Zhongda Lu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Zhongda Lu has a total of 41 co-authors. Here are the top co-authors by number of shared publications:
- Fengxia Xu
School of Mechanical and Electrical Engineering, Qiqihar University, Qiqihar, China
7 shared publications - Chuanshui Peng
Qiqihar Sida Railway Equipment Co., Ltd., Qiqihar 161006, China
3 shared publications - Shipeng Tian
School of Mechanical and Electrical Engineering, Qiqihar University, Qiqihar, China
3 shared publications - Shiwei Yan
Qiqihar Sida Railway Equipment Co., Ltd., Qiqihar 161006, China
3 shared publications Fengxia Xu
College of Mechatronic Engineering
2 shared publications- Guoliang Zhang
College of Mechatronic Engineering
2 shared publications
Latest Publications
- Open Access
- 23 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 2713-2735
Spatial–Frequency Feature Coupling Network for Semantic Segmentation of Remote Sensing Images
Semantic segmentation of high-resolution remote sensing images is a challenging task due to the diverse scale variations and significant intra-class variability of ground objects. Existing methods fail to establish an effective deep coupling mechanism between spatial- and frequency-domain features, resulting in limited capability for modeling fine-grained textures and periodic structures. To overcome these limitations, this paper proposes a novel spatial–frequency feature coupling network (SFFCNet), which dynamically decouples spatial and frequency features and establishes an cross-domain fusion mechanism to enhance both fine-grained texture and global context modeling. Specifically, the dual-domain feature coupling (DDFC) module adopts a dual-branch design, decoupling frequency- and spatial-domain features through spectral energy separation and gradient enhancement strategies, and then coupling them via a cross-domain feature fusion module to support fine-grained representation of ground objects. The cognitive state space (CSS) module captures sequential dependencies among targets through four designed scanning paths, effectively mitigating the impact of intra-class variability. The global–local feature integration (GLFI) module exploits the positional invariance and semantic similarity of extracted features to model relative spatial relationships between objects, thereby enhancing the completeness and discriminative power of semantic representations. Extensive experiments on four datasets demonstrate that the proposed SFFCNet exhibits superior performance compared to existing state-of-the-art methods.
- 23 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 19 (19391404) , 2713-2735
Open AccessSpatial–Frequency Feature Coupling Network for Semantic Segmentation of Remote Sensing Images
Semantic segmentation of high-resolution remote sensing images is a challenging task due to the diverse scale variations and significant intra-class variability of ground objects. Existing methods fail to establish an effective deep coupling mechanism between spatial- and frequency-domain features, resulting in limited capability for modeling fine-grained textures and periodic structures. To overcome these limitations, this paper proposes a novel spatial–frequency feature coupling network (SFFCNet), which dynamically decouples spatial and frequency features and establishes an cross-domain fusion mechanism to enhance both fine-grained texture and global context modeling. Specifically, the dual-domain feature coupling (DDFC) module adopts a dual-branch design, decoupling frequency- and spatial-domain features through spectral energy separation and gradient enhancement strategies, and then coupling them via a cross-domain feature fusion module to support fine-grained representation of ground objects. The cognitive state space (CSS) module captures sequential dependencies among targets through four designed scanning paths, effectively mitigating the impact of intra-class variability. The global–local feature integration (GLFI) module exploits the positional invariance and semantic similarity of extracted features to model relative spatial relationships between objects, thereby enhancing the completeness and discriminative power of semantic representations. Extensive experiments on four datasets demonstrate that the proposed SFFCNet exhibits superior performance compared to existing state-of-the-art methods.
- Open Access
- 30 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 16966-16982
Difference Perception Fusion Network for Remote Sensing Image Change Detection
Deep learning-based methods have achieved promising results in the field of remote sensing (RS) image change detection (CD). However, they have deficiencies in generating and utilizing difference features, which leads to inaccurate detection of changed regions with fine structures. To address the above problems, this paper proposes a difference perception fusion network (DPFNet). Firstly, multilevel bitemporal features are extracted through the Siamese backbone network. To generate discriminative difference features utilizing the extracted multilevel bitemporal features, the difference perception module (DPM) is designed, which consists of an adaptive denoising module (ADM) and an edge enhancement module (EEM). Specifically, ADM relies on frequency filters with learnable parameters to suppress the noise produced during difference generation; EEM uses pooled subtraction to extract edge information for maintaining the fine contours of changed regions. The hybrid attention difference fusion (HADF) module is constructed to realize the interaction between multilevel difference features, enriching the edge details and internal integrity of changed regions. In addition, the refined representation module (RRM) is developed to realize the sophisticated representation of difference features and enhance the detection effect of changed regions. The effectiveness of the proposed DPFNet is verified by extensive experiments on three RSCD datasets (WHU-CD, CDD-CD, SYSU-CD), and DPFNet has fewer false detections and missed detections compared to other state-of-the-art methods. The code can be found at: https://github.com/tsp138/DPFNet.
