基于BP神经网络的溜井堵塞率预测  

Prediction of ore-pass blockage rate based on BP neural network

作　　者：刘永涛[1] 刘艳章[1] 邹晓甜 陈小强[1] 潘世华[1] 吴恩桥 张奎[1] 张丙涛[1,2] 叶义成[1] 

机构地区：[1]武汉科技大学资源与环境工程学院,湖北武汉430081 [2]十堰市国土资源局矿山设计院,湖北十堰442000

出　　处：《化工矿物与加工》2017年第4期41-44,共4页Industrial Minerals & Processing

基　　金：湖北省自然科学基金重点项目(2015CFA142);国家自然科学基金面上项目(51074115;51574183)

摘　　要：针对某矿溜井频繁发生的堵塞问题,以该矿-480^-410m段主溜井为工程背景,将在相同放矿条件下发生堵塞故障的概率定义为溜井堵塞率,构建溜井放矿相似试验平台,选取贮矿高度、粉矿含量、贮矿时间和矿石含水率作为溜井堵塞的关键影响因素,进行4因素5水平的溜井放矿相似模拟试验,将实验数据作为训练样本构建BP神经网络模型,对溜井堵塞率进行预测,并对预测模型进行检验。研究结果表明,BP神经网络模型能够准确预测溜井矿石在不同放矿条件下的堵塞率,研究结果可为溜井堵塞预防提供指导。In response to the frequent occurrence of ore-pass blockage, -480- - 410 m main ore-pass in Jinshandian Iron Mine is taken as the engineering background and the ore-pass blockage rate is defined as the occurrence probability of the ore- pass blockage at the same ore drawing condition. The ore drawing similarity test plat is developed. The ore storage height, the powder ore content, the ore storage time and the water content are selected as key factors affecting the ore-pass blockage. Test of four factors and five levels ore drawing similarity simulation test is carried out. The BP neural network model that predicts the ore-pass blockage rate is built using the similarity test data as the training samples to predict the ore-pass blockage rate against prediction model.

关 键 词：BP神经网络 相似模拟试验 溜井堵塞率 预测模型 

分 类 号：TD853[矿业工程—金属矿开采]