A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Short-term building energy consumption prediction strategy based on modal decomposition and reconstruction algorithm
https://orcid.org/0000-0002-5777-0423
https://orcid.org/0000-0002-1215-9807
https://orcid.org/0000-0002-1415-855X
Highlights A density clustering-based method is proposed to optimize time-series datasets. Time-series decomposition method is applied to reduce the nonlinearity and non-stationary of the original data. A more engineering energy consumption prediction modeling strategy is provided. A neural network model optimized by a self-attention mechanism is presented for modeling. A hybrid deep-machine learning model is used for prediction to reduce the prediction error.
Abstract Building energy consumption prediction plays a vital role in building energy systems. However, the complexity of building energy use behavior and frequent fluctuations in demand pose significant challenges for accurate energy consumption prediction. Therefore, based on a time-series decomposition method, a hybrid energy consumption prediction model of Random Forest (RF) and combined deep learning method is proposed for accurate energy consumption prediction modeling. In the first stage of our method, Density Based Spatial Clustering of Applications with Noise (DBSCAN) and Lagrange interpolation method are respectively adopted to detect and process the abnormal energy consumption data, to reduce the impact of outliers on modeling. Then, the processed historical input series data are decomposed into Intrinsic Mode Functions (IMFs) using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Analysis (CEEMDAN) algorithm, and the Fuzzy Entropy (FuzzyEn) of each IMF component is calculated. The components are divided into high-frequency components and low-frequency components using the component partitioning method. In the last stage, the high-frequency components are predicted using RF, and the low-frequency components are predicted using the proposed hybrid deep learning model which combines a Convolutional Neural Network (CNN) layer and a Gated Recurrent Unit (GRU) layer optimized via a self-attention mechanism. Subsequently, the prediction results are superimposed and reconstructed to derive the ultimate prediction results. Additionally, the proposed method has been tested using real-world building energy consumption data from five public datasets of Building Data, and the experimental results demonstrated that the proposed method outperforms the state-of-the-art algorithms and could effectively control the prediction error within a small interval. Therefore, it is feasible to apply the hybrid model to building energy consumption prediction.
Short-term building energy consumption prediction strategy based on modal decomposition and reconstruction algorithm
https://orcid.org/0000-0002-5777-0423
https://orcid.org/0000-0002-1215-9807
https://orcid.org/0000-0002-1415-855X
Highlights A density clustering-based method is proposed to optimize time-series datasets. Time-series decomposition method is applied to reduce the nonlinearity and non-stationary of the original data. A more engineering energy consumption prediction modeling strategy is provided. A neural network model optimized by a self-attention mechanism is presented for modeling. A hybrid deep-machine learning model is used for prediction to reduce the prediction error.
Abstract Building energy consumption prediction plays a vital role in building energy systems. However, the complexity of building energy use behavior and frequent fluctuations in demand pose significant challenges for accurate energy consumption prediction. Therefore, based on a time-series decomposition method, a hybrid energy consumption prediction model of Random Forest (RF) and combined deep learning method is proposed for accurate energy consumption prediction modeling. In the first stage of our method, Density Based Spatial Clustering of Applications with Noise (DBSCAN) and Lagrange interpolation method are respectively adopted to detect and process the abnormal energy consumption data, to reduce the impact of outliers on modeling. Then, the processed historical input series data are decomposed into Intrinsic Mode Functions (IMFs) using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Analysis (CEEMDAN) algorithm, and the Fuzzy Entropy (FuzzyEn) of each IMF component is calculated. The components are divided into high-frequency components and low-frequency components using the component partitioning method. In the last stage, the high-frequency components are predicted using RF, and the low-frequency components are predicted using the proposed hybrid deep learning model which combines a Convolutional Neural Network (CNN) layer and a Gated Recurrent Unit (GRU) layer optimized via a self-attention mechanism. Subsequently, the prediction results are superimposed and reconstructed to derive the ultimate prediction results. Additionally, the proposed method has been tested using real-world building energy consumption data from five public datasets of Building Data, and the experimental results demonstrated that the proposed method outperforms the state-of-the-art algorithms and could effectively control the prediction error within a small interval. Therefore, it is feasible to apply the hybrid model to building energy consumption prediction.
Short-term building energy consumption prediction strategy based on modal decomposition and reconstruction algorithm
https://orcid.org/0000-0002-5777-0423
https://orcid.org/0000-0002-1215-9807
https://orcid.org/0000-0002-1415-855X
Highlights A density clustering-based method is proposed to optimize time-series datasets. Time-series decomposition method is applied to reduce the nonlinearity and non-stationary of the original data. A more engineering energy consumption prediction modeling strategy is provided. A neural network model optimized by a self-attention mechanism is presented for modeling. A hybrid deep-machine learning model is used for prediction to reduce the prediction error.
Abstract Building energy consumption prediction plays a vital role in building energy systems. However, the complexity of building energy use behavior and frequent fluctuations in demand pose significant challenges for accurate energy consumption prediction. Therefore, based on a time-series decomposition method, a hybrid energy consumption prediction model of Random Forest (RF) and combined deep learning method is proposed for accurate energy consumption prediction modeling. In the first stage of our method, Density Based Spatial Clustering of Applications with Noise (DBSCAN) and Lagrange interpolation method are respectively adopted to detect and process the abnormal energy consumption data, to reduce the impact of outliers on modeling. Then, the processed historical input series data are decomposed into Intrinsic Mode Functions (IMFs) using the Complete Ensemble Empirical Mode Decomposition with Adaptive Noise Analysis (CEEMDAN) algorithm, and the Fuzzy Entropy (FuzzyEn) of each IMF component is calculated. The components are divided into high-frequency components and low-frequency components using the component partitioning method. In the last stage, the high-frequency components are predicted using RF, and the low-frequency components are predicted using the proposed hybrid deep learning model which combines a Convolutional Neural Network (CNN) layer and a Gated Recurrent Unit (GRU) layer optimized via a self-attention mechanism. Subsequently, the prediction results are superimposed and reconstructed to derive the ultimate prediction results. Additionally, the proposed method has been tested using real-world building energy consumption data from five public datasets of Building Data, and the experimental results demonstrated that the proposed method outperforms the state-of-the-art algorithms and could effectively control the prediction error within a small interval. Therefore, it is feasible to apply the hybrid model to building energy consumption prediction.
Short-term building energy consumption prediction strategy based on modal decomposition and reconstruction algorithm
Jiao, Yinghao (author) / Tan, Zhi (author) / Zhang, De (author) / Zheng, Q.P. (author)
Energy and Buildings ; 290
2023-04-10
Article (Journal)
Electronic Resource
English
Short-Term Prediction of PV Power Based on Combined Modal Decomposition and NARX-LSTM-LightGBM
| DOAJ | 2023
Short-term office building elevator energy consumption forecast using SARIMA
| Taylor & Francis Verlag | 2020