Dynamic decoupling in motion systems using a gradient approximation-based algorithm

To reduce cross coupling in the control of multi-input multi-output (MIMO) motion systems having diagonal dominance, a dynamic decoupling design is studied. The dynamic decoupling is used in parallel to a nominal (and static) decoupling with the aim to support a straight-forward single-input single-output (SISO) control design. For a six degrees-of-freedom nano-accurate wafer stage, the off-diagonal entries of the dynamic decoupling matrix are given a fourth-order FIR filter structure as to counteract the coupled and dominant fourth-order plant behavior. The FIR filter coefficients are obtained using a gradient approximation-based algorithm which supports the means to obtain the optimized set of FIR coefficients through a limited set of perturbed-parameter experiments. In simulation, both performance as well as closed-loop stability are shown to be effectively dealt with.