膏体充填材料配比的神经网络预测研究  

Using Artificial Neural Network Model to Determine the Prescription of Paste Filling Materials

作　　者：常庆粮[1] 周华强[1] 秦剑云[1] 范军[1] 王玉禄 

机构地区：[1]中国矿业大学煤炭资源与安全开采国家重点实验室,矿业工程学院,江苏徐州221008 [2]阳泉煤业集团有限责任公司,山西阳泉045000

出　　处：《采矿与安全工程学报》2009年第1期74-77,共4页Journal of Mining & Safety Engineering

基　　金：国家自然科学基金项目(50490270,50774077,50574089,50490273);教育部新世纪优秀人才支持计划项目(NCET-06-0475);全国博士学位论文作者专项资金项目(200760)

摘　　要：充填材料配比的预测与设计在膏体充填工程中具有十分重要的意义,充填材料的配比预测是一个典型的多变量、非线性系统.为了准确可靠的掌握充填材料的强度,在综合分析充填材料质量影响因素的基础上,采用粒子群算法来优化改进的BP神经网络技术建立了充填材料质量的预测模型.并用基于正交试验获得的充填材料试验数据作为学习训练和测试样本,对模型预测结果和实际值进行对比分析.研究结果表明,用人工神经网络方法预测充填材料配比是科学、可行的,为煤矿膏体充填开采的充填材料配比预测和评价探索出了一种新的方法.The prescription prediction for and design of filling material is very significant in paste filling engineering.The prescription prediction is a typical multi-variable and nonlinear system.In order to forecast the strength of filling material exactly,we comprehensively analyzed the main factors affecting the strength of filling material and then established the model to predict the strength of filling material according to the theory of artificial neural network optimized by particle swarm.Based on some cases related to test data of filling material,the predicted results of model and measured values are compared and analyzed.The results show that to predict the strength of filling material by artificial neural network optimized by particle swarm is feasible and scientific.

关 键 词：采矿工程 充填材料 神经网络 配比预测 

分 类 号：TD353[矿业工程—矿井建设]