In this paper are discussed the dynamic characteristics and the control methods of a vacuum pan which is used as semi-batch vacuum crystallizer in a sugar refinery.
Some properties of the vacuum pan and its operation are presented showing that the vaccum pan can be satisfactorily controlled with three controlled variables although the necessary and sufficient number of the controlled variables is three as obtained by the authors' method of calculating degrees of freedom. The dynamic characteristics are analyzed, and the block diagram is shown. The predicted characteristics are verified with an actual large-scale pan installed in a suger refinery. In order to attain the optimal operation and control, the “mobility”, one of the controlled variables, has to be clearly understood. The authors give the meaning of this particular variable, and discuss its importance in relation to the operation of crystal growth. To control the mobility is to control the distance between surfaces of adjacent sugar crystals. The difficult operational problem here is that the requirement for obtaining a good boiling efficiency is incompatible with that for obtaining a good quality product except the case that a mechanical circulator is mounted on the pan. The authors suggest the solution based on the concept of “trouble control”, and give the optimal operational program for the mobility in the form of eq. (21).

View full abstract

Recently some attempts on the automatic color matching have been tried with the remarkable development of spectrophotometer. A few years ago, a commercial color mixture computer appeared in the U. S. A.. In this computer, however, human operations of trial and error are still necessary to obtain the concentration results.
This paper introduces a convenient equipment by which the color matching is automatically accomplished by the use of the learning machine technique. In this equipment, values of weighting elements in the learning machine indicate the concentration of each pigment. An input pattern of the learning machine is composed of the reflectances of pigments in a fixed sampled wavelength. The desired output of the pattern is composed of the reflectance of the standard color in the same sampled wavelength. Thus, the equipment is considered to be a special learning machine which has N input patterns and N analog desired outputs, where N is the number of sampling points in the spectrophotometric curves. The desired concentrations of pigments are given by the final values of weighting elements which the learnig machine indicates after finishing the learning process.
The equipmentis simple in mechanisms and low in cost The trial equipment has shown considerably good results.

View full abstract