Adaptive derivative estimation for DSP-based acceleration measurement

Differentiation by means of digital signal processing is needed for estimating the rate of change or slope of digitized signals in many motion control and automation applications, including acceleration measurement. Since the ideal differentiator is unrealizable, several kinds of approximations have been proposed in the literature. In this work, we propose a digital differentiation scheme which is flexible and in many cases avoids the unfavourable noise properties of wideband approximation approaches. This is achieved by cascading an integrator and a FIR filter in an adaptive loop such that the filter adapts as an application-specific differentiator. Examples of operation with different signal classes are given, including piecewise polynomials, frequently encountered in motion control applications.