A flight trajectory prediction method based on dynamic time-frequency fusion

By constructing a time-frequency fusion flight trajectory prediction method, and utilizing first-order differential coding, extended discrete Fourier transform, and dynamic weight fusion techniques, the problems of frequency misalignment and high computational complexity are solved, and high-precision flight trajectory prediction is achieved.

CN122174039APending Publication Date: 2026-06-09SHANGHAI SIJIN INTELLIGENT TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHANGHAI SIJIN INTELLIGENT TECH CO LTD
Filing Date
2026-02-12
Publication Date
2026-06-09

AI Technical Summary

Technical Problem

Existing flight trajectory prediction methods have limitations in time and frequency domain modeling, and cannot effectively integrate the complementary advantages of time and frequency domains, resulting in frequency asymmetry and high computational complexity. They are also difficult to capture high-frequency information and local motion details, which affects prediction accuracy.

Method used

A flight trajectory prediction method based on dynamic time-frequency fusion is constructed. The method captures local fluctuation features through a time-domain prediction module and represents global dependencies through a frequency-domain prediction module. It adopts first-order differential coding, extended discrete Fourier transform, low-rank approximation representation and frequency-domain attention mechanism to dynamically allocate weights to fuse time-domain and frequency-domain information and generate the final prediction result.

Benefits of technology

It significantly improves the accuracy and robustness of flight trajectory prediction, reduces computational complexity, and can adaptively adjust the contribution of dual-domain information to meet the high-precision requirements of modern aviation systems.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122174039A_ABST
    Figure CN122174039A_ABST
Patent Text Reader

Abstract

This invention provides a flight trajectory prediction method bridging the time and frequency domains. It constructs and preprocesses a flight trajectory dataset, dividing it into training, validation, and test sets. It captures local time-domain fluctuation features of the training set in the time domain to generate time-domain prediction results. Simultaneously, it generates frequency-domain prediction results by periodically learning the global dependencies of the flight trajectory in the training set under frequency-domain representation. Based on the spectral energy distribution, it calculates the energy proportion of dominant harmonic sequences, dynamically allocates the fusion weights of the time-domain and frequency-domain prediction results, and generates the final flight trajectory prediction result, constructing a preliminary flight trajectory prediction model. It constructs a loss function for the preliminary flight trajectory prediction model based on the training set and trains it. The model is then validated using a validation set, and parameters are fine-tuned according to evaluation metrics until the validation effect meets the requirements when using the test set, resulting in the final flight trajectory prediction model. This method not only reduces model complexity but also improves prediction accuracy.
Need to check novelty before this filing date? Find Prior Art