scientific article; zbMATH DE number 1289939

From MaRDI portal
Publication:4243952





The author considers the following dynamic system. The state and output equations have the form \[ x_{i+1}= Ax_i+ bu_i+ cv_i,\tag{1} \] \[ y_i= d^T x_i+ v_i,\tag{2} \] where \(x_i\), \(y_i\) and \(u_i\) denote state, observation and control, respectively; \(A\), \(b\), \(c\) and \(d\) are real matrices of appropriate dimensions; \(v_i\) is a discrete-time white noise with covariance \(E[v^2_i]= \sigma^2\); the initial condition \(x_0\) is a normal random vector variable independent of the disturbances \(v_i\). The criterion of optimality has the form \[ I_i= E[J_i]= E\Biggl[\sum^{i+ N-1}_{j= i} y^2_{j+ k}+ \lambda \sum^{i+ N_u-1}_{j= i} u^2_j\Biggr].\tag{3} \] To obtain an optimal control and a filtered state variable, standard techniques, namely LQG and Kalman filtering, have been applied, respectively. Since in both cases either the matrix Riccati equation or the so-called vector Chandrasekhar-type equations have to be solved, the author derives a fast Chandrasekhar-type algorithm to minimize the number of operations required to perform the necessary computations.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4243952)