Xiao Cai
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
Xiao Cai has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
Min Xie
3 shared publications- Min Xie
City University of Hong Kong
3 shared publications - Ajith Kumar Parlikad
University of Cambridge
2 shared publications - Anandarup Mukherjee
Institute for Manufacturing, University of Cambridge, Cambridge, U.K.
2 shared publications Arjun Parekh
2 shared publications- Naipeng Li
Xi’an Jiaotong University
2 shared publications
Latest Publications
- Open Access
- 17 September 2026
- Published by Elsevier in Computers in Industry
- Vol. 182
Inferring network-level alarm influence pathways for cellular infrastructure reliability
- 17 September 2026
- Published by Elsevier in Computers in Industry
- Vol. 182
Open AccessInferring network-level alarm influence pathways for cellular infrastructure reliability
- Open Access
- 21 February 2026
- Published by Elsevier in SSRN Electronic Journal
Topology-Agnostic Inference of Alarm Influence Pathways using Time-Aware Graph Attention Networks in Cellular Networks
Large-scale telecommunication infrastructures, such as radio access networks (RANs), constitute critical components of modern digital Industry. To ensure infrastructure intelligence and operational reliability, it is essential to understand how alarms propagate across interconnected base stations (BSs). Existing alarm analysis approaches mainly focus on isolated temporal patterns at individual sites, while overlooking cross-site alarm influence mechanisms at the infrastructure level. To address this gap, this paper proposes a co-occurrence and time-aware graph attention network (COTA-GAT) to infer alarm influence pathways in telecommunication infrastructure without relying on explicit physical or logical topology information. COTA-GAT constructs a directed propagation graph from alarm co-occurrence patterns and incorporates time-decay weighting to capture the intensity and temporal proximity of inter-site interactions. Dual time-aware graph attention layers are employed to learn heterogeneous inter-BS influences, while a long short-term memory (LSTM) module captures the temporal evolution of alarm dynamics. In addition, a unified feature encoding and aggregation process is developed to transform heterogeneous alarm records into structured BS-level representations suitable for topology-agnostic graph learning. Through joint short-term temporal and long-term feature correlation analysis, this study reveals consistent and non-random cross-site alarm dependencies, providing empirical evidence of genuine alarm propagation behavior in operational RANs. Experiments conducted on a large-scale real-world RAN dataset demonstrate that COTA-GAT significantly outperforms traditional machine learning methods and existing graph-based baselines in inferring cross-site alarm propagation links. It achieves AUC and average precision values of up to 0.97 and 0.95, respectively, under short-term propagation windows. The proposed framework offers new insights into network-level alarm influence pathways and provides a data-driven foundation for proactive reliability management in large-scale communication networks.
- 21 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessTopology-Agnostic Inference of Alarm Influence Pathways using Time-Aware Graph Attention Networks in Cellular Networks
Large-scale telecommunication infrastructures, such as radio access networks (RANs), constitute critical components of modern digital Industry. To ensure infrastructure intelligence and operational reliability, it is essential to understand how alarms propagate across interconnected base stations (BSs). Existing alarm analysis approaches mainly focus on isolated temporal patterns at individual sites, while overlooking cross-site alarm influence mechanisms at the infrastructure level. To address this gap, this paper proposes a co-occurrence and time-aware graph attention network (COTA-GAT) to infer alarm influence pathways in telecommunication infrastructure without relying on explicit physical or logical topology information. COTA-GAT constructs a directed propagation graph from alarm co-occurrence patterns and incorporates time-decay weighting to capture the intensity and temporal proximity of inter-site interactions. Dual time-aware graph attention layers are employed to learn heterogeneous inter-BS influences, while a long short-term memory (LSTM) module captures the temporal evolution of alarm dynamics. In addition, a unified feature encoding and aggregation process is developed to transform heterogeneous alarm records into structured BS-level representations suitable for topology-agnostic graph learning. Through joint short-term temporal and long-term feature correlation analysis, this study reveals consistent and non-random cross-site alarm dependencies, providing empirical evidence of genuine alarm propagation behavior in operational RANs. Experiments conducted on a large-scale real-world RAN dataset demonstrate that COTA-GAT significantly outperforms traditional machine learning methods and existing graph-based baselines in inferring cross-site alarm propagation links. It achieves AUC and average precision values of up to 0.97 and 0.95, respectively, under short-term propagation windows. The proposed framework offers new insights into network-level alarm influence pathways and provides a data-driven foundation for proactive reliability management in large-scale communication networks.
