基于最小二乘支持向量机的混合煤气加压过程在线预测  

Online prediction of gas mixing and pressuring process based on least squares support vector machine

作　　者：刘彰 刘振兴[1] 刘斌[1] 吴畏[2] 廖荣震 罗小龙 LIU Zhang;LIU Zhenxing;LIU Bin;WU Wei;LIAO Rongzhen;LUO Xiaolong(College of Information Science and Engineering,Wuhan University of Science and Technology,Wuhan 430081,China;Planning Department,Hunan Valin Xiangtan Iron and Steel Co.,Ltd.,Xiangtan 411101,China)

机构地区：[1]武汉科技大学信息科学与工程学院,湖北武汉430081 [2]湖南华菱湘潭钢铁有限公司企划部,湖南湘潭411101

出　　处：《冶金自动化》2022年第1期17-24,共8页Metallurgical Industry Automation

基　　金：国家自然科学基金重点项目(61333008);国家自然科学基金面上项目(51975433)。

摘　　要：针对混合煤气加压过程具有非线性、多变量耦合、大滞后及不确定参数众多等复杂特性,提出了一种基于最小二乘支持向量机(least squares support vector machine, LSSVM)模型的混合煤气加压过程在线预测方法。首先,基于某钢厂混合煤气加压工艺分析,选择以混合后煤气压力、风机出口总管压力以及工作机组电流为混合煤气加压过程的预测目标;然后,将局部多项式核函数与全局拉普拉斯核函数相结合,实现了一种兼顾学习能力和泛化能力的混合核函数,并将自适应时刻估计(adaptive moment estimation, Adam)算法用于基于LSSVM预测模型的参数在线优化;最后,基于现场实测数据对预测效果进行验证,结果表明,本文所设计方法可有效提高混合煤气加压过程预测模型的预测精度。In view of the complex characteristics of gas mixing and pressuring process, such as nonlinearity, multivariable coupling, large delay and many uncertain parameters, an online prediction approach based on least squares support vector machine(LSSVM) model is proposed for gas mixing and pressuring process.Firstly, the mixed gas pressure, the main pipe pressure of the fan outlet and the current of the service units are selected as the prediction objectives of the gas mixing and pressuring process with the analysis of the gas mixing and pressuring process in a steel plant.Secondly, a mixed kernel function combined a local polynomial kernel function with a global Laplace kernel function is developed to reach a compromise between learning ability and generalization ability, and the Adam algorithm is introduced to realize the parameters online-optimization of the designed prediction model.Finally, the prediction effect is verified based on the field measured data.The results showthat the method designed in this paper can effectively improve the prediction accuracy of the prediction model of gas mixing and pressuring process.

关 键 词：混合煤气加压 在线预测 最小二乘支持向量机 混合核函数 

分 类 号：TF702.7[冶金工程—钢铁冶金] TP181[自动化与计算机技术—控制理论与控制工程]