Neural network-based software sensor for feed identification in crude distillation unit

摘要

A predictive method of identifying the feed and product using neural network-based software sensor is proposed, which is able to provide the information for down stream scheduling, production, and control. Data preprocessing and post-processing methods are also utilized to enhance the predictability of the software sensor model. Implementation is conducted by using Borland C++(3.0) Builder. The simulation results demonstrate the effectiveness of the proposed neural network-based software sensor.