Spectral Swell Prediction Model for Longer Forecast Horizons
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing swell prediction methods, such as those based on AR models and Kalman filters, struggle to provide accurate predictions beyond half a swell period, especially when signal-to-noise ratio is low or non-linear effects are present, and fail to combine measurements from different sensors, limiting the efficiency of wave-energy converters and stability of floating systems.
Innovation Solution
A method for predicting swell characteristics using a spectral model that updates in real-time with sensor measurements, incorporating a transfer function to determine a swell prediction model, allowing for accurate predictions up to 5 minutes into the future, and includes filtering and confidence determination steps to enhance prediction accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If autoregressive models with adaptive filters are used for short-term wave prediction, then prediction accuracy is improved within half a swell period, but prediction accuracy deteriorates beyond half a swell period when signal-to-noise ratio is low or non-linear effects are present
Solution Approach 1:
The patent applies dynamics by making the spectral model adaptive and time-varying. The spectral density function is continuously updated using recursive least-squares estimation, allowing the prediction model to adapt to changing sea conditions. This dynamic adaptation enables accurate predictions beyond half a swell period by tracking changes in wave characteristics as they occur, rather than assuming stationary conditions.
Solution Approach 2:
The patent changes parameters by using a spectral approach that models wave characteristics in the frequency domain rather than time domain. The spectral density function parameters are continuously estimated and updated, allowing the system to capture non-linear effects and varying signal-to-noise ratios. This parameter transformation from time-series to spectral domain enables robust predictions under varying conditions.
2Device complexity
If a single time series AR model is used for prediction, then the model structure is simple, but the ability to combine measurements from different sensors is lost
Solution Approach 1:
The patent applies universality by creating a spectral model framework that can universally accommodate multiple types of sensors and measurement sources. The spectral density function is estimated from multiple time series inputs simultaneously, allowing integration of data from different sensors (e.g., wave elevation, wave force, system response) into a unified prediction model. This multi-functional approach maintains relatively simple model structure while gaining the versatility to process diverse sensor data.
3Power
If traditional prediction methods are used, then computational requirements are low, but prediction horizon is limited to half a swell period
Solution Approach 1:
The patent applies dimensionality change by transforming the prediction problem from the time domain to the spectral (frequency) domain. This dimensional transformation allows the model to capture long-range correlations and periodicities that are difficult to detect in the time domain. The spectral approach enables predictions beyond half a swell period by utilizing frequency-domain characteristics that reveal underlying wave patterns and structures.
Data Source
AI summary
The present invention is a method for predicting a swell-resulting characteristic for a floating system. The method updates (MAJ) a spectral model (MSH) of the swell to form a swell prediction model (MPR). This model is applied to real-time measurements (MES) to predict the swell-resulting characteristic (pred).