A Kalman-filtering derivation of input and state estimation for linear discrete-time systems with direct feedthrough (Q6119730)

In this paper, the authors study the simultaneous input and state estimator (SISE) problem for linear discrete-time systems with direct feedthrough by employing a Kalman filtering (KF) derivation approach. The stochastic systems are described by the discrete-time equations \[ \left\{ \begin{array}{ll} x(k + 1) = Ax(k) + Bu(k) + w(k), \\ y(k) = Cx(k) + Du(k) + v(k), \end{array}\right. \] where \(x(k)\in\mathbb R^{n_x} , y(k)\in\mathbb R^{n_y}\) and \(u(k)\in\mathbb R^{n_u}\) are the system state, the measurement output, and the unknown input to be estimated, respectively, \(A, B, C, D\) are system matrices. The noises \(w(k)\) and \(v(k)\) are zero-mean Gaussian random variables with covariances \(E{w(k)w^{\top} (j)} = Q\delta_{kj}\) and \(E{v(k)v^{\top} (j)} = R\delta_{kj}\), respectively. The initial state \(x(0)\) is a Gaussian random variable with mean \(\bar{x}(0)\) and covariance matrix \(S(0)\). The Gaussian random processes \(w(k), v(k)\) and the initial state \(x(0)\) are mutually independent for all \(k\). The authors derived a Kalman filter under condition that the unknown input \(u(k)\) in the system is a white Gaussian noise with finite mean and finite variance. For the considered system of equations the authors present the SISE design given in [\textit{H. Fang} and \textit{R. A. De Callafon}, Automatica 48, No. 12, 3183--3186 (2012; Zbl 1255.93134)] and [\textit{S. Gillijns} and \textit{B. De Moor}, Automatica 43, No. 5, 934--937 (2007; Zbl 1117.93366)]. It is demonstrated that the SISE can be obtained from a standard KF under some mild conditions. For linear discrete-time systems without direct feedthrough, a relationship between the Kalman filter and the simultaneous input and state estimator is established.