Multi-Resolution Fault Prediction via Time-Scale Model Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current fault prediction techniques for aircraft engines, which combine parametric and non-parametric data, still face challenges in accuracy and timing, often failing to detect faults sufficiently in advance, leading to maintenance disruptions and increased costs.
Innovation Solution
A method that utilizes multiple models associated with different time scales, where features are selected and processed based on their respective time scales, and outputs are fused to generate a more accurate prediction of impending faults, allowing for earlier detection and scheduling of maintenance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional fault prediction techniques combining parametric and non-parametric data are used, then some useful information is obtained, but the prediction accuracy and timing are insufficient, leading to late fault detection
Solution Approach 1:
The patent segments the fault prediction task into multiple models, each specialized for different time scales (e.g., short-term, medium-term, long-term predictions). This segmentation allows each model to optimize for specific temporal patterns, improving both accuracy and the timing of detections across different forecast horizons
Solution Approach 2:
The patent introduces a multi-dimensional approach by evaluating data across multiple time scales simultaneously. Instead of single-time-point analysis, the system creates a temporal dimension with multiple resolution levels, allowing faults to be detected at optimal advance times based on their development patterns
2Reliability
If multiple models with different time scales are used, then fault prediction accuracy and advance detection are improved, but the system complexity increases
Solution Approach 1:
The patent merges multiple specialized models into a unified multi-resolution classification system. By combining the outputs of different time-scale models through fusion techniques, the system achieves improved reliability while managing complexity through integrated architecture and coordinated model evaluation
3Productivity
If faults are detected well in advance, then maintenance can be scheduled efficiently, but immediate maintenance actions may cause service disruptions
Solution Approach 1:
The patent implements dynamic maintenance scheduling based on predicted fault timing and severity. The system adjusts maintenance plans according to the specific prediction outcomes, allowing flexible scheduling that optimizes efficiency while minimizing operational disruptions through adaptive decision-making
Data Source
AI summary
A method, apparatus and computer program product are provided to predict faults. Initially, a plurality of features are provided to a plurality of models. A subset of features is selected for each model. The plurality of features selected by a respective model is dependent upon a time scale associated with a respective model. As a result of their dependence upon different time scales, the plurality of selected features provided to a first model will differ from those provided to a second model. The plurality of models process the respective plurality of selected features. The outputs from the plurality of models are fused to generate a measure indicative of an impending fault. By providing different selected features to the models that are dependent upon the associated time scales and by then combining the outputs of the plurality of models, the resulting measure of an impending fault may accurately predict a fault well in advance of its occurrence.