摘要
针对电厂飞灰含碳量难以进行有效预测的问题,提出一种嵌套智能集成预测方法。首先,利用变学习率BP神经网络和主成分分析方法对输入变量进行降维处理;其次,为提高模型自适应能力,利用上述分析结果采用在线支持向量机建立飞灰含碳量预测模型;然后,为反映飞灰含碳量数据的时间相关性,采用灰色预测方法建立当前时刻飞灰含碳量预测模型;最后,在上述预测模型的基础上,利用信息熵的权值组合方法获得最终的飞灰含碳量预估值。仿真结果表明,该智能集成预测模型的预测精度要高于单一模型,能对电厂飞灰含碳量进行有效预测。
Based on the fact that the power plant carbon content in fly ash is hard to predict effectively, a nesting-intelligent-integrated prediction method was proposed from improving the prediction accuracy and adaptive ability. Firstly, the variable-learning-rate-based back propagation neural network and principal element analysis method were utilized to reduce the dimension of the input variables. Secondly, in order to improve the adaptive ability of the prediction model, the online support vector machine method was carried on to predict the carbon content in fly ash based on the above analysis results. Thirdly, for the purpose of reflecting the time correlation of carbon content in fly ash, the improved grey prediction model was used to predict the current value of carbon content in fly ash. Finally, the final carbon content in fly ash was obtained through the information entropy method for weight combination prediction. Simulation results show that the prediction precision of the intelligent integrated prediction model is higher than that of single model, and it can predict the power plant carbon content in fly ash effectively.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2013年第4期727-731,共5页
Journal of System Simulation
基金
湖南省高等学校科学研究项目(09C1020)
关键词
BP神经网络
在线支持向量机
灰色预测
信息熵
back propagation neural network
online support vector machine
grey prediction
information entropy
