中储式制粉系统出力在线监测软测量建模

Soft Sensor Modeling for On-line Monitoring the Capacity of Coal Pulverizing System

为降低球磨机制粉系统的制粉单耗,对制粉系统出力进行软测量建模研究。在建模过程中,针对软测量辅助变量选择难的问题,将混沌理论与灰熵关联理论相结合,提出一种新的适用于具有混沌特性的灰系统对象的软测量辅助变量选择方法,即混沌灰熵分析(GECA)法;为解决软测量模型在线校正问题,提出模型在线校正的新算法,以提高软测量模型的测量精度;在此基础上实现了基于混沌灰熵分析支成分分析支持向量机(PCA-ε-SVR)、主成分分析BP网络持向量机(GECA-ε-SVR)的制粉系统出力软测量建模。与主(PCA-BP)软测量模型相比较,GECA-ε-SVR模型具有较高精度。研究表明,文中所建立的软测量模型具有较高测量精度,具有重要的工程应用价值。

To achieve a minimal unit power consumption of the pulverizing system in power plant, a model of the capacity of coal pulverizing system was studied by using soft sensor technique. In the process of modeling, to deal with the trouble choosing secondary variables for soft sensor, a novel method named grey entropy and chaos analysis （GECA） was advised based on chaos theory and gray correlation degree theory, which is useful and adaptive to the chaos and gray system. Furthermore, an improved algorithm was advised to revise the parameters of the soft sensor model online to improve the precision of the model. Then the two novel points were combined to construct the model of pulverizing capacity which was named GECA ＆ ε-support vector regression （GECA-ε-SVR） model. Comparing with PCA-ε-SVR and PCA-BP respectively, the GECA-ε-SVR model was much more precise. The study indicates that the model of pulverizing capacity possesses high accurate estimating ability, and it is useful to engineering application.