Title:
|
Softly switched model predictive control : generic development and application to water supply and distribution systems
|
This thesis is concerned with the softly switched model predictive control (SS-MPC) theory
and its application to optimal operation of water supply and distribution systems.
For systems having control input saturation, a softly switched model predictive control
scheme with two specific methods is proposed in order to achieve better switching transients
than MPC controller hard switching. Stability properties of the SS-MPC are investigated and
sufficient asymptotic stabilization conditions are derived. Taking state constraints and system
uncertainty into account, a robustly feasible soft switching process is designed by bridging
the old and new MPC controllers through a sequence of linked invariant sets. An invariant set
based fast switching method is also developed in order to complete the switching process as
fast as possible in the case of infeasible MPC controller hard switching while sacrificing the
switching transients. The developed SS-MPC methodology is extended to the scope of hybrid
predictive control, where a class of nonlinear hybrid dynamical systems are considered.
For operational control purposes, a piecewise affine modelling method for water supply
and distribution systems is proposed. The proposed control model enables predictive control
strategies to be formulated as mixed integer linear inequalities and ensures that the obtained
optimization solutions are global optima. The effectiveness of the proposed control modelling
method and the softly switched model predictive control mechanism are finally verified by
applying them to a simulated benchmark water supply and distribution network.
|