基于多源时序数据的煤矿入井人员风险预警研究

打开文本图片集

中图分类号：TD67 文献标志码：A

Abstract：To address theproblemsof strong nonlinearcouplingand significant spatial heterogeneity in multivariate time series data fromcoal mines，ariskearly warning model forunderground coal mine personnel integrating multi-source time series data was proposed.A multimodal data synchronization method based on a codirectional dual-pointer sliding window wasadopted.Combined with Kalman filtering，a delaycompensation mechanism was introduced to improve interpolationaccuracy，therebyachieving high-precision time alignmentof signals with different sampling frequencies.A ten-dimensional feature vector was constructed，and the SHAP method wasutilized for globaland local importanceanalysis to eliminateredundant features，achieving efficient dimensionality reduction.This significantly improved the interpretabilityand robustnessof model decisionmaking while maintaining prediction performance.The Mixture of Attention Heads （MOA） mechanism was incorporated tocapture thenonlinear dependenciesand potentialcoupling features of multi-source signals.The MOA-Transformer model was constructed，where the Transformer encoder structure was used for featureengineering-based risk level classification.The MOA was then employed to construct featurerepresentations for classification. Field test results showed that the proposed model outperformed models such as recurrent neural networks and convolutional neural networks in terms of accuracy， precision， recall， and F1 -score. It could achieve high detection rates and low false alarm rates under conditions of few abnormal events，providing a feasible technical approach for risk identification and graded early warning for underground coal mine personnel.

Key words： underground coal mine personnel; risk early warning; time-series data synchronization; codirectional dual-pointer sliding window; feature importance analysis; SHAP method; MOA-Transformer

0引言

随着物联网、传感器和人工智能技术的发展，构建以人为中心、具备早期风险识别与预判能力的煤矿智能安全系统成为可能[1-2]。（剩余13233字）