Aircraft Engine Behavior Prediction for Exhaust Temperature Exceedance

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current methods for predicting exhaust gas temperature exceedances in aircraft engines are manual, time-consuming, and prone to errors, leading to substantial costs and safety concerns.

Innovation Solution

A system utilizing a probabilistic model platform that creates a predictive model based on historical aircraft data to calculate the likelihood of exhaust gas temperature exceeding a threshold, employing Gaussian mixture models to capture relationships between engine and flight parameters, and providing automatic recommendations for adjustments to prevent exceedances.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If manual methods are used to predict exhaust gas temperature exceedances based on operator knowledge and past experiences, then decisions can be made about engine assignment, but the process is very time consuming and error prone

Engineering Contradiction:
Improveprediction accuracyVSAvoiddecision making time
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent replaces the manual mechanical decision-making process with an automated computer-based system that uses probabilistic models and machine learning algorithms to predict exhaust gas temperature exceedances, eliminating human error and significantly reducing decision-making time

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system enables self-service by automatically analyzing historical engine data, calculating exceedance probabilities, and generating predictions without requiring operator intervention, allowing the system to serve itself in making engine assignment decisions

Inventive Principle:
Principle #25Self-service

2Ease of operation

If manual approaches are used for engine assignment decisions, then operator knowledge is utilized, but the process is error prone and substantial costs are incurred

Engineering Contradiction:
Improvedecision making simplicityVSAvoidprediction accuracy
Core Design Contradiction:
Ease of operationVSReliability

Solution Approach 1:

The patent replaces error-prone manual decision-making with an automated computational system that processes data objectively, eliminating human errors while maintaining operational simplicity through automated recommendations

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Measurement precision

If exhaustive analysis of all engine parameters and flight conditions is performed manually, then accurate predictions may be achieved, but the process becomes extremely time consuming

Engineering Contradiction:
Improveprediction precisionVSAvoiddecision making efficiency
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent uses automated computer-based probabilistic modeling and machine learning algorithms to perform exhaustive parameter analysis instantaneously, achieving high prediction precision without the time cost of manual analysis

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system performs preliminary analysis by pre-processing historical data and building probabilistic models in advance, so that when predictions are needed, the system can quickly query pre-computed results rather than analyzing all parameters from scratch each time

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3217242B1Using aircraft data recorded during flight to predict aircraft engine behavior
Publication Date: 2021.06.09 GENERAL ELECTRIC CO
  • EP3217242B1 patent drawingFigure 1
  • EP3217242B1 patent drawingFigure 2
  • EP3217242B1 patent drawingFigure 3

AI summary

According to some embodiments, historical aircraft data recorded during flight 110 associated with a plurality of aircraft engines and a plurality of prior aircraft flights may be accessed. Metrics from the aircraft data recorded during flight may be automatically calculated on a per-flight basis, and a probabilistic model 140 may be employed to capture and represent relationships based on the calculated metrics, the relationships including a plurality of engine parameters and flight parameters. Conditional probability distributions 160 may then be calculated for a particular aircraft engine during a potential or historical aircraft flight based on the probabilistic model 140, engine parameter values associated with the particular aircraft engine, and flight parameter values associated with the potential or historical aircraft flight, and indications associated with the calculated conditional probability distributions 140 may be displayed 180.