure 4).

Figure 4 Schematic diagram of manual automatic switching process

in order to solve the problem of multiple solutions of inverse solution, variables are introduced into the inverse solution module to read the real-time coordinates of the real axis (), which is compared with the calculated value of inverse solution (), Take the group of solutions closest to the real-time coordinates of the real axis as the output, and drive the real axis movement after processing. In order to save calculation time, this judgment process only takes effect when the single axis of the real axis is manually switched to the automatic state. After judgment, the analytical expression of the selected group of solutions is used to calculate the real axis coordinates until the next switch.

3.3 rotation axis "zero crossing" problem

when the fourth axis (rotation axis) passes through the 0 ° point, There will be sudden reversal: when the parallel robot is processing, it needs to move from 359.999 ° to 0.001 ° in some positions, and the processing process only needs the rotation axis to move 0.002 °, while the actual situation is that the rotation axis rotates a large arc of 359.998 °. The reason for this phenomenon is that the system treats the rotation axis as a linear axis, while the inverse solution calculation can only calculate the half closed and half open interval from 0 ° to 360 °, Rotate the axis to 360 ° and it will return to zero. This is not necessary for processing, and the system cannot complete a cycle of rotation in an interpolation cycle.

to solve this problem, subtract the calculated value of the reverse solution of the fourth axis from the current real-time coordinates, and then correct the calculated value according to the following formula:

Where is the corrected coordinate of the fourth axis, the calculated value of the coordinates of the fourth axis before correction, and the current real-time coordinates of the fourth axis, || for rounding up symbol.

3.4 machining curve generation problem

for 5-DOF machining, manual programming is very difficult, and the existing computer-aided manufacturing (CAM) software does not have a module for parallel mechanism. In order to use the existing CAM software for machining programming, a conversion module is added to the positive and negative solution module of pose, The structure given in cam and the input and output coordinates of the positive and negative solutions are transformed into each other. The operator can customize different conversion modules according to different cam software.

the displacement measurement of the domestic X and Y experimental machines with the five axis double pendulum structure of UG is a vernier structure, which adopts a mechanical positioning mechanism, and the measurement force is displayed as a pointer. Although some experimental machines adopt a digital display mode, its displacement positioning adopts a mechanical structure, with low accuracy, It is not applicable to high-precision measurement, Z, a, B as an example to illustrate the conversion process of the conversion module. First, in order to simplify the conversion process, set the offset of the 4th and 5th axes to 0, so that the tool length is not considered during the conversion of UG, so the x, y, Z in the UG output code is consistent with the requirements of the parallel structure, and there is no need to convert again. We only need to convert the tool axis attitude: set the projection of the unit tool axis vector on the X axis as I, and the projection on the Y axis as J The projection of the Z axis is k, a is the main swing angle, B is the auxiliary swing angle, and the tool axis is on the Z axis. Through mathematical derivation, the tool axis vector and a The relationship of B is as follows (where I, J and K are the modules of vectors I, J and K in turn, and a and B are the angular values of angles a and B in turn):

similarly, the conversion relationship of the positive solution module can be obtained as:

4 Design of the InterpolationAlgorithm

laser processing with parallel machine is mainly to control the movement of laser focus, And keep the laser in the normal direction of the current machining point of the workpiece. The parallel robot is composed of complex spatial mechanisms. Coordinate transformation is required between the laser optical axis pose parameters and the driving coordinate position of the control system, so as to obtain the geometric description required for control. Therefore, The interpolation method of parallel robot is also different from traditional numerical control.

the interpolation calculation of this system discretizes the motion trajectory at the end of the mechanism into a series of small line segments connected end to end: given an interpolation cycle (as a system parameter, which can be set by the user), according to the trajectory feed speed given in the instruction, calculate the displacement increment in each axis direction separately, and discretize the continuous trajectory into several discrete point sequences, Then the S-type acceleration and deceleration algorithm is applied to modify the interpolation step size. The discrete points generated by interpolation are mapped to the joint space by the inverse solution module, and the system drives the servo motor to move according to the calculation results of the inverse solution module.

5 design and realization of the laser processing module

the parallel laser processing NC system also includes the laser process module, This module is used to set and adjust laser processing parameters, including laser power control, defocus control, offset control, laser state and other parameters, as shown in Figure 5.

Figure 5 laser processing parameter interface

laser power setting parameters are sent to the laser through the axis control board, and the laser takes this parameter as the amplitude of the output power curve. Defocus refers to the distance between the focal plane and the upper surface of the weldment, Here, the defocusing amount is used as a correction parameter for the distance between the laser focus and the rotation center of the fifth axis, and is transmitted to the forward and reverse solution module. The distance from the beam center to the processing track is called the offset. The offset setting is used to promote the logistics industry to low emission, pollution-free, and recyclable processing in XY, YZ, ZX planes. Here, it is specified that the offset of the beam center on the right side of the processing track is positive, The left side is negative.

6 experiment and conclusion

portable coordinate measuring machine is used to detect the accuracy of the end of the parallel robot. The end of the parallel robot moves from (0; 400; 0; 0; 0) to (1000) along a straight line