基于ICA的动态测量误差分解重构方法

Dynamic Measurement Error Decomposition and Reconstruction Method Based on ICA

提出了利用独立成分分析(ICA)方法对动态测量误差进行分解重构的方法。详细介绍了利用ICA方法进行误差分解重构的基本算法,利用三坐标测量机和双频激光干涉仪设计了动态测量误差测量装置,并进行了实验。分别利用自适应滤波、小波分析和ICA方法对动态误差数据进行处理,分离重构线性变化、周期变化的系统误差和动态随机误差。研究结果表明,要精确分析动态测量误差和辨别误差组成成分,必须充分应用测量得到的误差总体样本进行分析。与自适应滤波、小波分析两种方法相比,利用ICA方法可以更有效地精确分离重构线性变化、周期变化的系统误差和动态随机误差。

The method that ICA was used to decompose and reconstruct dynamic measurement errors was researched.The basic algorithm of ICA to decompose and reconstruct dynamic errors was elabrated detailedly.A measuring device for dynamic measurement errors was built by using the coordinate measuring machine and dual-frequency laser interferometer.The dynamic error separating experiments were done.The dynamic error data was processed respectively by using the method of adaptive filtering,wavelet analysis and ICA,which was aimed to separate and reconstruct the systematic errors with linear change and with periodic change and the dynamic random errors.The experimental results prove that the overall sample of the measured dynamic errors must be fully applied for the precise analysis and error component indentification of the measured dynamic errors.Compared with the previous two methods,the systematic error with linear change and with periodic change and the dynamic random error can be separated and reconstructed more effectively and accurately by using ICA.