Digital Filter Threshold Algorithm Based on Clustering Analysis

A new digital filter algorithm is proposed. The concept of support and effective support quantity is introduced to survey the degree of measured data reliability. How to choose the filter threshold is key to determine support quantity. First measured data is classified, and then different filter thresholds are determined according to the different classification numbers. It can repress the proliferation of abnormal data influence on the case of not losing useful information. Because of the introduction of the effective support quantity, it can remove the remnant influence of abnormal data further. The data experiment confirms that when some abnormal data appear continuously and the true value jumps, it has good anti-interference and fast tracking ability comparing with same kinds of algorithms.