AN ONLINE ACTIVE SET STRATEGY FOR FAST ADJOINT BASED NONLINEAR MODEL PREDICTIVE CONTROL

Although nonlinear model predictive control (NMPC) has become a well-established control approach its application to fast systems with millisecond timescales is still a challenge. In order to reduce the online computational effort significantly, we propose to combine two recently developed techniques: the adjoint based real-time iteration scheme which allows treatment of nonlinear processes while keeping constraint sensitivities fixed and the online active set strategy for fast solution of the resulting parametric quadratic program. Considerable performance enhancement compared with the standard real-time iteration scheme is demonstrated by controlling a system with fast dynamics.