Aircraft Engine State Evaluation Using Temporal Flight Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for monitoring aircraft engine state do not effectively account for engine aging over time without requiring prior knowledge or expertise, and fail to explicitly use temporal information to predict improvements in engine condition.
Innovation Solution
A method that creates datasets from time series data recorded during flights, trains an embedding function and classifier to learn temporal order, and calculates an engine state indicator by minimizing a cost function, allowing for the explicit consideration of engine aging and maintenance intervals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional indicator-based methods are used to monitor engine condition, then industry expertise and knowledge of data values are required, but this increases the complexity and difficulty of operation
Solution Approach 1:
The system performs self-assessment by automatically learning engine degradation patterns from historical data without requiring external expertise. The embedding function and classifier automatically identify aging signals and assess engine condition, eliminating the need for human experts to interpret complex indicator sets.
Solution Approach 2:
The patent replaces manual expert analysis with an automated machine learning system. The embedding function transforms raw sensor data into meaningful representations, and the classifier automatically detects degradation patterns, substituting the mechanical process of expert evaluation with an automated computational system.
2Measurement precision
If physical or statistical models are used to fit engine data, then prior knowledge of engine physical models is required, but this increases device complexity and difficulty of implementation
Solution Approach 1:
The patent replaces complex physical modeling with a data-driven embedding approach. Instead of requiring detailed engine physics models, the embedding function learns relevant patterns directly from sensor data, substituting mechanical modeling complexity with automated feature extraction.
Solution Approach 2:
The system transforms the problem from modeling physical parameters to learning temporal patterns in data embeddings. The embedding function changes the representation of engine data into a form where degradation patterns emerge naturally, avoiding the need to model complex physical relationships.
3Ease of manufacture
If traditional trend extraction methods are used without temporal information, then calculation is simpler, but engine aging and monotonic degradation cannot be properly detected
Solution Approach 1:
The patent replaces simple trend calculation with an embedding-based temporal analysis system. The embedding function captures temporal patterns and the classifier detects monotonic degradation, substituting simple arithmetic with intelligent temporal pattern recognition that ensures physically consistent aging detection.
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
One aspect of the invention relates to a method for evaluating the state of an aircraft engine for a given flight, each flight of the aircraft being associated with one time-domain series, comprising the following steps: - Each time maintenance of the aircraft is performed, creating a dataset comprising the time-domain series associated with each flight carried out between the maintenance and the preceding maintenance; - Creating a set of datasets comprising each created dataset and dividing it into a training set and a validating set; - Conjointly training an embedding function and a classifier, comprising: - For each dataset of the training set: - Randomly drawing a pair comprising first and second time-domain series; - Applying the embedding function to the pair to obtain a pair of embedments; - Using the classifier on the pair of embedments to obtain a probability that the first time-domain series occurred before the second time-domain series; - Computing a cost function on the basis of the probability; - Optimising the embedding function and the classifier via minimisation of the cost function; - Selecting a reference time-domain series from the time-domain series of the validating set; and - Computing a distance between the embedment associated with the time-domain series of the given flight and the embedment of the reference time-domain series, with a view to computing an indicator of engine state.