Ling Bai
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
Ling Bai has a total of 32 co-authors. Here are the top co-authors by number of shared publications:
- Zheng Liu
University of British Columbia
4 shared publications - Yinguo Li
Department of Automation, Chongqing University of Posts and Telecommunications, Chongqing 400065, China
3 shared publications - Fen Zhao
School of computer science and technology, Chongqing University of Posts and Telecommunications, Chongqing, 400065, China
2 shared publications - Rakiba Rayhana
School of Engineering, University of British Columbia, Kelowna, BC, Canada
2 shared publications - Xinheng Wang
School of Communication and Information Engineering, Chongqing University of Posts and Telecommunications, Chongqing, 400065, China
2 shared publications - Zhen Tian
Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ)
2 shared publications
Latest Publications
- Open Access
- 21 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Instrumentation and Measurement
- Vol. 5, 8500111
ANEP: Adaptive Newton Ego Planner for Multiwaypoint UAV Trajectory Optimization in Constrained Environments
Roof gutters are highly susceptible to the accumulation of debris, such as fallen leaves, which can block drain outlets and pose a significant threat to the structural integrity of a building. Traditional high-altitude cleaning methods primarily rely on manual operations with ropes, ladders, and extended tools. These methods are not only inherently dangerous, but also inefficient, making them inadequate for building maintenance. The use of drones offers a promising solution for high-altitude cleaning. However, drones still face significant technical challenges in ensuring precise arrival at target points after detecting debris while simultaneously navigating multi-point obstacle avoidance during flight. In particular, autonomous roof gutter cleaning requires robust multi-stage trajectory planning in a cluttered environment where state estimation and waypoint accuracy can be degraded by the noises and inaccuracies from sensing and Global Navigation Satellite System (GNSS). This paper presents an Adaptive Newton Ego Planner (ANEP) framework for multi-waypoint trajectory optimization under constrained environments. The proposed ANEP integrates a novel Adaptive Newton-Raphson based optimizer (ANRBO) to enhance exploration and prevent premature convergence during online trajectory refinement. The optimizer leverages Newton-Raphson Search Rules (NRSR) and a Trap Avoidance Operator (TAO), while an online dynamic adaptation mechanism enables efficient multi-point exploration and real-time trajectory optimization. In this paper, we focus on trajectory planning and optimization. Perception, e.g., debris detection, and cleaning actuation were not evaluated in this simulation study. Experimental results demonstrate that the proposed trajectory planning method improves multi-stage flight missions by generating smoother trajectories, lower trajectory errors, and a higher mission completion rate. These results suggest that ANEP can serve as an effective trajectory planning component for UAV-based roof gutter maintenance scenarios.
- 21 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Instrumentation and Measurement
- Vol. 5, 8500111
Open AccessANEP: Adaptive Newton Ego Planner for Multiwaypoint UAV Trajectory Optimization in Constrained Environments
Roof gutters are highly susceptible to the accumulation of debris, such as fallen leaves, which can block drain outlets and pose a significant threat to the structural integrity of a building. Traditional high-altitude cleaning methods primarily rely on manual operations with ropes, ladders, and extended tools. These methods are not only inherently dangerous, but also inefficient, making them inadequate for building maintenance. The use of drones offers a promising solution for high-altitude cleaning. However, drones still face significant technical challenges in ensuring precise arrival at target points after detecting debris while simultaneously navigating multi-point obstacle avoidance during flight. In particular, autonomous roof gutter cleaning requires robust multi-stage trajectory planning in a cluttered environment where state estimation and waypoint accuracy can be degraded by the noises and inaccuracies from sensing and Global Navigation Satellite System (GNSS). This paper presents an Adaptive Newton Ego Planner (ANEP) framework for multi-waypoint trajectory optimization under constrained environments. The proposed ANEP integrates a novel Adaptive Newton-Raphson based optimizer (ANRBO) to enhance exploration and prevent premature convergence during online trajectory refinement. The optimizer leverages Newton-Raphson Search Rules (NRSR) and a Trap Avoidance Operator (TAO), while an online dynamic adaptation mechanism enables efficient multi-point exploration and real-time trajectory optimization. In this paper, we focus on trajectory planning and optimization. Perception, e.g., debris detection, and cleaning actuation were not evaluated in this simulation study. Experimental results demonstrate that the proposed trajectory planning method improves multi-stage flight missions by generating smoother trajectories, lower trajectory errors, and a higher mission completion rate. These results suggest that ANEP can serve as an effective trajectory planning component for UAV-based roof gutter maintenance scenarios.
