Adaptive shaper-based filters for fast dynamic filtering of load cell measurements

Abstract

This paper proposes a novel adaptive model-based filtering technique for dynamic mass measurement through load cells, named Adaptive Shaper-Based Filter (ASBF). Filtering is performed by convolving the oscillating load cell signal with a baseline of few impulses (usually 3) to rapidly compensate for its underdamped behaviour. The correct timings and amplitudes of the impulses are computed by means of a simplified dynamic model of the load cell (that requires the knowledge of the natural frequency and the damping ratio of the empty cell), and the filter output itself (i.e. the estimated mass). The load cell is modelled through the theory of systems with variable mass as a linear time-variant system and the variations of frequency and damping are predicted; in this way, the filter zeros track the load cell poles to cancel them. Robustness specifications are included in the filter design to account for the unavoidable uncertainty and estimation errors. Given the non-rational transfer function of the proposed Adaptive Shaper Based Filter, whose poles have an infinite and negative real part, rapid settling time is ensured. Experimental assessment is proposed by comparing the results with some benchmark filters.