With the vigorous development of intelligent sensing technology, large-scale sensor networks based on fiber Bragg gratings have been widely used in many fields. With its advantages of anti-electromagnetic interference and high sensitivity, it provides strong support for accurate monitoring. However, the massive heterogeneous data generated by large-scale sensor networks poses severe challenges in transmission, storage and processing. Traditional data fusion algorithms are difficult to meet the needs of efficient and accurate processing. This paper proposes a new hierarchical distributed data fusion algorithm, which first pre-processes the data of each node to remove outliers and noise interference; then uses an adaptive weighted fusion strategy at the cluster head node to dynamically adjust the weight according to data characteristics; and finally performs global fusion at the aggregation node. This algorithm significantly improves the accuracy and timeliness of data fusion. Experimental results show that the three-level fusion architecture proposed in this paper outperforms traditional methods in temperature (accuracy ± 0.2 °C), strain (±8με) and multi-field coupling (positioning accuracy 1 m) scenarios.