Multi-Sensor Measurement System Using Reconditioned UKF Filter
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing sensor validation methods based on voter-based logic require case-by-case adaptation and are costly, introducing risks due to the diversity of functions and the need for extensive unit testing, making them inefficient for determining the best measurement among multiple sensors.
Innovation Solution
A system that merges data from multiple sensors using a reconditioned Unscented Kalman Filter (UKF) to produce a single, error-free measurement signal by identifying and rejecting faulty sensor information, allowing for a unified evaluation procedure regardless of sensor type or number.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If voter-based logic is used to validate sensor measurements, then measurement validation can be performed, but the system requires case-by-case adaptation and extensive unit testing, increasing qualification cost and complexity
Solution Approach 1:
The patent implements a universal measurement validation system that works across multiple sensor types and measurement scenarios without requiring case-by-case adaptation. The system uses a standardized UKF-based approach that can handle diverse sensors (accelerometers, gyroscopes, magnetometers, barometers, GPS) and measurement types (position, velocity, orientation, altitude) through a single unified algorithm, eliminating the need for separate voter-based logic for each sensor combination.
Solution Approach 2:
The patent replaces the mechanical voter-based logic system with a mathematical UKF-based system. Instead of using discrete voting rules that require extensive programming and testing for different scenarios, the system uses continuous probabilistic modeling and matrix operations to automatically determine the best measurement, significantly reducing software complexity and qualification requirements.
2Measurement precision
If voter-based techniques are adapted for each case, then accurate measurement validation is achieved, but qualification cost increases due to extensive unit testing requirements
Solution Approach 1:
The UKF-based system provides a universal solution that maintains high measurement precision across all sensor types and scenarios without requiring separate validation procedures. The single unified algorithm handles diverse measurement scenarios (sensor fusion, fault detection, noise filtering) through consistent mathematical operations, eliminating the need for extensive unit testing of multiple voter-based implementations.
3Reliability
If multiple sensors are used to measure the same physical quantity, then measurement reliability improves, but determining the best measurement becomes complex due to the need for case-by-case voter-based logic
Solution Approach 1:
The patent replaces complex voter-based decision logic with a streamlined UKF mathematical model that automatically evaluates multiple sensor measurements. The system uses state estimation and covariance matrix operations to continuously determine the most reliable measurement without requiring case-by-case analysis, making the process simple and automated regardless of the number or type of sensors used.
Data Source
Figure 1
Figure 2~3
Figure 4~7
AI summary
The multi-sensor measuring system having multiple sensors of a same physical variable comprises at least one set of n redundant sensors (C1, C2,..., Cn) or models representing the same physical variable to deliver n measurement signals, a merging unit (10) for merging the n measurement signals after multiplexing in order to provide a single merged multiplexed output signal from the n measurement signals and a reconditioned UKF filter (20) receiving the multiplexed output signal to provide an output signal that constitutes the best estimation of the measurement of the physical variable, having rejected the signals that indicate a sensor fault.