Deep Learning Drift Velocity Prediction for Floating Objects
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for predicting the drift path of floating objects at sea do not accurately account for wave-induced drift velocity, leading to reduced accuracy in locating these objects during salvage or rescue operations.
Innovation Solution
A method utilizing a deep learning model to predict wave-induced drift velocity by inputting wave characteristic parameters, trained on sample drift data that includes water surface flow, wind, and wave parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If wave-induced drift velocity is not considered in drift path prediction, then the prediction model is simpler and easier to operate, but the accuracy of locating floating objects decreases
Solution Approach 1:
The patent replaces traditional mechanical/mathematical models for calculating wave-induced drift velocity with a deep learning model. The neural network learns complex wave- floating object interaction patterns from training data, automatically capturing non-linear relationships without requiring explicit physical equations. This substitution enables accurate prediction of wave-induced drift velocity while maintaining model usability through automated inference.
Solution Approach 2:
The patent transforms the approach from using fixed theoretical formulas to using data-driven parameter estimation. The deep learning model takes wave characteristic parameters (significant wave height, peak period, mean direction) as input and outputs predicted wave-induced drift velocity components. This parameter transformation enables the system to adapt to varying sea conditions while maintaining prediction accuracy across different environments.
2Measurement precision
If wave-induced drift velocity is included in drift path prediction, then the accuracy of locating floating objects improves, but the device complexity increases
Solution Approach 1:
The deep learning model serves multiple functions: it processes wave characteristic parameters, predicts wave-induced drift velocity components, and integrates with existing wind and current drift prediction systems. This multi-functionality allows a single model to handle the complex wave- floating object interaction while maintaining system efficiency and reducing overall computational complexity.
Solution Approach 2:
The patent introduces wave characteristic parameters (significant wave height, peak period, mean direction) as intermediary variables that mediate between observed sea conditions and drift velocity prediction. These intermediaries capture essential wave properties that influence floating object drift, enabling the model to accurately predict wave-induced drift without directly modeling complex hydrodynamic interactions.
3Measurement precision
If a comprehensive model considering all factors (wind, current, waves) is used, then the drift path prediction accuracy improves, but the computational resources and time required increase
Solution Approach 1:
The deep learning model is pre-trained offline using extensive training data containing wave characteristics and corresponding drift velocity observations. During online prediction, the model directly outputs wave-induced drift velocity from wave parameters without requiring real-time complex computations. This preliminary training action separates the computationally intensive learning phase from the efficient inference phase, reducing operational computational time while maintaining high prediction accuracy.
Data Source
AI summary
A method, device, computing equipment, and storage medium for predicting the drift of the floating object are provided. A method for predicting drift velocity includes: obtaining environmental characteristic parameters at the location of the floating object to be predicted, wherein the environmental characteristic parameters include wave characteristic parameters; inputting the wave characteristic parameters into a deep learning model for wave-induced drift velocity to obtain the wave-induced drift velocity; wherein the deep learning model for wave-induced drift velocity is trained based on first sample drift data, wherein the first sample drift data includes observed drift velocity of sample floating objects, corresponding sample water surface flow characteristic parameters, sample wind characteristic parameters, and sample wave characteristic parameters.