A unified approach for Composite filter and Controller fusion

A feedback control using both acceleration and displacement measured signal is very effective to let any machines trace to accurate motion. In ideal case, when the displacement output signal is equals to the desired displacement signal, the output acceleration signal is also equals to the desired acceleration signal. However, this is not true in anytime for real situations. Because, there are lucks of friction, model error and measurement noise in real controlled plant. Against this problem, Some methods are proposed to realize feedback control using both signals. Composite filter based control (Stoten, 2001) is an example. This algorithm reproduces acceleration signals and displacement signal from measured row displacement signal and acceleration signal. For high frequency signal, reproduced signals are mainly led from acceleration signal. For low frequency signal, they are mainly led from displacement signal. Composite filter method combine signals in sensor signal level. Then, controller uses reproduced signal to calculate control input for a plant. Another is Controller fusion (Shimono and Tagawa, 2013). This scheme also let control signal to be led from both acceleration signal and displacement signal. This method Also has a same character as Composite filter. However, Controller fusion combine signals in Control input level. So, Composite filter and Controller fusion have a same purpose and same character, but the approaches are different. In this paper, we discuss three points. First Composite filter and Controller fusion algorithms are introduced. Second, relationships between these methods are shown. And also possibility to unify these methods is discussed. Finally, the unified approach scheme is validated by hydraulic cylinder control tests.