The Dempster-Shafer evidence theory (DSET) is commonly applied in multi-sensor data fusion to address uncertainty, but handling highly conflicting evidence continues to be a major difficulty. To overcome this issue, we propose a novel divergence measure called the Symmetric Tsallis-Belief divergence (STBD), which quantifies discrepancies between belief functions while considering relationships among belief hypotheses. Numerical examples validate the theoretical properties and effectiveness of STBD in managing conflicting evidence. Building on this foundation, we develop a multi-sensor data fusion algorithm that adaptively assigns weights to evidence sources based on their credibility, improving fusion reliability. Finally, the algorithm is utilized to two practical scenarios: classification fusion and fault diagnosis, highlighting its advantages in real-world applications. These results emphasize the theoretical and practical contributions of the proposed method in addressing conflicting evidence and enhancing decision-making accuracy in multi-sensor data fusion.