基于多元线性回归-BP神经网络的自移式破碎机生产能力预测  

Prediction of self-moving crusher's production based on multiple linear regression-BP neural network

作　　者：张崇欣[1] 李克民[1] 肖双双[1] 

机构地区：[1]中国矿业大学矿业工程学院,煤炭资源与安全开采国家重点实验室,江苏徐州221116

出　　处：《中国矿业》2013年第10期137-140,共4页China Mining Magazine

基　　金："十一五"国家科技支撑计划项目资助(编号:2006BAB16B00);中央高校基本科研业务费专项资金资助(编号:2010QNA33;2010ZDP01A02)

摘　　要：本文在对自移式破碎机系统生产能力影响因素分析的基础上,选取设备运行时间、炸药单耗和电铲作业周期时间作为可量化自变量,以系统生产能力为因变量建立多元线性回归方程,得到系统生产能力的预测模型。对多元线性回归模型预测结果的残差建立BP神经网络模型,利BP神经网络非线性拟合能力对残差进行调整。以某露天煤矿自移式破碎机系统生产数据为样本进行计算,多元线性回归模型预测误差为7%,修正后的模型预测误差为1.42%,预测精度显著提高。The equipment running time, explosives consumption and shovel cycle time are selected to be the quantifiable argument, and the production capacity of the system is selected to be the dependent variable to establish a multiple linear regression equation, based on analyzing the factors affected the self-moving crusher system＇s production. The equation can predict system production. The BP neural network is established to adjust residuals of the multiple linear regression model, used with the feature of the model in nonlinear fitting. The prediction accuracy is significantly improved. The error of multiple linear regression model is 7% ,and the error of the modified model is 1.42%.

关 键 词：多元线性回归 BP神经网络 自移式破碎机 能力预测 

分 类 号：TD451[矿业工程—矿山机电]