圆度测量误差分离技术解决ICA的不确定性问题

独立分量分析方法(Independent Component Analysis，简称ICA)在国内尚属一门新型的方法。在介绍ICA的无噪声模型、原理、预处理、非高斯性量度以及ICA快速定点算法的基础上，重点讨论了ICA的不确定性在圆度误差分离中的处理方法。仿真结果表明，基于独立分量分析的圆度误差分离技术比传统的频域法和时域法均简单、实用、高效。同时，由ICA分离出的信号的不确定性问题得到了很好的解决。

Solution to ICA ambiguities in roundness measurement error separation

Independent Component Analysis (ICA) is still a new method in our country. The paper introduces ICA noiseless model, principles, pre-processing, measures of non-Gaussianity, and fast ICA algorithm. The paper stresses the method of overcoming the ambiguities of ICA in roundness error separation. The emulation result shows that the roundness error separation technique (EST) based on ICA presents a more simple, more practical and more effective method than the traditional frequency-domain and time-domain method. The method offers a better solution to ambiguities as discussed in this paper.