- 28 March 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 175
CrossRanker: Cross-dimensional vessel signals fusion for destination prediction
- 28 March 2026
- Published by Elsevier in Engineering Applications of Artificial Intelligence
- Vol. 175
CrossRanker: Cross-dimensional vessel signals fusion for destination prediction
- 23 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-18
A Survey of Water Pipeline Leak Detection Using Acoustic Sensing
Water transmission and distribution infrastructures are essential assets within urban communities and crucial for maintaining a reliable and efficient water supply. The aging of water mains can lead to water loss, environmental contamination, and service disruptions. Acoustic-based leak detection methods have become a key approach because they can detect small leaks and are cost-effective. This paper reviews both traditional and state-of-the-art acoustic detection techniques, focusing on the integration of machine learning approaches. Machine learning-based approaches have shown promising accuracy and efficiency in recent years. The survey covers several key areas, including the characterization of common leak types, the deployment of acoustic sensors, the general framework for acoustic leak detection processes (highlighting the role of conventional machine learning and deep learning in processing acoustic signals for leak detection), and the research challenges in this field. This article presents an overview of recent advances in acoustic leak detection technologies, providing a guide to select suitable approaches based on data availability and research requirements. It also identifies prospective research areas for water pipeline inspection studies.
- 23 February 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 75 (00189456) , 1-18
A Survey of Water Pipeline Leak Detection Using Acoustic Sensing
Water transmission and distribution infrastructures are essential assets within urban communities and crucial for maintaining a reliable and efficient water supply. The aging of water mains can lead to water loss, environmental contamination, and service disruptions. Acoustic-based leak detection methods have become a key approach because they can detect small leaks and are cost-effective. This paper reviews both traditional and state-of-the-art acoustic detection techniques, focusing on the integration of machine learning approaches. Machine learning-based approaches have shown promising accuracy and efficiency in recent years. The survey covers several key areas, including the characterization of common leak types, the deployment of acoustic sensors, the general framework for acoustic leak detection processes (highlighting the role of conventional machine learning and deep learning in processing acoustic signals for leak detection), and the research challenges in this field. This article presents an overview of recent advances in acoustic leak detection technologies, providing a guide to select suitable approaches based on data availability and research requirements. It also identifies prospective research areas for water pipeline inspection studies.
- 26 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Multiserver Federated Learning for Building Energy Prediction With Wind Speed
Energy consumption of buildings is correlated with environmental factors such as weather and temperature. Developing a reliable energy forecasting model requires a large-scale, high-quality dataset. However, privacy regulations mandate that data be stored locally, prohibiting data sharing. As a solution, a Federated Building Energy Forecasting framework has emerged, supporting machine learning using decentralized energy data. This framework facilitates the communication of deep learning models between a server and multiple clients. However, it remains vulnerable to server crashes, which can disrupt the entire network’s functionality. To address this issue, multi-server federated learning strategies have been proposed to reduce dependency on a single server and enhance reliability. Additionally, the meteorological variable of wind speed has been integrated to improve energy forecasting through multi-task learning. A multi-task learning method is proposed to forecast energy load and wind speed simultaneously, employing a mixture-of-experts (MoE) architecture. Furthermore, a new metaformer neural network has been designed using the Taylor expansion approximation to simplify self-attention mechanisms. Extensive experiments have evaluated the effectiveness of multi-task learning and multi-server federated training. The results demonstrate that forecasting accuracies have improved by 10% to 30% compared to existing state-of-the-art transformer forecasting models. Additionally, incorporating wind speed within the MoE framework benefits energy forecasting. Multi-server strategies also help mitigate data scarcity issues through strategic random clustering.
