Forecasting, hindcasting and feature selection of ocean waves via recurrent and sequence-to-sequence networks

Pirhooshyaran, Mohammad; Snyder, Lawrence

doi:10.1016/j.oceaneng.2020.107424

Cited by 62 publications

(25 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Table 11 indicates the results of the comparison. First two rows are reported from [3]. Furthermore, to be consistent with the mentioned study we used RMSE and HUBER losses besides MAAPE.…”

Section: Feature Selection In Wave Renewable Energymentioning

confidence: 95%

“…Therefore, in this section we aim to find the most important marine features affecting the wave output power. We use the case study established by [3] in the east coast of the U.S. 1 The case study explores the effect of different ocean features, recorded by several sensors of different distances, on the significant wave height and ocean power of specified location. They use an Elastic net (EN) regularizer for feature selection.…”

Section: Feature Selection In Wave Renewable Energymentioning

confidence: 99%

“…Machine learning (ML) approaches such as support vector regression (SVR) and neural networks are considered as the state-of-the-art methods for multivariate multistep forecasting in particular for renewable energy systems [1][2][3]. We argue there can be two possible improvements.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Feature engineering and forecasting via derivative-free optimization and ensemble of sequence-to-sequence networks with applications in renewable energy

Pirhooshyaran

Scheinberg

Snyder

2020

Energy

Self Cite

View full text Add to dashboard Cite

This study introduces a framework for the forecasting, reconstruction and feature engineering of multivariate processes along with its renewable energy applications. We integrate derivative-free optimization with an ensemble of sequence-to-sequence networks and design a new resampling technique called additive resampling, which, along with Bootstrap aggregating (bagging) resampling, are applied to initialize the ensemble structure. Moreover, we explore the proposed framework performance on three renewable energy sources-wind, solar and ocean wave-and conduct several short-to long-term forecasts showing the superiority of the proposed method compared to numerous machine learning techniques. The findings indicate that the introduced method performs more accurately when the forecasting horizon becomes longer. In addition, we modify the framework for automated feature selection. The model represents a clear interpretation of the selected features. Furthermore, we investigate the effects of different environmental and marine factors on the wind speed and ocean output power, respectively, and report the selected features. Finally, we explore the online forecasting setting and illustrate that the model outperforms alternatives through different measurement errors.

show abstract

“…Table 11 indicates the results of the comparison. First two rows are reported from [3]. Furthermore, to be consistent with the mentioned study we used RMSE and HUBER losses besides MAAPE.…”

Section: Feature Selection In Wave Renewable Energymentioning

confidence: 95%

Section: Feature Selection In Wave Renewable Energymentioning

confidence: 99%

See 1 more Smart Citation

Feature engineering and forecasting via derivative-free optimization and ensemble of sequence-to-sequence networks with applications in renewable energy

Pirhooshyaran

Scheinberg

Snyder

2020

Energy

Self Cite

View full text Add to dashboard Cite

show abstract

“…The LSTM does not require exceptional complexity to debug hyperparameters, and it can choose to remember or forget long-term information through a forgetting gate [ 11 ]. Many researchers have used LSTM as a practical algorithm for solving vanishing gradients in electricity pricing, stock prices, robot control and disease prediction [ 12 ]. For time series prediction and analysis, this algorithm outperforms other traditional machine learning algorithms.…”

Section: Introductionmentioning

confidence: 99%

A New Machine Learning Algorithm for Numerical Prediction of Near-Earth Environment Sensors along the Inland of East Antarctica

Wang

Dou

Yang

et al. 2021

Sensors

View full text Add to dashboard Cite

Accurate short-term small-area meteorological forecasts are essential to ensure the safety of operations and equipment operations in the Antarctic interior. This study proposes a deep learning-based multi-input neural network model to address this problem. The newly proposed model is predicted by combining a stacked autoencoder and a long- and short-term memory network. The self-stacking autoencoder maximises the features and removes redundancy from the target weather station’s sensor data and extracts temporal features from the sensor data using a long- and short-term memory network. The proposed new model evaluates the prediction performance and generalisation capability at four observation sites at different East Antarctic latitudes (including the Antarctic maximum and the coastal region). The performance of five deep learning networks is compared through five evaluation metrics, and the optimal form of input combination is discussed. The results show that the prediction capability of the model outperforms the other models. It provides a new method for short-term meteorological prediction in a small inland Antarctic region.

show abstract

“…More and more attention has been received on the predicting ocean wave parameters using deep learning network by now. The RNN and sequence-to-sequence neural networks have been introduced to predict short and long-term significant wave height and output power of the ocean waves, and the case studies proved that the Adam and AMSGrad optimization algorithms are the most robust ones to optimize the sequence-to-sequence network based on real data obtained from NOAA buoy measurements [16]. A two-step wind-wave prediction model was explored to predict wind speed and wave height based on deep RNNs with a lower prediction error being produced when compared with shallower MLP [17].…”

Section: Introductionmentioning

confidence: 99%

An integrated long-short term memory algorithm for predicting polar westerlies wave height

2020

Ocean Engineering

View full text Add to dashboard Cite

The improved knowledge of wave height and period conditions has considerably influenced on ocean navigation, marine fishery and engineering, especially in the polar regions. The methods of predicting ocean wave height which involve field measurements, numerical simulation, physical models and analytical solutions have been gradually developed with intelligent functions. Despite numerical wave models being dominant for recent decades, wave forecasting is still facing many challenges such as small region forecasting and large amounts of data needed. This paper presents a novel deep learning algorithm, namely Long Short Term Memory (LSTM), incorporating with Principal Component Analysis (PCA) to predict the wave height by using data from four wave buoys as deployed in the polar westerlies for two and half months. The PCA method is used to extract principal components from a set of input signals while LSTM is adopted to avoid long term independences during the forecasting. The novelty of this paper is to investigate an artificial intelligence (AI) based model in the field of time sequence forecasting in order to determine the performance of wave conditions by using AI technology. The result from this integrated method demonstrates that the LSTM model has the potential to better predict wave height in the polar condition based on time-space domain information. The PCA is proved essential for selection of input signals and for correlation analysis. For comparison, different data-driven models are applied and the results also show the purposed model achieves the highest scores in terms of R-squared value. Besides, the paper also discusses the challenges for long term and high-value prediction which needs to be optimized in the future work.

show abstract

Forecasting, hindcasting and feature selection of ocean waves via recurrent and sequence-to-sequence networks

Cited by 62 publications

References 47 publications

Feature engineering and forecasting via derivative-free optimization and ensemble of sequence-to-sequence networks with applications in renewable energy

Feature engineering and forecasting via derivative-free optimization and ensemble of sequence-to-sequence networks with applications in renewable energy

A New Machine Learning Algorithm for Numerical Prediction of Near-Earth Environment Sensors along the Inland of East Antarctica

An integrated long-short term memory algorithm for predicting polar westerlies wave height

Contact Info

Product

Resources

About