Soft sensor modeling of steel pickling concentration based on IGEP algorithm

Accurate measurement of acid concentration is paramount for ensuring the quality of strip steel pickling. Online measurement, a method that reduces operational complexity and lags effectively, is gradually replacing offline measurement of acid concentration. In this study, an indirect soft sensor model based on the improved gene expression programming (IGEP) algorithm has been constructed, leveraging easily measurable indexes from a large-scale dataset. The IGEP-based model predicted that the mean absolute errors for H⁺ and Fe²⁺ concentrations were 1.72 and 1.98 g/L, respectively. Additionally, the goodness of fit values for the H⁺ and Fe²⁺ prediction models were 0.945 and 0.933, respectively. Compared with the model based on support vector regression (SVR), which is suitable for small samples, it was demonstrated that the IGEP-based model achieved better predictive performance. Taken together, our study has designed a more effective and practical model for determining the acid concentration of strip steel pickling, providing a new ideal choice for the steel industry, which is of profound significance in the concentration control of pickling solution and the production of strip steel.

精准测量酸浓度对于保障带钢酸洗的质量至关重要。在线测量作为一种可有效降低操作复杂度与测量滞后的方法，正逐步取代酸浓度的离线测量方式。本研究依托大规模数据集内的易测指标，构建了基于改进基因表达式编程（IGEP）算法的间接软测量模型。基于IGEP的模型对氢离子与亚铁离子浓度的预测平均绝对误差分别为1.72 g/L与1.98 g/L；两款预测模型的拟合优度分别为0.945与0.933。相较于适用于小样本场景的支持向量回归（SVR）模型，本研究所提出的IGEP基模型展现出更优异的预测性能。综上，本研究设计了一款更高效实用的带钢酸洗酸浓度测定模型，为钢铁行业提供了全新的理想选择，其在酸洗溶液浓度调控与带钢生产领域具有深远意义。