Abstract: In the fine detection of water-bearing trap columns in coal mines, the fusion processing method combining seismic wave fields and transient electromagnetic fields has a good development prospect. To this end, taking Sunjiagou coal mine as an example, we optimize eight seismic attributes sensitive to the target geological structure by extracting the inverse resistivity of the transient electromagnetic method, choose the particle swarm optimization (PSO-BPNN)-based back propagation neural network (BPNN) as the fusion method for model testing and engineering application, and verify the feasibility and validity of PSO-BPNN in predicting the different water-bearing trap columns of coal mines (WBCC) by the experimental model, and in the engineering application, the drill hole attributes are used to predict the water-bearing trap columns (WBCC) of different WBCCs.) feasibility and validity, and in the engineering application, the borehole attributes were used for PSO-BPNN training, and a better fusion effect was achieved. The results show that the eight attributes can be used as the main feature set for the fusion of WBCC information based on seismic and transient electromagnetic methods, which in turn can be used to detect the boundary and abundance of coal mine trap columns by PSO-BPNN.