Sensor Fault Detection Using Estimator Networks and Residuals
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for monitoring critical systems using sensor redundancy are costly and inefficient, as they do not adequately address the reliability of data from a single sensor without hardware redundancy.
Innovation Solution
A fault determination method using a sensor array and estimator network, including artificial neural networks, calculates residuals from different data points to identify faulty sensors, providing software redundancy and reducing hardware requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If hardware redundancy of sensors is implemented, then system reliability is improved, but cost and weight increase
Solution Approach 1:
The patent creates virtual copies of sensor data through estimators that simulate sensor behavior using mathematical models and neural networks. These software-based estimators replicate the function of physical sensors without the associated hardware cost and weight, resolving the contradiction between reliability and quantity of substance
Solution Approach 2:
The patent replaces the mechanical sensor hardware system with a software-based estimator system that uses computational models and neural networks to generate sensor data. This substitution eliminates the need for redundant physical sensors while maintaining system reliability through software redundancy
2Quantity of substance
If hardware redundancy is reduced, then cost and weight decrease, but data reliability from remaining sensors deteriorates
Solution Approach 1:
The patent introduces estimators as intermediary components that bridge the gap between reduced hardware redundancy and maintained data reliability. These estimators process available sensor data and generate additional data points, ensuring reliable data availability even when physical sensors are minimized
Solution Approach 2:
The patent creates virtual sensor data through estimators that copy the functional behavior of physical sensors. This software-based copying allows the system to maintain data reliability from multiple sources without actually having multiple physical sensors, thus reducing hardware while preserving reliability
3Quantity of substance
If software redundancy is implemented, then hardware requirements are reduced, but system complexity increases
Solution Approach 1:
The patent creates a universal estimator framework that can serve multiple sensors and multiple physical quantities simultaneously. The neural network estimators are designed to handle various sensor types and measurement parameters through a unified architecture, reducing overall system complexity despite the introduction of software redundancy
Solution Approach 2:
The patent transforms the sensor system from hardware-based redundancy to software-based redundancy by changing the fundamental parameter of how sensor data is generated. This parameter change from physical to virtual sensor data enables hardware reduction while managing complexity through computational methods
Data Source
Figure 1~2
AI summary
A method for determining a fault for a set of sensors (10), comprising: - obtaining data, said data comprising a plurality of measurements returned respectively by the sensors (A, B, C) and a plurality of estimates determined respectively by a plurality of estimators (D, E) taking different subsets of said measurements at input, at least one of the estimators (E) comprising an artificial neural network; - computing residuals (30) representative of the difference between data representing one and the same physical quantity; - determining (40) whether there is a fault on at least one of the sensors (A, B, C) based on the residuals computed on the basis of data obtained from measurements returned by said sensor.