ML Watercraft Classification via Acoustic Propeller Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current sonar analysis systems for watercraft classification are prone to errors, leading to collisions and near misses due to the complexity and ambiguity of acoustic signals, particularly in high-pressure environments where subtle aural markers require experienced analysts to distinguish between normal and abnormal sounds under time constraints.
Innovation Solution
A machine learning algorithm trained using audio signals from watercraft, incorporating labels related to propeller, propulsor, prime mover, and submerged equipment, is implemented to improve classification by mimicking the cognitive constructs of sonar operators, enabling more accurate detection, identification, and monitoring of watercraft.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual sonar analysis is used by human operators, then classification can be performed with interpretability, but errors occur due to complexity and ambiguity of acoustic signals under time constraints
Solution Approach 1:
The patent replaces the manual mechanical analysis process with an automated machine learning system. The ML algorithm processes acoustic signals and classifies watercraft types automatically, substituting human operators' manual interpretation with computational analysis that operates faster and without fatigue, thereby reducing analysis time while maintaining or improving accuracy
Solution Approach 2:
The patent introduces an intermediary ML-based decision support system between the raw acoustic signals and the final classification decision. This intermediary processes the complex acoustic data, provides structured classifications and confidence scores, and assists human operators in making faster, more accurate decisions without completely replacing human judgment
2Measurement precision
If manual sonar analysis is used by human operators, then classification can be performed with interpretability, but errors occur due to complexity and ambiguity of acoustic signals
Solution Approach 1:
The patent replaces complex manual analysis with an ML system that handles the complexity internally. The ML model is trained on diverse acoustic data to recognize subtle patterns and variations in watercraft sounds, providing precise classifications without requiring human operators to manually analyze complex acoustic features
Solution Approach 2:
The patent implements feedback mechanisms where the ML system provides confidence scores and classification results that can be validated by human operators. This feedback loop allows the system to learn from corrections and improve over time, while operators benefit from the ML system's precise pattern recognition capabilities
Data Source
AI summary
A method of training a machine learning, ML, algorithm, is described. The method is implemented, at least in part, by a computer comprising a processor and a memory. The method comprises: providing training data comprising a set of audio signals, including a first audio signal, as respective bitstreams, corresponding with respective sets of watercraft (S501); training the ML algorithm using the provided training data comprising detecting the respective sets of watercraft and classifying the detected respective sets of watercraft according to a set of classes, including a first class, based, at least in part, on a set of labels, including a first label, wherein the set of labels relates to propeller, propulsor, prime mover and/or submerged equipment associated with watercraft (S502).