Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Hao Bai has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
Changwu Huang
1 shared publication- Changwu Huang
Southern University of Science and Technology
1 shared publication - Christophe Gautrelet
Normandie Université, Laboratoire de Mécanique de Normandie (LMN)/INSA ROUEN Normandie
1 shared publication Didier Lemosse
1 shared publication- Didier Lemosse
LMN INSA Rouen Avenue de l'Université 76800 Saint-Étienne du Rouvray France
1 shared publication - Jean-Marc Cherfils
Laboratory of Mechanics of Normandy (LMN), INSA Rouen Normandie, 76000 Rouen, France
1 shared publication
Latest Publications
- Open Access
- 6 July 2026
- Published by Elsevier in SSRN Electronic Journal
SMC-Rec: Spectral Semantic Memory Convolution with Uncertainty-Guided Learning for Sequential Recommendation
Sequential recommendation in information-rich platforms requires models that can exploit behavior, attributes, context, and external semantic representations while remaining efficient for long user histories. Existing item-sequence encoders usually fuse such information as static side features and compress all histories with fixed rules, which can distort semantic signals and discard useful local evidence under uncertainty. This study proposes SMC-Rec, a spectral semantic memory compression framework for sequential recommendation. Each historical interaction is represented as a semantic memory unit that integrates item identity, attributes, context, temporal interval, and calibrated external semantics. A spectral calibration mechanism reduces anisotropy in pretrained semantic embeddings before they are used in memory construction. An uncertainty-adaptive pyramid convolution then compresses semantic memories at multiple temporal scales and adjusts compression strength according to prediction confidence. Training is further enhanced by pyramid-level future supervision and intra-pyramid divergence-consensus learning. Experiments on Amazon Beauty, Sports, Toys, and MovieLens-1M show that SMC-Rec consistently improves Recall@20 and NDCG@20 over representative sequential, attribute-aware, efficient, and semantic-enhanced baselines. Ablation, semantic-geometry, uncertainty-group, robustness, sparse-user, and cold-start analyses confirm that the gains arise from calibrated semantic memory compression rather than from a single auxiliary component.
- 6 July 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSMC-Rec: Spectral Semantic Memory Convolution with Uncertainty-Guided Learning for Sequential Recommendation
Sequential recommendation in information-rich platforms requires models that can exploit behavior, attributes, context, and external semantic representations while remaining efficient for long user histories. Existing item-sequence encoders usually fuse such information as static side features and compress all histories with fixed rules, which can distort semantic signals and discard useful local evidence under uncertainty. This study proposes SMC-Rec, a spectral semantic memory compression framework for sequential recommendation. Each historical interaction is represented as a semantic memory unit that integrates item identity, attributes, context, temporal interval, and calibrated external semantics. A spectral calibration mechanism reduces anisotropy in pretrained semantic embeddings before they are used in memory construction. An uncertainty-adaptive pyramid convolution then compresses semantic memories at multiple temporal scales and adjusts compression strength according to prediction confidence. Training is further enhanced by pyramid-level future supervision and intra-pyramid divergence-consensus learning. Experiments on Amazon Beauty, Sports, Toys, and MovieLens-1M show that SMC-Rec consistently improves Recall@20 and NDCG@20 over representative sequential, attribute-aware, efficient, and semantic-enhanced baselines. Ablation, semantic-geometry, uncertainty-group, robustness, sparse-user, and cold-start analyses confirm that the gains arise from calibrated semantic memory compression rather than from a single auxiliary component.
- 6 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Machine learning assisted probabilistic prediction of long-term fatigue damage and vibration reduction of wind turbine tower using active damping system
Utilisation de l'apprentissage automatique pour la prédiction probabiliste à long-terme du dommage par fatigue et la réduction des vibrations d'un mât d'éolienne par un contrôle actif: Cette thèse est consacrée au développement d'un système d'amortissement actif pour la réduction des vibrations du mât d'éoliennes en cas de vent avec rafales et de vent avec turbulence. La présence de vibrations entraîne souvent soit une déflexion ultime au sommet du mât d'éolienne, soit une défaillance due à la fatigue du matériau près du bas du mât d'éolienne. De plus, étant donné la nature aléatoire de l'état du vent, il est indispensable d'examiner ce problème d'un point de vue probabiliste. Dans ce travail, un cadre probabiliste d'analyse de la fatigue est développé et amélioré en utilisant le réseau de neurones résiduels. Un système d'amortissement utilisant un amortisseur actif, le Twin Rotor Damper, est conçu pour l'éolienne référentielle NREL 5MW. La conception est optimisée par un algorithme évolutionniste avec une méthode de réglage automatique des paramètres basée sur l'exploitation et l'exploration.
