Analysis of feature matrix in machine learning algorithms to predict energy consumption of public buildings

Ding, Yong; Fan, Lingxiao; Liu, Xue

doi:10.1016/j.enbuild.2021.111208

Cited by 69 publications

(16 citation statements)

References 31 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In this paper, a large general hospital located in Binzhou City, Shandong Province, China, is selected as a case to verify the applicability of NSGA-II in the field of large public buildings, and to demonstrate the advantages of the algorithm in practical scenarios. Among many construction projects in China, large public buildings have always been one of the building types with high energy consumption [31,9]. Some scholars have deeply discussed the energy consumption of public buildings such as commercial complexes [43], hotels [40] and office buildings [19].…”

Section: 3mentioning

confidence: 99%

Research and application of energy consumption optimization design of electromechanical system based on genetic algorithms

Guo,

Zhang,

Peng

2024

JIMO

View full text Add to dashboard Cite

In recent years, most of the new buildings in China are high-energy buildings, and electromechanical system, as a construction project, occupies a large proportion of energy consumption, which is an important way to improve the green design of buildings. The components of electromechanical system are complex, which have dialectical relationship between performance indexes under the multi-objective optimization of low energy consumption and low cost. Aiming at the common problem of high energy consumption in China's new buildings, this paper comprehensively applied NSGA-II method in genetic algorithm and building information technology (BIM), combined with Matlab, Revit, Ecotect and other computing tools. Seven indexes of electromechanical energy consumption have been analyzed for multi-objective optimization design, which namely ACE, REE, CE, HE, LE, BMS and EE The case study of Shandong Province hospital project is selected to reveal the high share of building building electromechanical system in energy consumption through the optimised energy saving effect, and at the same time, there is a huge space for energy saving optimisation. On the basis of BIM technology, genetic algorithm is used to find the optimal solution set of multi-factors of building electromechanical system, so as to provide data support for solving multi-objective and multi-variable problems in green building design. The main conclusions are as follows: (1) The NSGA-II method is suitable for the energy consumption optimization calculation of electromechanical equipment in green building projects, and the efficiency and quality of index optimization analysis can be improved by quickly searching Pareto optimal solution set through multiple iterations, which provides an important reference for scheme optimization in the design stage; (2) Pareto optimal solution sets of seven indexes of electromechanical energy consumption are obtained. The highest variation rates are CE(3.23%) and HE(2.67%), and the lowest variation rate is BMS(0.16%). (3) Through optimization simulation calculation, it is found that ACE, REE, HE and LE have a large contribution to reducing the overall energy consumption, while CE, BMS and EE have certain fluctuations before and after optimization.

show abstract

Section: 3mentioning

confidence: 99%

Research and application of energy consumption optimization design of electromechanical system based on genetic algorithms

Guo,

Zhang,

Peng

2024

JIMO

View full text Add to dashboard Cite

show abstract

“…The Filter algorithm, which is separate from the classifier and is typically used as a preprocessing step to provide the classifier with a reduced feature set, can quickly eliminate a large number of irrelevant features from the original set [ 23 , 24 ]. The Wrapper algorithm combines with the classifier in the selection process, uses the classification accuracy as the evaluation criterion to measure the quality of the feature subset, and the final retained feature subset has very good classification performance [ 9 , 25 ]. In fact, any search strategy can be used as a guide to the search direction of the Wrapper algorithm, so the search process has a strong generalization capability, but it also results in an extremely high time complexity.…”

Section: Motivation and Literature Reviewmentioning

confidence: 99%

A Hybrid Feature Selection and Multi-Label Driven Intelligent Fault Diagnosis Method for Gearbox

Liu

Zhang

et al. 2023

Sensors

View full text Add to dashboard Cite

Gearboxes are utilized in practically all complicated machinery equipment because they have great transmission accuracy and load capacities, so their failure frequently results in significant financial losses. The classification of high-dimensional data remains a difficult topic despite the fact that numerous data-driven intelligent diagnosis approaches have been suggested and employed for compound fault diagnosis in recent years with successful outcomes. In order to achieve the best diagnostic performance as the ultimate objective, a feature selection and fault decoupling framework is proposed in this paper. That is based on multi-label K-nearest neighbors (ML-kNN) as classifiers and can automatically determine the optimal subset from the original high-dimensional feature set. The proposed feature selection method is a hybrid framework that can be divided into three stages. The Fisher score, information gain, and Pearson’s correlation coefficient are three filter models that are used in the first stage to pre-rank candidate features. In the second stage, a weighting scheme based on the weighted average method is proposed to fuse the pre-ranking results obtained in the first stage and optimize the weights using a genetic algorithm to re-rank the features. The optimal subset is automatically and iteratively found in the third stage using three heuristic strategies, including binary search, sequential forward search, and sequential backward search. The method takes into account the consideration of feature irrelevance, redundancy and inter-feature interaction in the selection process, and the selected optimal subsets have better diagnostic performance. In two gearbox compound fault datasets, ML-kNN performs exceptionally well using the optimal subset with subset accuracy of 96.22% and 100%. The experimental findings demonstrate the effectiveness of the proposed method in predicting various labels for compound fault samples to identify and decouple compound faults. The proposed method performs better in terms of classification accuracy and optimal subset dimensionality when compared to other existing methods.

show abstract

“…On the other hand, the quadratic regression model was better for shorter intervals, being efficient for hour‐ahead electricity prediction. Several other machine learning methods have been applied for short‐term electricity consumption forecasting at the building level: a data‐driven swarm intelligence‐based ensemble model by Li et al, 30 a method based on rough set theory and deep learning algorithms by Lei et al, 31 linear kernel‐based algorithms and tree model‐based algorithms by Ding et al 32 and a Bayesian regression model by Dab et al 33 Finally, a deep generative model based on a latent stochastic recurrent neural network for predicting the electricity generation demand of large‐scale hydropower stations has been presented by Zhou et al 34 It also leverages generative flows for approximating the time series' distribution.…”

Section: Related Workmentioning

confidence: 99%

Electricity production and consumption modeling through fuzzy logic

Olaru

Gellért

Fiore

et al. 2022

Int J of Intelligent Sys

View full text Add to dashboard Cite

This paper proposes a prediction model based on fuzzy logic applied to anticipate electricity production and consumption in a building equipped with photovoltaics and connected to the grid. The goal is a smart energy management system able to make decisions and to adapt the consumption to the actual context and to the future electricity levels. The interest is to use as much electricity as possible from own production. The surplus is captured by an energy storage system or is sent to the grid. When no electricity is available from self-production, the grid is used to cover the necessities. The evaluations are performed on a data set collected in a real household. The proposed method is compared in terms of mean absolute error with other existing methods. The method developed based on fuzzy logic has an error of about 67 W, which places it among the most efficient models.

show abstract

Analysis of feature matrix in machine learning algorithms to predict energy consumption of public buildings

Cited by 69 publications

References 31 publications

Research and application of energy consumption optimization design of electromechanical system based on genetic algorithms

Research and application of energy consumption optimization design of electromechanical system based on genetic algorithms

A Hybrid Feature Selection and Multi-Label Driven Intelligent Fault Diagnosis Method for Gearbox

Electricity production and consumption modeling through fuzzy logic

Contact Info

Product

Resources

About