Current Projects
Digital Twin of Pedestrian Bridges for Impact Detection and Rapid Condition Assessment
NSF Small Business Innovation Research (Collaborated with Dr. Kirill Mechitov and Prof. B.F. Spencer, Jr. )
A digital twin of pedestrian bridge is built to represent the real bridge in terms of dynamic responses. It starts with an initial FE model based on the drawings in ABAQUS. It is then updated using field test data to match the dynamic properties with the real bridge. In particular, a full-scale dynamic test should be performed to obtain the modal properties. Model updating is then applied to optimize the dimensions and other parameters of the FE model. On the other hand, the updated digital twin model is then help to serves two purposes: 1) to build the learning database for neural network modeling and training, with the objective of impact localization and impact force estimation, and 2) to obtain maximum allowable impact forces at different bridge locations through nonlinear analysis, using as the reference for rapid bridge impact condition assessment. The proposed system is demonstrated through a full-scale field test.
-
Fu, Y., Zhu, Y., Hoang, T., Mechitov, K., & Spencer, B. F. (2022). "xImpact: Intelligent Wireless System for Cost-Effective Rapid Condition Assessment of Bridges under Impacts", Sensors, 22(15), 5701.
A Smart Digital Twin Framework for Monitoring and Management of Underground Transportation Infrastructure
MOE Tier 1 Project
The aims of the project will be achieved by developing and implementing a smart digital-twin framework for the railroad tunnels, with heterogenous digital modelling and hierarchy data fusion. The conceptual framework is a closed cyber-physical loop, including physical structures (i.e., tunnels and physical testbed), cyber counterparts (i.e., surrogate models and physical-based deep learning models), and two interfaces (i.e., information and intervention). The information interface is extracted by smart sensing systems and multi-fidelity model data from physical tunnels and fed into digital models. The intervention is generated by cyber counterparts and applied on physical tunnels. It is evaluated and selected by engineers for optimized predictive maintenance and rail traffic management.
NTU Testbed for Digital Construction
NTU Start-up Grant
This project is aimed to develop a multi-purpose reconfigurable testbed (aka, NTU Testbed) to support the design, analysis, deployment, and validation of interdisciplinary solutions for digital construction and automation in Singapore. Particularly, built upon the facilities at PE Lab, we will collect spaces/slopes for both building construction demo and underground construction demo. In addition, we will setup a framework for data collection and visualization, built upon a network of wireless/wired sensor/cameras surrounding the lab for both situational awareness and project demonstration. It is designed as a focal point among research, industry and education fields, attracting top researchers, students, and engineers in Singapore for both current and future (next-generation) digital construction development.
A novel framework for occupant comfort and building energy management to accelerate decarbonization
Imperial-NTU Collaboration Fund (Collaborated with Dr Aruna Sivakumar, link)
The objective of this proposal is to develop a novel framework to evaluate the thermal comfort, awareness, and attitude of occupants toward smart applications in buildings, in order to reduce operational carbon emissions as part of global efforts for decarbonisation. To achieve the objective, we will combine the strengths from ICL and NTU teams and develop the framework in two aspects: (i) the NTU team will develop an energy data collection and visualisation platform to collect, transmit and analyse office environment data; (ii) the ICL team will develop models of complex occupant behaviours including occupant preferences of manual/automated HVAC system, thermal comfort prediction with indoor sensors. The data collected from the platform (by the NTU team) will feed into the occupant models for calibration, evaluation, and upgrading, while the occupant models (by the ICL team) will be further integrated into the platform operating in the cloud to establish a novel framework for operational carbon reduction in buildings.
Novel Context-Aware Multivariate Time Series Modelling for Underground Transportation Infrastructure Monitoring
AI Singapore (Collborated with Prof. Yang Yaowen and Prof. Cong Gao, link)
The objective of this project is to develop and implement novel artificial intelligence (AI) techniques on smart sensing systems to detect, diagnose and predict potential faults in high-risk underground transportation infrastructure. We focus on railroad tunnels, as they are critical components of the underground infrastructure, and their main faults include cracks, seepages, concrete spalling, and tunnel joint failures. The proposed AI-based smart sensing system will address the needs of tunnel owners and engineering services companies by providing actionable and timely condition assessment of tunnels under both sudden events (e.g., ground tremors or nearby construction) and long-term deterioration (e.g., ground deformation). Achieving accurate and reliable diagnosis/prognosis of various source fault information from a massive amount of low-quality multivariate data is a primary challenge. Our key innovation lies in context-aware data imputation to reconstruct the missing/noisy data and explainable deep learning models to diagnose/predict tunnel faults.
Real-time Construction Inspection and Quality Assurance using Mixed Reality and Deep Learning.