- 6 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Machine learning assisted probabilistic prediction of long-term fatigue damage and vibration reduction of wind turbine tower using active damping system
Utilisation de l'apprentissage automatique pour la prédiction probabiliste à long-terme du dommage par fatigue et la réduction des vibrations d'un mât d'éolienne par un contrôle actif: Cette thèse est consacrée au développement d'un système d'amortissement actif pour la réduction des vibrations du mât d'éoliennes en cas de vent avec rafales et de vent avec turbulence. La présence de vibrations entraîne souvent soit une déflexion ultime au sommet du mât d'éolienne, soit une défaillance due à la fatigue du matériau près du bas du mât d'éolienne. De plus, étant donné la nature aléatoire de l'état du vent, il est indispensable d'examiner ce problème d'un point de vue probabiliste. Dans ce travail, un cadre probabiliste d'analyse de la fatigue est développé et amélioré en utilisant le réseau de neurones résiduels. Un système d'amortissement utilisant un amortisseur actif, le Twin Rotor Damper, est conçu pour l'éolienne référentielle NREL 5MW. La conception est optimisée par un algorithme évolutionniste avec une méthode de réglage automatique des paramètres basée sur l'exploitation et l'exploration.
- 17 October 2022
- Published by Elsevier in Probabilistic Engineering Mechanics
- Vol. 70
A non-parametric evaluation method in staircase test for improving fatigue limit assessment
- 17 October 2022
- Published by Elsevier in Probabilistic Engineering Mechanics
- Vol. 70
A non-parametric evaluation method in staircase test for improving fatigue limit assessment
- 3 January 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (8) , 9193-9211
Online algorithm configuration for differential evolution algorithm
- 3 January 2022
- Published by Springer Nature in Applied Intelligence
- Vol. 52 (8) , 9193-9211
Online algorithm configuration for differential evolution algorithm
- Open Access
- 11 November 2021
- Published by MDPI AG in Infrastructures
- Vol. 6 (11) , 162
Design of an Active Damping System for Vibration Control of Wind Turbine Towers
The vibration of wind turbine towers is relevant to the reliability of the wind turbine structure and the quality of power production. It produces both ultimate loads and fatigue loads threatening structural safety. This paper aims to reduce vibration in wind turbine towers using an active damper named the twin rotor damper (TRD). A single degree of freedom (SDOF) oscillator with the TRD is used to approximate the response of wind turbines under a unidirectional gusty wind with loss of the electrical network. The coincidence between the wind gust and the grid loss is studied to involve the maximum loading on the structure. The performance of the proposed damping system under the maximum loading is then evaluated on the state-of-the-art wind turbine NREL 5 MW. The effectiveness of the TRD is compared to a passive tuned mass damper (TMD) designed with similar requirements. The numerical results reveal that, at the 1st natural mode, the TRD outperforms the passive TMD by three to six times. Moreover, the results show that the TRD is effective in reducing ultimate loads on wind turbine towers.
- 11 November 2021
- Published by MDPI AG in Infrastructures
- Vol. 6 (11) , 162
Open AccessDesign of an Active Damping System for Vibration Control of Wind Turbine Towers
The vibration of wind turbine towers is relevant to the reliability of the wind turbine structure and the quality of power production. It produces both ultimate loads and fatigue loads threatening structural safety. This paper aims to reduce vibration in wind turbine towers using an active damper named the twin rotor damper (TRD). A single degree of freedom (SDOF) oscillator with the TRD is used to approximate the response of wind turbines under a unidirectional gusty wind with loss of the electrical network. The coincidence between the wind gust and the grid loss is studied to involve the maximum loading on the structure. The performance of the proposed damping system under the maximum loading is then evaluated on the state-of-the-art wind turbine NREL 5 MW. The effectiveness of the TRD is compared to a passive tuned mass damper (TMD) designed with similar requirements. The numerical results reveal that, at the 1st natural mode, the TRD outperforms the passive TMD by three to six times. Moreover, the results show that the TRD is effective in reducing ultimate loads on wind turbine towers.