Iterative active-set method for efficient on-line MPC design
Consider the LQ discrete-time control design problem as follows: \(J(x_0, \delta u_k)=\sum_{k=0}^\infty x^T_kQx_k+\delta u^T_kR\delta u_k \rightarrow\min\), s.t. \(x_{k+1}=Ax_k+B\delta u_k\), \(z_k=\delta u_k+Ex_k+d\), \(||z_k||_{\infty}\leq 1,\forall k\geq 0\), where \(k\in{\mathbb Z}=\{0,1,2,\dots{}\}\), \(x_k\in{\mathbb R}^n\) is the state vector, \(\delta u_k=u_{k+1}-u_k\in{\mathbb R}^m\) is the input increment, \(z_k\in{\mathbb R}^m\) is the vector subject to the pointwise-in-time constraints, \(Q=Q^T\geq 0\), \(R=R^T>0\), \((A,B,Q^{1\over 2})\) is stabilizable-detectable. A fast computational algorithm is presented solving the problem in the state-space by using the active set method and dynamic programming (d.p.). Updates require only \(pn^2\) operations after inclusion/deletion of an active set constraint by using a proposed simple d.p. procedure, where \(p\) is the time of the contraint inclusion/deletion. For the entire collection see [Zbl 0934.00030].
- Fast algorithm for a constrained infinite horizon LQ problem
- An online active set strategy to overcome the limitations of explicit MPC
- The explicit linear quadratic regulator for constrained systems
- Efficient active set optimization in triple mode MPC
- A fast MPC algorithm using nonfeasible active set methods
This page was built for publication: Iterative active-set method for efficient on-line MPC design
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4952764)
