Robust data-driven Kalman filtering for unknown linear systems using maximum likelihood optimization (Q6933801)

The authors deal with the problem of state estimation in linear Gaussian stochastic systems with a pre-defined state but unknown system matrices described by the discrete-time equations \N\[\N\left\{ \begin{array}{ll}{x}_{k + 1} ={A}{x}_{k} + {B}{u}_{k} + {w}_{k}, \\\N{y}_{k} = {C}{x}_{k} + {v}_{k}, \, k\in\mathbb{N}, \end{array}\right. \N\]\Nwhere \({x}_{k}\in\mathbb R^{n}\) is the vector-valued system state to be estimated, \({u}_{k}\in\mathbb R^{m}\) is the input vector, \({y}_{k}\in\mathbb R^{p}\) is the output vector, \({w}_{k}\in\mathbb R^{n}\) is the state noise, \({v}_{k}\in\mathbb R^{p}\) is the measurement noise. The state noise and the measurement noise are assumed to be uncorrelated zero mean Gaussian noises with the covariance matrices \(Q\) and \(R\). The matrices \({A}\), \({B}\), \({C}\) are unknown system state, input, and output matrices, respectively. The authors propose a robust data-driven Kalman filter (RDKF) that integrates model identification with state estimation for the system. The proposed RDKF algorithm is applied to estimate the state of a given state-space model. A numerical simulation example is presented to verify the effectiveness of the developed algorithm.