- 12 February 2026
- Published by Elsevier in Mechanical Systems and Signal Processing
- Vol. 248
Order-adaptive subspace scale learning for unsupervised anomaly detection under time-varying rotational speed conditions
- 12 February 2026
- Published by Elsevier in Mechanical Systems and Signal Processing
- Vol. 248
Order-adaptive subspace scale learning for unsupervised anomaly detection under time-varying rotational speed conditions
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 40 (11) , 17040-17050
Small-Sample Prediction Validation Testing: Uncertainty-Aware Design and Robust Maintenance Strategy for Power Electronic Converters
Reliable degradation prediction is paramount for ensuring operational stability and minimizing failure risks in power electronic converters (PECs). Effective maintenance strategies, and thus system longevity and safety, depend on the accuracy of these predictions. However, verifying prediction accuracy with limited data and test durations presents a persistent challenge. This article introduces a test design methodology for degradation prediction verification in PECs, coupled with a preventive maintenance strategy that explicitly incorporates prediction uncertainties. Our approach provides a systematic means of determining the minimum test duration and a principled basis for selecting small sample sizes, enabling the derivation of robust maintenance plans that account for all potential prediction error scenarios. The methodology is validated through a case study of a three-phase inverter, demonstrating its efficacy with sample sizes of fewer than ten. The results highlight the practicality and effectiveness of the proposed methodology in assessing degradation prediction accuracy and, critically, illustrate the superior performance of the uncertainty-aware maintenance strategy in mitigating the risk of catastrophic failures.
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Power Electronics
- Vol. 40 (11) , 17040-17050
Small-Sample Prediction Validation Testing: Uncertainty-Aware Design and Robust Maintenance Strategy for Power Electronic Converters
Reliable degradation prediction is paramount for ensuring operational stability and minimizing failure risks in power electronic converters (PECs). Effective maintenance strategies, and thus system longevity and safety, depend on the accuracy of these predictions. However, verifying prediction accuracy with limited data and test durations presents a persistent challenge. This article introduces a test design methodology for degradation prediction verification in PECs, coupled with a preventive maintenance strategy that explicitly incorporates prediction uncertainties. Our approach provides a systematic means of determining the minimum test duration and a principled basis for selecting small sample sizes, enabling the derivation of robust maintenance plans that account for all potential prediction error scenarios. The methodology is validated through a case study of a three-phase inverter, demonstrating its efficacy with sample sizes of fewer than ten. The results highlight the practicality and effectiveness of the proposed methodology in assessing degradation prediction accuracy and, critically, illustrate the superior performance of the uncertainty-aware maintenance strategy in mitigating the risk of catastrophic failures.
