Spatial forecasting of solar radiation using ARIMA model

Remote Sensing Applications Society and Environment, Journal Year: 2020, Volume and Issue: 20, P. 100427 - 100427

Published: Oct. 20, 2020

Language: Английский

---

Efficient energy consumption prediction model for a data analytic-enabled industry building in a smart city

Building Research & Information, Journal Year: 2020, Volume and Issue: 49(1), P. 127 - 143

Published: Sept. 17, 2020

The fast development of urban advancement in the past decade requires reasonable and realistic solutions for transport, building infrastructure, natural conditions, personal satisfaction smart cities. This paper presents explores predictive energy consumption models based on data-mining techniques a small-scale steel industry South Korea. Energy data is collected using IoT systems used prediction. Data include lagging leading current reactive power, power factor, carbon dioxide emissions, load types. Five statistical algorithms are prediction:(a) General linear regression, (b) Classification regression trees, (c) Support vector machine with radial basis kernel, (d) K nearest neighbours, (e) CUBIST. Root mean squared error, Mean absolute error Coefficient variation to measure prediction efficiency models. results show that CUBIST model provides best lower values this can be efficient structural design which helps optimize policy making

Language: Английский

---

Hybrid deep CNN-SVR algorithm for solar radiation prediction problems in Queensland, Australia

Engineering Applications of Artificial Intelligence, Journal Year: 2022, Volume and Issue: 112, P. 104860 - 104860

Published: April 13, 2022

This study proposes a new hybrid deep learning (DL) model, the called CSVR, for Global Solar Radiation (GSR) predictions by integrating Convolutional Neural Network (CNN) with Support Vector Regression (SVR) approach. First, CNN algorithm is used to extract local patterns as well common features that occur recurrently in time series data at different intervals. Then, SVR subsequently adopted replace fully connected layers predict daily GSR six solar farms Queensland, Australia. To develop CSVR we adopt most pertinent meteorological variables from Climate Model and Scientific Information Landowners database. From pool of Models ground-based observations, optimal are selected through metaheuristic Feature Selection algorithm, an Atom Search Optimization method. The hyperparameters proposed optimized mean HyperOpt method, overall performance objective benchmarked against eight alternative DL methods, some other Machine Learning approaches (LSTM, DBN, RBF, BRF, MARS, WKNNR, GPML M5TREE) methods. results obtained shows model can offer several predictive advantages over models, conventional ML models. Specifically, note recorded root square error/mean absolute error ranging between ≈ 2.172–3.305 MJ m2/1.624–2.370 m2 tested compared 2.514–3.879 m2/1.939–2.866 algorithms. Consistent this predicted error, correlation measured GSR, including Willmott's, Nash-Sutcliffe's coefficient Legates & McCabe's Index was relatively higher methods all sites. Accordingly, advocates merits provide viable accurately renewable energy exploitation, demand or forecasting-based applications.

Language: Английский

---

A novel structural adaptive discrete grey prediction model and its application in forecasting renewable energy generation

Expert Systems with Applications, Journal Year: 2021, Volume and Issue: 186, P. 115761 - 115761

Published: Aug. 17, 2021

Language: Английский

---

Deep learning based long-term global solar irradiance and temperature forecasting using time series with multi-step multivariate output

Renewable Energy, Journal Year: 2023, Volume and Issue: 206, P. 135 - 147

Published: Feb. 7, 2023

Language: Английский

---

A Novel Machine Learning Approach for Solar Radiation Estimation

Sustainability, Journal Year: 2023, Volume and Issue: 15(13), P. 10609 - 10609

Published: July 5, 2023

Solar irradiation (Rs) is the electromagnetic radiation energy emitted by Sun. It plays a crucial role in sustaining life on Earth providing light, heat, and energy. Furthermore, it serves as key driver of Earth’s climate weather systems, influencing distribution heat across planet, shaping global air ocean currents, determining patterns. Variations Rs levels have significant implications for change long-term trends. Moreover, represents an abundant renewable resource, offering clean sustainable alternative to fossil fuels. By harnessing solar energy, we can actively reduce greenhouse gas emissions. However, utilization comes with its own challenges that must be addressed. One problem variability, which makes difficult predict plan consistent generation. Its intermittent nature also poses difficulties meeting continuous demand unless appropriate storage or backup systems are place. Integrating large-scale into existing power grids present technical challenges. influenced various factors; understanding these factors applications, such planning, modeling, environmental studies. Overcoming associated requires advancements technology innovative solutions. Measuring applications achieved using devices; however, expense scarcity measuring equipment pose accurately assessing monitoring levels. In order address this, methods been developed estimate Rs, including artificial intelligence machine learning (ML) models, like neural networks, kernel algorithms, tree-based ensemble methods. To demonstrate impact feature selection predictions, propose Multivariate Time Series (MVTS) model Recursive Feature Elimination (RFE) decision tree (DT), Pearson correlation (Pr), logistic regression (LR), Gradient Boosting Models (GBM), random forest (RF). Our article introduces novel framework integrates models incorporates overlooked factors. This offers more comprehensive integrations different multivariate forecasting. research delves unexplored aspects theories related results show reliable predictions based essential criteria. The ranking may vary depending used, RF Regressor algorithm selecting features maximum temperature, minimum precipitation, wind speed, relative humidity specific months. DT yield slightly set selected features. Despite variations, all exhibit impressive performance, LR demonstrating outstanding performance low RMSE (0.003) highest R2 score (0.002). other promising results, scores ranging from 0.006 0.007 0.999.

