|
| 基于信号分类的金属矿床采空区稳定性预测研究 |
| Stability Prediction of Metal Mine Goafs Based on Signal Classification |
| 投稿时间：2025-12-06 修订日期：2026-02-25 |
| DOI： |
| 中文关键词: 金属矿床 采空区 稳定性 信号分类 提前量预警 |
| 英文关键词: metal mine goaf stability signal classification lead-time early warning |
| 基金项目:安徽省教育厅 2019年安徽高校自然科学研究项目,基于C-ALS精准探测与建模技术的采空区稳定性分析研究,课题编号:KJ2019A1227 |
|
| 摘要点击次数: 29 |
| 全文下载次数: 0 |
| 中文摘要: |
| 以地震信号分类为基础的金属矿区采空区稳定性预测方法逐渐受到关注，然而振动信号中混杂大量非矿震干扰源，直接影响模型输入的可信度。针对以上不足，研究设计了基于时频能量图卷积与通道注意力机制的信号分类算法，构建融合静态工程因子与动态信号证据的金属矿床采空区稳定性预测模型。实验在自建数据集上开展，结果显示该模型在生存函数为0.5时，提前量中位数为33.9分钟，其他关键指标均优于对比模型，为金属矿床采空区实时风险预警提供了一种具备落地条件的模型框架。 |
| 英文摘要: |
| Seismic-signal-based stability prediction methods for metal mine goafs have attracted increasing attention; however, vibration records are heavily contaminated by non-mining-induced disturbances, which directly undermines the reliability of model inputs. To address this limitation, this study develops a signal classification algorithm based on convolution over time–frequency energy maps combined with a channel attention mechanism, and further designs a stability prediction model for metal mine goafs that fuses static engineering factors with dynamic signal evidence. Experiments were conducted on a self-constructed dataset, and the results show that when the survival function equals 0.5, the proposed model achieves a median lead time of 33.9 minutes, with all other key performance indicators outperforming those of the comparison models, thereby providing a practically deployable modeling framework for real-time risk early warning in metal mine goafs. |
|
View Fulltext
查看/发表评论 下载PDF阅读器 |
| 关闭 |
