大尺寸测量中多传感器的融合

Multiple sensor fusion in large scale measurement

为了实现大型工件的几何量测量,建立了多测量站的测量网络,研究了相应的多传感器信息融合方法以充分融合多测量站的信息。基于奇异值分解的解析形式将多测量站的测量信息进行空间数据配准,统一到全局坐标系下。再建立各测量站的测量模型,得到各测量站的协方差矩阵和雅可比矩阵,按矩阵加权线性最小方差最优融合准则,求取权重矩阵,对多测量站的测量数据有效融合。仿真实验给出了某一空间的平均测量结果,全站仪、经纬仪测量系统和激光跟踪仪的测量误差分别为0.7451 mm、0.0449 mm和0.0327 mm,融合测量误差降低为0.0128 mm。最后,将该方法实际应用于隧道构件的测量网络,理论、仿真和实际实验都验证了本文提出的融合规则可有效提高融合精度,融合精度优于各局部精度,增强了测量网络的冗余性和可靠性。

In order to realize the geometric measurement of large scale workpiece, a measuring net for multiple measurement stations was established. With the aim of the information fusion of multiple measurement station, a relevant multi-sensor information fusion method was researched. The spatial data of multiple measurement stations were registrated based on singular value decomposition to unify as global coordinate system. Then, the measuring model of each measurement station was established to get its covariance matrix and Jacobian matrix, so the weight matrix was obtained by the optimum fusion criterion of matrix weighting linearity minimum variance for fusing effectively measured data of multiple measurement station. A simulation experiment was carried out and the average measuring results were given as follow,the measurement errors of total station, theodolite and laser tracker are 0. 745 1 mm, 0. 044 9 mm and 0. 032 7 mm, respectively, and the fusion measurement error is 0. 012 8 mm. Finally, the fusion method was applied to measure a net in tunnel component. The validity of fusion criterion for precision is proved by theory, simulation and practical experiments, and the fusion precision is superior to each partial precision.