Real-time Prediction of Styrene Production Volume based on Machine Learning Algorithms

Due to wide application of styrene and complex process of modern dehydrogenation of ethylbenzene, traditional methods usually spend much more time on chemical examinations and tests for identification of the production volume. Generally, there are several hours or days of time lag for the information to be made available. In this article, the whole ethylene cracking plants are investigated. The generalized regression neural network model is designed to timely predict the styrene output after the high-dimensional reduction. The usefulness of the model will be demonstrated by specific cases. The appropriate data mining techniques and implementation details will also be depicted. Finally, the simulation results show that this model can monitor the styrene output per hour with high accuracy.