Comparison of soft computing techniques for a three-phase oil field centrifuge

In this work we compare fuzzy techniques to neural network techniques for building a soft sensor for a three-phase oil field centrifuge. The soft sensor is used in a feed-forward control system that augments a feedback control system. Two approaches were used to develop the soft sensor. The first approach was to use a fuzzy rule based system based upon the experience of an expert operator. The expert operator's experience was supplemented using a computer model of the system. The second approach was to use a neural network to build the inverse of the computer model. The pros and cons of both techniques are discussed.