Minimax identification with uncertain initial information

A method of identification is proposed for the case of uncertainty where only the range of possible process input and output values is known. Reasonably general examples are given where the method reduces the identification to a linear programming procedure.