Improved predictive control and optimization based on modified BP neural network model in diesel oil blending

In the light of the multi-inputs and multi-outputs objects of the diesel oil blending production in the oil refinery, this paper proposed an optimization method of continuous diesel oil blending production with the constraints of the quality indices such as the 50% distillation range, acidity and the Cetane value. It also provides a predictive control model based on BP neural network. The Simulation results show that the linear optimization method proposed by this paper can compute the optimal formula of the diesel oil blending production. With the constraints of the production capability and the indices of the product quality, the predictive control model can realize the optimal utilization of the component oils and finally raise the profit of production.