基于CFOA-PNN网络的采场顶板沉降量预测  

Prediction of Stope Roof Subsidence Based on CFOA-PNN Network

作　　者：唐鸣东 杨宁 熊晓辉 曹易恒 盛佳 TANG Mingdong;YANG Ning;XIONG Xiaohui;CAO Yiheng;SHENG Jia(Changsha Institute of Mine Research Co.,Ltd,Changsha 410012,China;National Engineering Research Center for Metal Mining,Changsha 410012,China;Hunan University of Science and Technology,Xiangtan 411201,China)

机构地区：[1]长沙矿山研究院有限责任公司,长沙410012 [2]国家金属采矿工程技术研究中心,长沙410012 [3]湖南科技大学,湖南湘潭411201

出　　处：《有色金属工程》2021年第7期107-113,共7页Nonferrous Metals Engineering

基　　金：湖南省重点实验室开放基金项目(E21833)。

摘　　要：针对环境再造采场顶板沉降影响因素复杂、数据离散等特点,选用一种仿生智能算法改进的概率神经网络(CFOA-PNN),避免模型陷入局部极值,增加网络预测精度,建立采场顶板沉降量预测模型。模型选取岩体强度、充填体抗拉强度等8个主要影响因素。以广东某铅锌矿的29个代表性样本对模型进行训练、预测,并对比传统预测模型结果准确性,利用局部敏感性分析法评价模型影响因素对预测精度的影响。结果表明:最主要影响因素为充填体抗拉强度,其次为采场暴露面积。所构建的CFOA-PNN模型正确率为88.9%,明显优于传统预测模型。In view of the complex influencing factors and discrete data of the roof subsidence in the environment reconstruction stope,using a bionic intelligent algorithm improved probabilistic neural network(CFOA-PNN)to avoid falling into local extremum and increase the prediction accuracy of network,and establishing the prediction model of roof subsidence.8 main influencing factors,such as the strength of rock mass and tensile strength of filling body,are selected in the model.29 representative samples of a lead-zinc mine in Guangdong province were used to train and predict the model,and compared with the accuracy of the traditional prediction model.The local sensitivity analysis method was used to evaluate the influence of the model influencing factors on the prediction accuracy.The results show that the main influencing factor is tensile strength of filling body,followed by the exposed area of stope.The accuracy of CFOA-PNN model is 88.9%,which is obviously better than other traditional prediction models.

关 键 词：环境再造采场 顶板沉降量 果蝇算法 概率神经网络 

分 类 号：TD327[矿业工程—矿井建设]