摘要
煤气消耗预测是钢铁企业中能源管理重要组成部分之一,以炼钢过程煤气消耗为研究对象,将灰色理论与径向基函数(RBF)神经网络进行组合,建立了基于灰色RBF神经网络的炼钢煤气消耗预测模型,利用灰色理论累加求和特性对样本数据进行预处理,减小了数据的随机性,增强了数据变化的规律;利用RBF神经网络逼近这种数据变化的规律,通过预测误差,动态调整RBF神经网络的结构,使得预测误差在允许的范围内。通过仿真表明,提出的模型预测精度较BP神经网络预测精度高,均方差为2.02%。
Gas consumption prediction is an important component of energy management in iron and steel corporation. To predict gas consumption in the steelmaking process, a grey radial basis function (RBF) neural network forecasting model was proposed by combining grey theory with RBF neural network. Grey accumulated generating operation was used for data preprocessing, which could reduce data randomness and enhance changes of data. RBF neural network was trained to predict these changes. Parameters of RBF network were modified on-line by the prediction error. The approach of BP neural network model was also investigated to provide a comparison with model. The results of simulation show that prediction accuracy is very high and mean square deviation is less than 2.02%.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2011年第11期2460-2464,共5页
Journal of System Simulation
基金
国家杰出青年科学基金资助项目(60425310)