基于分布式稀疏LS的热轧过程质量相关故障检测

Quality-related fault detection for hot rolling processes based on distributed sparse LS

质量相关故障检测是保障热轧过程安全运行和质量稳定的重要手段,是当前流程工业过程控制领域的研究热点。针对传统最小二乘(LS)方法求解参数存在过拟合及故障检测实时性不强等问题,该文将分布式优化与1-范数正则化思想引入到过程变量与质量变量相关关系建模中,提出了基于分布式稀疏LS(DSLS)方法,通过设计合理的监测统计量和控制限,实现了质量相关故障检测。通过热轧过程现场数据进行仿真验证,并与传统方法对比,验证了新算法的有效性。

Quality-related fault detection is an important mean to ensure safe operation and stable quality for hot rolling processes,which,thus,have recently become hotspots in the process industrial control domain.In this paper,a DSLS-based quality-related fault detection method is developed by distributed optimization and L1 norm regularization,and the associated statistics and thresholds are designed,which aims at addressing the issues on parameter overfitting and poor real-time performance for the traditional LS.Moreover,a case study on hot rolling process is finally given to compare with other methods to demonstrate the advantages of the new approach.