基于GA-KPLSR的转炉终点碳含量的预测研究  

Prediction of Carbon Content at End Point Based on GA-KPLSR in Converters

作　　者：严良涛[1] 李鸣[2] 杨大勇[2] 

机构地区：[1]南昌大学机电工程学院,南昌330031 [2]南昌大学信息工程学院,南昌330031

出　　处：《控制工程》2017年第5期923-926,共4页Control Engineering of China

基　　金：江西省科技厅科技计划项目(20123BBG70217)

摘　　要：终点碳含量是决定钢的种类和质量的关键因素,是转炉炼钢过程中最难控制的变量之一。建立了基于遗传算法的核偏最小二乘回归(GA-KPLSR)方法的终点碳含量的预测模型。数据仿真结果表明,基于GA-KPLSR的预测模型,不仅能高效的处理变量之间的非线性关系,而且能快速收敛至最优解,得出预测结果的均方误差比主元回归(PCR)和偏最小二乘回归(PLSR)分别降低了25.77%、23.27%;相对误差降低了29.55%、26.83%;绝对误差降低了27.22%、24.84%。该方法可为实际生产中的终点控制提供参考,提高生产效益。Carbon content of end point is the key factor determining the types of steel and steel quality. End point carbon content is one of the most difficult variables to control in the process of converter steel-making. A predicting model of carbon content at end point based on kernel partial least squares regression of genetic algorithm （GA-KPLSR） in converter steel-making is established. Data simulation results show that the predicting model based on GA-KPLSR can efficiently handle nonlinear relationship between variables, and furthermore, this method can converge to the optimal solution quickly. Compared with principal component regression （PCR） and partial least squares regression （PLSR）, the prediction results of mean square error reduce 25.77 % and 23.77 %; the relative errors reduce 29.55 %, 26.83 %; the absolute errors reduce 27.22 % and 24.84 %. This method can provide a reference for end point control in the actual production to improve the production efficiency.

关 键 词：终点控制 终点碳含量 遗传算法 核偏最小二乘回归 

分 类 号：TP27[自动化与计算机技术—检测技术与自动化装置]