基于支持向量机的电弧炉炉况判断方法

Furnace Condition Diagnostic System of Electric Arc Furnace Based on SVM

由于判断电弧炉炉况的直接条件不好测得,导致电弧炉各个炉况不容易确定。本文针对此类问题,提出一种基于支持向量机(SVM)理论的炉况判断系统。通过传感器采集数据,获得废铁量、废铁温度、炉渣量和炉渣温度等参数作为训练样本,离线训练SVM,得到分离器,然后选取测试的特征数据作为测试样本,根据本文提出的分类方法,通过离线分类器对炉况类型进行判断,从而为调整电极升降提供依据。实验表明SVM在有限的训练样本情况下可以有效地判断炉况。

To determine all status of the electric arc furnace not easily due to the direct conditions of judging the states of the electric furnace are measured not easily. Based on those, this paper proposes a SVM theory-based diagnostic system to determine the furnace condition of electric arc furnace. Taking the data of weight of scrap iron, temperature of scrap iron, weight of slag and temperature of slag obtained by sensors as training samples, this system can train Support Vector Machine （SVM） off-line, and gain a segregator, then make the characteristics of the test data as test samples. According to the classification method proposed in this paper, the judgments of the furnace off-line status of classifier types provide the basis for electrode adjusting. The test shows the generalization ability of SVM under the conditions of limited training samples with furnace condition.