Structural Impact Classification via Multi-Characteristic Signal Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current structural health monitoring systems rely on threshold-based methods that often result in false positives or false negatives, failing to accurately classify vibrations caused by various events, leading to delayed detection of damage or debris on structures.
Innovation Solution
A method that classifies accelerometer data by determining event characteristics such as slope-to-peak, slope-from-peak, standard deviation, and event period, comparing them to historical event ranges, and weighing classifications to determine event types, with the option to transmit notifications and update historical data for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If strict threshold-based monitoring is used to detect impact events, then the system can identify events when vibration levels exceed a certain threshold, but the system produces false positives and artificially inflates event lengths when the threshold is set too low
Solution Approach 1:
The system changes from using a single threshold parameter to using multiple event characteristics (slope-to-peak, slope-from-peak, standard deviation, event period) that are compared against historical ranges. This multi-parameter approach allows for more nuanced discrimination between true impact events and false positives, resolving the contradiction between detection sensitivity and false positive rate.
Solution Approach 2:
The system performs preliminary classification of event characteristics before making a final impact determination. By pre-defining characteristic ranges from historical events and classifying current events against these ranges, the system can filter out false positives before they propagate through the monitoring system, improving both accuracy and reliability.
2Reliability
If the threshold is set too high in threshold-based monitoring, then the system reduces false positives, but the system misses lower magnitude impacts such as vehicles scraping along the underside of a bridge
Solution Approach 1:
The system transitions from a single threshold parameter to multiple event characteristics including slope-to-peak, slope-from-peak, standard deviation, and event period. This allows the system to detect lower magnitude impacts by analyzing the shape and temporal characteristics of the vibration signal rather than relying solely on amplitude thresholding, thus improving detection accuracy without increasing false positives.
Solution Approach 2:
The system adds temporal and morphological dimensions to event detection by analyzing slope characteristics and event periods in addition to amplitude. This multi-dimensional approach enables detection of low-magnitude impacts that have distinctive temporal signatures, resolving the contradiction between reducing false positives and detecting subtle events.
3Productivity
If current threshold-based systems are used, then the system can identify events, but the systems cannot determine the nature of an event and therefore produce false results
Solution Approach 1:
The system segments the event analysis into distinct classification steps: determining event characteristics, comparing to historical ranges, classifying individual characteristics, weighing classifications, and determining final event type. This segmented approach maintains fast detection while improving classification accuracy by systematically evaluating multiple characteristics rather than relying on a single threshold crossing.
Solution Approach 2:
The system uses historical event data to establish characteristic ranges and continuously refines classification by comparing current events against this historical baseline. The weighing mechanism for multiple characteristic classifications provides feedback that improves event nature determination, resolving the contradiction between rapid detection and accurate classification.
4Device complexity
If no notification system is implemented, then the system structure remains simple, but the structure may go uninspected for hours or days until debris or new damage is noticed
Solution Approach 1:
The system performs preliminary classification of event characteristics and determines event types before triggering notifications. By pre-classifying events using historical data and characteristic comparison, the system can immediately notify authorities of verified impact events rather than waiting for manual inspection, reducing inspection delay without significantly increasing system complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enhances the accuracy of event classification, reducing false positives and negatives, enabling timely detection and notification of structural impacts, and improving the reliability of structural health monitoring systems.
Implementation Method 1
The measurements include a first measurement, a second measurement, and a plurality of intermediate measurements between the first measurement and the second measurement
Implementation Method 2
structures are generally subject to outside forces that can cause vibrations within a structure
Data Source
AI summary
A method for classifying accelerometer data of a structure includes obtaining acceleration data from an accelerometer positioned to monitor a structure and receive vibrations from the structure. The acceleration data includes a plurality of measurements. The method includes selecting a subset of the plurality of measurements as an event signal. The subset of the plurality of measurements have a magnitude that exceeds a noise floor and includes a first measurement, a second measurement, and a plurality of intermediate measurements between the first measurement and the second measurement. The plurality of intermediate measurements exceed an event threshold. The event threshold is greater than the noise floor. The method includes comparing the event signal to a set of historical events and classifying the event signal as an event type.