Optimal bilinear observers for bilinear state-space models by interaction matrices

This paper formulates a bilinear observer for a bilinear state-space model. Relationship between the bilinear observer gains and the interaction matrices are established and used in the design of such observer gains from input-output data. In the absence of noise, the question of whether a deadbeat bilinear observer exists that would cause the state estimation error to converge to zero identically in a finite number of time steps is addressed. In the presence of noise, an optimal bilinear observer that minimizes the state estimation error in the same manner that a Kalman filter does for a linear system is presented. Numerical results illustrate both the theoretical and computational aspects of the proposed algorithms.