Feihu Hu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Feihu Hu has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Xinhao Liang
School of Electrical Engineering, Xi'an Jiaotong University, Xi'an, China
3 shared publications - Hui Cao
State Key Laboratory of Electrical Insulation and Power Equipment, School of Electrical Engineering, Xian Jiaotong University, Xian, China
2 shared publications - Huifeng Xu
School of Electrical Engineering, Xi'an Jiaotong University 2 , Xi'an, 710049 Shaanxi
2 shared publications - Mohammad Abu Gunmi
School of Electrical Engineering, Xian Jiao tong University, Xian, China
2 shared publications - Xuan Feng
School of Electrical Engineering, Xi'an Jiaotong University, Xi'an, China
2 shared publications - Diana Abu-Ghunmi
Department of Finance, The University of Jordan, Amman 11942, Jordan
1 shared publication
Latest Publications
- 1 July 2026
- Published by AIP Publishing in Journal of Renewable and Sustainable Energy
- Vol. 18 (4)
A model-based feature selection and load forecasting framework
Accurate power load forecasting plays a crucial role in grid scheduling and stable operation. As power systems develop, load data has become more extensive and detailed. While this provides extensive data support for deep learning models to predict power loads, the challenge lies in identifying the optimal input features within these data. To tackle this problem, this paper presents a model-based feature evaluation and selection method designed to evaluate the importance of input features for deep learning models in predicting power loads, thereby enabling the selection of the best input features for prediction models. Experimental results from the Nordic electricity market and the New England region of the United States demonstrate the superiority of this approach over methods like the Pearson correlation coefficient, the Spearman correlation coefficient, the maximum information coefficient, and gray correlation analysis. It facilitates the selection of necessary input features for deep learning models, thereby enhancing the accuracy of power load forecasting. The proposed method proves beneficial for grid scheduling and provides more reliable load predictions for related power enterprises.
- 1 July 2026
- Published by AIP Publishing in Journal of Renewable and Sustainable Energy
- Vol. 18 (4)
A model-based feature selection and load forecasting framework
Accurate power load forecasting plays a crucial role in grid scheduling and stable operation. As power systems develop, load data has become more extensive and detailed. While this provides extensive data support for deep learning models to predict power loads, the challenge lies in identifying the optimal input features within these data. To tackle this problem, this paper presents a model-based feature evaluation and selection method designed to evaluate the importance of input features for deep learning models in predicting power loads, thereby enabling the selection of the best input features for prediction models. Experimental results from the Nordic electricity market and the New England region of the United States demonstrate the superiority of this approach over methods like the Pearson correlation coefficient, the Spearman correlation coefficient, the maximum information coefficient, and gray correlation analysis. It facilitates the selection of necessary input features for deep learning models, thereby enhancing the accuracy of power load forecasting. The proposed method proves beneficial for grid scheduling and provides more reliable load predictions for related power enterprises.
- 26 September 2025
- Published by Springer Nature in Annals of Operations Research
- p. 1-45
A stable service composition and optimal selection considering uncertain quality in cloud remanufacturing
- 26 September 2025
- Published by Springer Nature in Annals of Operations Research
- p. 1-45
A stable service composition and optimal selection considering uncertain quality in cloud remanufacturing
- 3 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Sustainable Energy
- Vol. 16 (3) , 1816-1826
Ultra-Short-Term Spatio-Temporal Wind Speed Prediction Based on OWT-STGradRAM
Taking into account the orientation and distance characteristics of wind turbine stations in wind farms can improve the accuracy of wind power prediction. This paper proposed a deep learning spatio-temporal prediction method named orthogonal wind direction transformation spatio-temporal gradient Regression Activation Mapping (OWT-STGrad-RAM) for wind speed prediction. The model encodes the wind farm using an image, and each wind turbine is encoded as a point in the image. The spatio-temporal data related to wind turbines, such as wind speed, temperature, and air pressure, are integrated into fusion features through spatio-temporal fusion convolutional networks model for pre training to obtain a feature dataset. OWT is used to eliminate the effects of different prevailing winds, and STGrad-RAM is used to characterize the orientation and distance between wind turbine nodes and make the spatial features interpretable. The feature dataset is used for wind speed prediction. The experimental results show that the proposed method has achieved a significant improvement in wind speed prediction accuracy compared to the comparative models.
