DFOS Rain Detection via Domain Generalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing DFOS systems face a domain-shifting issue that leads to a significant decrease in rain detection performance when new data is collected, as the variance in features between source and target data hampers accurate rain intensity detection.
Innovation Solution
The implementation of a domain generalization method using DAS technology and machine learning, which enriches source domain distributions by disturbing the frequency domain and transferring ambient noise patterns, allows for real-time rain intensity detection across different fiber routes and dates without requiring additional target data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If transfer learning and domain adaption methods are used to improve rain detection performance, then detection accuracy is improved, but computational cost increases and real-time detection capability is lost
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model on source domain data (rain events from specific fiber routes) before deployment. The domain generalization technique pre-adjusts the model to handle variations in ambient noise patterns, so when deployed to target domains, it can detect rain events without requiring computationally expensive retraining or fine-tuning on target data, thus achieving real-time detection capability while maintaining accuracy
Solution Approach 2:
The patent changes parameters by modifying the training process to include domain generalization techniques that adjust the model's sensitivity to different ambient noise characteristics. By altering how the model learns from source domain data (using techniques like mixing different ambient noise conditions during training), the model becomes adaptable to target domains without requiring parameter re-optimization, reducing computational time while preserving detection accuracy
2Adaptability or versatility
If domain adaption with target data is used to adapt to new environments, then model generalization is improved, but additional data collection and processing requirements increase system complexity
Solution Approach 1:
The patent applies self-service by enabling the model to adapt to target domains autonomously without requiring external intervention for data collection or model retraining. The domain generalization technique embedded in the training process allows the model to self-adjust to different ambient noise patterns in target domains, eliminating the need for additional target data collection, labeling, and processing infrastructure, thus reducing system complexity while maintaining high adaptability
3Productivity
If the trained model is applied to target domain data with different ambient noise patterns, then real-time detection is achieved, but detection performance decreases due to domain shift
Solution Approach 1:
The patent changes parameters by modifying the training process to include domain generalization techniques that adjust the model's sensitivity to different ambient noise characteristics. By altering how the model learns from source domain data (using techniques like mixing different ambient noise conditions during training), the model becomes adaptable to target domains without requiring parameter re-optimization, reducing computational time while preserving detection accuracy
Solution Approach 2:
The patent applies universality by training the model to handle multiple ambient noise patterns and environmental conditions simultaneously. The domain generalization technique makes the model universal across different fiber routes and locations by incorporating variations in wind, traffic, and human activity noise during training, allowing a single model to maintain high detection performance across diverse target domains without requiring domain-specific customization
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 significantly reduces computational time and achieves real-time weather detection, enhancing the accuracy and reliability of rain intensity measurements across varying environmental conditions.
Implementation Method 1
distributed acoustic sensing (DAS) technology... DAS can capture acoustic signals from every location along an aerial fiber cable simultaneously... the vibration is aroused by the raindrops hitting the aerial fiber cable
Data Source
AI summary
Disclosed are systems, methods, and structures that provide superior DFOS rain intensity measurements and introduce a universal solution for rain intensity detection based on the data collected by distributed acoustic sensing (DAS) technology and a designed domain generalization method. As a result, systems and methods according to the present disclosure distinguish the rain intensity of a large area through which the fiber optic cables traverse and address the domain shift issue, by employing a domain generalization technique based on machine learning technology in which newly collected target domain inference data may be distributed differently from the previously captured training source domain data. To generalize the trained model to different target domains, source domain distributions are enriched by disturbing the distribution in the frequency domain. Algorithms specifically designed to transfer the noise pattern under ambient noise environments are used to further augment the source domain distributions.