- 26 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 74 (00189456) , 1-14
Multiserver Federated Learning for Building Energy Prediction With Wind Speed
Energy consumption of buildings is correlated with environmental factors such as weather and temperature. Developing a reliable energy forecasting model requires a large-scale, high-quality dataset. However, privacy regulations mandate that data be stored locally, prohibiting data sharing. As a solution, a Federated Building Energy Forecasting framework has emerged, supporting machine learning using decentralized energy data. This framework facilitates the communication of deep learning models between a server and multiple clients. However, it remains vulnerable to server crashes, which can disrupt the entire network’s functionality. To address this issue, multi-server federated learning strategies have been proposed to reduce dependency on a single server and enhance reliability. Additionally, the meteorological variable of wind speed has been integrated to improve energy forecasting through multi-task learning. A multi-task learning method is proposed to forecast energy load and wind speed simultaneously, employing a mixture-of-experts (MoE) architecture. Furthermore, a new metaformer neural network has been designed using the Taylor expansion approximation to simplify self-attention mechanisms. Extensive experiments have evaluated the effectiveness of multi-task learning and multi-server federated training. The results demonstrate that forecasting accuracies have improved by 10% to 30% compared to existing state-of-the-art transformer forecasting models. Additionally, incorporating wind speed within the MoE framework benefits energy forecasting. Multi-server strategies also help mitigate data scarcity issues through strategic random clustering.
- 3 January 2023
- Published by Springer Nature in Environmental Science and Pollution Research
- Vol. 30 (13) , 39107-39120
Ecological water replenishment to the Yongding River, China: effects of different water sources on inorganic ions and organic matter characteristics
Ecological water replenishment is considered to be an important project to adjust river ecosystems with water shortages and degradation, and its impact on the water quality of the target river section deserves attention. By investigating the characteristics of inorganic ions and organic components of the Beijing section of the Yongding River (YDR) from upstream to downstream, the sources of inorganic ions and dissolved organic matter (DOM) during an ecological water replenishment event were analysed and discussed. This study illustrated the hydrochemical response to different supplemental water sources in three sections of the YDR (mountain gorge section (MGS), urban plain section (UPS), and suburb plain section (SPS)). The results showed that the total dissolved solids (TDS) and ion concentrations were significantly different (p < 0.001) in the three river sections due to different supplemental water sources. The highest concentration of TDS was found in the UPS (870.92 mg/L) replenished by reclaimed water, while the lowest concentration of TDS was found in the SPS (306.95 mg/L) replenished by the water of the South-to-North Water Diversion Project (SNWD). Despite the differences in the water sources of replenishment, the optical parameters of DOM and fluorescent components showed similar protein-like dominated endogenous source characteristics in the three river sections of the YDR, which was due to the influence of reservoir water (MGS and SPS) and reclaimed water (UPS). Our study emphasizes the importance of understanding the impact of different water sources on the water replenishment process, which provides a scientific reference for the management of ecological water replenishment.
