Concurrent Learning Adaptive Model Predictive Control with Pseudospectral Implementation

摘要: This paper presents a control architecture in which direct adaptive technique is used within the model predictive framework, using concurrent learning based approach, to compensate for uncertainties. At each time step, sequences and parameter estimates are both as optimization arguments, thereby undermining need switching between phase phase, case with hybrid-direct-indirect architectures. The state derivatives approximated pseudospectral methods, vastly numerical optimal problems. Theoretical results simulation examples establish effectiveness of architecture.