神经网络在膏体充填质量模型中的应用  

Application of Neural Network to Paste Backfill Quality Model

作　　者：常庆粮[1] 周华强[1] 赵才智[1] 

机构地区：[1]中国矿业大学能源与安全工程学院,江苏徐州市221008

出　　处：《矿业研究与开发》2007年第5期8-9,56,共3页Mining Research and Development

基　　金：国家自然科学基金重大项目资助(50490270)

摘　　要：煤矿膏体充填质量受多因素影响,且具有非线性特性,用数理统计的方法直接建立充填质量模型很困难。为了减少试验次数、降低试验费用,通过神经网络建立的膏体材料充填质量模型明显优于传统的回归分析法,利用膏体充填材料塌落度与主要影响因素浓度、胶结料用量、细集料用量的关系模型,可以有效预测膏体充填材料的塌落度。The quality of paste backfill is a muhiple factor - influenced problem with nonlinear characteristic. It is very difficuh to establish paste backfill quality model by straight use of the mathematical statistics method. In order to reduce the number and cost of test, the model of backfill quality is established by the neural network and is obviously superior to the traditional re- gression analysis method. The relationship model of slump of paste backfill with such main influencing factors as slurry concentration, dosages of cement and minute granule material is established, it is used to effectively forecast slump of paste backfill material.

关 键 词：神经网络 膏体充填 充填质量 塌落度 

分 类 号：TD823.7[矿业工程—煤矿开采]