Lei Zhang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Lei Zhang has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Difen Shi
School of Electronic Engineering, Nanjing Xiaozhuang University, Nanjing 211171, China
2 shared publications - Gang Lei
University of Technology Sydney
2 shared publications - Ming Yao
School of Automotive and Traffic Engineering, Jiangsu University, Zhenjiang, China
2 shared publications - Xiaodong Sun
School of Electrical and Information Engineering, Jiangsu University, Zhenjiang, China
2 shared publications - Dinghui Wang
School of Materials Science and Physics, China University of Mining and Technology, Xuzhou 221116, China
1 shared publication - Guo Xu
School of Electrical Engineering and Information, Northeast Petroleum University, Ranghu Road, Daqing 163318, China
1 shared publication
Latest Publications
- Open Access
Co-Optimization of Capacity and Operation for Battery-Hydrogen Hybrid Energy Storage Systems Based on Deep Reinforcement Learning and Mixed Integer Programming
The hybrid energy storage system (HESS) that combines battery with hydrogen storage exploits complementary power/energy characteristics, but most studies optimize capacity and operation separately, leading to suboptimal overall performance. To address this issue, this paper proposes a bi-level co-optimization framework that integrates deep reinforcement learning (DRL) and mixed integer programming (MIP). The outer layer employs the TD3 algorithm for capacity configuration, while the inner layer uses the Gurobi solver for optimal operation under constraints. On a standalone PV–wind–load-HESS system, the method attains near-optimal quality at dramatically lower runtime. Relative to GA + Gurobi and PSO + Gurobi, the cost is lower by 4.67% and 1.31%, while requiring only 0.52% and 0.58% of their runtime; compared with a direct Gurobi solve, the cost remains comparable while runtime decreases to 0.07%. Sensitivity analysis further validates the model’s robustness under various cost parameters and renewable energy penetration levels. These results indicate that the proposed DRL–MIP cooperation achieves near-optimal solutions with orders of magnitude speedups. This study provides a new DRL–MIP paradigm for efficiently solving strongly coupled bi-level optimization problems in energy systems.
Open AccessCo-Optimization of Capacity and Operation for Battery-Hydrogen Hybrid Energy Storage Systems Based on Deep Reinforcement Learning and Mixed Integer Programming
The hybrid energy storage system (HESS) that combines battery with hydrogen storage exploits complementary power/energy characteristics, but most studies optimize capacity and operation separately, leading to suboptimal overall performance. To address this issue, this paper proposes a bi-level co-optimization framework that integrates deep reinforcement learning (DRL) and mixed integer programming (MIP). The outer layer employs the TD3 algorithm for capacity configuration, while the inner layer uses the Gurobi solver for optimal operation under constraints. On a standalone PV–wind–load-HESS system, the method attains near-optimal quality at dramatically lower runtime. Relative to GA + Gurobi and PSO + Gurobi, the cost is lower by 4.67% and 1.31%, while requiring only 0.52% and 0.58% of their runtime; compared with a direct Gurobi solve, the cost remains comparable while runtime decreases to 0.07%. Sensitivity analysis further validates the model’s robustness under various cost parameters and renewable energy penetration levels. These results indicate that the proposed DRL–MIP cooperation achieves near-optimal solutions with orders of magnitude speedups. This study provides a new DRL–MIP paradigm for efficiently solving strongly coupled bi-level optimization problems in energy systems.
- Open Access
Electricity Data Quality Enhancement Strategy Based on Low-Rank Matrix Recovery
Electricity consumption data form the foundation for the efficient and reliable operation of smart grids and are a critical component for ensuring effective data mining. However, due to factors such as meter failures and extreme weather conditions, anomalies frequently occur in the data, which adversely impact the performance of data-driven applications. Given the near full-rank nature of low-voltage distribution area electricity consumption data, this paper employs clustering to enhance the low-rank property of the data. Addressing common issues such as missing data, sparse noise, and Gaussian noise in electricity consumption data, this paper proposes a multi-norm optimization model based on low-rank matrix theory. Specifically, the truncated nuclear norm is used as an approximation of matrix rank, while the -norm and F-norm are employed to constrain sparse noise and Gaussian noise, respectively. The model is solved using the Alternating Direction Method of Multipliers (ADMM), achieving a unified framework for handling missing data and noise processing within the model construction. Comparative experiments on both synthetic and real-world datasets demonstrate that the proposed method can accurately recover measurement data under various noise contamination scenarios and different distributions of missing data. Moreover, it effectively separates principal components of the data from noise contamination.Open AccessElectricity Data Quality Enhancement Strategy Based on Low-Rank Matrix Recovery
