Sonobuoy Acoustic Source Identification Using Wavelet Denoising and CNN
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Identifying submarine signals in noisy marine environments is challenging due to sound distortion and interference from dense water layers, subsurface features, and background noise, making it difficult to detect and locate submarines using sonobuoys.
Innovation Solution
The method employs wavelet denoising to minimize background noise, followed by autocorrelation for signal enhancement, and uses a convolutional neural network (CNN) for signal source isolation and classification, enabling accurate and automated detection and geolocation of submarines in noisy environments.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If traditional noise reduction methods are used, then background noise can be reduced, but the processing time increases and transient signals are lost
Solution Approach 1:
The patent replaces traditional mechanical signal averaging methods with wavelet transform-based denoising. Wavelet transform provides a mathematical framework that can separate signal and noise in the time-frequency domain, achieving noise reduction without requiring multiple signal acquisitions or lengthy processing times. This substitution of mathematical transformation for mechanical averaging resolves the contradiction between noise reduction and processing time.
Solution Approach 2:
The patent changes the parameter domain from time-domain averaging to time-frequency domain analysis using wavelet transforms. By transforming the signal into different resolution levels, the method can selectively preserve transient features while removing noise, achieving both noise reduction and preservation of transient signals without increased processing time.
2Measurement precision
If signal processing is performed to identify submarine signals, then detection accuracy improves, but the complexity of the processing system increases
Solution Approach 1:
The patent segments the signal processing into distinct stages: wavelet denoising, autocorrelation enhancement, and CNN classification. Each stage handles a specific aspect of signal processing, making the overall complex system manageable through modular organization. This segmentation allows high detection accuracy to be achieved while controlling system complexity through structured processing steps.
Solution Approach 2:
The patent introduces autocorrelation as an intermediary processing step between wavelet denoising and CNN classification. This intermediary enhances the signal characteristics by emphasizing periodic patterns while suppressing random noise, bridging the gap between raw signal processing and machine learning classification, and improving overall detection accuracy without requiring direct complex interactions between all processing components.
3Measurement precision
If multiple sonobuoys are used for geolocation, then submarine location accuracy improves, but the cost and system complexity increase
Solution Approach 1:
The patent makes each sonobuoy multi-functional by equipping it with both passive acoustic monitoring capabilities and onboard signal processing (wavelet denoising and autocorrelation). This allows individual sonobuoys to perform multiple functions independently, reducing the need for complex centralized processing and making the deployment of multiple units more cost-effective while maintaining geolocation accuracy through distributed processing.
Data Source
AI summary
A method to detect the presence and location of submarines in a complex marine environment by wavelet denoising, wavelet signal enhancement, by autocorrelation and signal source identification a convolutional neural network.