Prediction and Intelligent Control of Si Element in Blast Furnace Smelting Process Based on NARX Dynamic Neural Network

Tao WU,Chong WU,Jia-wang CAO,Yi-yu WANG,Shao-jie ZHANG,Yuan ZHU
DOI: https://doi.org/10.3969/j.issn.1671-0460.2017.09.003
2017-01-01
Abstract:NARX dynamic neural network was used to establish the step-ahead prediction network and two-step prediction network of Si element in blast furnace smelting process, and the feasibility of dynamic control in process industry using NARX neural network modeling was analyzed. The experimental results showed that the hit rate (error ±0.1) of the single step was almost 100%, and the correct rate of the direction was 80%. At the same time, the hit rate of the two-step prediction was nearly 100%, and the accuracy of the forecasting direction was 90%. Therefore, the NARX dynamic neural network can be applied in the dynamic process control to get a good performance of prediction.
What problem does this paper attempt to address?