【作者】 高延峰；

【导师】 张华；

【作者基本信息】 南昌大学 ， 机械电子工程， 2008， 博士

【摘要】 随着工业技术的不断发展,大型结构件的应用越来越多。在这些大型结构件的焊接生产中存在大量的弯曲角焊缝和折线角焊缝,实现这些焊缝的自动化焊接对于提高生产效率和保证产品质量具有非常重要的意义。这些工件结构庞大,很多焊接作业必须在现场进行,难以采用手臂式机器人进行自动焊接,也难以采用编程或示教的方式进行焊缝跟踪;另外在对这些焊缝进行自动焊接时,不仅要控制焊枪跟踪焊缝移动,同时还要调整焊枪的倾角,以保证焊接质量。为此,本文以轮式移动焊接机器人为平台,解决大范围移动焊接问题;同时采用旋转电弧作为传感器,进行焊枪偏差识别与倾角检测,从而实现大型构件角焊缝自动焊接。研究内容主要包括:焊接电流信号的滤波处理;焊枪偏差与倾角检测;水平弯曲角焊缝、具有直角转弯的角焊缝和水平折线角焊缝跟踪及焊枪倾角调整控制器的设计。针对焊接电流信号易受外界噪声干扰影响的问题,本文提出以软阈值小波滤波为核心的组合滤波算法,对旋转电弧传感器采集到的电流信号进行滤波处理,使电流波形得到了明显地改善,提高了电流信号的信噪比,为焊枪的偏差和倾角检测奠定了基础。对特征谐波法在焊枪偏差和倾角检测中的应用问题进行了理论研究并给出了试验验证,发现该方法可以用来对焊枪偏差和倾角进行同时检测,扩展了特征谐波法的应用范围。指出在焊枪倾角不为零时,传统的采用一次谐波幅值检测焊枪偏差大小的方法会产生较大误差。为了提高焊枪偏差的检测精度,首次提出特征平面法对焊枪偏差和倾角同时进行检测。该方法充分利用旋转电弧传感器采集得到的焊接电流信息,采用最小二乘原理在三维空间构建特征平面,通过求特征平面与坐标平面交线的斜率,将焊枪偏差和焊枪倾角信息分别投影到两个正交的平面上,实现了二者的解耦。针对焊接过程难以建模的问题,采用分段控制策略设计控制器对水平滑块进行控制,该控制器在大偏差时采用比例控制,在小偏差时采用参数自调整模糊控制,并利用免疫反馈规律对比例因子进行修正,实现了直线焊缝、小曲率焊缝的跟踪。针对水平弯曲角焊缝跟踪的特点,设计模糊控制器对水平滑块和车轮进行协调控制,采用焊枪偏差信息获得机器人运动方向和焊缝走向之间的角度偏差,控制车轮转弯,并采用预测控制原理对控制量进行修正,实现水平弯曲角焊缝的平滑跟踪。在船舱格子形角焊缝焊接中,存在90度的直角转弯,给跟踪控制带来了很大难度。文中详细介绍了如何利用焊枪倾角信息检测拐角点,利用超声波传感器测量前方焊缝位置的方法。并对机器人的运动学模型进行推导,设计控制器实现此类焊缝的跟踪焊接。对于变化角度较大的折线角焊缝跟踪问题,设计带有转动关节的焊枪,并将该焊枪置于移动机器人平台上,设计控制器利用焊枪偏差和倾角信息,首次实现在焊缝跟踪的同时对焊枪的倾角进行调整。最后通过实际焊接试验证明了本文研究工作的有效性。另外,本文所设计的焊缝跟踪系统在九江同方江新造船厂进行了生产现场实际试用,取得了预期的效果。更多还原

【Abstract】 With the development of modern industry, there is a growing application of large welding structures, so there are many curved fillets and broken-lined fillets need to be welded. For improvement of production efficiency and guarantee of product quality, it is essential to realize arc welding automation in the manufacturing of these workpieces. Because of the large scale, these structures must been welded in out of workshop, so the arm welding robot can not be used and the welding paths are not easily pre-programmed. Moreover it is necessary to control welding torch moving along with the seams and adjust the torch’s inclination simultaneously for obtain highly welding quality when tracking these seams.In this dissertation, a wheeled mobile robotic arc welding seam tracking system is developed, and the system uses a rational arc sensor to detect the deviation and inclination of welding torch. The main research contents are: the filtering of welding currents; the identification of welding torch deviation and inclination; the control methods designing for automatic tracking curved fillets, lattice box fillets and broken-lined fillets.In consideration of the fact that welding current signals are often disturbed by outside noises, soft threshold wavelet filtering method is applied to process the welding current signals, that make the welding current shape is obviously smoothed and the signal-to-noise ratio is much improved.It is researched in theory and experiments that the character harmonic method can be used to detect deviation and inclination of welding torch simultaneously. The results show that the traditional method of using the amplitude of one-order harmonic to detect deviation would result in big error when the inclination is not zero. For improve the detection precition, the character plane method is developed in this dissertation, which fits the arc currents to a plane in three dimensions using a least-square fitting method. The deviation of welding torch is obtained through calculating the intersection line slope of the fitting plane with the YZ plane, and the inclination is calculated through the intersection line slope of the fitting plane with the XZ plane. Because of the deviation and inclination is projected to two orthogonal planes, so they can be decoupled.A multi-segment controller is designed to track lined fillet and small curvature fillet, which is composed of a proportion controller and a self-turning fuzzy controller to control horizontal slider. When the deviation is large the proportion controller will be used to quickly reduce the deviation, otherwirse the self-turning fuzzy controller will be used to avoid overshoot and achieve smooth tracking, and the immunity feedback method is applied to modify proportion gene of fuzzy controller. For the curved seam tracking problem a predictive fuzzy controller is designed to coordinate control cross-slider and wheels. This method uses the information of weld torch deviation to estimate the robot’s orientation errors, and uses predictive control theories to offset the control outputs. The experiment results show that the presented method is valid to track curved fillets.In the welding of lattice box fillets, there are many right-angle corners that make tracking seam become very difficult. The method of using welding torch inclination to detect the corner and using ultrasonic sensor to detect the position of frontage seam is detail discussed in this dissertation. The mathematical model of mobile robot is built and the controller is designed to tracking these seams.For the broken-lined welding seams tracking problem, a welding torch with rotation axis is designed which is placed in the flat of mobile robot. A controller is proposed that use of the welding torch deviation and inclination information as input values, meantime use of the mathematical model of mobile robot to control rotation axis, horizontal silder and wheels to track welding seam and adjust torch inclination.In the last though experiment results show that the feasibility and validity of thisresearch work. In addition, the designed arc welding seam tracking system was triedout in Jiujiang Tongfang Jiangxin Shipyard Co., Ltd, and anticipant result is acquired.更多还原