AI Mass Estimation for Aircraft Using Flight Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating an aircraft's mass during flight are either inaccurate due to reliance on crew input, limited by the absence of instrumented landing gear, or dependent on limited and representative training data for machine-learning models, which affects flight performance and maintenance precision.
Innovation Solution
A method for learning an automatic machine-learning model that acquires and validates flight data using multiple sensors to calculate a reliable reference mass, which is then used to train a self-learning AI model capable of estimating the aircraft's mass in real-time during flight or on the ground, without requiring additional instrumentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If machine learning models are trained using traditional flight test data, then the model training process is simple, but the training data is limited in quantity and representativeness
Solution Approach 1:
The system uses the aircraft's existing sensors and flight data systems to automatically collect and store flight data without requiring external instrumentation or complex data acquisition systems. The aircraft serves itself by utilizing its own operational data for model training.
Solution Approach 2:
The flight data collected during normal operational flights serves dual purposes: it is both used for monitoring aircraft performance and for training machine learning models. This eliminates the need for separate dedicated training flights and instrumented landing gear.
2Adaptability or versatility
If pressure sensors are arranged on the landing gear to estimate mass, then mass estimation can be performed, but not all aircraft are equipped with instrumented landing gear
Solution Approach 1:
The method applies to all aircraft types by utilizing existing flight data systems and sensors that are standard equipment on every aircraft. No additional instrumentation or instrumented landing gear is required, making the solution universally applicable while maintaining accuracy through the use of multiple flight parameters.
3Measurement precision
If crew enters payload information manually, then the process is simple, but the accuracy is insufficient and prone to oversights
Solution Approach 1:
The system automatically collects and processes payload information from multiple sensors and flight data sources without requiring manual intervention from the crew. The aircraft's existing data systems automatically gather the necessary information for accurate mass calculation.
Solution Approach 2:
The system continuously monitors and validates flight data from multiple sources, providing feedback mechanisms that ensure data consistency and accuracy. This automated feedback loop eliminates human errors and oversights in payload data entry.
Data Source
Figure 1~2
Figure 3~4
AI summary
The present invention relates to a method for learning (20) at least one artificial intelligence model for estimating the mass of an aircraft in flight from usage data, said at least one artificial intelligence model being developed to be implemented during at least one predetermined flight phase of at least one aircraft of the same type. The method (20) comprises an embodiment (21) of a plurality of flights and in that, for at least one flight among said plurality of flights, said method (20) comprises an in-flight acquisition (22) of at least one flight data set (J1, J2), an embodiment (23) of at least one consistency test to verify that a reliable reference mass (Mref) is calculated or computable, a calculation (24) of at least one calculated mass (Mc1, Mc2) of said aircraft and a storage (25) of said at least one flight data set (J1, J2) and of said at least one calculated mass (Mc1, Mc2).