- 3 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Sustainable Energy
- Vol. 16 (3) , 1816-1826
Ultra-Short-Term Spatio-Temporal Wind Speed Prediction Based on OWT-STGradRAM
Taking into account the orientation and distance characteristics of wind turbine stations in wind farms can improve the accuracy of wind power prediction. This paper proposed a deep learning spatio-temporal prediction method named orthogonal wind direction transformation spatio-temporal gradient Regression Activation Mapping (OWT-STGrad-RAM) for wind speed prediction. The model encodes the wind farm using an image, and each wind turbine is encoded as a point in the image. The spatio-temporal data related to wind turbines, such as wind speed, temperature, and air pressure, are integrated into fusion features through spatio-temporal fusion convolutional networks model for pre training to obtain a feature dataset. OWT is used to eliminate the effects of different prevailing winds, and STGrad-RAM is used to characterize the orientation and distance between wind turbine nodes and make the spatial features interpretable. The feature dataset is used for wind speed prediction. The experimental results show that the proposed method has achieved a significant improvement in wind speed prediction accuracy compared to the comparative models.
- 24 September 2024
- Published by Springer Nature
- p. 306-316
Predictive Modeling of Transmission Line Parameters for Different Operating Conditions
- 24 September 2024
- Published by Springer Nature
- p. 306-316
Predictive Modeling of Transmission Line Parameters for Different Operating Conditions
- 15 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Systems
- Vol. 39 (5) , 6277-6289
Attention Mechanism Multi-Size Depthwise Convolutional Long Short-Term Memory Neural Network for Forecasting Real-Time Electricity Prices
Real-time electricity price forecasting affects both the interests of power companies and the stability of power systems. Although deep learning models have achieved rich results in forecasting, due to the variable temporal characteristics and numerous influencing factors of real-time electricity prices, it is difficult for general deep learning models to extract electricity price features with obvious regularity, which affects forecasting accuracy. To solve this problem, this paper proposes an attention mechanism multi-size depthwise convolutional long short-term memory neural network (AM-MDC-LSTM) for predicting real-time electricity prices. The model improves prediction capability in the following aspects. 1) Using an attention mechanism to adaptively assign weights to electricity price time series and electricity price exogenous variables (production, consumption, electricity prices in neighboring regions) to improve electricity price feature extraction efficiency. 2) Using convolution kernels of different sizes to convolve individual electricity price exogenous variables one by one to extract local burst and global periodic electricity price features with obvious regularity. This is then combined with long short-term memory networks to extract temporal features reflected in electricity prices. Experimental results conducted in the Nordic and PJM electricity markets demonstrate that the proposed model outperforms other models discussed in the paper, exhibiting higher prediction accuracy.
- 15 January 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Systems
- Vol. 39 (5) , 6277-6289
Attention Mechanism Multi-Size Depthwise Convolutional Long Short-Term Memory Neural Network for Forecasting Real-Time Electricity Prices
Real-time electricity price forecasting affects both the interests of power companies and the stability of power systems. Although deep learning models have achieved rich results in forecasting, due to the variable temporal characteristics and numerous influencing factors of real-time electricity prices, it is difficult for general deep learning models to extract electricity price features with obvious regularity, which affects forecasting accuracy. To solve this problem, this paper proposes an attention mechanism multi-size depthwise convolutional long short-term memory neural network (AM-MDC-LSTM) for predicting real-time electricity prices. The model improves prediction capability in the following aspects. 1) Using an attention mechanism to adaptively assign weights to electricity price time series and electricity price exogenous variables (production, consumption, electricity prices in neighboring regions) to improve electricity price feature extraction efficiency. 2) Using convolution kernels of different sizes to convolve individual electricity price exogenous variables one by one to extract local burst and global periodic electricity price features with obvious regularity. This is then combined with long short-term memory networks to extract temporal features reflected in electricity prices. Experimental results conducted in the Nordic and PJM electricity markets demonstrate that the proposed model outperforms other models discussed in the paper, exhibiting higher prediction accuracy.
