柔性坐标测量机建模的泛函网络研究  

Research on functional networks of flexible coordinate measuring machine modeling

作　　者：郑大腾[1,2] 费业泰[1] 张梅[3] 

机构地区：[1]合肥工业大学仪器科学与光电工程学院,合肥230009 [2]皖西学院机械与电子工程系,六安237012 [3]安徽大学电子科学与技术学院,合肥230039

出　　处：《电子测量与仪器学报》2009年第4期33-37,共5页Journal of Electronic Measurement and Instrumentation

基　　金：国家自然科学基金(编号:50775062)资助项目

摘　　要：柔性坐标测量机的测量模型若用D-H方法和传统的神经网络建模,则对测量精度有一定的影响。为了进一步提高测量精度,该文对泛函网络的基函数、拓扑结构和学习算法进行了研究,并应用泛函网络建立柔性坐标测量机的测量模型。利用高精度的正交三坐标测量机进行单点坐标标定,采用标准锥窝法在测量空间上随机取点采样,实验结果表明,泛函网络建模结果与采样数据误差等级为0.1μm,这说明基于泛函网络建立柔性坐标测量机的测量模型是可行的。If the measurement model of flexible coordinate measuring machine（FCMM） is constructed with D-H methods and traditional neural networks, the measurement accuracy could be influenced. To improve the measurement accuracy, functional networks＇ basis functions, topological structure and learning algorithms are researched, and the measurement model based on the FCMM with functional networks is constructed. The single-point calibration model based on high-accuracy FCMM and the random sample method based on single-point cone socket in the measuring space are used. The experimental results show that the error from functional networks modeling results and sampled data reaches 0.1 μm. It shows that the measurement model of FCMM based on functional networks is feasible.

关 键 词：柔性坐标测量机 泛函网络 基函数 拓扑结构 

分 类 号：TH721[机械工程—仪器科学与技术]