Sensor Data Fusion for Aircraft Anomaly Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for merging sensor measurements in aircraft do not reliably determine the probability of false alarms, which can lead to incorrect identification of sensor anomalies.
Innovation Solution
A method that calculates detection deviations for each sensor, compares them to predetermined thresholds, and determines the probability of false alarms to accurately identify sensor anomalies, allowing for a more reliable estimation of flight parameters by merging sensor data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If a tolerance band method is used to detect sensor anomalies, then the searching for possible anomalies is simplified, but the probability of false alarm cannot be reliably determined
Solution Approach 1:
The patent transforms the anomaly detection problem from a simple tolerance band comparison to a statistical hypothesis testing framework by changing the parameters to include detection deviations, predetermined thresholds, and probability calculations. This allows reliable determination of false alarm probabilities while maintaining operational simplicity through automated statistical computations.
2Reliability
If multiple sensors are used to improve estimation accuracy, then the reliability of flight parameter estimation is improved, but the complexity of measurements merging increases
Solution Approach 1:
The patent implements a feedback mechanism where the detection deviation of each sensor is computed based on the difference between its measurement and the estimated parameter derived from other sensors. This feedback loop enables automatic identification and exclusion of anomalous sensors, improving estimation reliability while managing complexity through systematic statistical processing.
3Measurement precision
If sensor measurements are merged to provide a consolidated value, then the estimation of flight parameters is improved, but the ability to detect actual sensor anomalies decreases due to false alarms
Solution Approach 1:
The patent performs preliminary anomaly detection and classification before final parameter estimation by computing detection deviations and comparing them to predetermined thresholds. This preliminary action identifies potentially anomalous sensors, allowing the merging process to either exclude them or apply appropriate weighting, thereby maintaining measurement precision while reducing false alarm impacts on anomaly detection reliability.
Data Source
AI summary
A method for merging measurements of a flight parameter of an aircraft, from measurements (y1, y2, y3, y4) of this parameter supplied respectively by a plurality of sensors (C1, C2, C3, C4), comprising: for each sensor (C1; C2; C3; C4), computing a deviation (T1; T2; T3; T4), proportional to the absolute value of a difference between a measurement (y1, y2, y3; y4) supplied by this sensor, and an estimation of the parameter computed from the measurements supplied by the other sensors; comparing each deviation to a corresponding threshold (Td1; Td2; Td3; Td4); based on the comparisons, determining the presence or not of an anomaly on one of the sensors with a determined total probability of false alarm; and merging measurements to provide a final estimation of the parameter ({circumflex over (x)}).