- 30 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 16966-16982
Open AccessDifference Perception Fusion Network for Remote Sensing Image Change Detection
Deep learning-based methods have achieved promising results in the field of remote sensing (RS) image change detection (CD). However, they have deficiencies in generating and utilizing difference features, which leads to inaccurate detection of changed regions with fine structures. To address the above problems, this paper proposes a difference perception fusion network (DPFNet). Firstly, multilevel bitemporal features are extracted through the Siamese backbone network. To generate discriminative difference features utilizing the extracted multilevel bitemporal features, the difference perception module (DPM) is designed, which consists of an adaptive denoising module (ADM) and an edge enhancement module (EEM). Specifically, ADM relies on frequency filters with learnable parameters to suppress the noise produced during difference generation; EEM uses pooled subtraction to extract edge information for maintaining the fine contours of changed regions. The hybrid attention difference fusion (HADF) module is constructed to realize the interaction between multilevel difference features, enriching the edge details and internal integrity of changed regions. In addition, the refined representation module (RRM) is developed to realize the sophisticated representation of difference features and enhance the detection effect of changed regions. The effectiveness of the proposed DPFNet is verified by extensive experiments on three RSCD datasets (WHU-CD, CDD-CD, SYSU-CD), and DPFNet has fewer false detections and missed detections compared to other state-of-the-art methods. The code can be found at: https://github.com/tsp138/DPFNet.
- Open Access
- 13 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 5656-5672
Spatial–Spectral and Channel–Spectral Differences Integration Network for Remote Sensing Image Change Detection
Deep learning methods have shown promising performance in remote sensing (RS) image change detection (CD), largely due to their excellent ability to explore local and global contexts. However, existing methods fail to effectively integrate spectral information into difference features, which makes CD results more susceptible to pseudo-changes. To address the above challenges, this paper proposes a spatial-spectral and channelspectral differences integration network (SCDI-Net), which enriches the discriminative representation of change regions. First, the spatial-spectral and channel-spectral differences generation (SSCS-DG) module is designed to explore valuable difference clues from the spatial-channel-spectral perspective for mitigating the effect of pseudo-changes. The deformable feature pyramid network (DFPN) is constructed to aggregate multiscale spatialspectral and channel-spectral differences. Then, the encoderdecoder (ED) is used to model the global context of the difference features. The most critical tokens in spatial-spectral and channelspectral differences are captured by the top-k token selection and reinforcement (TTSR) module, which eliminates irrelevant tokens and reinforces the feature representation. Finally, the multidirectional attention (MA) module facilitates interaction between spatial-spectral and channel-spectral differences, improving the understanding of change regions. Extensive experiments on three datasets show that SCDI-Net effectively detects changes in complex RSCD tasks.
- 13 January 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- Vol. 18 (19391404) , 5656-5672
Open AccessSpatial–Spectral and Channel–Spectral Differences Integration Network for Remote Sensing Image Change Detection
Deep learning methods have shown promising performance in remote sensing (RS) image change detection (CD), largely due to their excellent ability to explore local and global contexts. However, existing methods fail to effectively integrate spectral information into difference features, which makes CD results more susceptible to pseudo-changes. To address the above challenges, this paper proposes a spatial-spectral and channelspectral differences integration network (SCDI-Net), which enriches the discriminative representation of change regions. First, the spatial-spectral and channel-spectral differences generation (SSCS-DG) module is designed to explore valuable difference clues from the spatial-channel-spectral perspective for mitigating the effect of pseudo-changes. The deformable feature pyramid network (DFPN) is constructed to aggregate multiscale spatialspectral and channel-spectral differences. Then, the encoderdecoder (ED) is used to model the global context of the difference features. The most critical tokens in spatial-spectral and channelspectral differences are captured by the top-k token selection and reinforcement (TTSR) module, which eliminates irrelevant tokens and reinforces the feature representation. Finally, the multidirectional attention (MA) module facilitates interaction between spatial-spectral and channel-spectral differences, improving the understanding of change regions. Extensive experiments on three datasets show that SCDI-Net effectively detects changes in complex RSCD tasks.
