【作者】 牟鲁西；

【导师】 熊有伦； 尹周平；

【作者基本信息】 华中科技大学 ， 机械电子工程， 2012， 博士

【摘要】 多传感器集成的在机测量技术是实现复杂曲面零件设计-加工-测量一体化制造的关键支撑技术之一,它涵盖了传感器的标定、规划、数据处理等单一传感器技术内容,同时又引入了多传感器系统全局标定、组合测量策略与数据处理等新问题,并且需要考虑在机测量的高效率的要求。本学位论文针对叶片等复杂曲面的多轴加工中的在机测量技术,重点探讨了接触式测头-关节臂激光扫描仪集成的在机测量系统的全局标定、接触式测点规划、曲面误差分析及组合测量等关键技术,主要研究工作和创新之处体现在：(1)提出了接触式测头-关节臂激光扫描仪集成的复杂曲面在机组合测量系统全局标定新方法,分析了已有的全局标定方法的优缺点,选择简单六面体零件为标定靶标,建立了基于单位四元数表达的全局标定模型,根据数据的聚类特性建立了单位四元数球面上的点数据和二值图像的映射关系,应用数字图像处理中灰度图像矩的方法进行标定数据的优化,得到最优的标定矩阵。仿真和实验结果表明,该方法选计算量小、标定速度快精度高,克服了复杂特征靶标和自标定两类方法需要大量的数据处理的不足,更适合于接触测头-关节臂激光扫描仪在机组合的全局标定；(2)提出了基于曲面质量的截面相似曲面接触式测点规划算法,定义了曲线质量、曲面质量和截面相似曲面,建立了曲面质量坐标系下的曲面表达模型,通过将曲面质量均匀化实现曲面按照曲率的展开。在此基础上,应用扩展的Hammersley序列方法规划出接触式测量点的位置。仿真和实验的结果与常用的等参数法、随机法进行了比较,结果表明该算法可以在相同的测点数目下,取得了比其他方法更高的测量精度,体现了用较少的测点数目获得较高的测量精度的特点,非常适合快速高精度的在机测量；(3)研究了基于误差分析的复杂曲面零件接触式测头-关节臂激光扫描仪组合测量方法,建立了曲面误差和轮廓误差的计算方法,并通过分析轮廓误差等值线确定了重采样区域,利用形态学骨架线算法抽取出采样路径并由接触式测头完成二次测量,两次测量结果集成起来为加工参数的选取提供依据。实验表明,根据本文测量结果对曲面误差不同区域选择不同的精加工参数,可以显著提高曲面的加工质量。最后,在单元技术研究的基础上,初步建立了面向复杂曲面加工的接触式测头-关节臂激光扫描仪集成的在机组合测量系统,开发了功能模块的人机交互界面,并在该系统下实现了测量系统全局标定、截面相似曲面测点规划和曲面零件误差分布分析等研究内容的工程应用。更多还原

【Abstract】 As one of key technology of design, manufacturing and measurement integration manufacturing for complex surfaces, multisensor integration on-machine metrology not only includes many traditional contents for single sensor such as sensors calibration, sensor planning and measurement data processing, but the metrology system global calibration, combination measurement strategy and multisensor data processing etc. This dissertation address some important problems of developing an on multisensor on-machine metrology system in NC machining process. These problems include sensors global calibration, tactile sensor sample planning and surface error distribution analysis based multisensor combination measurement. The major research works and contributions of the dissertation are introduced as follows:(1) A global calibration method for tactile-laser scanning metrology system is proposed. After analyzing merits and drawbacks of present methods, a mathematical model based on unit quaternion is constituted. To get the optimal calibration data, the mathematical morphological image processing method is involved by clustering. In this method, the calibration target is a simple hexahedral part. Because of simple geometric features, short computing time and rapid calibration speed, this method is a good choice for tactile-laser scanning calibration.(2) A tactile sample planning method for similar section surfaces based on surface mass is proposed. Firstly, curve mass and surface mass are defined. Based on this, similar section surface is discussed and expressed with surface mass coordinate system. Then surface can be flattened by averaging surface mass. The sample point position is planed by Hammersley sequence with the flattened surface. Simulation and cased study are shown that this sample method can acquire higher measurement accuracy with same sample points number.(3) Most machined surface is with nonuniform error, and surface with isotonic error need to be assigned different parameters in finish machining process, to ensure machining quality. However, as requirement of high accuracy and rapid speed, single sensor is impossible to handle this problem. A tactile-laser scanner integration metrology strategy applied to analyze surface error distribution is proposed in this dissertation. The strategy starts with scanning the whole surface using laser scanner, then profile error is derived by processing point cloud. As introducing error contour map, the zone out of tolerance is extracted, which need to be re-inspected with tactile probe. The probe path is planned by applying morphological skeleton algorithm, when point cloud is transformed to 3D binary image. Experiment is performed on blade surface grinding process and validates the method effect.At last, based on these works and the present software and hardware platforms in State Key Laboratory of Digital Manufacturing Equipment and Technology, the mutisensor-OIMS (multisensor on-machine intelligent measuring system) is basically developed. To validate performance, some experiments about global calibration, sample planning and surface error analysis are implemented.更多还原