Distributed Acoustic Sensing Event Model Training via Geospatial Calibration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for automatically analyzing distributed acoustic sensing (DAS) data to identify events of interest, such as footsteps or vehicle movements, are labor-intensive and prone to user errors, making it difficult to generate accurate and sufficient training data for machine learning models.
Innovation Solution
The method involves calibrating a sensing optical fibre using a geospatial reference system to automatically define training data subsets within DAS signals, allowing for the creation of accurate training data without human intervention, using a calibration vibration source and position detectors to map acoustic signals to specific positions along the fibre, enabling the training of machine learning models to detect events of interest.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual or semi-automated methods are used to label DAS data for training, then human intervention can correct errors, but the process becomes slow, cumbersome, and prone to user error
Solution Approach 1:
The system uses automatically generated labels from machine learning models to train itself, eliminating the need for manual human intervention in the data labelling process. The model generates its own training data through simulated events and automatically labels them, creating a self-sufficient training pipeline that is both fast and scalable.
Solution Approach 2:
The system pre-generates synthetic DAS signal data containing simulated events of interest before actual field deployment. By creating and labeling training data in advance through simulation, the system prepares comprehensive training datasets without requiring manual intervention during actual event detection, significantly accelerating the training process.
2Productivity
If semi-automated generation of labels is used, then productivity increases, but frequent errors occur requiring human correction
Solution Approach 1:
The system creates synthetic copies of real DAS signals through simulation, generating training data that replicates actual field conditions without requiring manual labelling. By copying the physical characteristics and event patterns of real signals through computational simulation, the system produces accurate labels automatically without human intervention while maintaining high fidelity to real-world scenarios.
3Reliability
If extensive user analysis is performed to characterize events, then detection accuracy improves, but the process becomes onerous and requires extensive manual effort
Solution Approach 1:
The system replaces manual mechanical analysis processes with automated computational simulations. Instead of requiring users to manually characterize events through graphical interfaces and iterative analysis, the system uses computational models to automatically generate and label training data, significantly reducing operational complexity while maintaining or improving detection accuracy through systematic simulation of diverse event scenarios.
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 allows for efficient and accurate generation of training data subsets, reducing human error and increasing the volume of labeled data, enabling machine learning models to reliably detect events of interest in DAS signals with minimal false positives.
Implementation Method 1
probe light backscattered within the native material of a sensing optical fibre extending through an environment, typically using coherent Rayleigh backscatter, is used to detect acoustic vibration at the sensing optical fibre
Data Source
AI summary
There is disclosed a method of training one or more event models for use in identifying events of interest proximal to the sensing optical fibre, from a distributing acoustic sensing signal representing acoustic vibration at positions along a sensing optical fibre. A distributed acoustic sensor is provided and arranged to form a distributed acoustic sensing signal. A calibration defining the mapping between the distributed acoustic sensing signal and positions along the sensing optical fibre is obtained. Events of interest of one or more different event categories are implemented at measured positions along the sensing optical fibre. The distributed sensing signal is formed during the implemented events of interest. For each event of interest, one or more training data subsets of the distributed acoustic sensing signal are defined. These training data sets are defined to be contemporary with the implemented event and spatially positioned within the signal, using the measured positions and the calibration, so as to include the implemented event. The training data subsets are then used to train one or more event models to detect the events of interest.