As mining engineering develops towards deeper and more intelligent operations, the complexity and danger of the mining environment have significantly increased. Accurate and real-time safety monitoring has become a core requirement for ensuring the safety of mining operations. Multi-source mapping technologies (such as UAV remote sensing, laser scanning, and inertial measurement) can acquire safety-related data on mining area topography, geology, and mining faces from different dimensions. However, current multi-source data suffers from problems such as heterogeneous formats, large differences in accuracy, and excessive redundant noise, resulting in low data utilization and difficulty in supporting efficient safety decision-making. This paper first outlines the characteristics and fusion requirements of multi-source mapping data for mining safety, and constructs a multi-dimensional data preprocessing system. Second, it proposes a big data fusion algorithm based on an improved Transformer and adaptive weight allocation to achieve accurate alignment and effective fusion of heterogeneous data. Finally, it designs and develops an intelligent decision support system integrating data acquisition, fusion analysis, risk warning, and decision output. Experimental results show that the proposed algorithm achieves the highest fusion accuracy, reaching 98%, which is 8 and 12 percentage points higher than the traditional weighted average fusion algorithm and wavelet transform fusion algorithm, respectively. This research provides reliable technical support for the efficient utilization and intelligent decision-making of multi-source mapping data for mining safety.
Li et al. (Thu,) studied this question.