2-D Convolutional Deep Neural Network for the Multivariate Prediction of Photovoltaic Time Series

Rosato, Antonello; Araneo, Rodolfo; Andreotti, Amedeo; Succetti, Federico; Panella, Massimo

doi:10.3390/en14092392

Cited by 15 publications

(9 citation statements)

References 47 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep learning, on the other hand, can be effective in tackling challenges such as noise reduction and predicting for multiple outputs, given sufficient training dataset. Examples include the use of convolutional neural network 25 or recurrent neural network. 12,26…”

Section: Case Study 3: Pem Fuel Cellsmentioning

confidence: 99%

Data science accelerates energy device development

Cheng,

Cui,

Chueh

et al. 2024

Preprint

View full text Add to dashboard Cite

Data science has become increasingly prevalent in the development of energy devices, offering significant advancements in predicting future behaviors and identifying optimal process parameters in a resource-saving manner. This perspective begins by examining the role of data science and ML in enhancing accelerated aging tests across solar, battery and fuel cells. We present a generalizable data-driven workflow for processing aging test data and predicting the lifespan of different device types. In this perspective, we discuss two strategies to improve our understanding of device failures: integrating physics-based parameters and utilizing interpretable machine learning (ML) techniques. Following a brief review on ML-assisted process optimization, we propose an interpretable closed-loop platform towards digital manufacturing for thin-film solar and Li-ion battery production. Finally, we discuss the current challenges and research gaps in applying data science for accelerated energy device development, aiming to spark further investigation in this field.

show abstract

Section: Case Study 3: Pem Fuel Cellsmentioning

confidence: 99%

Data science accelerates energy device development

Cheng,

Cui,

Chueh

et al. 2024

Preprint

View full text Add to dashboard Cite

show abstract

“…However, these approaches fail to implement long-term dependencies into the prediction [12]. Modern architectures, therefore, focus on using complex gated recurrent Networks with memory storage [13][14][15]. The results clearly indicate superior performance on multiple datasets which contain long-term dependencies in the data [15].…”

Section: Related Workmentioning

confidence: 99%

Data Driven prediction of forced nonlinear vibrations using stabilised Autoregressive Neural Networks

Westmeier

Kreuter

Bäuerle

et al. 2023

Proc Appl Math and Mech

View full text Add to dashboard Cite

In this work, we propose a novel approach to the data‐driven prediction of vibration responses of nonlinear systems. The main idea is based on Autoregressive Neural Networks (ARNN) to model the nonlinear transfer behaviour between an external excitation and the system response. We propose an autoregressive network architecture with embedded symmetry using bias‐free tanh activation and guarantee Input‐to‐State‐Stability (ISS) by enforcing a special penalty term to the weights. The resulting training procedure is analysed for the example of a DUFFING oscillator with white noise excitation. In a BAYESian optimisation, it is found that beyond enforcing input‐to‐state‐stability, the stabilising penalty term also decreases sensitivity with respect to other training parameters compared to other classical techniques. Furthermore, we show that the stabilised ARNN is able to give excellent approximations of the nonlinear response of the DUFFING oscillator for a wide range of excitation intensities. In contrast, linear models, such as autoregressive models with exogenous input (ARX) in time domain or linear transfer functions in frequency domain, will only find some linear approximation. In particular, by construction, they will not be able to capture nonlinear effects for arbitrary amplitudes and excitation levels.

show abstract

“…Thus, it seems logical to assume that they can also provide accurate predictions in time series forecasting, as a great amount of data, with strong time dependency, is usually used to obtain future values. Therefore, they have been used to predict electric energy load [ 1 , 4 , 5 , 14 , 35 , 36 , 37 , 38 ], electricity prices [ 29 ] energy production in photovoltaic plants [ 39 ], consumption in residential areas [ 7 ] and buildings [ 8 , 9 ] or CO 2 emission allowance prices [ 40 ].…”

Section: Introductionmentioning

confidence: 99%

“…Although LSTMs are able to provide accurate predictions [ 1 , 5 , 9 , 35 , 37 , 38 ], they have been used along with other tools to improve performance, in the same way as was done with the statistical tools. Thus, in [ 2 ], the structure was combined with a Convolutional Neural Network (CNN, another class of Deep Learning networks) to provide short-term electric load forecasting, while in [ 39 ], this same structure was used to forecast power generation in a photovoltaic plant; in [ 41 ] it was combined with a CNN and an ARIMA to predict future prices of CO 2 allowance prices in the European Union, and in [ 14 ], a Genetic Algorithm was used to optimize an LSTM.…”

Section: Introductionmentioning

confidence: 99%

Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building with Recurrent Neural Networks and Empirical Mode Decomposition

Fernández-Martínez

Jaramillo-Morán

2022

Sensors

View full text Add to dashboard Cite

Short-term forecasting of electric energy consumption has become a critical issue for companies selling and buying electricity because of the fluctuating and rising trend of its price. Forecasting tools based on Artificial Intelligence have proved to provide accurate and reliable prediction, especially Neural Networks, which have been widely used and have become one of the preferred ones. In this work, two of them, Long Short-Term Memories and Gated Recurrent Units, have been used along with a preprocessing algorithm, the Empirical Mode Decomposition, to make up a hybrid model to predict the following 24 hourly consumptions (a whole day ahead) of a hospital. Two different datasets have been used to forecast them: a univariate one in which only consumptions are used and a multivariate one in which other three variables (reactive consumption, temperature, and humidity) have been also used. The results achieved show that the best performances were obtained with the multivariate dataset. In this scenario, the hybrid models (neural network with preprocessing) clearly outperformed the simple ones (only the neural network). Both neural models provided similar performances in all cases. The best results (Mean Absolute Percentage Error: 3.51% and Root Mean Square Error: 55.06) were obtained with the Long Short-Term Memory with preprocessing with the multivariate dataset.

show abstract

2-D Convolutional Deep Neural Network for the Multivariate Prediction of Photovoltaic Time Series

Cited by 15 publications

References 47 publications

Data science accelerates energy device development

Data science accelerates energy device development

Data Driven prediction of forced nonlinear vibrations using stabilised Autoregressive Neural Networks

Multi-Step Hourly Power Consumption Forecasting in a Healthcare Building with Recurrent Neural Networks and Empirical Mode Decomposition

Contact Info

Product

Resources

About