Filter design using data fusion for a pneumatic control valve

Abstract

This paper presents a filter design technique for a pneumatic valve using data fusion techniques. The objective of this paper is to examine the suppression of the effect of parameters causing deviation from normal system performance using the technique of data fusion over time. The output of a system affected by inherited noise is processed by applying operations such as finding the statistical variance, time warping, interpolation, and extrapolation. These techniques are used to compute the transfer function of the filter, which when cascaded with the system will suppress the effect of noise on the process. The operation of the control valve is affected by characteristics such as stiction, structural deformation, etc. The characteristics of the system are studied and data for multiple time instances are extracted to carry out fusion across time by dynamic time warping. Tests show that the filter presented here can suppress the effects of stiction and mechanical deformation on the output signal.