NGMV Control Toolbox

Basic Steps in NGMV Control Design

The typical NGMV control design procedure involves the following steps:

1. Define the system model. This will often be given as a Simulink subsystem but can also have an equation-based or other structured form. Split the model into a series connection of Nonlinear Black-box - Linear Transfer - Output Delay subsystems. For stability, the blackbox component must be open-loop stable, however the linear part can include unstable modes.
2. If known, specify the stochastic disturbance models representative of the actual disturbances affecting the system. In case the models are not known, set the models to low-pass filters and treat as design parameters.
3. Repeat step 2 for the reference models. For example, if a sequence of set-point step changes is expected, this can be modelled by an integrator driven by white noise. In general, use the "generating function" for any particular signal of interest.
4. Build a Simulink model of the system and insert the NGMV controller block from the NGMV Control Blockset. Look under mask and replace the NLSS model with the delay-free model of the system. Alternatively, for simpler or equation-based plant models, you can define the NLSS object using the nlss routine.
5. If a linear controller C0 stabilizing the delay-free plant is known, set the initial weighting choice as Pc = C0 and Fck = -I. Otherwise, start the design by defining Pc as a lag term and Fck as a constant or as a lead term, experimenting with the gains.
6. Perform the controller synthesis and run the simulation. If the performance is satisfactory, then stop; otherwise go back to step 5 and modify the design. You may also need to update the models in steps 2 and 3.

Common Terms Used in NGMV Control

Reading More About NGMV Control