Research on aircraft positioning algorithm based on multimodal data fusion

Intelligent algorithms in the field of aircraft positioning are often limited by the insufficient accuracy of a single data source and reliability fluctuations caused by complex environmental interference. To address these issues, an aircraft positioning algorithm based on multimodal data fusion is proposed. A unified coordinate transformation model is established to align multimodal sensor data. An improved Residual Network Squeeze-and-Excitation Blocks (ResNeSt) architecture is introduced to address multimodal data heterogeneity and extract aircraft features. Finally, geometric constraints and statistical optimization principles are integrated with multi-source observation data to calculate the precise aircraft position based on the coordinate transformation information and feature fusion results, thereby achieving aircraft positioning. Experiments show that this method leads to a steeper upward trend in the cumulative positioning accuracy curve and a higher cumulative percentage in the higher positioning accuracy range. The average positioning error in all scenarios is reduced to 2.3 meters, representing reductions of up to 59.6%, 56.6%, and 53.1% compared to comparison methods 1, 2, and 3, respectively. Experiments demonstrate that the multimodal data fusion positioning algorithm significantly improves the accuracy and reliability of aircraft positioning, providing solid technical support for the safe and efficient operation of aircraft in complex environments