- Open Access
- 28 March 2023
- Published by MDPI AG in Sustainability
- Vol. 15 (7) , 5871
Spatio-Temporal Wind Speed Prediction Based on Improved Residual Shrinkage Network
Considering the massive influx of new energy into the power system, accurate wind speed prediction is of great importance to its stability. Due to the influence of sensor accuracy and harsh natural environments, there is inevitable noise interference in original wind speed data, which adversely affects the accuracy of wind speed prediction. There are some problems associated with traditional signal processing methods when dealing with noise such as signal loss. We propose the use of a deep residual shrinkage unit based on soft activation (SDRSU) in order to reduce noise interference and ensure the integrity of original wind speed data. A deep network is constructed by stacking multiple SDRSUs to extract useful features from noisy data. Considering the spatio-temporal coupling relationship between wind turbines in a wind farm, a ST-SDRSN (soft-activation based deep spatio-temporal residual shrinkage network) will be used to model the wind speed series neighboring time property and daily periodic property. An accurate wind speed prediction can be achieved by extracting the spatial correlations between the turbines at each turbine along the time axis. We designed four depth models under the same spatio-temporal architecture to verify the advantages of the soft-activation block and the proposed ST-SDRSN model. Two datasets provided by the National Renewable Energy Laboratory (NREL) were used for our experiments. Based on different kinds of evaluation criteria in different datasets, ST-SDRSN was shown to improve prediction accuracy by 15.87%.
- 28 March 2023
- Published by MDPI AG in Sustainability
- Vol. 15 (7) , 5871
Open AccessSpatio-Temporal Wind Speed Prediction Based on Improved Residual Shrinkage Network
Considering the massive influx of new energy into the power system, accurate wind speed prediction is of great importance to its stability. Due to the influence of sensor accuracy and harsh natural environments, there is inevitable noise interference in original wind speed data, which adversely affects the accuracy of wind speed prediction. There are some problems associated with traditional signal processing methods when dealing with noise such as signal loss. We propose the use of a deep residual shrinkage unit based on soft activation (SDRSU) in order to reduce noise interference and ensure the integrity of original wind speed data. A deep network is constructed by stacking multiple SDRSUs to extract useful features from noisy data. Considering the spatio-temporal coupling relationship between wind turbines in a wind farm, a ST-SDRSN (soft-activation based deep spatio-temporal residual shrinkage network) will be used to model the wind speed series neighboring time property and daily periodic property. An accurate wind speed prediction can be achieved by extracting the spatial correlations between the turbines at each turbine along the time axis. We designed four depth models under the same spatio-temporal architecture to verify the advantages of the soft-activation block and the proposed ST-SDRSN model. Two datasets provided by the National Renewable Energy Laboratory (NREL) were used for our experiments. Based on different kinds of evaluation criteria in different datasets, ST-SDRSN was shown to improve prediction accuracy by 15.87%.
- Open Access
- 1 January 2019
- Published by Institute of Electronics and Computer in Electrical Engineering and Computer Science (EECS)
- Vol. 3, 87-91
Component, formatting, style, styling, insert
The precision air conditioner accounts for 10% of energy consumption in data center. It is important to study how to adjust precision air conditioner to reduce energy consumption. Computational Fluid Dynamics (CFD) model is the most common method to simulate the temperature field in computer room, but the simulation time of existing software is large and the efficiency is low. In this paper, the proper orthogonal decomposition (POD) method is adopted. By taking the temperature field of the computer room under discrete configuration parameters as input, the proper orthogonal decomposition of each temperature field is performed to obtain the proper orthogonal basis of the temperature field, and then the approximate value of the temperature field of the computer room is obtained. Simulink is used to build air conditioning model and computer room model, and the temperature field under random load level is simulated. The results show that POD method has a great advantage in computing speed under the premise of little difference in accuracy.
