Engine Thermal Anomaly Detection With VQVAE Reconstruction

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing methods for detecting thermal anomalies in gas turbine engines are inefficient, time-consuming, and prone to false positives, particularly due to the inflexibility of fixed IR cameras and the computational complexity of image comparison algorithms.

Innovation Solution

Utilizing a vector quantized variational autoencoder (VQVAE) trained on baseline infrared images to generate reconstructed images and detect thermal anomalies, combined with AI camera control systems for real-time temperature mapping and anomaly identification.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Stability of the object's composition

If fixed IR cameras are used to monitor thermal conditions, then the monitoring coverage is stable, but the system flexibility and adaptability to different locations are reduced

Engineering Contradiction:

Improvemonitoring coverage stabilityVSAvoidsystem flexibility

Core Design Contradiction:

Stability of the object's compositionVSAdaptability or versatility

Solution Approach 1:

The patent employs movable or reconfigurable sensor platforms that can dynamically adjust their positions and orientations to monitor different locations within the turbine engine. This allows the system to maintain stable monitoring of specific areas while adapting to different inspection requirements and locations, resolving the contradiction between coverage stability and system flexibility.

Inventive Principle:

Principle #15Dynamics

2Ease of operation

If traditional image comparison algorithms are used for thermal anomaly detection, then the detection process is straightforward, but the computational complexity and time consumption increase

Engineering Contradiction:

Improvedetection process simplicityVSAvoidcomputation time

Core Design Contradiction:

Ease of operationVSLoss of time

Solution Approach 1:

The system performs preliminary actions by pre-processing thermal images to extract key features and characteristics before comparison. This includes normalizing temperature data, identifying regions of interest, and pre-computing reference profiles. By preparing data in advance, the actual anomaly detection requires less computational time while maintaining simplicity in the detection process.

Inventive Principle:

Principle #10Preliminary action

3Measurement precision

If sensors are placed closer to hot gas components for better monitoring, then the measurement precision improves, but the sensors are exposed to higher temperatures that can damage them

Engineering Contradiction:

Improvethermal anomaly detection accuracyVSAvoidsensor damage from high temperature

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent introduces intermediary cooling mechanisms and protective barriers between the sensors and the hot gas components. These intermediaries allow sensors to operate at safe temperatures while still capturing high-quality thermal data from nearby components, thus maintaining measurement precision without exposing sensors to damaging heat levels.

Inventive Principle:

Principle #24Intermediary (Mediator)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

Enables efficient, autonomous detection of thermal anomalies with reduced false positives, facilitating timely corrective actions and reducing engine downtime.

Implementation Method 1

a plurality of infrared cameras to capture a baseline image set, the baseline image set including at least two thermal images

Methodology Applied

Scientific EffectThermal radiation: Thermal Radiation

Implementation Method 2

providing the baseline image set and the emissivity data to an artificial intelligence model, the artificial intelligence model to generate a reconstructed image set

Methodology Applied

Scientific EffectVector quantized variational autoencoder (VQVAE):

Data Source

PatentUS12422331B2Methods and apparatus to autonomously detect thermal anomalies

Publication Date: 2025.09.23 GENERAL ELECTRIC CO

AI summary

Methods, apparatus, systems, and articles of manufacture are disclosed to autonomously detect thermal anomalies. Disclosed examples include an example apparatus to detect engine anomalies comprising: at least one memory; instructions in the apparatus; and processor circuitry to execute the instructions to: control a plurality of infrared cameras to capture a baseline image set, the baseline image set including at least two thermal images; generate emissivity data based on the baseline image set; provide the baseline image set and the emissivity data to an artificial intelligence model, the artificial intelligence model to generate a reconstructed image set; determine a difference between the baseline image set and the reconstructed image set; and in response to the difference exceeding a threshold, generate an alert indicating detection of an engine anomaly.