Abstract: Seam tracking system is one of the key technology to realize intelligent welding. Aiming at the problems of welding seam tracking in narrow gap, such as the difficulty of real time seam tracking weld seam of sensor, a new multipole array capacitive sensor based on fringe electric field technology is designed. And the welding seam surface reconstruction is realized through the capacitive signal processing technology based on wavelet filtering and nonlinear mapping technology. Firstly, a mathematical model of multipole array capacitive sensing is established, and the capacitive sensor structure is optimized through theory calculation results and finite element simulation analysis. Subsequently,The capacitive signal of the sensor is extracted and optimized with the help of capacitance signal processing technology, the welding gun offset,and weld deviation are both obtained. The results show that the reconstructed weld surface is basically the same as the narrow gap weld in the experiment, and it is feasible for the multipole array capacitive sensor applying to narrow gap weld tracking, which is of great significance for narrow gap weld tracking.