Fault accommodation using model predictive methods

In the paper we first present several failure models, and derive the model predictive control (MPC) algorithm for the no-failure case when the objective is to track the outputs of a reference model. Following that, the multiple model switching and tuning methodology, as applied in the context of failure detection and identification (FDI) and adaptive reconfigurable control, is presented and described in detail, and its stability and convergence properties are discussed. Multiple model predictive control (MMPC) approach is presented and the MMPC algorithm is derived for the case of lock-in-place control effector failures. Combining the online FDI system with the MPC approach results in an efficient MMPC algorithm that is well suited for different types of failures.