Description
Hello and a very happy new year everyone,
I am developing MPC control using ACADO in Simulink. The first 3 inputs to the ACADO MPC S-function are int32 data type values (0 or 1) for signal names 'Control', 'Shifting' and 'Initialization'. While the 'Control' signal name enables the controller when value is 0 and disables it when the value is 1, I am unable to understand what is the definition and purpose of the other two signals ('Shifting' and 'Initialization'). Can someone help me with this ?
I was trying to find it in internet and could find this ACADO university assignment on 'inverted pendulum' where it is mentioned:
RTI: remember that ACADO generates a custom RTI algorithm to solve each OCP
instance, without iterating the SQP scheme each time until convergence (see literature for more information on the Real-Time Iteration scheme itself). As a result,
the sampling time of the NMPC scheme needs to be small enough to let the solver
converge over time and result in a stabilizing controller. Also the choice for the
initialization and shifting procedure is typically very important. Let ACADO shift
the optimization trajectories after each call to the solver by setting
input.shifting.strategy = 1;
which means that the last state will be reused at the end of the horizon.
I do understand that both the options are somehow related to how they affect the optimization problem being solved but I am failing to understand it in simpler terms.
Hope you all can help me out. Looking forward to your replies and kind and much appreciated help.
Cheers,
Nishant