摘要
为了降低人工神经网络(ANN)在静电除尘器(ESP)运行参数预测中的计算量,提出了一种改进的ANN预测模型。在分析了ESP运行参数的特点的前提下,运用K-means聚类算法确定其聚类中心,然后把所有的聚类中心作为神经网络的预处理层,并在此基础上建立了ESP放电信号的预测模型。仿真结果表明,该预测模型能够有效地对电除尘器放电信号进行预断,是其状态预判的一种有效手段。
To reduce the prediction model' s calculation for the operating parameters of electrostatic precipitator (ESP), a improved prediction model based on artificial neural network (ANN) is presented. Firstly, the model get the cluster centers of ESP' s operating parameters by K-means cluster algorithm after analyzed their feature, then these cluster centers are used as the pretreatment layer of ANN, and the spark signal prediction model is constructed. The simulation results demonstrated that the model can accurately predict ESP spark signal, is an effective means for spark signal pre-sentence.
出处
《计算机工程与设计》
CSCD
北大核心
2010年第21期4682-4685,共4页
Computer Engineering and Design
基金
科技部中小企业创新基金项目(04C26216200902)