Grey‐Box Model‐Based Vibration Isolation Using a Dielectric Elastomer Actuator

Abstract

AbstractThis contribution reviews the use of a grey‐box model of a tubular dielectric elastomer actuator for active vibration isolation of a payload, subject to ground vibrations. Two different scenarios, measurable ground vibration plus varying payload and unmeasurable ground vibration with constant payload, are examined. The choice of control approach is tailored to the two scenarios. The overall aims of the work are to (i) provide insight into the possible usefulness of the grey‐box model for active vibration isolation (AVI) for the two different scenarios, and (ii) to provide a qualitative assessment of the practicalities of each control approach examined. The initial part of the paper provides a validation of the grey‐box model for three different sizes of tubular DE actuator indicating that the model is good over the frequency range of interest. For the first scenario, measurable ground vibration with time‐varying payload, adaptive feedforward control, which is an accepted approach for AVI with a measureable disturbance, is used. Here the grey‐box model is only used within the cancellation path, which filters the reference signal by the plant dynamics in order to provide stable estimator performance. As the plant dynamics change with the changing payload the possibility of the estimator becoming unstable arises. Auto‐tuning of the grey‐box model as the payload changes, and hence the cancellation path, is shown to maintain stable estimator performance, though the auto‐tuning procedure is currently taking too long. For the second scenario the ground vibrations are assumed unmeasurable though the payload is assumed constant. The internal model control (IMC) approach was chosen not only because its design is model dependent but also to generally consider the appropriateness of IMC for active vibration applications. Process/model mismatch studies provided a choice of robustness filter time constant to satisfy stability considerations for all three actuators. Compared to adaptive feedforward control the vibration attenuation performance is likely to be slightly inferior because there is always a tendency to be conservative with the choice of robustness filter. Overall, the IMC approach tends to provide a degree of transparency in the design process as well as providing greater flexibility, which implies it is a useful approach for active vibration control applications, as long as a good model is available.