Electricity consumption data form the foundation for the efficient and reliable operation of smart grids and are a critical component for ensuring effective data mining. However, due to factors such as meter failures and extreme weather conditions, anomalies frequently occur in the data, which adversely impact the performance of data-driven applications. Given the near full-rank nature of low-voltage distribution area electricity consumption data, this paper employs clustering to enhance the low-rank property of the data. Addressing common issues such as missing data, sparse noise, and Gaussian noise in electricity consumption data, this paper proposes a multi-norm optimization model based on low-rank matrix theory. Specifically, the truncated nuclear norm is used as an approximation of matrix rank, while the -norm and F-norm are employed to constrain sparse noise and Gaussian noise, respectively. The model is solved using the Alternating Direction Method of Multipliers (ADMM), achieving a unified framework for handling missing data and noise processing within the model construction. Comparative experiments on both synthetic and real-world datasets demonstrate that the proposed method can accurately recover measurement data under various noise contamination scenarios and different distributions of missing data. Moreover, it effectively separates principal components of the data from noise contamination. - Open Access
- 6 February 2025
- Published by MDPI AG in Nanomaterials
- Vol. 15 (3) , 249
Deriving High-Energy-Density Polymeric Nitrogen N10 from the Host–Guest ArN10 Compound
Discovering stable polymeric nitrogen phases and exploring their properties are crucial for energy storage and conversion, garnering significant attention. In this study, we investigate the formation possibility of a stable compound between Ar and N2 through ab initio calculations under low-pressure conditions (0–100 GPa). The novel super nitride, Imm2 ArN10, is designed to demonstrate robust thermodynamic stability under high pressures (91 GPa) and showcase the unique host–guest structure, in which guest atoms (Ar) are trapped inside the host polymeric N10. Significantly, given the weak interaction between Ar and N atoms and a channel parallel to the c-crystallographic axis in ArN10, we propose a novel method to stabilize the previously unknown polymeric nitrogen structure, Imm2-N10, by removing the guest argon atoms from the natural channels of ArN10. Imm2 ArN10 and N10 are thermodynamically and dynamically stable, with energy densities of 9.1 kJ g−1 and 12.3 kJ g−1, respectively—more than twice that of TNT. Additionally, ArN10 and N10 stand out as leading green energetic materials, boasting a superior explosion velocity of 17.56 km s−1 and a detonation pressure of 1712 kbar, surpassing that of TNT. These findings significantly impact on the creation of pure nitrogen frameworks through chemical reactions involving inert elements under high pressure.
- 6 February 2025
- Published by MDPI AG in Nanomaterials
- Vol. 15 (3) , 249
Open AccessDeriving High-Energy-Density Polymeric Nitrogen N10 from the Host–Guest ArN10 Compound
Discovering stable polymeric nitrogen phases and exploring their properties are crucial for energy storage and conversion, garnering significant attention. In this study, we investigate the formation possibility of a stable compound between Ar and N2 through ab initio calculations under low-pressure conditions (0–100 GPa). The novel super nitride, Imm2 ArN10, is designed to demonstrate robust thermodynamic stability under high pressures (91 GPa) and showcase the unique host–guest structure, in which guest atoms (Ar) are trapped inside the host polymeric N10. Significantly, given the weak interaction between Ar and N atoms and a channel parallel to the c-crystallographic axis in ArN10, we propose a novel method to stabilize the previously unknown polymeric nitrogen structure, Imm2-N10, by removing the guest argon atoms from the natural channels of ArN10. Imm2 ArN10 and N10 are thermodynamically and dynamically stable, with energy densities of 9.1 kJ g−1 and 12.3 kJ g−1, respectively—more than twice that of TNT. Additionally, ArN10 and N10 stand out as leading green energetic materials, boasting a superior explosion velocity of 17.56 km s−1 and a detonation pressure of 1712 kbar, surpassing that of TNT. These findings significantly impact on the creation of pure nitrogen frameworks through chemical reactions involving inert elements under high pressure.
- 19 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (7) , 6831-6839
FCS-MPC With Improved Prediction for Suppressing Torque and Current Pulsations of Switched Reluctance Motors
This article proposes an improved finite control set model predictive control strategy to suppress the dual pulsations of torque and current of three-phase switched reluctance motors. It enhances delay compensation to strengthen predictive capabilities and predicts the motor torque and current under all possible switching states to construct a cost function. The working condition that minimizes the cost function is considered the optimal drive signal and is applied to control the power converter, thereby simultaneously reducing torque and current ripples. The delay compensation is enhanced by including a Kalman filter stage. The algorithm is validated through experiments using a three-phase SRM. The results demonstrate that the proposed method significantly suppresses both torque and current dual pulsations, verifying its effectiveness.
