基于QPSO-BP神经网络的矿井突水水源判识模型研究  

Research on an Identification Model of Mine Water Inrush Sources Based on the QPSO-BP Neural Network

作　　者：李迎雪 郑禄林 杨爱莲[1] 曾艳 石鑫 冉浪 LI Yingxue;ZHENG Lulin;YANG Ailian;ZENG Yan;SHI Xin;RAN Lang(Mining College,Guizhou University,Guiyang 550025,China;College of Resources and Environmental Engineering,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学矿业学院,贵州贵阳550025 [2]贵州大学资源与环境工程学院,贵州贵阳550025

出　　处：《贵州大学学报(自然科学版)》2025年第1期114-124,共11页Journal of Guizhou University:Natural Sciences

基　　金：2023年度贵州省大学生创新创业训练计划资助项目(gzusc2023080);国家自然科学基金资助项目(52164006);贵州省科技支撑计划资助项目(黔科合支撑[2022]一般248);贵州大学青年教师国家自然科学基金培育资助项目(贵大培育[2020]81号)。

摘　　要：为厘清矿井水化学成分与矿井突水水源之间的非线性关系,实现突水来源的快速、准确判别。本研究提出了一种基于量子粒子群算法(quantum particle swarm optimization,QPSO)优化反向传播(back propagation,BP)神经网络的矿井突水水源判识模型,并将该判识模型运用于黔北煤田龙凤矿区以验证其实用性。通过与BP模型、遗传算法(genetic algorithm,GA)优化的BP神经网络模型GA-BP、粒子群算法(particle swarm optimization,PSO)优化的BP神经网络模型PSO-BP、量子粒子群算法优化的支持向量机(support vector machine,SVM)模型QPSO-SVM和量子粒子群算法优化的随机森林(random forests,RF)模型QPSO-RF判识结果进行对比,结果表明,QPSO算法有效优化了BP神经网络模型性能,提升了模型收敛速度和分类精度;QPSO-BP模型相较于以上5种模型分类性能更佳,对突水水源分类判识的准确率达到了93.75%。以上结果表明,QPSO-BP模型在矿井突水水源判识上有更好的优越性和应用前景。To elucidate the nonlinear relationship between the chemical composition of mine water and the sources of mine water inrush,so as to identify the sources of mine water inrush effectively.A model for identifying mine water inrush sources,which utilizes a quantum particle swarm optimization(QPSO)back propagation neural network(BP)is proposed.This identification model was applied to the Longfeng mining area of the northern Guizhou coalfield to assess its practical applicability.In comparison to the BP model,the genetic algorithm(GA)-optimized BP neural network model(GA-BP),the particle swarm optimization(PSO)-optimized BP neural network model(PSO-BP),the support vector machine(SVM)model optimized by quantum particle swarm optimization(QPSO-SVM),and the random forests(RF)model also optimized by quantum particle swarm optimization(QPSO-RF).The results indicate that the QPSO algorithm effectively enhances the performance of the BP neural network model,leading to improved convergence speed and classification accuracy.The QPSO-BP model exhibits superior classification performance compared to the aforementioned five models,achieving an impressive accuracy of 93.75%in the classification and identification of water inrush sources.These findings suggest that the QPSO-BP model not only demonstrates significant advantages but also offers promising applications in the identification of mine water inrush sources.

关 键 词：矿井突水 水源识别 量子粒子群算法 BP神经网络 机器学习 

分 类 号：TD713[矿业工程—矿井通风与安全]