论文部分内容阅读
针对邯钢集团邯宝钢铁有限公司西区炼钢厂转炉的冶炼工艺特点和生产数据,建立了基于PCA-GA-BP神经网络的转炉终点磷含量预测模型。通过主成分分析(PCA)将终点磷含量的影响因素降维,并采用遗传算法(GA)对BP神经网络的初始权重进行优化。用Java语言开发了转炉终点磷含量预测模型的软件,在炼钢厂进行了现场使用。结果表明:转炉终点钢水w(P)控制精度在±0.007%时,命中率达到96.67%;控制精度在±0.005%时,命中率达到93.33%;控制精度在±0.004%时命中率达到86.67%。
Aiming at the smelting process characteristics and production data of converter in West Zone of Handan Iron & Steel Co., Ltd. of Handan Iron and Steel Group Co., Ltd., the phosphorus end-point prediction model of converter based on PCA-GA-BP neural network was established. The principal component analysis (PCA) was used to reduce the influence factors of end-point phosphorus content, and genetic algorithm (GA) was used to optimize the initial weight of BP neural network. Software that developed a predictive model for converter end-point phosphorus content in Java was used on-site in steel mills. The results show that the hit rate reaches 96.67% when the w (P) control accuracy of molten steel at the end of the converter reaches ± 0.007%; the hit rate reaches 93.33% when the control accuracy is within ± 0.005%; the hit rate reaches 86.67% when the control accuracy is ± 0.004% .