Language: Английский

---

Forecasting Network Traffic: A Survey and Tutorial With Open-Source Comparative Evaluation

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 6018 - 6044

Published: Jan. 1, 2023

This paper presents a review of the literature on network traffic prediction, while also serving as tutorial to topic. We examine works based autoregressive moving average models, like ARMA, ARIMA and SARIMA, well Artifical Neural Networks approaches, such RNN, LSTM, GRU, CNN. In all cases, we provide complete self-contained presentation mathematical foundations each technique, which allows reader get full understanding operation different proposed methods. Further, perform numerical experiments real data sets, comparing various approaches directly in terms fitting quality computational costs. make our code publicly available, so that readers can readily access wide range forecasting tools, possibly use them benchmarks for more advanced solutions.

Language: Английский

---

An optimized hybrid methodology for short‐term traffic forecasting in telecommunication networks

Transactions on Emerging Telecommunications Technologies, Journal Year: 2023, Volume and Issue: 34(12)

Published: Sept. 14, 2023

Abstract With the rapid development of telecommunication networks, predictability network traffic is significant interest in analysis and optimization, bandwidth allocation, load balancing adjustment. Consequently, recent years, research attention has been paid to forecasting traffic. Telecommunication problems can be considered a time‐series problem, wherein periodic historical data fed as input model. Time‐series approaches are broadly categorized statistical machine learning (ML) methods their combinations. Statistical forecast linear characteristics data, unable capture nonlinear complex patterns. ML‐based model data. In hybrid combining have widely used characteristics. However, performance these highly depends on feature selection techniques hyper‐parameter tuning ML methods. A novel method proposed for short‐term based hyperparameter optimization address this problem. It combines components First, technique, modified mutual information combination targets, find candidate variables. Next, vector auto regressive moving average (VARMA), long memory (LSTM), multilayer perceptron (MLP), called VARMA‐LSTM‐MLP forecaster, suggested metaheuristic algorithm, composed firefly BAT, employed optimal set values. The assessed by real‐world dataset containing Tehran city's daily IRAN. evaluation results demonstrate that outperforms existing terms mean squared error absolute error.

Language: Английский

---

Improving Solar PV Prediction Performance with RF-CatBoost Ensemble: A Robust and Complementary Approach

Renewable energy focus, Journal Year: 2023, Volume and Issue: 46, P. 207 - 221

Published: June 28, 2023

Language: Английский

---

Prediction of Solar Irradiance One Hour Ahead Based on Quantum Long Short-Term Memory Network

IEEE Transactions on Quantum Engineering, Journal Year: 2023, Volume and Issue: 4, P. 1 - 15

Published: Jan. 1, 2023

The short-term forecasting of photovoltaic (PV) power generation ensures the scheduling and dispatching electrical power, helps design a PV-integrated energy management system, enhances security grid operation. However, due to randomness solar energy, output PV system will fluctuate, which affect safe operation grid. To solve this problem, high-precision hybrid prediction model based on variational quantum circuit (VQC) long memory (LSTM) network is developed predict irradiance 1 hour in advance. VQC embedded LSTM iteratively optimize weight parameters four gates (forgetting gate, input cell state, gate) improve accuracy. evaluate performance model, five radiation observatories located China are selected, together with widely used models including seasonal autoregressive integrated moving average, convolution neural network, recurrent (RNN), gate unit, (GRU), LSTM; comparisons made under different seasons months. experimental results show that annual average root mean square error 61.756 $\text{W/m}^{2}$ , reduced by 10.7%, 13.9%, 8.1%, 3.8%, 3.4%, respectively, compared other models; absolute 24.257 28.1%, 28.9%, 24.1%, 12.2%, 12.8%, R-Square ( notation="LaTeX">$R^{2}$ ) 0.946, improved 1.5%, 1.9%, 1.2%, 0.4%, models.

Language: Английский

---