Underwater Acoustic Target Ranging via Neural Network Feature Extraction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current underwater acoustic target recognition methods are inadequate due to reliance on subjective judgment, high noise interference, and high costs associated with complex device setups, leading to inaccurate and inefficient positioning in underwater environments.
Innovation Solution
An underwater acoustic target ranging method based on feature extraction and a neural network, which acquires and processes acoustic signals using a single hydrophone, extracts specific features, and trains a neural network for accurate distance estimation with reduced environmental susceptibility and device requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional acoustic signal theories and modern spectrum theories are adopted for underwater acoustic target recognition, then recognition accuracy and efficiency are improved, but device complexity and cost increase due to requiring multiple array element transducers, depth sensors, and complex signal processing systems
Solution Approach 1:
The patent extracts and utilizes only the essential acoustic signal features (amplitude, frequency, time difference of arrival) from the complex acoustic environment, discarding unnecessary information. This allows accurate target recognition using a single hydrophone instead of complex multi-element arrays, directly resolving the contradiction between measurement precision and device complexity
Solution Approach 2:
The patent replaces complex mechanical/acoustic systems (multi-element transducer arrays, depth sensors, synchronous emission systems) with an intelligent software-based neural network system that processes acoustic signals. This substitution maintains high recognition accuracy while dramatically reducing hardware complexity and cost
2Measurement precision
If multiple array element transducers and depth sensors are mounted on targets for positioning, then positioning accuracy is improved, but the cost and ease of operation deteriorate due to high device requirements and installation complexity
Solution Approach 1:
The patent enables targets to perform self-positioning by emitting acoustic signals that are detected by a single fixed hydrophone. The neural network processes the acoustic characteristics to determine target distance and position, eliminating the need for targets to carry complex positioning equipment like depth sensors or multiple transducers, thus improving ease of operation while maintaining positioning accuracy
Solution Approach 2:
The patent creates a virtual model of the acoustic environment and target characteristics through neural network training, allowing the system to infer position information from acoustic signal patterns alone. This copying approach replaces the need for physical sensors on targets, simplifying deployment and operation
3Reliability
If fixed band sound signal emission systems with multiple components are deployed for underwater acoustic positioning, then positioning capability is improved, but cost increases due to requiring multiple hydrophones, AD processors, signal amplifiers, and FPGA control chips
Solution Approach 1:
The patent makes a single hydrophone perform multiple functions: detecting acoustic signals, measuring time difference of arrival, and providing input for neural network-based target identification and positioning. This universal approach eliminates the need for separate depth sensors, multiple transducers, and complex signal processing hardware, reducing device quantity while maintaining positioning reliability
Solution Approach 2:
The patent changes the approach from using multiple sensors measuring different physical parameters to using a single sensor measuring acoustic signal parameters (amplitude, frequency, timing) that are then processed through neural network parameter transformations to derive position information. This parameter-based approach reduces hardware quantity while maintaining positioning capability
4Measurement precision
If navigational baseline node arrays with synchronous signal emission are arranged on the water surface for passive positioning, then positioning accuracy is improved, but real-time performance and adaptability deteriorate due to inability to position arbitrary targets and lack of real-time capability
Solution Approach 1:
The patent inverts the traditional active positioning approach by using a single fixed hydrophone to passively detect acoustic signals emitted by targets at various positions. The neural network processes these passive detections to determine target position, achieving both high accuracy and full adaptability to any target location without requiring pre-configured baseline arrays or synchronous emission protocols
Data Source
AI summary
The present invention provides an underwater acoustic target ranging method based on feature extraction and a neural network, including: acquiring underwater acoustic signals transmitted by an underwater acoustic target at different distances, dividing data by seconds, and using data of one second as one sample; performing framing on each sample; and separately calculating a zero-crossing rate of a time domain waveform, the second, fifth, and eighth coefficients of MFCC, a spectral centroid, a spectral skewness, a spectral entropy, and a spectral sharpness for each frame of data of each sample. In the underwater acoustic target ranging method based on feature extraction and a neural network provided in the present invention, the received underwater acoustic signal data is directly processed, so that the real-time performance is high and the reaction speed is fast.