Cabin Air Compressor AI Surge Detection for Early Intervention

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Aircraft cabin air compressors are vulnerable to compressor surge, which causes unstable performance, mechanical damage, and potential catastrophic failure, with existing built-in tests only detecting fully developed surges, leading to costly repairs and reduced aircraft usability.

Innovation Solution

A machine learning model, trained on vibration and speed data, is deployed on a microcontroller to detect incipient compressor surge through spectrogram analysis, allowing for real-time detection and corrective actions to prevent full surge.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If built-in tests are used to detect compressor surge, then surge detection capability is provided, but only fully developed surges are detected leading to delayed intervention

Engineering Contradiction:
Improvesurge detection capabilityVSAvoiddetection timing
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The machine learning model is trained on vibration data collected during built-in tests, including data from before surge events occur. This preliminary training enables the model to recognize early signs of surge (incipient surge) before full surge develops, allowing detection at an earlier stage and enabling preventive action before damage occurs.

Inventive Principle:
Principle #10Preliminary action

2Speed

If machine learning model is deployed on microcontroller, then real-time detection capability is achieved, but computational resources and model size are constrained

Engineering Contradiction:
Improvedetection speedVSAvoidmodel size
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The patent extracts and implements only the essential inference functionality of the machine learning model on the microcontroller, while training and extensive processing are performed offline. The model is converted to a reduced form suitable for embedded deployment, taking out only the necessary computational elements needed for real-time spectrogram analysis on resource-constrained hardware.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The machine learning model undergoes parameter optimization and conversion to adapt to the microcontroller's computational constraints. This includes converting the model to appropriate data formats, optimizing numerical precision, and adjusting model architecture parameters to fit within the embedded system's memory and processing capabilities while maintaining detection accuracy.

Inventive Principle:
Principle #35Parameter changes

3Measurement precision

If comprehensive training data is collected during built-in tests, then model accuracy is improved, but data storage and processing requirements increase

Engineering Contradiction:
Improvemodel accuracyVSAvoiddata volume
Core Design Contradiction:
Measurement precisionVSQuantity of substance

Solution Approach 1:

The system extracts and stores only the essential vibration data features during built-in tests, rather than storing complete raw datasets. By focusing on capturing relevant vibration signatures and operational parameters during testing, the system reduces data volume while maintaining sufficient information for effective machine learning model training and accurate surge detection.

Inventive Principle:
Principle #2Taking out (Extraction)

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

The system effectively prevents compressor surge by detecting incipient events, reducing damage and improving aircraft reliability and usability by enabling early intervention.

Implementation Method 1

a vibration sensor associated with the cabin air compressor to generate a plurality of operating spectrograms

Methodology Applied
Scientific EffectVibration: Vibration

Implementation Method 2

generating the training spectrogram comprises applying a fast Fourier transform to the training data

Methodology Applied
Scientific EffectFast Fourier transform:

Data Source

PatentEP4216013B1Incipient compressor surge detection using artificial intelligence
Publication Date: 2025.11.12 HAMILTON SUNDSTRAND CORP
  • EP4216013B1 patent drawingFigure 1
  • EP4216013B1 patent drawingFigure 2
  • EP4216013B1 patent drawingFigure 3

AI summary

Examples described herein provide a computer-implemented method that includes receiving training data indicative of incipient compressor surge for cabin air compressors (112). The method further includes generating, using the training data, a training spectrogram. The method further includes training, by a processing system (100), a machine learning model to detect incipient compressor surge events for the cabin air compressors using the spectrogram. The method further includes receiving, at a microcontroller (118) associated with a cabin air compressor, operating data associated with the cabin air compressor. The method further includes generating, at the microcontroller and using the operating data, an operating spectrogram. The method further includes detecting, by the microcontroller associated with the cabin air compressor, an incipient compressor surge event by applying the machine learning model to the operating spectrogram. The method further includes implementing a corrective action to correct the incipient compressor surge event.