- 1 January 2019
- Published by Institute of Electronics and Computer in Electrical Engineering and Computer Science (EECS)
- Vol. 3, 87-91
Open AccessComponent, formatting, style, styling, insert
The precision air conditioner accounts for 10% of energy consumption in data center. It is important to study how to adjust precision air conditioner to reduce energy consumption. Computational Fluid Dynamics (CFD) model is the most common method to simulate the temperature field in computer room, but the simulation time of existing software is large and the efficiency is low. In this paper, the proper orthogonal decomposition (POD) method is adopted. By taking the temperature field of the computer room under discrete configuration parameters as input, the proper orthogonal decomposition of each temperature field is performed to obtain the proper orthogonal basis of the temperature field, and then the approximate value of the temperature field of the computer room is obtained. Simulink is used to build air conditioning model and computer room model, and the temperature field under random load level is simulated. The results show that POD method has a great advantage in computing speed under the premise of little difference in accuracy.
- Open Access
A Short-Term Decision Model for Electricity Retailers: Electricity Procurement and Time-of-Use Pricing
This paper establishes a short-term decision model, based on robust optimization, for an electricity retailer to determine the electricity procurement and electricity retail prices. The electricity procurement process includes purchasing electricity from generation companies and from the spot market. The selling prices of electricity for the customers are based on time-of-use (TOU) pricing which is widely employed in modern electricity market as a demand response program. The objective of the model is to maximize the expected profit of the retailer through optimizing the electricity procurement strategy and electricity pricing scheme. A price elasticity matrix (PEM) is adopted to model the demand response. Also, uncertainty in spot prices is modeled using a robust optimization approach, in which price bounds are considered instead of predicted values. Using a robust optimization approach, the retailer can adjust the level of robustness of its decisions through a robust control parameter. A case study is presented to illustrate the performance of the model. The simulation results demonstrate that the developed model is effective in increasing the expected profit of the retailer and flattening the load profiles of customers.
Open AccessA Short-Term Decision Model for Electricity Retailers: Electricity Procurement and Time-of-Use Pricing
This paper establishes a short-term decision model, based on robust optimization, for an electricity retailer to determine the electricity procurement and electricity retail prices. The electricity procurement process includes purchasing electricity from generation companies and from the spot market. The selling prices of electricity for the customers are based on time-of-use (TOU) pricing which is widely employed in modern electricity market as a demand response program. The objective of the model is to maximize the expected profit of the retailer through optimizing the electricity procurement strategy and electricity pricing scheme. A price elasticity matrix (PEM) is adopted to model the demand response. Also, uncertainty in spot prices is modeled using a robust optimization approach, in which price bounds are considered instead of predicted values. Using a robust optimization approach, the retailer can adjust the level of robustness of its decisions through a robust control parameter. A case study is presented to illustrate the performance of the model. The simulation results demonstrate that the developed model is effective in increasing the expected profit of the retailer and flattening the load profiles of customers.
- 1 December 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 26-30
A Network Model Research for Emergency Supply Database
For the question that how to show the emergency supply in the software, a mathematical model is developed in this paper. The mathematical model distinguishes connecting sides, nodes from different networks, which includes modeling events as material information, information of transporting vehicles, scheduling schemes, and disaster points. Then according to requirements of the modeling elements and relationships between different modeling elements, the constraints are established. At the end, we developed software which can show the mathematical model is reasonable.
- 1 December 2015
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 26-30
A Network Model Research for Emergency Supply Database
For the question that how to show the emergency supply in the software, a mathematical model is developed in this paper. The mathematical model distinguishes connecting sides, nodes from different networks, which includes modeling events as material information, information of transporting vehicles, scheduling schemes, and disaster points. Then according to requirements of the modeling elements and relationships between different modeling elements, the constraints are established. At the end, we developed software which can show the mathematical model is reasonable.