Generally, it is difficult to control processes with dead time by means of a conventional PIDcontroller. In order to control such processes without upsetting the stability, Prof. Otto J. M.Smith proposed a control method, which utilized a dead time element in the controller. However, it is known that, when this control method is applied to processes with dead time and having no.self-regulation, large offset is left to disturbance, even though a quick and stable response to referenceinput can be obtained.
In this paper, the author thoroughly studies the characteristics of the Smith's method, makes clearthe causes of its defects, and proposes a new modified control method which retains the merits ofthe original one and makes up for its imperfection. The characteristics of this modified controlmethod are shown in comparison with the Smith's method and the PID control method. When thesystem is controlled by this modified method, no large offset is left to a step disturbance, unlike thesystem controlled by the Smith's method. And the modified method is superior to the PID controlmethod in rapidity of the response and in stability.

View full abstract

The usual statistical synthesis method of servomechanisms has two principal defects; the firstis that the network is not realizable physically which is determined under the condition of theleast mean-square error, and the second is that when the power of noise is negligibly small comparedwith that of signal, an unreasonably bulky actuator is required, although the difficulty ofphysical realizability of the network vanishes in that special case.
We propose several servomotors for the specific problem. The network is determined so as tominimize the mean-square error under the limitation of the allowable maximum mean-square torqueof the respective servomotor. We should choose the servomotor as optimum one which reveals theminimum least mean-square error. By this method, the servomechanism is evidently physicallyrealizable and the actuator is not unreasonably bulky, and we will have the optimum servomechanismin the strict meaning.

View full abstract