L. Magni

A stabilizing model-based predictive control algorithm for nonlinear systems

Using distinct prediction and control horizons, nonlinear model-based predictive control can guarantee: (i) computational efficiency, (ii) enlargement of the stability domain and (iii) local optimality.

Stability and Robustness of Nonlinear Receding Horizon Control

The main design strategies for ensuring stability and robustness of nonlinear RH (Receding-Horizon) control systems are critically surveyed. In particular, the following algorithms with guaranteed closed-loop stability of the equilibrium are considered: the zero-state terminal constraint, the dual-mode RH controller, the infinite-horizon closed-loop costing, the quasi-infinite method, and the contractive constraint. For each algorithm, we analyse and compare feasibility, performance, and implementation issues. For what concerns robustness analysis and design, we consider: monotonicity-based robustness, inverse optimality robustness margins, nonlinear H ∞ RH design, and a new nonlinear RH design with local H ∞ recovery.

Brief Output feedback and tracking of nonlinear systems with model predictive control

This paper presents an output feedback Receding Horizon (RH) control algorithm for nonlinear discrete-time systems which solves the problem of tracking exogenous signals and asymptotically rejecting disturbances generated by a properly defined exosystem. The regulator is composed by an internal model of the exosystem and a stabilizing RH regulator. Some robustness results are also achieved in the case of constant references.

Reduction of Hyper- and Hypoglycemia during Two Months with a Wearable Artifi cial Pancreas from Dinner to Breakfast in Patients with Type 1 Diabetes

929-P Patient Responses to Interim Data from Cardiovascular Outcomes Trials: Results from an Online Patient Survey MANU V. VENKAT, RICHARD S. WOOD, ADAM S. BROWN, PHIN YOUNGE, LISA S. ROTENSTEIN, KELLY L. CLOSE, San Francisco, CA, Boston, MA The disclosure of interim data from ongoing clinical trials is usually discouraged, as it can alter participant behavior and threaten trial integrity. Current U.S. regulatory guidance requires long-term cardiovascular outcomes trials (CVOTs) for new T2DM drugs, but allows interim data to be disclosed to support approval. The purpose of this study was to examine how such disclosure could influence enrollment dynamics in a CVOT. An online survey from the diabetes market research company dQ&A was distributed to a panel of adult T2DM patients. Of the 1,984 total respondents with T2DM, 1,542 reported a history of CVD and/or being told by their healthcare provider that they are at elevated CVD risk. This represents a patient subgroup that is targeted for enrollment in most diabetes CVOTs. In the survey, respondents were described a hypothetical CVOT. Next, they were randomized to receive scenarios in which evidence of either an increase or decrease in CVD incidence was disclosed during the trial. In both scenarios, the drug was approved. All participants selected one of four choices regarding their subsequent actions (see table below).

Robust Model Predictive Control Algorithms for Nonlinear Systems: an Input-to-State Stability Approach

This paper presents and compares two robust MPC controllers for constrained nonlinear systems based on the minimization of a nominal performance index. Under suitable modifications of the constraints of the Finite Horizon Optimization Control Problems (FHOCP), the derived controllers ensure that the closed loop system is Input-to-State Stable (ISS) with a robust invariant region, with relation to additive uncertainty/disturbance. Assuming smoothness of the model function and of the ingredients of the FHOCP, the effect of each admissible disturbance in the predictions is considered and taken into account by the inclusion in the problem formulation of tighter state and terminal constraints. A simulation example shows the potentiality of both the algorithms and highlights their complementary aspects.

Robust Control of Uncertain Systems with Nominal Tracking Performance

Abstract This paper addresses the design of quadraucally stabilizing regulators for linear discrete-time systems subject to parameter uncertainty. The controller is designed in two steps. First, uncertainty is neglected and a controller ensuring nominal tracking performance is worked out. e.g. by means of H 2 techniques. Then, based on an internal model scheme, a robustifying regulator is synthesized using the standard H ∞ machinery. Provided that a feasible solution exists, quadratic stability is guaranteed without affecting the tracking performance in nominal conditions.