Aircraft Engine Sensor Validation Using AI for Redundant Mismatch
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing aircraft electronic control systems lack the capability to establish the integrity and validity of redundant sensor signals, leading to uncertainty and potential non-optimal engine performance due to mismatch errors between redundant sensors.
Innovation Solution
Utilizing an artificial intelligence model trained on historical parameter data to predict the accurate sensor value from mismatched redundant sensors, enabling the engine control unit to select the most appropriate value for engine control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple redundant sensors are used to sense the same parameter, then reliability is improved, but mismatch errors between sensors create uncertainty about signal integrity
Solution Approach 1:
An AI model serves as an intermediary between the redundant sensors and the control system. The model receives signals from multiple sensors, processes them to detect mismatch errors, and outputs a determined sensor value that represents the accurate measurement. This intermediary resolves the uncertainty created by sensor mismatches while preserving the reliability benefits of redundancy.
Solution Approach 2:
The patent replaces traditional mechanical/electronic sensor validation methods with an AI-based computational approach. Instead of using simple threshold comparisons or voting mechanisms, the system uses machine learning models trained on historical sensor data to intelligently determine which sensor signals are accurate, thereby substituting conventional signal processing with intelligent algorithms.
2Difficulty of detecting and measuring
If traditional control system logic is used to detect sensor failures, then out-of-range signals can be identified, but mismatch errors between in-range signals cannot be detected
Solution Approach 1:
The AI model transforms the approach to sensor validation by changing from fixed threshold parameters to dynamic, learned parameters. The model learns normal sensor relationships and variations from training data, enabling it to detect subtle mismatch errors that fall within traditional in-range thresholds. This parameter transformation allows detection of previously undetectable error conditions.
Solution Approach 2:
The system performs preliminary action by training the AI model offline using historical sensor data before deployment. This pre-training establishes baseline expectations for sensor behavior under various operating conditions, enabling the model to quickly and accurately detect mismatches during real-time operation without requiring complex real-time analysis algorithms.
3Reliability
If redundant sensors are implemented to mitigate failure risk, then system reliability improves, but system complexity and cost increase
Solution Approach 1:
The AI model performs multiple functions within a single computational framework: it processes signals from multiple sensors, detects mismatch errors, determines accurate sensor values, and adapts to different operating conditions. This multi-functionality consolidates what would otherwise require separate validation systems, reducing overall complexity despite the presence of redundant sensors.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method and system for processing parameter values from a redundant sensor (34) configured to sense a parameter used in the control of an aircraft engine (10) is provided. The method includes: a) receiving a plurality of parameter values from a redundant sensor (34) by sensing the same parameter at the same time; b) identifying mismatched parameter values; c) producing a predicted parameter value using an artificial intelligence (AI) model having a database of parameter values representative of the sensed parameter; d) providing the predicted parameter value to a control unit (36); and e) operating the control unit (36) to select a first parameter value or a second parameter value using the predicted parameter for use in the control of the aircraft engine (10).