- 24 February 2025
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 259
Knowledge embedded spatial–temporal graph convolutional networks for remaining useful life prediction
- 24 February 2025
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 259
Knowledge embedded spatial–temporal graph convolutional networks for remaining useful life prediction
- 4 January 2024
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 244
RUL prediction for two-phase degrading systems considering physical damage observations
- 4 January 2024
- Published by Elsevier in Reliability Engineering & System Safety
- Vol. 244
RUL prediction for two-phase degrading systems considering physical damage observations
- Open Access
- 6 September 2020
- Published by Wiley in Orthopaedic Surgery
- Vol. 12 (6) , 1627-1634
Balser Plate Stabilization for Traumatic Sternoclavicular Instabilities or Medial Clavicle Fractures: A Case Series and Literature Review
Objective: This study was performed to observe the effect of internal Balser plate fixation for treating unstable sternoclavicular joints (SCJ) and displaced medial clavicle fractures.Methods: From April 2009 to September 2016, 17 consecutive patients who underwent open reduction and internal Balser plate fixation for SCJ dislocations or medial clavicle fractures were retrospectively reviewed. There were 11 male and six female patients, with a mean age of 45.6 ± 15.5 years. Standardized treatment procedures consisted of reduction, creating a space posterior dorsal osteal face of the sternal manubrium, an inverted Balser plating, and postoperative immobilization. At follow‐up, plain radiographs were assessed for fracture union, implant loosening, degenerative changes, and joint congruity. Clinical evaluation included: completion of the Disability of the Arm, Shoulder, and Hand (DASH) questionnaire; determination of the Constant and Murley score and visual analog scale (VAS) score; and assessment of intraoperative and postoperative complications.Results: All patients were followed up, at a mean follow‐up of 20.1 ± 7.9 months, each fracture had a solid union, and each dislocation showed no sign of recurrent dislocation. The mean shoulder forward flexion was 162.9° ± 8.1°. The mean DASH score was 5.2 ± 5.2 points. The mean Constant and Murley joint function score was 93.7 ± 7.9 points, with 15 excellent cases and two good cases. The mean VAS score was 1.1 ± 1.4 points, showing significant improvement compared with the VAS score preoperatively. Postoperative complications included one wound hematoma which was healed after a debridement and one recurrent instability due to hook migration, which underwent revision reconstruction. All patients were satisfied with their treatment outcome at the final follow‐up.Conclusion: Sternoclavicular joints dislocation or medial clavicle fractures can be treated successfully with Balser plate fixation. This technique permits early functional exercise while preserving the SCJ.
- 6 September 2020
- Published by Wiley in Orthopaedic Surgery
- Vol. 12 (6) , 1627-1634
Open AccessBalser Plate Stabilization for Traumatic Sternoclavicular Instabilities or Medial Clavicle Fractures: A Case Series and Literature Review
Objective: This study was performed to observe the effect of internal Balser plate fixation for treating unstable sternoclavicular joints (SCJ) and displaced medial clavicle fractures.Methods: From April 2009 to September 2016, 17 consecutive patients who underwent open reduction and internal Balser plate fixation for SCJ dislocations or medial clavicle fractures were retrospectively reviewed. There were 11 male and six female patients, with a mean age of 45.6 ± 15.5 years. Standardized treatment procedures consisted of reduction, creating a space posterior dorsal osteal face of the sternal manubrium, an inverted Balser plating, and postoperative immobilization. At follow‐up, plain radiographs were assessed for fracture union, implant loosening, degenerative changes, and joint congruity. Clinical evaluation included: completion of the Disability of the Arm, Shoulder, and Hand (DASH) questionnaire; determination of the Constant and Murley score and visual analog scale (VAS) score; and assessment of intraoperative and postoperative complications.Results: All patients were followed up, at a mean follow‐up of 20.1 ± 7.9 months, each fracture had a solid union, and each dislocation showed no sign of recurrent dislocation. The mean shoulder forward flexion was 162.9° ± 8.1°. The mean DASH score was 5.2 ± 5.2 points. The mean Constant and Murley joint function score was 93.7 ± 7.9 points, with 15 excellent cases and two good cases. The mean VAS score was 1.1 ± 1.4 points, showing significant improvement compared with the VAS score preoperatively. Postoperative complications included one wound hematoma which was healed after a debridement and one recurrent instability due to hook migration, which underwent revision reconstruction. All patients were satisfied with their treatment outcome at the final follow‐up.Conclusion: Sternoclavicular joints dislocation or medial clavicle fractures can be treated successfully with Balser plate fixation. This technique permits early functional exercise while preserving the SCJ.