摘要
介绍了适宜支持向量机处理大规模数据回归问题的序列最小优化(SMO)学习算法,针对SVR进行二次规划处理大规模数据时计算复杂度高和学习机参数选择方法复杂的问题,从算法结构和参数选择两个方面对SMO算法进行了改进,使运算速度和建模效率得到了进一步提高。结合灰色理论进行辅助变量选取,并应用改进的SMO算法建立了火电厂烟气含氧量软仪表,通过电厂的实测历史数据仿真表明,改进的算法较传统的SMO算法在计算速度和性能上有较大提高,建立的软仪表模型具有更高的精度,能满足应用要求。
In order to solve the problem that normal SVM algorithm can not deal with large scale data,the improved SVR algorithm of sequential minimal optimization（SMO） is introduced in this paper.The improved algorithm makes improvements in two aspects of structure and parametric selection to increase operational speed.It uses grey theory to select the auxiliary variables and build a model of soft instrument for the flue gas oxygen content in power plant.The simulation with historical data measured by plant shows that compared with the normal SMO algorithm,the operating speed and the soft instrument precise of the improved SMO algorithm is better.
出处
《自动化技术与应用》
2010年第10期4-6,18,共4页
Techniques of Automation and Applications
关键词
序列最小优化(SMO)
灰色关联分析
氧量
软测量
Sequential Minimal Optimization（SMO）
grey relational analysis
uolume of oxygen
soft-sensing
