コューラルネットワークと遺伝的アルゴリズムを用いた非線形モデリングによる酸素濃度推定

抄録

Residue Fluid Catalytic Cracking Unit (RFCCU) is the heart of the modem petroleum refinery. There has been a strong demand for decreasing the electric power and manpower in operating RFCCU. A combined control system has been constructed for satisfying this demand by controlling the reactor/regenerator differential pressure of RFCCU. The problem of this control system was that the analyzer of oxygen component, which is one of the important input variables of this control system, lacked reliability. This paper presents a nonlinear modeling method for identification of predictor of oxygen component using a neural network and a linear regression equation. This method uses genetic algorithm to select predictor variables of the linear regression equation. The obtained predictor showed a satisfactory performance even in the case where the analyzer was malfunctioning. While the measured value changed drastically and became wrong, the predicted value stayed in a valid range. The combined controller was possible to be implemented with this predictor. The differential pressure was stabilized, and a considerable amount of electric power was reduced. The implementation of this control system could also reduce the manpower needed for the operation of RFCCU.