基于频率分析的实时滤波器研究

Study of a Real Time Filter Based on Frequency Analysis

大射电望远镜FAST关键技术研究的模型实验中,采用莱卡全站仪构建了馈源位姿测量系统。针对测量数据的噪声,提出了一种基于频率分析的实时滤波方法,能够有效地跟踪大范围机动目标的运动。该滤波器在传统多项式拟合滤波器的基础上,采用动态频率分析和预测技术,使得滤波时滞只有采样周期的三分之一左右,适合高实时性要求的应用。在FAST馈源精调平台的振动控制实验中,应用该滤波器显著提高了控制效果。

In the model test addressing key techniques of a Large Radio Telescope known as FAST, the feedback measurement system is constructed using Leica total stations. For the measurement noise problem, this note proposes a real time filter based on frequency analysis, which is capable of tracking large-scale maneuvering target. Derived from traditional least square polynomial fit filter, the new filter employs dynamic frequency analysis and prediction technique, making the filter-introduced-delay only one third of the sampling period and particularly suits applications where real time is heavily demanded. Application of the proposed filter improves control effect remarkably in the model test of the fine feed-adjustment platform for FAST.