基于EEMD-BAS-RBF的高炉煤气利用率预测

Prediction for utilization rate of blast furnace gas based on EEMD-BAS-RBF

随着人工智能的发展,通过数字赋能助力高炉智能化转型、实现“30·60”目标已经成为当前主要研究目标。作为高炉生产的重要指标之一,合理预测高炉煤气利用率变化行为有助于生产现场及时了解炉况运行状态和降低生产能耗。因此,提出了一种基于EEMD-BAS-RBF的高炉煤气利用率预测模型。首先,针对高炉数据存在时滞性长、异常值难以判断等特点,采用3σ准则(拉依达准则)、数据标准化处理以及工艺分析的方法完成了对样本数据的处理。其次,运用工艺与大数据技术互补选参的方式对模型的输入参数进行筛选,提高了模型输入变量的解释性。最后,在模态分解算法(EEMD)的基础上建立了天牛须搜索(BAS)优化RBF(Radial Basis Function)的BAS-RBF预测模型,并采用5个对比指标对该模型的模型预测能力进行了分析。分析结果显示,使用该模型对高炉煤气利用率的准确率高达87.55%。

With the development of artificial intelligence,the intelligent transformation of blast furnace could be facilitated by the digital empowerment to realized the"30·60"goal.The reasonable prediction for the gas utilization rate that is one of the important indexes to grasp the operation status of blast furnace in time and reduce the energy consumption.Therefore,a prediction model for utilization rate of blast furnace gas based on EEMD-BAS-RBF was proposed.In response to the characteristics of long time delay and difficulty in identifying outliers in blast furnace data,the sample data was processed using criteria(Laida criteria),data standardization processing,and process analysis methods.Then,by using the complementary selection method of process and big data technology,the input parameters of this model were screened to improve the interpretability of the input variables.Finally,based on the Ensemble Empirical Mode Decomposition(EEMD),a BAS-RBF prediction model that uses Beetle Antennae Search(BAS)to optimize Radial Basis Function(RBF)was established.And the prediction ability was analyzed by using five comparative indicators.The prediction results show that the accuracy rate of gas utilization rate is up to 87.55%.