Aircraft Performance Modeling for Robust Flight Trajectory Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for optimizing aircraft operations are limited by the lack of publicly available performance data from aircraft manufacturers, leading to dependence on proprietary models that are not royalty-free and are prone to convergence issues with non-linear parametric estimation methods, resulting in poor estimation of aircraft behavior.
Innovation Solution
The method partitions the aircraft flight trajectory into segments governed by distinct sets of equations based on thrust mode and vertical guidance, using a parametric optimization engine with least squares calculations to determine aircraft performance models independently of manufacturer data, ensuring convergence in high-dimensional hybrid multi-model problems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If non-linear parametric estimation methods (e.g., least squares with gradient descent) are used to model aircraft performance, then the model can handle complex non-linear relationships, but the method may converge to local minima leading to poor estimation accuracy
Solution Approach 1:
The patent segments the aircraft performance model into multiple linear sub-models, each representing a specific flight condition or operational mode. By dividing the complex non-linear performance characteristics into manageable linear segments, the system avoids convergence to local minima while maintaining modeling accuracy for non-linear relationships.
Solution Approach 2:
The system dynamically selects and switches between different linear sub-models based on current flight conditions (altitude, speed, weight, etc.). This dynamic adaptation allows the model to accurately represent non-linear performance characteristics across the entire flight envelope without relying on gradient-based optimization that can get trapped in local minima.
2Measurement precision
If proprietary aircraft performance data from manufacturers is used, then the model can achieve high accuracy, but the system becomes dependent on non-royalty-free data and loses independence
Solution Approach 1:
The system performs self-calibration by using actual flight data from the specific aircraft to automatically determine and update its performance parameters. Instead of relying on proprietary manufacturer data, the model learns and adapts to the unique characteristics of each aircraft through self-service calibration processes, achieving both accuracy and independence.
Solution Approach 2:
The system implements feedback mechanisms where actual flight performance data is continuously compared with model predictions. This feedback loop allows the model to automatically adjust and refine its parameters based on real-world observations, eliminating dependence on manufacturer data while maintaining or improving accuracy through empirical validation.
3Ease of manufacture
If existing aircraft performance models (BADA, EUCASS) are used, then the system can leverage established frameworks, but the models are limited in thrust/drag representation or engine type applicability
Solution Approach 1:
The patent creates a universal performance modeling framework that can accommodate multiple engine types (turbojet, turbofan, propeller) and various thrust models through a common architecture. The system uses standardized input parameters and unified calculation methods that work across different aircraft configurations, eliminating the limitations of engine-specific models like EUCASS while maintaining ease of implementation.
Data Source
Figure 1
Figure 2
AI summary
Methods and systems for optimizing aircraft flight are disclosed. The trajectory is partitioned into segments, each governed by distinct sets of equations that are functions of the engine thrust mode and vertical guidance (climb, cruise, or descent). Assuming two models—aerodynamic and engine speed—data from flight recordings are received, and a number of parameters for a parametric optimization engine are iteratively determined by applying a least-squares calculation until a predefined minimum criterion is met. The parametric optimization engine is then used to predict the trajectory point following a given point. Software and system aspects (e.g., FMS and/or EFB) are described.