Unknown Input Detection Using Receding Horizon Approach

The paper offers the possibility of the design of unknown input detection for dynamic systems under external noise effect. The presented geometric based fundamental problem in residual generation (FPRG) method uses on the one hand the Kalman filtering and on the other hand the moving horizon estimation (MHE) when stochastic noise on the input and on the output, with additive failure directions, are presents. The paper combines the optimal Kalman and MHE method with geometric based unknown input observer strategy. The MHE solution makes to treat constraints during the estimation process possible. A numerical example supports the necessity of constrained unknown input estimation