Thermal Hydraulic Sensor Assignment for Fault Diagnosis Under Uncertainty
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current computational diagnostic frameworks for thermal hydraulic systems face challenges in accurately detecting small equipment degradations over long timescales, often resulting in false positives and failing to uniquely identify faults due to measurement uncertainties and changes in operating conditions, while also being costly and inefficient in terms of sensor installation and maintenance.
Innovation Solution
The development of a method to determine an optimal sensor set by identifying possible faults and diagnostic objectives, generating descriptions of sensor sets, calculating scores based on diagnostic capability and cost, and selecting the lowest-scoring optimal sensor set, using physics-based models and residual analysis to diagnose faults and account for uncertainty, thereby improving fault detection and reducing costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the sensitivity of fault detection algorithm is increased to detect small equipment degradations, then detection capability is improved, but false positives increase
Solution Approach 1:
The patent introduces residual analysis as an intermediary mechanism between sensor measurements and fault detection. By computing residuals (differences between expected and actual sensor readings based on physics models) and analyzing their patterns over time, the system can detect subtle equipment degradations while filtering out random noise that would otherwise cause false positives. This intermediary layer enables sensitive detection without sacrificing reliability.
2Measurement precision
If redundant sensors are installed to improve diagnostic capability, then measurement coverage is improved, but installation and maintenance costs increase
Solution Approach 1:
The patent creates virtual sensor readings through physics-based models and residual analysis. Instead of installing physical redundant sensors, the system generates virtual measurements by comparing actual sensor data against expected behavior from thermal-hydraulic models. These virtual sensor copies provide additional diagnostic information without the cost of physical hardware installation and maintenance.
Solution Approach 2:
The patent makes existing sensors serve multiple diagnostic functions through advanced signal processing and physics-based analysis. A single sensor reading is analyzed through multiple physics models and residual calculations to extract various diagnostic information, enabling one sensor to perform the work of multiple sensors would traditionally be required.
3Measurement precision
If physics-based models and residual analysis are used to account for uncertainty, then diagnostic accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the computational diagnostic framework into modular components: physics model evaluation, residual calculation, uncertainty analysis, and decision-making layers. Each module handles a specific aspect of the diagnostic process independently. This segmentation allows the system to incorporate complex physics-based models and uncertainty analysis while maintaining manageable computational complexity through modular architecture and selective application of computational methods.
Data Source
AI summary
A method for determining an optimal sensor set includes identifying a set of possible faults of a thermal hydraulic system and a set of diagnostic objectives. The method also includes obtaining descriptions of sensor sets, which includes: receiving a first description of a first sensor set, and generating, based on the first description, a second description of a second sensor set. The method further includes, for each sensor set: determining a diagnostic capability of the sensor set, and calculating a score of the particular sensor set based on the particular description and the diagnostic capability, wherein a score of a sensor set is increased by a monetary cost of the sensor set and decreased by the sensor set meeting a diagnostic objective. The method also includes identifying the optimal sensor set that has a lowest score of the sensor sets, and displaying an indication of the optimal sensor set.