A Hierarchical Prediction Method and System for Civil Aviation Passenger Demand Based on Multi-Source Heterogeneous Data

By constructing a hierarchical forecasting method for civil aviation passenger demand based on multi-source heterogeneous data, and using LSTM and GAT models for multi-level forecasting, the problems of logical consistency and fine granularity in civil aviation demand forecasting are solved, and high-precision multi-level forecasting results are achieved, supporting the operation and management of airlines and airports.

CN121526176BActive Publication Date: 2026-07-17CHINA ACAD OF CIVIL AVIATION SCI & TECH
View PDF 9 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
CHINA ACAD OF CIVIL AVIATION SCI & TECH
Filing Date
2025-11-13
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

Existing civil aviation demand forecasting technologies have failed to effectively address the logical consistency problem of multi-level forecasts, have failed to fully utilize multi-source heterogeneous data, and lack fine-grained and highly consistent forecast results, making it difficult to meet the operational needs of airlines.

Method used

A hierarchical forecasting method for civil aviation passenger demand based on multi-source heterogeneous data is constructed. The hierarchical prediction sub-model is based on the LSTM model as the core architecture. The spatial correlation network between routes is constructed by combining the GAT model. The dynamic consistency loss function is used for logical constraints to achieve multi-level forecasting from routes, airports, cities to countries.

Benefits of technology

It enables multi-level daily passenger flow forecasting by route, airport, city, and country, improving forecast accuracy and consistency, and supporting the operation management and scheduling decisions of airlines and airports.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121526176B_ABST
    Figure CN121526176B_ABST
Patent Text Reader

Abstract

This invention discloses a hierarchical prediction method and system for civil aviation passenger demand based on multi-source heterogeneous data. The method includes: constructing a civil aviation passenger demand-driven feature library, which is built according to hierarchical items and hierarchical feature items at different levels; constructing a hierarchical combined prediction model for civil aviation passenger demand, including several hierarchical item prediction sub-models; extracting causal correlation time-series data from a subset of historical feature data, with daily passenger flow as the dependent variable and hierarchical feature item data as the independent variable, and inputting it into the hierarchical item prediction sub-model for feature time-series data prediction training; and the hierarchical combined prediction model for civil aviation passenger demand outputting the daily passenger flow corresponding to each level sequentially from bottom to top according to the hierarchical items. This invention enables daily passenger flow prediction at multiple hierarchical levels, including routes, airports, cities, and countries, and can obtain passenger flow demand data with macro-to-micro linkage constraints, providing rich hierarchical data for all parties in civil aviation management to improve service capabilities.
Need to check novelty before this filing date? Find Prior Art