The main parameter of the finished workpiece E spinning blank is the thinning rate: (cylinder): to? blank thickness, degree. The thinning rate is different depending on the type of material. Generally, the shape of the boss tube f ~ 190) is calculated in detail. The pressing part is combined with the loading and unloading robot to complete the problems of the package, local diffusion, and poor surface quality. The specific value is related to the raw material characteristics of the workpiece, the thickness of the spinning, the shape of the processing, and the like. For the coarse pressure roller a (see), the thickness of the spinning is large (accounting for about 85%), and the spinning resistance is large. We have modified the head wheel position relative to the precision wheel 6 to increase its strength and make it Conducive to the flow of metal in the spinning process, the middle master is also divided into two parts: the rough pressing mold 3 and the precision pressing mold 8. When designing the size, the influence of the rebound of the workpiece should be considered.

2 Spinning system structure and principle The main parameters of the machine tool are: () the center height of the spindle is 850mm; -j longitudinally fast a|longitudinal slow-moving i-profiling model tracking longitudinal fast-forward i value is greater than G cavity liquid pressure still, due to piston The rod 2 is fixed, and the cylinder 1 is also moved to the left as the contact 9, and due to the pressing action of the spring 8, the contact 9 abuts against the contour surface of the template, achieving a leftward contouring.

To the right, the opposite is true. When the contact 9 is raised to the limit position, the solenoid valve is closed, and the spool 6 is quickly moved to the left to move the cylinder 1 to the left end position. The master tracking device (shown) is composed of a one-way throttle control slide valve and a differential cylinder, and the coarse and fine master molds are fixed on the oil cylinder 1, and the two ends of the differential cylinder piston rod 2 are respectively fixed on the bed head and the tail top. On the seat. As the workpiece is stretched during the machining process, the lower end of the lever 6 swings to the left to move the spool 5 to the left, and the gap of the throttle groove gradually becomes smaller to zero. Since /A, >/A2, the cylinder 1 is connected to the valve body 4 to the left. When the workpiece is finished, (down to page 150) 4 Welding trajectory planning The control compensation amount obtained by the fuzzy neural network is perpendicular to the trajectory of the robot. Real-time path correction is performed in Cartesian space. Due to the limited image processing speed, the welding trajectory motion calculation consists of three levels. The highest level is the path planning level. At this level, the position vector at the interpolation point of teaching and programming is summed with the control compensation amount at regular intervals to obtain the target motion point. The second level is the conventional robot interpolation calculation, and the interpolation calculation is also performed in the Cartesian space. The third level is to transform the spatial interpolation point coordinates obtained by the second stage into the robot joint space, which is obtained by the normal robot kinematics inverse solution. In the joint space, the motion between the two points is linear interpolation.

Such a three-level motion calculation strategy has an interpolation period from coarse to fine, which can be called a compensation level, a spatial interpolation level, and an joint interpolation level, respectively.

5 Experimental results The robotic automatic arc welding system consists of a six-degree-of-freedom welding robot, a master control system and a vision system. The vision system consists of a narrowband filter, a CCD camera, and a TMS32Q20 real-time image processor.

The vision system takes eight available images per second. Therefore, the calculation period of the compensation stage is 125ms. The interpolation period of the robot space trajectory is 25ms, and the interpolation rate of the robot joint space is the servo update period of the DMC controller, that is, 1ms. Median filtering and Laplace in image processing. The operator's handler uses a 3X3 template to improve image processing speed. The time-consuming process in the image processing operation is the fast-sorting sub-range in the median filtering. If the 9X9 operation template is adopted, the processing speed is greatly slowed down, and the requirements of real-time control cannot be basically satisfied. In this paper, the median filtering process has been deleted, and after the binarization process, the processing speed of the isolated point elimination sub-process is greatly improved, and a satisfactory effect can also be obtained.

A large number of welding experiments show that the fuzzy neural network welding path correction system proposed in this paper is effective, and the system has good robustness and learning ability. Moreover, the fuzzy neural network has a strong generalization ability.

6 Conclusion Robot arc welding path correction can overcome the deviation of the actual weld from the pre-teaching and programming path caused by various uncertain factors.

Fuzzy neural network has the advantages of non-modeling and neural network self-learning ability of fuzzy reasoning. The three-level motion calculation strategy can effectively achieve real-time weld path correction.

(Finish)

Gear Oil Additive Package

It is a Gear Oil Additive Package which contains Antiwear EP Additive and other Lubricant Additive Component. In addition to providing excellent EP/AW performance, rust and corrosion resistance, it offers outstanding antifoaming and demulsifying properties. T4810 is suitable for blending with group I, II, III base oil and synthetic base stock includin PAO, ester and alkylnaphthalene etc.