- 3 January 2023
- Published by Springer Nature in Environmental Science and Pollution Research
- Vol. 30 (13) , 39107-39120
Ecological water replenishment to the Yongding River, China: effects of different water sources on inorganic ions and organic matter characteristics
Ecological water replenishment is considered to be an important project to adjust river ecosystems with water shortages and degradation, and its impact on the water quality of the target river section deserves attention. By investigating the characteristics of inorganic ions and organic components of the Beijing section of the Yongding River (YDR) from upstream to downstream, the sources of inorganic ions and dissolved organic matter (DOM) during an ecological water replenishment event were analysed and discussed. This study illustrated the hydrochemical response to different supplemental water sources in three sections of the YDR (mountain gorge section (MGS), urban plain section (UPS), and suburb plain section (SPS)). The results showed that the total dissolved solids (TDS) and ion concentrations were significantly different (p < 0.001) in the three river sections due to different supplemental water sources. The highest concentration of TDS was found in the UPS (870.92 mg/L) replenished by reclaimed water, while the lowest concentration of TDS was found in the SPS (306.95 mg/L) replenished by the water of the South-to-North Water Diversion Project (SNWD). Despite the differences in the water sources of replenishment, the optical parameters of DOM and fluorescent components showed similar protein-like dominated endogenous source characteristics in the three river sections of the YDR, which was due to the influence of reservoir water (MGS and SPS) and reclaimed water (UPS). Our study emphasizes the importance of understanding the impact of different water sources on the water replenishment process, which provides a scientific reference for the management of ecological water replenishment.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 552-559
A Supervisory Mask Attentional Network for Person Re-Identification in Uniform Dress Scenes
Person re-identification (Re-ID) aims at retrieving a person's identification across multiple non-overlapping cameras. While recent Re-ID methods have achieved significant success on a number of benchmark datasets, most of them are still insufficient in the scenes with unified or highly similar dressing like construction sites, schools and factories. To address this problem, we propose a supervisory mask attention network (SMA-Net). Our approach combines two key components: (1) ROI mask mapping (RMM) is a supervisory branch to provide ROI mappings that divide a person region into several parts; (2) Partial mask attention (PMA) integrates channel and space attention mechanisms that focus on local features and different accessories in each ROI. Therefore, the network can pay more attention to the local features and different accessories herein. Compared with the state of the art methods, SMA-Net demonstrates excellent performance on our dataset of construction scenes, with improvement of 6.65% in mAP and 3.8% in Rank-1 accuracy.
- 1 November 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 552-559
A Supervisory Mask Attentional Network for Person Re-Identification in Uniform Dress Scenes
Person re-identification (Re-ID) aims at retrieving a person's identification across multiple non-overlapping cameras. While recent Re-ID methods have achieved significant success on a number of benchmark datasets, most of them are still insufficient in the scenes with unified or highly similar dressing like construction sites, schools and factories. To address this problem, we propose a supervisory mask attention network (SMA-Net). Our approach combines two key components: (1) ROI mask mapping (RMM) is a supervisory branch to provide ROI mappings that divide a person region into several parts; (2) Partial mask attention (PMA) integrates channel and space attention mechanisms that focus on local features and different accessories in each ROI. Therefore, the network can pay more attention to the local features and different accessories herein. Compared with the state of the art methods, SMA-Net demonstrates excellent performance on our dataset of construction scenes, with improvement of 6.65% in mAP and 3.8% in Rank-1 accuracy.
