Deep learning-based feature engineering methods for improved building energy prediction

Applied Energy, Journal Year: 2019, Volume and Issue: 240, P. 35 - 45

Published: Feb. 13, 2019

Language: Английский

---

Tree-based ensemble methods for predicting PV power generation and their comparison with support vector regression

Energy, Journal Year: 2018, Volume and Issue: 164, P. 465 - 474

Published: Aug. 30, 2018

The variability of renewable energy resources, due to the characteristic weather fluctuations, introduces uncertainty in generation output that are greater than conventional reserves grid uses deal with relatively predictable uncertainties demand. high makes forecasting critical for optimal balancing and dispatch plants a smarter grid. challenge is improve accuracy confidence level forecasts at reasonable computational cost. Ensemble methods such as random forest (RF) extra trees (ET) well suited predicting stochastic photovoltaic (PV) they reduce variance bias by combining several machine learning techniques while improving stability; i.e. generalisation capabilities. This paper investigated accuracy, stability cost RF ET hourly PV output, compared their performance support vector regression (SVR), supervised technique. All developed models have comparable predictive power equally applicable output. Despite power, outperformed SVR terms algorithmic efficiency ETs make them an ideal candidate wider deployment forecasting.

Language: Английский

---

Machine learning applications in urban building energy performance forecasting: A systematic review

Renewable and Sustainable Energy Reviews, Journal Year: 2020, Volume and Issue: 133, P. 110287 - 110287

Published: Sept. 2, 2020

Language: Английский

---

Implementing ensemble learning methods to predict the shear strength of RC deep beams with/without web reinforcements

Engineering Structures, Journal Year: 2021, Volume and Issue: 235, P. 111979 - 111979

Published: Feb. 27, 2021

Language: Английский

---

Forecasting Energy Use in Buildings Using Artificial Neural Networks: A Review

Energies, Journal Year: 2019, Volume and Issue: 12(17), P. 3254 - 3254

Published: Aug. 23, 2019

During the past century, energy consumption and associated greenhouse gas emissions have increased drastically due to a wide variety of factors including both technological population-based. Therefore, increasing our efficiency is great importance in order achieve overall sustainability. Forecasting building important for applications planning, management, optimization, conservation. Data-driven models forecasting grown significantly within few decades their performance, robustness ease deployment. Amongst many different types models, artificial neural networks rank among most popular data-driven approaches applied date. This paper offers review studies published since year 2000 which use demand, with particular focus on reviewing applications, data, performance metrics used model evaluations. Based this review, existing research gaps are identified presented. Finally, future directions area highlighted.

Language: Английский

---

Artificial Intelligence Evolution in Smart Buildings for Energy Efficiency

Applied Sciences, Journal Year: 2021, Volume and Issue: 11(2), P. 763 - 763

Published: Jan. 14, 2021

The emerging concept of smart buildings, which requires the incorporation sensors and big data (BD) utilizes artificial intelligence (AI), promises to usher in a new age urban energy efficiency. By using AI technologies consumption can be reduced through better control, improved reliability, automation. This paper is an in-depth review recent studies on application (AI) buildings building management system (BMS) demand response programs (DRPs). In addition elaborating principles applications AI-based modeling approaches widely used use prediction, evaluation framework introduced for assessing research conducted this field across major domains, including energy, comfort, design, maintenance. Finally, includes discussion open challenges future directions buildings.

Language: Английский

---

Tuning machine learning models for prediction of building energy loads

Sustainable Cities and Society, Journal Year: 2019, Volume and Issue: 47, P. 101484 - 101484

Published: March 16, 2019

Language: Английский

---

A novel improved model for building energy consumption prediction based on model integration

Applied Energy, Journal Year: 2020, Volume and Issue: 262, P. 114561 - 114561

Published: Feb. 8, 2020

Language: Английский

---

Statistical investigations of transfer learning-based methodology for short-term building energy predictions

Applied Energy, Journal Year: 2020, Volume and Issue: 262, P. 114499 - 114499

Published: Jan. 11, 2020

Language: Английский

---

Short-Term Prediction of Residential Power Energy Consumption via CNN and Multi-Layer Bi-Directional LSTM Networks

IEEE Access, Journal Year: 2019, Volume and Issue: 8, P. 123369 - 123380

Published: Dec. 30, 2019

Excessive Power Consumption (PC) and demand for power is increasing on a daily basis, due to advancements in technology, the rise electricity-dependent machinery, growth of human population. It has become necessary predict PC order improve management co-operation between energy used building grid. State-of-the-art Energy Prediction (ECP) methods are limited terms predicting effectively, various challenges such as weather conditions dynamic behaviour occupants. Thus, overcome drawbacks these methods, we present an intelligent hybrid technique that combines Convolutional Neural Network (CNN) with Multi-layer Bi-directional Long-short Term Memory (M-BDLSTM) method using three steps. When applied short-term ECP, this approach helps provide efficient i.e. it can assist supplier produce optimum amount power. The first step our proposed integrates pre-processing data organisation mechanisms refine remove abnormalities. second employs deep learning network, where sequence refined fed into CNN via M-BDLSTM network learn pattern effectively. third generates ECP/PC by comparing actual predicted series evaluates prediction error metrics. achieves better results than existing techniques, thus demonstrating its effectiveness. Furthermore, achieved smallest value Mean Square Error (MSE) Root (RMSE) individual household dataset 10-fold Cross Validation (CV) hold-out method.

Language: Английский

---