Remaining Life Estimation Using Reduced-Order Sensor Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for estimating the remaining life of equipment, such as wind turbines, face challenges due to variable operating conditions and high costs associated with detailed models or data-driven approaches, which are often based on assumptions of near-constant load parameters and do not account for complex fault modes effectively.
Innovation Solution
A system and method that utilize sensors to measure operating parameters like temperature, load, and energy consumption, correlating these with established upper and lower limit values to estimate remaining life by tracking strain and fatigue, providing a reduced-order signal indicative of consumed life, and accounting for uncertainties in sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If detailed finite element modeling based on materials knowledge is used, then prediction accuracy for remaining life is improved, but development cost and complexity increase significantly
Solution Approach 1:
The patent creates a simplified virtual model that copies the essential behavior of the physical system without replicating its full complexity. Instead of using detailed finite element models, the invention uses a reduced-order model that captures the dominant failure mechanisms and operational characteristics, providing accurate predictions at a fraction of the computational and development cost
Solution Approach 2:
The patent extracts only the critical elements necessary for predicting remaining life from the complex physical system. By identifying and isolating the key failure modes and operational parameters that drive degradation, the model eliminates unnecessary complexity while maintaining prediction accuracy for the specific application
2Reliability
If data-driven approaches with voluminous sensor data are used, then comprehensive equipment behavior tracking is achieved, but processing cost and algorithm complexity increase
Solution Approach 1:
The patent extracts only the essential features and trends from voluminous sensor data that are relevant to predicting remaining life. Instead of processing all available data, the method identifies and focuses on critical parameters and degradation patterns, significantly reducing computational requirements while maintaining prediction reliability
Solution Approach 2:
The patent creates a simplified representation of equipment behavior by copying only the essential degradation patterns and operational characteristics from the full sensor dataset. This reduced-order representation captures the key information needed for accurate predictions without requiring complex processing of the complete data volume
3Ease of operation
If manufacturer's predefined cycles to failure are used, then simple prediction methodology is provided, but accuracy decreases under variable operating conditions
Solution Approach 1:
The patent transforms static, predefined cycle counts into a dynamic prediction model that adapts to actual operating conditions. The model continuously updates remaining life estimates based on real-time sensor data and actual operational patterns, allowing it to maintain accuracy when operating conditions vary from manufacturer specifications
Solution Approach 2:
The patent changes the fundamental parameters used for prediction from fixed manufacturer specifications to dynamic, condition-based parameters. By incorporating actual operational data such as temperature, load, and cycling patterns, the model adjusts prediction accuracy to reflect real-world conditions rather than idealized test conditions
4Reliability
If margin is provided for the most demanding turbine, then reliability under severe conditions is improved, but cost increases for the majority of turbines
Solution Approach 1:
The patent applies local quality by providing enhanced prediction accuracy and reliability only where needed - for individual turbines experiencing severe or variable operating conditions. The model identifies which turbines require closer monitoring and provides tailored predictions, rather than applying uniform conservative margins to all turbines, thus avoiding excessive costs for the majority operating under normal conditions
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A system and method for estimating a remaining life period for a device. The system (10) includes a sensor (14) coupled to the device (12), wherein the sensor is configured to facilitate measuring an operating parameter (28) of the device and generate a measurement signal (30) that is representative of the operating parameter. The system includes a database (24) comprising an upper limit value (32), a lower limit value (34) and a reference parameter (36). A processor (18) of system includes circuitry (40) coupled to the sensor and coupled to the database. The processor is configured to correlate measurements of the operating parameter with the reference parameter to facilitate estimating a remaining life period (RLP) of the device.