Asset Degradation Path Classification for Remaining Useful Life Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current prognostic methods fail to accurately estimate the remaining useful life (RUL) of assets due to the assumption of a clear failure threshold, which is often vague in real-world applications with complex failure modes, and lack flexibility in handling various degradation signals and expert opinions.
Innovation Solution
The Path Classification and Estimation (PACE) method and system classify asset degradation based on exemplar data and estimate RUL by combining class memberships with expected lifetimes, allowing for flexibility in incorporating multiple degradation signals and expert opinions, and does not require a predefined failure threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If a clear failure threshold is assumed for degradation signals, then prognostic methods can provide simple RUL estimates, but the accuracy deteriorates when failure modes are complex and thresholds are vague
Solution Approach 1:
The patent transforms the degradation problem from threshold-based classification to a continuous similarity measurement problem. Instead of asking whether degradation exceeds a threshold, the method computes similarity metrics between current and historical degradation paths, allowing accurate RUL estimation without requiring precise failure threshold definitions.
Solution Approach 2:
The patent uses historical degradation paths as exemplars or templates. By copying and comparing against these stored degradation trajectories, the system can estimate RUL based on similarity to known failure patterns, eliminating the need to define explicit failure thresholds for each new asset.
2Device complexity
If traditional prognostic methods use single degradation signals and fixed failure thresholds, then the model is simple, but it lacks flexibility in handling multiple degradation signals and expert opinions
Solution Approach 1:
The patent creates a universal prognostic framework that can handle multiple types of degradation signals (vibration, temperature, pressure, etc.) and multiple sources of information (sensor data, expert opinions, historical records) through a single similarity-based comparison mechanism. The same exemplar degradation paths can be compared against different signal types using consistent similarity metrics.
Solution Approach 2:
The patent segments the prognostic problem into independent similarity comparisons between current degradation paths and historical exemplars. Each degradation signal can be analyzed separately and then integrated through the similarity framework, allowing modular incorporation of multiple signal sources without requiring a completely integrated complex model.
3Ease of operation
If prognostic methods require predefined failure thresholds, then the estimation process is straightforward, but it cannot dynamically track changes in asset behavior
Solution Approach 1:
The patent implements a dynamic prognostic approach where similarity is continuously computed between current degradation paths and historical exemplars. As new degradation data arrives, the system dynamically updates similarity measurements and RUL estimates, automatically adapting to changes in asset behavior without requiring manual threshold adjustments or model reconfiguration.
Data Source
AI summary
Path classification and estimation method and system used in combination with a computer and memory for prognosticating the remaining useful life of an asset by classifying a current degradation path of a current asset as belonging to one or more of previously collected degradation paths of exemplary assets and using the resulting classifications to estimate the remaining useful life of the current asset.