- Open Access
- 26 November 2024
- Published by MDPI AG in Electronics
- Vol. 13 (23) , 4663
Research on Non-Intrusive Load Disaggregation Technology Based on VMD–Nyströmformer–BiTCN
Non-intrusive load disaggregation is a technique that monitors the total electrical load of an entire building or household. It uses a single power metering device to measure the total load. Then, it employs algorithms to break it down into the individual usage of different electrical devices. To address issues in load disaggregation models such as long training times, feature interference caused by the activation of other loads, and accuracy deficiencies caused by behavioral interference from users’ electricity usage habits, this paper proposes a VMD–Nyströmformer–BiTCN network architecture. The variational mode decomposition (VMD) filters the raw power data, reducing errors caused by noise and enhancing the accuracy of decomposing the load. A deep learning network utilizes a modified attention model, Nyströmformer, to reduce feature entanglement and accuracy degradation caused by habitual behavior interference during load disaggregation, while ensuring precise accuracy and improving network operational speed. The training network uses a bidirectional temporal convolutional network (BiTCN) and incorporates a residual network to expand the receptive field, allowing it to receive longer load sequence data and acquire more effective load information, thereby improving the disaggregation effectiveness for target appliances.
- 26 November 2024
- Published by MDPI AG in Electronics
- Vol. 13 (23) , 4663
Open AccessResearch on Non-Intrusive Load Disaggregation Technology Based on VMD–Nyströmformer–BiTCN
Non-intrusive load disaggregation is a technique that monitors the total electrical load of an entire building or household. It uses a single power metering device to measure the total load. Then, it employs algorithms to break it down into the individual usage of different electrical devices. To address issues in load disaggregation models such as long training times, feature interference caused by the activation of other loads, and accuracy deficiencies caused by behavioral interference from users’ electricity usage habits, this paper proposes a VMD–Nyströmformer–BiTCN network architecture. The variational mode decomposition (VMD) filters the raw power data, reducing errors caused by noise and enhancing the accuracy of decomposing the load. A deep learning network utilizes a modified attention model, Nyströmformer, to reduce feature entanglement and accuracy degradation caused by habitual behavior interference during load disaggregation, while ensuring precise accuracy and improving network operational speed. The training network uses a bidirectional temporal convolutional network (BiTCN) and incorporates a residual network to expand the receptive field, allowing it to receive longer load sequence data and acquire more effective load information, thereby improving the disaggregation effectiveness for target appliances.
- Open Access
Detection and Recognition of Tilted Characters on Railroad Wagon Wheelsets Based on Deep Learning
The quality of railroad wheelsets is an important guarantee for the safe operation of wagons, and mastering the production information of wheelsets plays a vital role in vehicle scheduling and railroad transportation safety. However, when using objection detection methods to detect the production information of wheelsets, there are situations that affect detection such as character tilting and unfixed position. Therefore, this paper proposes a deep learning-based method for accurately detecting and recognizing tilted character information on railroad wagon wheelsets. It covers three parts. Firstly, we construct a tilted character detection network based on Faster RCNN for generating a wheelset’s character candidate regions. Secondly, we design a tilted character correction network to classify and correct the orientation of flipped characters. Finally, a character recognition network is constructed based on convolutional recurrent neural network (CRNN) to realize the task of recognizing a wheelset’s characters. The result shows that the method can quickly and effectively detect and identify the information of tilted characters on wheelsets in images.
Open AccessDetection and Recognition of Tilted Characters on Railroad Wagon Wheelsets Based on Deep Learning
The quality of railroad wheelsets is an important guarantee for the safe operation of wagons, and mastering the production information of wheelsets plays a vital role in vehicle scheduling and railroad transportation safety. However, when using objection detection methods to detect the production information of wheelsets, there are situations that affect detection such as character tilting and unfixed position. Therefore, this paper proposes a deep learning-based method for accurately detecting and recognizing tilted character information on railroad wagon wheelsets. It covers three parts. Firstly, we construct a tilted character detection network based on Faster RCNN for generating a wheelset’s character candidate regions. Secondly, we design a tilted character correction network to classify and correct the orientation of flipped characters. Finally, a character recognition network is constructed based on convolutional recurrent neural network (CRNN) to realize the task of recognizing a wheelset’s characters. The result shows that the method can quickly and effectively detect and identify the information of tilted characters on wheelsets in images.
