AHRS Sensor Weighting for Fault-Tolerant Strapdown Attitude
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Inertial strapdown systems face challenges with sensor inaccuracies and failures, leading to invalid or inaccurate attitude and heading measurements, and existing Built-In-Test (BIT) solutions struggle to balance between trusting inaccurate data and generating false failure reports.
Innovation Solution
A method involving multiple redundant sensors, applying Gaussian curves to sensor outputs, and weighting them based on their positions on the curves to determine combined roll, pitch, and heading values, ensuring reliable navigation by combining valid outputs and ignoring outliers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If Built-In-Test (BIT) is used to detect sensor failures, then sensor failure detection capability is improved, but attitude/heading accuracy is lost when a sensor fails
Solution Approach 1:
The system changes the parameter of sensor output weighting dynamically. When a sensor fails BIT, its output is not completely discarded but reweighted with reduced confidence. The weighting parameter adjusts based on BIT status and agreement with other sensors, allowing the system to maintain attitude/heading accuracy by blending sensor inputs rather than abruptly switching to backup sensors.
Solution Approach 2:
The sensor weighting scheme is dynamic rather than static. The system continuously evaluates sensor agreement and BIT status to adjust weights in real-time. This dynamic adaptation allows smooth transitions between sensor configurations and maintains measurement precision even when sensors fail, avoiding the abrupt accuracy loss associated with traditional BIT approaches.
2Reliability
If BIT threshold is set to be sensitive, then false failure reports increase, but if set to be lenient, then inaccurate data is trusted
Solution Approach 1:
The system implements feedback through continuous monitoring of sensor agreement. When one sensor's output diverges from others, the system detects this discrepancy and adjusts weighting accordingly. This feedback mechanism allows the system to trust accurate data while identifying inaccurate readings without relying solely on fixed BIT thresholds, thereby reducing both false failures and trust in inaccurate data.
Solution Approach 2:
Instead of completely accepting or completely rejecting sensor data based on BIT status, the system applies partial weighting. Even sensors that fail BIT can contribute to the solution with reduced weight, while sensors that agree well receive higher weight. This partial action approach balances the trade-off by allowing the system to be sensitive to failures without completely discarding potentially useful data.
3Reliability
If multiple redundant sensors are used, then system reliability is improved, but device complexity increases
Solution Approach 1:
The system merges multiple sensor inputs into a unified attitude/heading solution through weighted combination. Rather than maintaining separate processing chains for each sensor, the system combines outputs from multiple sensors with dynamically adjusted weights, achieving high reliability while managing complexity through consolidation of the processing architecture.
Data Source
AI summary
An inertial navigation system is provided including sensors and a controller coupled to the sensors. A method of the system includes, at each of multiple iterations: receiving first outputs and second outputs from the sensors; applying a first Gaussian curve to the first outputs, and a second Gaussian curve to the second outputs; weighting each of the first outputs based on a position on the first Gaussian curve of each of the first outputs, and each of the second outputs based on a position on the second Gaussian curve of each of the second outputs; determining a combined first output based on the weighting of the first outputs and determining a combined second output based on the weighting of the second outputs; and calculating at least two of a roll, a pitch, and a heading based on the combined first output and combined second output.