ORACLE PHM Platform Anomaly Detection Neural Network Motifs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In manned space exploration missions beyond Low Earth Orbit, there is a challenge in timely and feasible resupply of consumables and replacement components, requiring innovative solutions for monitoring life support and mission critical systems, particularly in detecting anomalies in space habitats where regular maintenance is difficult.
Innovation Solution
A Prognostics and Health Management (PHM) test & validation platform, ORACLE, uses artificial neural networks to detect anomalies in biotic sensor data, such as ECG signals, by transitioning between motifs to simulate environmental changes and generate training data for deep learning-based frameworks, integrating crew health status as a supplementary indicator for habitat health monitoring.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If regular resupply and maintenance operations are implemented in deep space missions, then system reliability can be maintained, but mission feasibility and timing become compromised due to the inability to easily access resupply vessels
Solution Approach 1:
The system enables self-service through autonomous anomaly detection and diagnosis capabilities. The neural network continuously monitors sensor data from life support and mission critical systems, automatically detecting anomalies and diagnosing potential failures without human intervention, allowing the system to maintain itself in the isolated deep space environment
Solution Approach 2:
The system performs preliminary action by providing early warning of failures through continuous monitoring and prediction. The neural network analyzes sensor data trends to predict potential failures before they occur, enabling proactive maintenance planning and preventing system failures that would compromise mission reliability
2Measurement precision
If deep learning frameworks are trained with sufficient data for accurate anomaly detection, then detection precision improves, but data availability and training time are limited in space mission contexts
Solution Approach 1:
The system performs preliminary action by pre-training the neural network with simulated sensor data and historical patterns before deployment. This allows the model to be ready for immediate use with limited real-world data, achieving accurate anomaly detection without requiring extensive training time during the mission
Solution Approach 2:
The system uses copying by creating synthetic training data that replicates normal and anomalous sensor patterns. The neural network is trained on these copied datasets that mimic real mission conditions, enabling accurate anomaly detection without requiring actual mission data for training
Data Source
AI summary
Various examples are provided related to anomaly detection in sensor data (e.g., biotic or abiotic sensor data). In one example, a method includes applying data from portions of a real-time sensor signal to an artificial neural network trained to identify motifs associated with the real-time sensor signal; detecting a transition from a first motif to a second motif based upon changes in output signals providing an indication of correlation of the sensor signal to the motifs; and identifying a change in an environmental condition based upon the transition between the motifs. In another example, a method includes selecting a plurality of motifs associated with a desired training signal; generating the desired training signal by transitioning between different motifs in a pseudo-random basis; and generating training data sets from the desired training signal, which can then be utilized to train a network or other machine learning system.