摘要
为解决污水生化处理过程中的水质参数BOD5(5天生化需氧量)难以在线监测的难题,在充分考虑污水处理过程非线性和多变量耦合的基础上,结合Jolliffe参数和数据选择算法提出了鲁棒最近相关性算法,并将其与RPLS(迭代偏最小二乘)和线性偏差补偿等算法相结合,对JIT(Just-in-Time)在线学习算法进行了改进,最后将改进的JIT算法用于建立BOD5软测量模型.对现场数据的仿真结果表明,与传统JIT算法和RPLS法相比,文中方法提高了软测量的在线预测精度、自适应性和鲁棒性.
In order to overcome the difficulty in on-line measurement of five-day biochemical oxygen demand（BOD5）during the wastewater treatment,by taking into consideration the nonlinearity and multivariant coupling characteristics of wastewater treatment process,a robust nearest correlation algorithm based on Jolliffe parameters and correlation data selection algorithm is proposed,which is then combined with the recursive partial least square algorithm and the linear bias compensation algorithm to improve the conventional JIT（Just-in-Time） algorithm.Finally,the enhanced JIT algorithm is used to build an on-line soft-sensing model of BOD5.The results of simulation show that the enhanced JIT algorithm outperforms the conventional JIT and RPLS algorithms in terms of on-line prediction accuracy,adaptability and robustness of soft sensing.
出处
《华南理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2011年第5期55-60,67,共7页
Journal of South China University of Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(60704012)
华南理工大学中央高校基本科研业务费资助项目(2009ZM0161)
关键词
软测量
JIT算法
生化需氧量
污水处理
迭代偏最小二乘
soft sensing
JIT algorithm
biochemical oxygen demand
wastewater treatment
recursive partial least square
