仿型测量数据的预处理

Data Preprocessing of Profile Measurment

对仿型复杂的物体进行数控加工,可通过测量系统对物体几何形状进行仿型测量从而完成几何形状的数字化,测量后得到的数据往往有各种各样的误差,此类数字化数据的预处理是提高最终加工精度和成品质量的重要手段。以封闭曲面实物模型为对象,采用弦高差方法对数据异常点进行剔除,提出了一种改进的减测量头半径算法用于离散数据的处理,将奇异值分解降噪方法应用到测量数据的平滑处理,并提出了一种有效的基于曲率的特征点提取和离散数据简化方法,仿真结果证明,该方法对处理测量数据有很好的效果,有效地降低了数据噪声和曲线拐点处的轨迹误差。

In order to process objects with complex profil by conduct numerical control machine,digitization of objects＇geometric shape can be completed by profile measuring objects＇geometry with measuring system.However there are always various errors of the data acquired by the method;therefore preprocessing such digitalized data is an important way to improve the eventual processing accuracy and end product quality.For objects with closed-curve surface,data exception was eliminated with chordal deviation method.A modified reduced measuring probe-radius was proposed to deal with scattered data.Method of noise reduction in singular value decomposition was applied to smooth processing measuring data.A method of scattered data simplication was put forward based on curvature characteristic extraction.And profile measuring results showed that this method works well in dealing with profile measuring,and it effectively reduced the data noise and trajectory error at the curve inflexion.