摘要
制粉出力目前尚无法直接在线测量,众多球磨机制粉系统改造中都采用了软测量的方法,其中神经网络模型运用最为广泛。由于对神经网络输入节点(辅助变量)数的确定缺乏一个科学的指导原则,往往造成所构造的软测量模型效果不佳。运用分形理论的相空间重构算法及描述非线性特性的特征参数——关联维数对球磨机出力特性进行分析,找到球磨机差压信号的关联维数,从而得到软测量模型的辅助变量数目,为软测量模型的构建提供了有力的帮助。
As charging ration can't be directly dectected online yet, soft-sensing technique, especially the ANN,is applied in many rebuilding processes of the ball mill pulverizing system. However,due to the lack of a scientific method to determine the number of ANN input node,the effect of the soft-sensing model is usually unsatisfactory. In this paper, the output characteristics of ball mill is analyzed based on phase space reconstruction theory,and the correlation dimension of the stress difference is found to determine the number of secondary parameters to construct the soft computing model.
出处
《电力系统及其自动化学报》
CSCD
北大核心
2008年第2期117-120,共4页
Proceedings of the CSU-EPSA
关键词
球磨机
料位
混沌理论
软测量
人工神经网络
ball mill
charging ratio
chaos theory
soft-measurement
ANN(artificial neural network)