Day-ahead prediction of plug-in loads using a long short-term memory neural network

Markovic, Romana; Azar, Elie; Annaqeeb, Masab Khalid; Frisch, Jérôme; Treeck, Christoph van

doi:10.1016/j.enbuild.2020.110667

Cited by 19 publications

(10 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Deep learning models are neural networks with learned feature representation over multiple hidden layers [17]. In the related building research, these methods have been extensively used for energy consumption and occupant behavior (OB) modeling applications [5,[18][19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37]. Qian et al [23] explored the potential of ANNs for HVAC load forecasting, when applied to small amount of data.…”

Section: Deep Learning Methods For Buildings' Controlmentioning

confidence: 99%

“…The results proved that without the implemented model, classification accuracy could hardly reach 90 %. Conclusively, Markovic et al [37] developed a LSTM neural network for day-ahead prediction of miscellaneous electric loads (MELs). The proposed implementation outperformed benchmark approaches based on Weibull distribution and Gaussian mixture methods when MELs and occupancy information data were used as input parameters to the model.…”

Section: Deep Learning Methods For Buildings' Controlmentioning

confidence: 99%

“…Network saturation was identified to be a major modeling complexity in case of both feed-forward and convolutional denoising autoencoders. Formally, neural network saturation could be described as an impediment to gradient propagation [37,57]. Its main effect was to produce always the same output, no matter how different the input sequence was.…”

Section: General Observations During Model Developmentmentioning

confidence: 99%

“…In this study, since no other formal metrics were available at the current state of the research, network saturation was evaluated using the saturation performance metric (SAT) proposed by Markovic et al [37], calculated over the normalized reconstructed values.…”

Section: General Observations During Model Developmentmentioning

confidence: 99%

“…The saturation performance results of the proposed models for different CR are summarized in Table 3. Here, network saturation is defined as the SAT lower than 0.1 [37]. Every autoencoder was well above the previous defined limit, meaning that the adopted architecture strategies could efficiently overcome saturation issues.…”

Section: General Observations During Model Developmentmentioning

confidence: 99%

See 4 more Smart Citations

Indoor environment data time-series reconstruction using autoencoder neural networks

Liguori,

Markovic,

Dam

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

As the number of installed meters in buildings increases, there is a growing number of data time-series that could be used to develop data-driven models to support and optimize building operation. However, building data sets are often characterized by errors and missing values, which are considered, by the recent research, among the main limiting factors on the performance of the proposed models. Motivated by the need to address the problem of missing data in building operation, this work presents a data-driven approach to fill these gaps. In this study, three different autoencoder neural networks are trained to reconstruct missing indoor environment data time-series in a data set collected in an office building in Aachen, Germany. The models are applicable for different time-series obtained from room automation, such as indoor air temperature, relative humidity and CO 2 data streams. The results prove that the proposed methods outperform classic numerical approaches and they result in reconstructing the corresponding variables with average RMSEs of 0.42 C, 1.30 % and 78.41 ppm, respectively.

show abstract

Section: Deep Learning Methods For Buildings' Controlmentioning

confidence: 99%

Section: Deep Learning Methods For Buildings' Controlmentioning

confidence: 99%

Section: General Observations During Model Developmentmentioning

confidence: 99%

Section: General Observations During Model Developmentmentioning

confidence: 99%

Section: General Observations During Model Developmentmentioning

confidence: 99%

See 3 more Smart Citations

Indoor environment data time-series reconstruction using autoencoder neural networks

Liguori,

Markovic,

Dam

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Validation of virtual sensor-assisted Bayesian inference-based in-situ sensor calibration strategy for building HVAC systems

Xiong

Sun

et al. 2022

Build. Simul.

View full text Add to dashboard Cite

A Novel Approach for Forecasting and Scheduling Building Load through Real-Time Occupant Count Data

Rafiq,

Mahmood,

Ahmed

et al. 2024

Arab J Sci Eng

View full text Add to dashboard Cite

The smart buildings’ load forecasting is necessary for efficient energy management, and it is easily possible because of the data availability based on widespread use of Internet of Things (IoT) devices and automation systems. The information of buildings’ occupancy is directly associated with energy consumption. Therefore, we present a hybrid model consisting of a Long Short-Term Memory (LSTM) network, Extreme Gradient Boosting (XgBoost), Random Forest (RF) and Linear Regression (LR) for commercial and academic buildings’ load forecasting. The correlation between occupants’ count and total load of the building is calculated using Pearson Correlation Coefficient (PCC). The comparative analysis of the proposed approach with LSTM, XgBoost, RF and Gated Recurrent Unit (GRU) is also performed. Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Square Error (MSE) and Normalized Root Mean Square Error (NRMSE) are used as performance indicators for evaluating performance. Findings indicate that the proposed hybrid approach outperforms other models. The RMSE and MAE of 2.99 and 2.18, respectively, are recorded by the proposed model for commercial building dataset while for academic building the RMSE and MAE are 4.48 and 2.85, respectively. Occupancy and load consumption have a positive correlation as evident from PCC analysis. Therefore, we have scheduled the forecasted load based on occupancy patterns for two different cases. Cost is reduced by 17.42% and 33.40% in case 1 and case 2, respectively. Moreover, the performance of the proposed hybrid approach is compared with different techniques presented in literature for buildings load forecasting.

show abstract

Day-ahead prediction of plug-in loads using a long short-term memory neural network

Cited by 19 publications

References 46 publications

Indoor environment data time-series reconstruction using autoencoder neural networks

Indoor environment data time-series reconstruction using autoencoder neural networks

Validation of virtual sensor-assisted Bayesian inference-based in-situ sensor calibration strategy for building HVAC systems

A Novel Approach for Forecasting and Scheduling Building Load through Real-Time Occupant Count Data

Contact Info

Product

Resources

About