Recursive Observation Evidence Fusion Method for Acoustic Resonance-Based Level Detection

Abstract

Acoustic resonance-based level measurement principle needs to extract a sequence of resonance frequencies (RFs) from the synthesis wave and then calculate level height via this RF sequence. However, in practice, the uncertain disturbances in the measurement environment usually lead to the signal distortion of the collected synthesis wave. In this case, some RF points in the sequence are inevitably missed which causes the nonnegligible calculation error. Hence, based on the Dempster-Shafer evidence theory (DST), this paper presents a recursive evidence fusion method to combine multiple RF sequences in a row. It provides a natural way to supplement the missed RF points and also significantly improve the measurement accuracy even if the observed RF sequences are all intact. That is to say, regardless of the missing case or intact case, the proposed fusion method always has high performance. Finally, the comparative experiments of level detection show the level gauge using this method is robust for the sequence with the missing RF points and can further provide higher measurement accuracy than the single RF sequence-based and digital filteringbased level detection methods.

A Note on Incomplete Information Modeling in the Evidence Theory

“…Among the aforementioned methods, DST is an effective and a typical method for uncertain information processing in decision making [3], pattern recognition [18], [20] and so on. In DST framework, the uncertain information can be modeled as mass function, the uncertainty can be measured by belief entropy [2], [5], [12], [15] or other factor [10], the uncertain information can be fused with Dempster's rule of combination as well as the other extended rules [28], [31]. To address the incomplete information processing, the uncertain information processing model in the classical closed world scopes will be extended to the open world assumption in the evidence theory.…”