- Open Access
Double-Layer SOC and SOH Equalization Scheme for LiFePO4 Battery Energy Storage System Using MAS Blackboard System
26650 LiFePO4 battery, as an ideal energy storage battery for the smart grid system, has the shortcomings of fast aging speed and large dispersion of aging trend, which is the reason for accelerating the 26650 battery system aging. However, it is noted that the 26650 LiFePO4 battery with high aging trend dispersion shows the characteristics of grouping. Therefore, to prolong the 26650 battery system life, this paper proposes a state-of-charge (SOC) and state-of-health (SOH) double-layer equalization scheme for 26650 LiFePO4 batteries based on a multi-agent blackboard system (MABS) that utilizes these characteristics. Based on MABS, the battery units with similar SOH are clustered into a group by a hierarchical-clustering algorithm. Then, SOH balancing is performed by changing the depth of charge and discharge (DOD) between groups, and SOC balancing is performed within the group. MATLAB simulation results verify the effectiveness of the proposed scheme in ensuring the balance of battery charge and discharge and prolonging the life of the battery energy storage system (BESS).
Open AccessDouble-Layer SOC and SOH Equalization Scheme for LiFePO4 Battery Energy Storage System Using MAS Blackboard System
26650 LiFePO4 battery, as an ideal energy storage battery for the smart grid system, has the shortcomings of fast aging speed and large dispersion of aging trend, which is the reason for accelerating the 26650 battery system aging. However, it is noted that the 26650 LiFePO4 battery with high aging trend dispersion shows the characteristics of grouping. Therefore, to prolong the 26650 battery system life, this paper proposes a state-of-charge (SOC) and state-of-health (SOH) double-layer equalization scheme for 26650 LiFePO4 batteries based on a multi-agent blackboard system (MABS) that utilizes these characteristics. Based on MABS, the battery units with similar SOH are clustered into a group by a hierarchical-clustering algorithm. Then, SOH balancing is performed by changing the depth of charge and discharge (DOD) between groups, and SOC balancing is performed within the group. MATLAB simulation results verify the effectiveness of the proposed scheme in ensuring the balance of battery charge and discharge and prolonging the life of the battery energy storage system (BESS).
- Open Access
- 13 December 2022
- Published by Springer Nature in Research Square
Double-layer SOC and SOH Equalization Scheme for LiFePO4 Battery Energy Storage System using MAS Blackboard System
26650 LiFePO4 battery, as an ideal energy storage battery, has the shortcomings of fast aging speed and large dispersion of aging trend. However, it is noted that the 26650 LiFePO4 battery with high aging trend dispersion shows the characteristics of grouping. Therefore, this paper proposes a SOC and SOH double-layer equalization scheme for such batteries based on a multiagent blackboard system(MABS) by utilizing such characteristics. On the basis of MABS, the battery units with similar SOH are clustered into a group by a hierarchical clustering algorithm. Then SOH balancing is performed by changing the depth of charge and discharge (DOD) between groups, and SOC balancing is performed within the group. MATLAB simulation results verify the effectiveness of the proposed scheme in ensuring the balance of battery charge and discharge and prolonging the life of the battery energy storage system(BESS).
- 13 December 2022
- Published by Springer Nature in Research Square
Open AccessDouble-layer SOC and SOH Equalization Scheme for LiFePO4 Battery Energy Storage System using MAS Blackboard System
26650 LiFePO4 battery, as an ideal energy storage battery, has the shortcomings of fast aging speed and large dispersion of aging trend. However, it is noted that the 26650 LiFePO4 battery with high aging trend dispersion shows the characteristics of grouping. Therefore, this paper proposes a SOC and SOH double-layer equalization scheme for such batteries based on a multiagent blackboard system(MABS) by utilizing such characteristics. On the basis of MABS, the battery units with similar SOH are clustered into a group by a hierarchical clustering algorithm. Then SOH balancing is performed by changing the depth of charge and discharge (DOD) between groups, and SOC balancing is performed within the group. MATLAB simulation results verify the effectiveness of the proposed scheme in ensuring the balance of battery charge and discharge and prolonging the life of the battery energy storage system(BESS).
- Open Access
- 14 July 2022
- Published by MDPI AG in Electronics
- Vol. 11 (14) , 2202
Economic Dispatch of Microgrid Based on Load Prediction of Back Propagation Neural Network–Local Mean Decomposition–Long Short-Term Memory
To plan the work of power generation equipment, it is necessary to ensure that the power supply is sufficient and to achieve the minimum cost to ensure the safety and economy of the microgrid. Based on back propagation neural network–local mean decomposition–long short-term memory (BPNN–LMD–LSTM) load prediction, the design is based on a fixed-time consistency algorithm with random delay to predict the economic dispatch of microgrids. Firstly, the initial power load prediction sequence is obtained by continuous training of the back propagation neural network (BPNN); the residual sequence with other influencing factors is decomposed by local mean decomposition (LMD); and the long short-term memory neural network (LSTM) is used to predict the output prediction residual sequence, and the final short-term power load prediction is obtained. Based on predicting load, the fixed-time consistency algorithm with random delay is used to add supply and demand balance constraints to optimize the power distribution of the power generation units of the distributed microgrid and reduce the power generation cost of the microgrid. The results show that the prediction model has better prediction accuracy, and the scheduling algorithm based on the prediction model has a faster convergence rate to reach the lowest power generation cost.