- 1 October 2021
- Published by Science China Press., Co. Ltd. in Chinese Science Bulletin
- Vol. 66 (30) , 3785-3786
Exploring scientific frontiers and serving regional developments: The 16th Tibetan Plateau Earth Science Academic Annual Meeting
由中国科学院青藏高原研究所主办,中国青藏高原研究会、中国地质学会协办的“第16届青藏高原地球科学学术年会”于2021年6月2-3日在京召开。本次年会主题为:青藏高原地球系统科学。学术交流包括31个大会特邀报告、34个分组研讨报告及39个展板交流报告,涵盖了地质学、地球物理学、地理学、大气科学、生态学、考古学等多个研究领域。来自近五十家高校、科研单位的院士、专家、学者、管理人员等300余人参加本次年会,与会者就此展开了多方位的学术交流,进一步明确了青藏高原重大科学问题的研究方向, 并为今后地球系统科学的发展献计献策。
- 1 October 2021
- Published by Science China Press., Co. Ltd. in Chinese Science Bulletin
- Vol. 66 (30) , 3785-3786
Exploring scientific frontiers and serving regional developments: The 16th Tibetan Plateau Earth Science Academic Annual Meeting
由中国科学院青藏高原研究所主办,中国青藏高原研究会、中国地质学会协办的“第16届青藏高原地球科学学术年会”于2021年6月2-3日在京召开。本次年会主题为:青藏高原地球系统科学。学术交流包括31个大会特邀报告、34个分组研讨报告及39个展板交流报告,涵盖了地质学、地球物理学、地理学、大气科学、生态学、考古学等多个研究领域。来自近五十家高校、科研单位的院士、专家、学者、管理人员等300余人参加本次年会,与会者就此展开了多方位的学术交流,进一步明确了青藏高原重大科学问题的研究方向, 并为今后地球系统科学的发展献计献策。
- Open Access
- 20 November 2020
- Published by Association for Computing Machinery (ACM)
- p. 121-125
Research on Belt and Road Big Data Visualization Based on Text Clustering Algorithm
In the era of big data, people's visual needs for data expression are increasing. In order to achieve better big data display effects, this article introduced the use of text clustering algorithms to achieve data crawling and Echarts technology to realize big data visualization. This system used mvvm's architecture and vue framework development platform, ThinkPHP was used as the background framework, and ES6 related technologies and specifications were used for application development. This system used Echarts, IView, GIS technology and JavaScript development methods to demonstrate economic big data module functions on the web side; Applied CSS3, HTML5, GIS technology to implement project achievement module and university alliance module; Applied Echarts, HTML5, JS function library technology to achieve national information module. This system used stored procedure, database index optimization technology to achieve rapid screening of massive data, and dynamically update and displayed related data through two-way data binding. This system combined real-time location technology with GIS technology to measure the distance between the user and the destination, and automatically plan the tour route to provide related services. This system can provide feasibility suggestions for strategic researchers or experts in related areas of the “Belt and Road”, and provide theoretical basis and technical support.
- 20 November 2020
- Published by Association for Computing Machinery (ACM)
- p. 121-125
Open AccessResearch on Belt and Road Big Data Visualization Based on Text Clustering Algorithm
In the era of big data, people's visual needs for data expression are increasing. In order to achieve better big data display effects, this article introduced the use of text clustering algorithms to achieve data crawling and Echarts technology to realize big data visualization. This system used mvvm's architecture and vue framework development platform, ThinkPHP was used as the background framework, and ES6 related technologies and specifications were used for application development. This system used Echarts, IView, GIS technology and JavaScript development methods to demonstrate economic big data module functions on the web side; Applied CSS3, HTML5, GIS technology to implement project achievement module and university alliance module; Applied Echarts, HTML5, JS function library technology to achieve national information module. This system used stored procedure, database index optimization technology to achieve rapid screening of massive data, and dynamically update and displayed related data through two-way data binding. This system combined real-time location technology with GIS technology to measure the distance between the user and the destination, and automatically plan the tour route to provide related services. This system can provide feasibility suggestions for strategic researchers or experts in related areas of the “Belt and Road”, and provide theoretical basis and technical support.
- Open Access
- 21 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 98168-98180
Lithium-Ion Batteries State of Charge Prediction of Electric Vehicles Using RNNs-CNNs Neural Networks
Capsule network is a promising model in computer vision. It has achieved excellent results on simple datasets such as MNIST, but the performance deteriorates as data becomes complicated. In order to address this issue, we propose a deep capsule network in this paper. To deepen the capsule network, we present a new tensor capsule based routing algorithm and the corresponding convolution operation. Compared to vector capsules, tensor capsules can capture more instance-level information. Together, the relevant convolution operation is beneficial for reducing the amount of parameters in the routing process. Furthermore, we propose a dropout mechanism for vectors and tensors in order to alleviate the potential overfitting problem. Finally, we also inject the multi-scale capsules of the middle layers into a multi-scale decoder to pursue more details of an image and more clear reconstructed image. Experimental results on CIFAR10, Fashion-MNIST, and SVHN demonstrate that the proposed deep tensor network can achieve very competitive performance compared to other state-of-the-art capsule networks.
