Aircraft Engine Sensor Training for Fault-Tolerant Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Aircraft engine control systems face challenges in identifying and accommodating sensor malfunctions without increasing cost and system complexity, as existing methods rely on redundant sensors to mitigate failures.
Innovation Solution
An aircraft engine control system uses an AI model trained with real-time sensor data from multiple sensors to derive parameter values, allowing the system to continue operating by providing derived parameter values to the control unit, thereby reducing the need for redundant sensors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If redundant sensors are used to mitigate sensor failures, then system reliability is improved, but system complexity and cost increase
Solution Approach 1:
The patent creates a virtual copy of the failed sensor's functionality by training an AI model on historical sensor data. This digital twin or virtual sensor replicates the measurement capabilities of the physical sensor without requiring additional hardware, thereby maintaining reliability while avoiding the complexity of redundant physical sensors.
Solution Approach 2:
The patent replaces the mechanical/physical sensor system with an artificial intelligence-based software system. Instead of using additional physical sensors to detect parameters, the system uses machine learning models that process data from other sensors to derive the same information, substituting mechanical detection with computational analysis.
2Reliability
If redundant sensors are used to mitigate sensor failures, then system reliability is improved, but cost increases
Solution Approach 1:
The patent creates a virtual copy of the failed sensor's functionality by training an AI model on historical sensor data. This digital twin or virtual sensor replicates the measurement capabilities of the physical sensor without requiring additional hardware, thereby maintaining reliability while avoiding the complexity of redundant physical sensors.
Solution Approach 2:
The patent employs computationally lightweight AI models that can be trained and deployed efficiently using existing sensor data infrastructure. The approach uses standard machine learning techniques that leverage existing computational resources, avoiding the need for expensive specialized hardware or complex redundant sensor installations.
3Device complexity
If AI model is trained using stored sensor data to produce derived parameter values, then the need for redundant sensors is reduced, but measurement precision may be affected
Solution Approach 1:
The patent performs preliminary training of the AI model using extensive historical sensor data before deployment. This pre-training phase allows the model to learn accurate relationships between sensor parameters, ensuring that when the model operates in real-time, it can produce derived values with high precision comparable to physical sensors.
Solution Approach 2:
The patent implements a feedback mechanism where the AI model's derived sensor values are continuously evaluated against actual sensor readings when available. This feedback loop allows the system to validate measurement precision and potentially retrain or adjust the model to maintain or improve accuracy over time.
Data Source
AI summary
A system and method for controlling an aircraft engine is provided. The method includes a) producing first sensor data using a first sensor sensing a first parameter during operation of the aircraft engine on a flight mission; b) producing other sensor data using a plurality of second sensors sensing a plurality of other parameters, during operation of the aircraft engine; c) providing the first and other sensor data to a control unit during operation of the aircraft engine; d) storing the first and other sensor data during operation of the aircraft engine; e) using an artificial intelligence (AI) model that is trained using the stored first and other sensor data produced during operation of the aircraft engine, to produce one or more derived first parameter values; and f) selectively providing the one or more derived first parameter values to the control unit for use in controlling the aircraft engine.