- 19 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 72 (7) , 6831-6839
FCS-MPC With Improved Prediction for Suppressing Torque and Current Pulsations of Switched Reluctance Motors
This article proposes an improved finite control set model predictive control strategy to suppress the dual pulsations of torque and current of three-phase switched reluctance motors. It enhances delay compensation to strengthen predictive capabilities and predicts the motor torque and current under all possible switching states to construct a cost function. The working condition that minimizes the cost function is considered the optimal drive signal and is applied to control the power converter, thereby simultaneously reducing torque and current ripples. The delay compensation is enhanced by including a Kalman filter stage. The algorithm is validated through experiments using a three-phase SRM. The results demonstrate that the proposed method significantly suppresses both torque and current dual pulsations, verifying its effectiveness.
- Open Access
- 17 October 2024
- Published by Springer Nature in Research Square
A Load Forecasting Model for Smart Grid Based on RF-BiLSTM and Federated Learning
Accurate load forecasting is essential for improving the efficiency and reliability of smart grids. Traditional forecasting models face challenges in integrating high-dimensional data and protecting privacy. To address these issues, a novel load forecasting model is presented in this paper. The proposed approach combines a random forest (RF) and a bi-directional long short-term memory (BiLSTM) network based on a federated learning framework. To effectively process high-dimensional heterogeneous data in smart grids, the proposed model exploits the feature extraction capability of radio frequency. In addition, the sequential data processing capability of BiLSTM provides a means of processing time-varying load data in smart grids. The federated learning method protects user privacy and achieves the goal of global model training adopting distributed data. Experimental results demonstrate that the proposed model is superior to traditional models in accuracy and efficiency.
- 17 October 2024
- Published by Springer Nature in Research Square
Open AccessA Load Forecasting Model for Smart Grid Based on RF-BiLSTM and Federated Learning
Accurate load forecasting is essential for improving the efficiency and reliability of smart grids. Traditional forecasting models face challenges in integrating high-dimensional data and protecting privacy. To address these issues, a novel load forecasting model is presented in this paper. The proposed approach combines a random forest (RF) and a bi-directional long short-term memory (BiLSTM) network based on a federated learning framework. To effectively process high-dimensional heterogeneous data in smart grids, the proposed model exploits the feature extraction capability of radio frequency. In addition, the sequential data processing capability of BiLSTM provides a means of processing time-varying load data in smart grids. The federated learning method protects user privacy and achieves the goal of global model training adopting distributed data. Experimental results demonstrate that the proposed model is superior to traditional models in accuracy and efficiency.
- 4 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 30 (3) , 2367-2378
Augmented Continuous-Control-Set Model Predictive Current Control for Dual Three-Phase PMSM Drives
This article introduces an augmented continuous-control-set model predictive current control (ACCS-MPCC) for dual three-phase permanent magnet synchronous machines (DTP-PMSMs) to achieve comprehensive 4-D predictive current control. First, an optimization problem is formulated based on a cost function incorporating parameter disturbances and digital delays as disturbance terms to enhance the drive system's robustness. The complete voltage plane, regular switching characteristics, and improved cycle-by-cycle tracking control contribute to superior steady-state and transient performance, alongside excellent harmonic suppression capabilities within the x–y plane. Second, a moving horizon estimator (MHE) is employed to estimate the disturbance term, with its stability being mathematically substantiated. The utilization of MHE resolves the dependence of control performance on parameter accuracy, as commonly encountered in model predictive control, and effectively addresses the issue of digital delay. Finally, this article thoroughly investigates the parameter sensitivity of the continuous-control-set model predictive current control via experimental analysis of a DTP-PMSM system. The effectiveness of the proposed method is affirmed through comparative analysis, which demonstrates that the proposed ACCS-MPCC significantly mitigates the nonoptimal susceptibility issues inherent in multiphase motor systems.
- 4 September 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE/ASME Transactions on Mechatronics
- Vol. 30 (3) , 2367-2378
Augmented Continuous-Control-Set Model Predictive Current Control for Dual Three-Phase PMSM Drives
This article introduces an augmented continuous-control-set model predictive current control (ACCS-MPCC) for dual three-phase permanent magnet synchronous machines (DTP-PMSMs) to achieve comprehensive 4-D predictive current control. First, an optimization problem is formulated based on a cost function incorporating parameter disturbances and digital delays as disturbance terms to enhance the drive system's robustness. The complete voltage plane, regular switching characteristics, and improved cycle-by-cycle tracking control contribute to superior steady-state and transient performance, alongside excellent harmonic suppression capabilities within the x–y plane. Second, a moving horizon estimator (MHE) is employed to estimate the disturbance term, with its stability being mathematically substantiated. The utilization of MHE resolves the dependence of control performance on parameter accuracy, as commonly encountered in model predictive control, and effectively addresses the issue of digital delay. Finally, this article thoroughly investigates the parameter sensitivity of the continuous-control-set model predictive current control via experimental analysis of a DTP-PMSM system. The effectiveness of the proposed method is affirmed through comparative analysis, which demonstrates that the proposed ACCS-MPCC significantly mitigates the nonoptimal susceptibility issues inherent in multiphase motor systems.