Helicopter Transmission Anomaly Detection Under Varying Conditions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing helicopter transmission system monitoring systems face challenges in accurately detecting anomalies due to high false positive rates and inflexibility in threshold-based methods, which are affected by varying operating conditions.
Innovation Solution
A method utilizing autoencoder and unsupervised classifiers, such as isolation forest, angle-based outlier detection, K-nearest neighbors, and local outlier factor, to analyze health indexes and flight parameters, trained on data from functioning helicopters, for anomaly detection in transmission systems.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If threshold-based methods are used for anomaly detection, then the detection process is simple and fast, but the false positive rate increases due to inability to adapt to varying operating conditions
Solution Approach 1:
The patent transforms static threshold parameters into dynamic adaptive parameters. The autoencoder continuously learns from operational data to adjust its internal parameters, enabling the system to adapt to varying operating conditions automatically. This resolves the contradiction by making the detection system intelligent and self-adjusting rather than relying on fixed thresholds.
Solution Approach 2:
The patent replaces traditional mechanical threshold-based detection with an intelligent system using autoencoders and unsupervised classifiers. This substitution of mechanical/simple logic with intelligent algorithms enables the system to handle complex patterns and adapt to changing conditions, thereby reducing false positives while maintaining computational efficiency.
2Adaptability or versatility
If unsupervised classifiers are used to improve adaptability, then the system becomes more flexible and accurate, but the computational requirements and processing time increase
Solution Approach 1:
The patent segments the anomaly detection process into distinct stages: training phase where the autoencoder learns patterns, and detection phase where unsupervised classifiers operate. This segmentation allows the computationally intensive learning to occur offline during training, while online detection uses more efficient classification algorithms, thereby reducing real-time computational energy consumption.
Solution Approach 2:
The patent performs preliminary training of the autoencoder and classification models using historical operational data before actual anomaly detection begins. This preliminary action captures the computational burden during training when data is available, allowing the system to operate more efficiently during actual use with reduced real-time computational requirements.
3Measurement precision
If multiple sensors and comprehensive monitoring are implemented, then the detection capability improves, but the system complexity and cost increase
Solution Approach 1:
The patent creates a universal monitoring framework where a single autoencoder model can detect multiple types of anomalies across different transmission system components. The unsupervised classifiers are designed to handle various anomaly patterns generically, making the system multi-functional and reducing the need for separate specialized monitoring systems for each component type.
Solution Approach 2:
The patent merges multiple monitoring functions into a unified system. Instead of separate monitoring systems for different components, the patent combines sensor data from multiple sources into a single integrated analysis framework using the autoencoder and unsupervised classifiers, thereby reducing overall system complexity while maintaining comprehensive monitoring capability.
Data Source
AI summary
A computer-implemented method is described for detecting anomalies in a transmission system of an aircraft equipped with a monitoring system, which includes a number of sensors coupled to the transmission system and determines, for each flight of the aircraft a number of respective time intervals and acquires through each sensor, for each of the time intervals, a corresponding primary signal indicative of a corresponding dynamic quantity dependent on the operation of the transmission system during at least part of the time interval. For each sensor, the monitoring system determines, starting from each primary signal acquired through the sensor during a corresponding time interval, a corresponding set of values of at least one corresponding group of synthetic indexes. The method includes detecting anomalies of the transmission system on the basis of the groups of synthetic indexes.