摘要
针对水泥熟料质量指标的测量,提出一种基于最小二乘支持向量机的软测量建模方法;对于建模数据,提出了基于模糊聚类的数据预处理方法。实验研究表明,该数据预处理方法明显优于传统的拉依达准则方法,能够有效地去除现场测量数据中存在的异常数据;最小二乘支持向量机建模相比于RBF神经网络也具有明显优势,建立的软测量模型对于整个窑系统优化控制具有重要意义。
Aiming at the measure of cement clinker quality index,a soft-sensor modeling method is presented based on least square support vector machine （ LSSVM ）. Then a data preprocessing method is proposed based on fuzzy cluster. The experiments show that the data preprocessing method is superior to traditional Pauta criterion, because this method can eliminate effectively exceptional data in the actual measurement data. In addition, the LSSVM-based soft sensor also has more ohvious advantage than the RBFNN-based soft sensor, and the soft-sensor modeling is of important significance for the optimal control of the whole kiln system.
出处
《化工自动化及仪表》
EI
CAS
2006年第6期53-54,58,共3页
Control and Instruments in Chemical Industry
基金
国家高技术产业化示范工程项目(G050320)
