摘要
针对RPCL聚类算法存在的缺点,提出一种改进算法,并在此基础上得到了一种分布式支持向量机(DSVM).针对SVM算法中阈值难以确定的问题,提出了一种两段学习算法.最后将DSVM应用于氧化铝高压溶出过程苛性比值的软测量,现场数据的仿真结果表明该方法具有较高的精度,能满足实际生产的需要.
An improved rival penalized competitive learning (IRPCL) clustering algorithm is proposed. A distributed support vector machine (DSVM) is constructed. Aiming at the difficulty of computing bias of SVM, a two-phase algorithm is proposed. DSVM is applied in soft sensing for ratio of soda to alumina (RSA) in the process of high-pressure digestion of alumina. Simulation result shows that the method possesses high precision and can meet actual demands.
出处
《控制与决策》
EI
CSCD
北大核心
2004年第8期852-856,共5页
Control and Decision
基金
国家863计划项目(2001AA411040)
国家973计划资助项目(2002CB312200).
关键词
支持向量机
RPCL聚类算法
软测量
苛性比值
Alumina
Computer simulation
Learning algorithms
Predictive control systems