- 14 July 2022
- Published by MDPI AG in Electronics
- Vol. 11 (14) , 2202
Open AccessEconomic Dispatch of Microgrid Based on Load Prediction of Back Propagation Neural Network–Local Mean Decomposition–Long Short-Term Memory
To plan the work of power generation equipment, it is necessary to ensure that the power supply is sufficient and to achieve the minimum cost to ensure the safety and economy of the microgrid. Based on back propagation neural network–local mean decomposition–long short-term memory (BPNN–LMD–LSTM) load prediction, the design is based on a fixed-time consistency algorithm with random delay to predict the economic dispatch of microgrids. Firstly, the initial power load prediction sequence is obtained by continuous training of the back propagation neural network (BPNN); the residual sequence with other influencing factors is decomposed by local mean decomposition (LMD); and the long short-term memory neural network (LSTM) is used to predict the output prediction residual sequence, and the final short-term power load prediction is obtained. Based on predicting load, the fixed-time consistency algorithm with random delay is used to add supply and demand balance constraints to optimize the power distribution of the power generation units of the distributed microgrid and reduce the power generation cost of the microgrid. The results show that the prediction model has better prediction accuracy, and the scheduling algorithm based on the prediction model has a faster convergence rate to reach the lowest power generation cost.
- Open Access
Fault Detection for Interval Type-2 T-S Fuzzy Networked Systems via Event-Triggered Control
This paper investigates the event-triggered fault diagnosis (FD) problem for interval type-2 (IT2) Takagi–Sugeno (T-S) fuzzy networked systems. Firstly, an FD fuzzy filter is proposed by using IT2 T-S fuzzy theory to generate a residual signal. This means that the FD filter premise variable needs to not be identical to the nonlinear networked systems (NNSs). The evaluation functions are referenced to determine the occurrence of system faults. Secondly, under the event-triggered mechanism, a fault residual system (FRS) is established with parameter uncertainty, external disturbance and time delay, which can reduce signal transmission and communication pressure. Thirdly, the progressive stability of the fault residual system is guaranteed by using the Lyapunov theory. For the energy bounded condition of external noise interference, the performance criterion is established using linear matrix inequalities. The matrix parameters of the target FD filter are obtained by the convex optimization method. A less conservative fault diagnosis method can be obtained. Finally, the simulation example is provided to illustrate the effectiveness and the practicalities of the proposed theoretical method.
Open AccessFault Detection for Interval Type-2 T-S Fuzzy Networked Systems via Event-Triggered Control
This paper investigates the event-triggered fault diagnosis (FD) problem for interval type-2 (IT2) Takagi–Sugeno (T-S) fuzzy networked systems. Firstly, an FD fuzzy filter is proposed by using IT2 T-S fuzzy theory to generate a residual signal. This means that the FD filter premise variable needs to not be identical to the nonlinear networked systems (NNSs). The evaluation functions are referenced to determine the occurrence of system faults. Secondly, under the event-triggered mechanism, a fault residual system (FRS) is established with parameter uncertainty, external disturbance and time delay, which can reduce signal transmission and communication pressure. Thirdly, the progressive stability of the fault residual system is guaranteed by using the Lyapunov theory. For the energy bounded condition of external noise interference, the performance criterion is established using linear matrix inequalities. The matrix parameters of the target FD filter are obtained by the convex optimization method. A less conservative fault diagnosis method can be obtained. Finally, the simulation example is provided to illustrate the effectiveness and the practicalities of the proposed theoretical method.
- 1 May 2020
- Published by Springer Nature in Control Theory and Technology
- Vol. 18 (2) , 168-181
Finite-time non-fragile filtering for nonlinear networked control systems via a mixed time/event-triggered transmission mechanism
- 1 May 2020
- Published by Springer Nature in Control Theory and Technology
- Vol. 18 (2) , 168-181
Finite-time non-fragile filtering for nonlinear networked control systems via a mixed time/event-triggered transmission mechanism