用于采空区稳定性分析的FOA-GRNN模型研究及其应用  

Study and Application on Fruit Fly Optimization Algorithm Optimized General Regression Neural Network in Mined-out area Stability Analysis

作　　者：阳亦青 罗周全[1] 谢承煜[1] 

机构地区：[1]中南大学资源与安全工程学院,长沙410083

出　　处：《世界科技研究与发展》2015年第3期230-234,共5页World Sci-Tech R&D

基　　金：国家自然科学基金(51274250);中南大学中央高校基本科研业务费专项(2013zzts057)资助

摘　　要：为了更合理有效地分析采空区稳定性问题,引入一种新型的全局寻优智能算法(果蝇算法)对广义回归神经网络进行优化,构建采空区稳定性分析模型(FOA-GRNN模型)。对影响采空区稳定性的样本进行归一化处理,将标准化后的数据代入模型程序中,建立了由9个影响因素组成的采空区稳定性分析模型,研究结果表明:采空区稳定性主要影响因素权重为:矿柱面积比>RQD值>矿柱高宽比>岩石的质量指标;所建分析模型收敛速度快。并以某金属矿山采空区为例进行稳定性分析实际应用:采空区原始数据进行归一化处理后并输入所建立的FOAGRNN模型中,对照分级标准,得出结果:采空区不稳定,分析结果与实际情况相符,具有实用价值。The method of the general regression neural network model was optimized in a new type of global optimization intelligent algorithm（ Fruit Fly Optimization Algorithm）,and was used to construct a model（ FOA-GRNN model） about underground goaf stability analysis,to analyze the underground goaf stability more accurately and reasonably,which change samples of the stability in mined-out area into normalization ones. Then put the data into the program of FOA-GRNN model,the study result reflect that this model is composed of nine factors model of goaf stability influence factors,and the major influence factors are the area ratio of pillar,RQD value,ratio of width to height of pillar and the quality index of the rock. And the model has high convergence speed and accuracy. The analysis model was applied to the domestic metal mine empty district stability analysis,established the FOA-GRNN model after the raw data of goaf normalization processing,and make sure that the mined-out area is not stable by contrasting classification standard,the model accords with the actual situation,and the engineering application value is obvious.

关 键 词：采空区 稳定性分析 果蝇算法 归一化处理 广义回归神经网络 

分 类 号：TD325.3[矿业工程—矿井建设]