Contextual AI Flight Warning Prediction for Stable Approach Safety

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing technologies lack an effective means to identify probable causes of aircraft performance events and provide real-time predictions for stable flight conditions using machine learning techniques.

Innovation Solution

A contextual artificial intelligence model is developed to analyze flight data, incorporating aircraft operator background, environmental factors, and airport data, using machine learning approaches like logistic regression to predict safety-related events and provide proactive warnings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If machine learning techniques are applied to flight data analysis, then prediction accuracy for safety events is improved, but system complexity increases

Engineering Contradiction:
Improveprediction accuracyVSAvoidsystem complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The system segments flight data analysis into multiple specialized machine learning models, each targeting specific safety events (e.g., unstable approach prediction, runway overrun prediction, go-around prediction). This segmentation allows complex prediction tasks to be divided into manageable components while maintaining high accuracy for each specific event type.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent introduces contextual models as intermediary layers between raw flight data and prediction outputs. These contextual models process and interpret flight parameters, environmental factors, and aircraft state data before feeding them into prediction algorithms, thereby managing system complexity while preserving prediction accuracy.

Inventive Principle:
Principle #24Intermediary (Mediator)

2Reliability

If contextual data from multiple sources is integrated, then prediction reliability is improved, but data processing time increases

Engineering Contradiction:
Improveprediction reliabilityVSAvoiddata processing time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The system performs preliminary processing and filtering of contextual data from multiple sources (flight management systems, weather data, airport information) before predictions are required. Flight parameters are continuously monitored and pre-processed during normal flight operations, so that when prediction is needed, the data is already prepared and validated, reducing actual prediction processing time.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The contextual models continuously process flight data throughout the flight, maintaining an ongoing analysis of aircraft state and environmental conditions. This continuous processing ensures that prediction reliability is improved through comprehensive data integration, while the system is already prepared for rapid prediction when safety events are detected.

Inventive Principle:
Principle #20Continuity of useful action

3Loss of time

If real-time predictions are provided during flight, then operator response time is improved, but computational load increases

Engineering Contradiction:
Improveoperator response timeVSAvoidcomputational load
Core Design Contradiction:
Loss of timeVSUse of energy by moving object

Solution Approach 1:

The system implements partial real-time prediction by continuously monitoring flight parameters and providing predictions only when specific triggering conditions are met (e.g., when approach parameters indicate potential instability). This selective prediction approach reduces computational load during normal flight while maintaining rapid response capability when safety events are likely, balancing operator response time needs with energy consumption constraints.

Inventive Principle:
Principle #16Partial or excessive action

Data Source

PatentEP3657472B1Systems and methods for creating and using a contextual model applicable to predicting aircraft warning conditions and analyzing aircraft performance
Publication Date: 2026.02.25 HONEYWELL INTERNATIONAL INC
  • EP3657472B1 patent drawingFigure 1
  • EP3657472B1 patent drawingFigure 2
  • EP3657472B1 patent drawingFigure 3~4

AI summary

A method for creating and using a contextual Artificial Intelligence (Al) model to analyze flight data for one or more aircraft, by a central computer system, is provided. The method obtains a set of aggregate contextual data comprising at least aircraft and flight-specific data, airport and air traffic control (ATC) data, weather data, and human factor data associated with flight crew members of the one or more aircraft; creates the contextual AI model using the set of aggregate contextual data, by the at least one processor; applies the contextual AI model to a set of flight data, to perform a statistical analysis; generates a set of results based on the statistical analysis, by the at least one processor, wherein the set of results comprises at least one of probable causes of aircraft performance events and probable aircraft performance events resulting from current conditions; and presents the set of results.