热风炉煤气消耗量中期预测模型

在钢铁生产过程中，副产煤气占钢铁企业总能耗的40%，因此，准确预测副产煤气的消耗量可以为钢铁企业煤气系统的优化调度提供科学的指导。热风炉是副产煤气系统的最大用户之一，由于工作周期频繁调整导致副产煤气消耗量波动剧烈，预测难度较大。针对现有预测模型预测提前量较短的问题，建立了基于时间序列的BP神经网络预测模型，在保证较高的预测精度的前提下将预测提前量延长至30 min。以现场采集的热风炉煤气数据作为数据样本进行实例分析，发现训练样本为2 000组、预测样本为30组时预测效果最好，平均误差绝对值可达4.04%。此外，还对不同预测模型进行对比，结果表明本模型最适合热风炉煤气消耗量的中期预测。

Medium- term prediction model for byproduct gas consumption in hot blast stove

Byproduct gases account for 40% of total energy consumption in steel production process. Therefore, the accurate prediction of byproduct gas consumption can provide scientific guidance for the optimal scheduling of the gas system in steel enterprises. As one of the biggest byproduct gas consumers, the hot blast stoves suffer from huge and frequent fluctuations in gas consumption, and thus it is difficult to predict. Based on the shortage of previous prediction models whose prediction time is short, a BP neural network based time series prediction model was proposed and the prediction time was increased to 30 minutes with little influence on the prediction accuracy. The results of the case study indicated that the optimal volume of the training sample and the prediction sample were 2 000 and 30, respectively, and the absolute percentage error could reach 4.04%. In addition, different prediction models were compared and the results demonstrated that the proposed model was more suitable for the medium term prediction of byproduct gases.