Minimax linear dynamic filters of minimum dimension of coordinates of linear dynamic systems

Minimax linear filtering of phase coordinates in linear dynamic plants is performed in continuous time and with only partial information available on the distrubing processes effecting the plant and the observation channel. This information on the disturbing processes are constraints on the variances and various conditions under which the disturbing processes and noise in the plant and in the observation channel are uncorrelated. The solution is sought in the class of plant dimension filters, or filters of minimal dimension. General conditions of optimality (saddle point conditions) are found. Case studies of solutions are provided.