Influence Factor Analysis and Prediction Model of End-Point Carbon Content Based on Artificial Neural Network in Electric Arc Furnace Steelmaking Process

In this manuscript, we consider the accuracy of end-point carbon content prediction affected by oxygen injection in multiple stages electric arc furnace (EAF) melting process. Such a would help to further evaluate process control strategies and optimize overall operation furnace. Principal component analysis (PCA) was used normalize 13 input variables affecting endpoint content. log-sigmoid tan-sigmoid functions were verify same sample, it found that Mean squared error(MSE) model under logsig + function smaller, indicating more stable. At time, different hidden layer nodes tried, finally structure determined as × 10 8 1, activation logsig. Using historical smelting data train test neural network model, correlation coefficient (R) verified is 0.7632, range ±0.03%, hit rate 64.5%, 42% ±0.02%. Combining verification basis with metallurgical reaction principle EAF steelmaking process, pretreatment method phased total proposed. The divided into three stages, which are consumption volume 0–5 min, 5–30 min than 30 other kept unchanged. data. After verification, R staged 0.8274. 78.5%, 58% Finally, an on-line system based on artificial developed applied actual production. Running results illustrated 96.67%, 93.33% 86.67%, respectively when errors within ±0.05%, ±0.03% ±0.01%, improved can effectively predict content, provides good for at end point