人工神经网络在优化动力配煤中应用的理论研究  

THEORETICAL RESEARCH ON APPLICATION OF ARTIFICIAL NEURAL NETWORKS METHOD IN THE OPTIMUM STEAM COAL BLENDING

作　　者：骆仲泱 殷春根[1] 姚强[1] 周俊虎[1] 吴晓蓉[1] 曹欣玉[1] 汤龙华[1] 岑可法[1] 

机构地区：[1]浙江大学

出　　处：《煤炭学报》1997年第4期337-342,共6页Journal of China Coal Society

基　　金：国家"八五"重点科技攻关项目

摘　　要：论述了人工神经网络方法和非线性混合离散变量的优化设计方法应用于配煤中的必要性和可行性．对实际应用中所用到的神经网络ＢＰ算法及其一些改进和非线性混合离散变量的优化设计方法———ＭＤＯＤ法作了简要介绍．针对用加权平均或某些经验公式对配煤过程中所遇到的一些指标进行预测时效果不很理想这一现状，采用了神经网络技术对发热量、挥发分、水分、灰分及煤灰软化温度等指标进行建模以实现非线性映射，同时根据现有配煤系统的实际生产情况，把每种煤的配比取为离散变量，从而使ＭＤＯＤ算法在动力配煤中得以应用．The necessity and possibility of artificial neural networks method and non linear mixed discrete variables optimum design method used in coal blending are described.The BP algorithm of neural networks used in operation and non linear mixed discrete variables optimum design method (MDOD method) are briefly introduced.In view of non ideal effect of some indexes prediction encountered in the process of coal blending by weighted average method or some empirical formulas,neural networks technique is adopted to establish model for indexes of heating value,volatile matter,moisture content, ash content and softening temperature of coal ash to achieve non linear mapping,meanwhile,based on practical operation of current coal blending system,taking mixture ratio of every type of coal as discrete variable,thus the MDOD algorithm can be used in steam coal blending.

关 键 词：神经网络 优化设计 动力配煤 

分 类 号：TQ520.62[化学工程—煤化学工程]