- 21 May 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 98168-98180
Open AccessLithium-Ion Batteries State of Charge Prediction of Electric Vehicles Using RNNs-CNNs Neural Networks
Capsule network is a promising model in computer vision. It has achieved excellent results on simple datasets such as MNIST, but the performance deteriorates as data becomes complicated. In order to address this issue, we propose a deep capsule network in this paper. To deepen the capsule network, we present a new tensor capsule based routing algorithm and the corresponding convolution operation. Compared to vector capsules, tensor capsules can capture more instance-level information. Together, the relevant convolution operation is beneficial for reducing the amount of parameters in the routing process. Furthermore, we propose a dropout mechanism for vectors and tensors in order to alleviate the potential overfitting problem. Finally, we also inject the multi-scale capsules of the middle layers into a multi-scale decoder to pursue more details of an image and more clear reconstructed image. Experimental results on CIFAR10, Fashion-MNIST, and SVHN demonstrate that the proposed deep tensor network can achieve very competitive performance compared to other state-of-the-art capsule networks.
- Open Access
- 2 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 68000-68012
Semi-Supervised Multi-Granularity CNNs for Text Classification: An Application in Human-Car Interaction
With the rapid growth of advanced driving assistance technology, how to implement humancar interaction humanely and personally becomes increasingly important. Existing methods typically rely on hand-crafted data to conduct training, which cannot fully extract information hidden in text. In this paper, we tackle the problem of label data acquisition difficulties and inadequate feature extraction via semisupervised learning and multi-granularity learning. To fulfill the goal of label data acquisition and feature extraction, we present a novel semi-supervised multi-granularity convolutional neural networks (CNNs)- based (SSMGCNNs-based) model for an application in human-car interaction, which consists of the twoview- embedding (TVE) module and the multi-granularity CNNs (MGCNNs) module. The TVE module learns embeddings of text regions from the unlabeled user command data set and then integrate the learned tv-embeddings into MGCNNs, so that the learned tv-embedded regions are used as an additional input into MGCNNs’ convolution layer to solve the problem of data annotation. MGCNNs can fully extract information hidden in text by multiple convolution kernels of the same convolution layer. We compared our model with some state-of-the-art machine learning models. On the car operation command data set, the simulation results demonstrated that, compared with CNNs, our method respectively improved 5.13%, 5.64%, 3.60% and 5.34% on precision, recall, F-1 and training loss.
- 2 April 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 68000-68012
Open AccessSemi-Supervised Multi-Granularity CNNs for Text Classification: An Application in Human-Car Interaction
With the rapid growth of advanced driving assistance technology, how to implement humancar interaction humanely and personally becomes increasingly important. Existing methods typically rely on hand-crafted data to conduct training, which cannot fully extract information hidden in text. In this paper, we tackle the problem of label data acquisition difficulties and inadequate feature extraction via semisupervised learning and multi-granularity learning. To fulfill the goal of label data acquisition and feature extraction, we present a novel semi-supervised multi-granularity convolutional neural networks (CNNs)- based (SSMGCNNs-based) model for an application in human-car interaction, which consists of the twoview- embedding (TVE) module and the multi-granularity CNNs (MGCNNs) module. The TVE module learns embeddings of text regions from the unlabeled user command data set and then integrate the learned tv-embeddings into MGCNNs, so that the learned tv-embedded regions are used as an additional input into MGCNNs’ convolution layer to solve the problem of data annotation. MGCNNs can fully extract information hidden in text by multiple convolution kernels of the same convolution layer. We compared our model with some state-of-the-art machine learning models. On the car operation command data set, the simulation results demonstrated that, compared with CNNs, our method respectively improved 5.13%, 5.64%, 3.60% and 5.34% on precision, recall, F-1 and training loss.