Sensor Data Anomaly Detection via Dynamic Range Validation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In aircraft systems, determining reliable data sources during abnormal events is challenging when redundant sensors output faulty data, leading to difficulties in identifying anomalies without compromising situational awareness or safe flight operations.
Innovation Solution
A method and system that monitor sensor outputs by determining a probable range based on historical data, identifying anomalies through current behavior comparisons, and providing graphical indications on a display device to alert pilots, while also initiating remedial actions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If logical comparisons are used to compare data outputs against one another to identify discrepancies, then potential anomalies can be detected, but it becomes difficult to determine which data source is reliable when redundant sensors are each outputting faulty data
Solution Approach 1:
The patent introduces a data validation system that acts as an intermediary between redundant sensors and pilots. This system uses multiple validation techniques including range checking, rate-of-change analysis, and cross-sensor comparison to determine the reliability of each data source. When anomalies are detected, the system automatically identifies which sensor is unreliable and switches to the reliable data source, preventing loss of information about data source reliability.
Solution Approach 2:
The system implements continuous feedback loops that monitor sensor outputs and automatically adjust data source selection based on detected anomalies. The validation system provides feedback to the flight management system about the reliability of each sensor, enabling dynamic switching between data sources. This feedback mechanism ensures that reliable information is always available to pilots even when some sensors fail.
2Reliability
If redundant sensors are used to detect anomalies through logical comparisons, then potential discrepancies can be identified, but the assessment time available to pilots is insufficient to determine which sensor is unreliable without compromising situational awareness
Solution Approach 1:
The patent implements preliminary validation of sensor data continuously in the background, even during normal flight operations. The system pre-assesses the reliability of each redundant sensor by monitoring their outputs and comparing them against expected ranges and relationships. This preliminary action ensures that when an anomaly occurs, the system has already identified the reliable data source and can immediately switch without requiring pilot assessment time.
Solution Approach 2:
The data validation system operates autonomously to monitor, validate, and switch between redundant sensors without requiring pilot intervention. The system self-identifies anomalies, determines which sensor is unreliable, and automatically switches to the reliable data source. This self-service capability frees pilots from the time-consuming task of assessing sensor reliability, allowing them to maintain situational awareness while the system handles data validation.
3Reliability
If conventional anomaly detection methods are used, then some discrepancies can be detected, but valid data may be incorrectly identified as anomalous when redundant sensors are frozen or stuck in a persistent state
Solution Approach 1:
The patent implements dynamic validation thresholds and ranges that adapt to changing flight conditions and sensor behavior patterns. Instead of using fixed thresholds, the system continuously adjusts validation criteria based on historical data, current flight parameters, and the observed behavior of redundant sensors. This dynamic approach allows the system to distinguish between legitimate sensor variations and actual anomalies, preventing false identification of valid data as anomalous even when sensors are frozen or stuck.
Data Source
AI summary
Methods and systems are provided for monitoring sensors and other data sources and detecting data anomalies. One exemplary method involves determining a probable range for a metric influenced by a behavior a sensor based at least in part on historical data associated with the sensor, identifying an anomalous condition with respect to the sensor based on a relationship between a current value for the metric indicative of a current behavior of the sensor and the probable range, and providing a graphical indication of the anomalous condition on a display device.