NTU Start-up Grant
This study unveils a groundbreaking pipeline termed as "Spatially Enhanced Geometric Segmentation for Augmented Measurement (SegSAM)". This framework embodies the convergence of Augmented Reality (AR) and Human-Machine Interaction (HMI), revolutionizing on-site quantification processes. The SegSAM pipeline empowers the AR device to discern the targeted surfaces, isolate the targeted elements from their background, quantify their dimensions accurately, and ultimately, superimpose virtual contour holograms onto the precise locations within a 3D space to offer a visualization of dimensions. A cornerstone of this pipeline is the integration and further enhancement of a latest deep learning-based image segmentation algorism, Fast Segment Anything Model (FastSAM). This inclusion significantly bolsters the pipeline's ability to distinguish expected targets from surrounding elements seamlessly. This innovation mitigates existing hardware limitations, paving the way for a broader application of our methodology.
System of Systems Modeling and Testbed for the Resilient Design of Deep Space Habitats
Space Technology Research Institutes Grant (Collaboration with Prof. Shirley Dyke) (Media Link)
As NASA, ESA, and private sectors (e.g., SpaceX) make plans for space settlement on the Moon or Mars, developing smart habitats is of vital importance to enable sustainable long-term presence in deep space. The futuristic infrastructure must be resilient, functioning as intended under continuous extreme conditions (e.g., meteoroid impacts, vibrations, etc.). The mission of the Resilient ExtraTerrestrial Habitats institute (RETHi) is to develop next-generation technologies, that will anticipate and adapt to extreme conditions, actively detect and diagnose habitat faults, and rapidly recover from disruptions using autonomous robots. As one of key personnel, my main responsibilities are to simulate the integrated deep space habitat as a system of systems, capturing the emergent dynamic behavior and cascading events subjected to multi-hazard environment. In parallel, I am coordinating the development of a multi-physics cyber-physical testbed to enable unprecedented testing of resilient strategies at scale.
-
Fu, Y., Montoya, H., Maghareh, A., Dyke, S. “Modular Coupled Virtual Testbed: A Real-time Platform for Cyber-physical Testing of Extraterrestrial Habitat Systems”, (in preparation).
Fault Detection and Diagnostics for Meteorite Impacts on Deep Space Habitats
Due to their unpredictable nature, many impact events (e.g., overheight vehicles striking on bridges) go unnoticed or get reported many hours later. However, they can induce structural failures or hidden damage that accelerates the structure's long-term degradation. Therefore, prompt impact detection and localization strategies are essential for early warning of impact events and rapid maintenance of structures. In particular, due to the harsh environment, structural impact localization must be robust to a limited number of sensors and multi-source errors (e.g., measurement errors). In this study, an effective impact localization strategy is proposed to identify impact locations using a limited number of vibration measurements, especially in harsh environments (e.g. in deep space). Convolutional neural networks are trained for each sensor node and are fused using Bayesian theory to improve the accuracy of impact localization.
-
Fu, Y., Wang, Z., Maghareh, A., Dyke, S., Jahanshahi, M., Shahriar, A. (2022). "Scalable Impact Detection and Localization Using Deep Learning and Information Fusion", Proc. 13th IWSHM, Stanford, CA, USA.
Smart IoT System for Rapid Damage Assessment of Aging Infrastructure under Multiple Hazards
Many of the civil infrastructure damage scenarios involve unpredictable natural disasters (e.g., earthquakes) or anthropogenic hazards (e.g., blasts). An efficient monitoring system is thus critical for both early warning of hazards and rapid condition assessment for engineers to make informed decisions for maintenance, thus maximizing their contribution to environmental and social needs. We developed a smart wireless IoT system. When deployed on in-service structures, hazards of interest will be automatically assessed and reported in real-time. The key component is the demand-based wireless smart sensors, which can capture high-fidelity transient structural responses, with minimal power budget and zero latency. In addition, it addresses the challenges of remote data retrieval by integrating 4G-LTE functionality into the sensor network and completes the data pipeline with cloud-based data management. The versatile systems were installed on over ten railroad bridges of the Canadian National Railway for over one month.
-
Fu, Y., Hoang, T., Mechitov, K., Spencer Jr, B.F., Kim, J. (2018). “Sudden Event Monitoring of Civil Infrastructure using Demand-based Wireless Smart Sensors”, Sensors, 18(12), 4480.
-
Fu, Y., Hoang, T., Mechitov, K., Spencer Jr, B.F. “xShake: Intelligent Wireless System for Cost-effective Real-time Seismic Monitoring”, Smart Structures and Systems, Smart Structures and Systems, 28(4), 483-497.
Instability Monitoring of Space Grid Structures under Blizzards
Space grid structures have been widely used as important large-scale public buildings, which are now increasingly applied for railway stations in China, such as Hangzhou East Railway Station. However, in history, a series of space grid structures collapsed under blizzards, initiating from the sudden buckling of individual members and developing into progressive collapse in seconds. Therefore, the conditions of space grid structures should be timely monitored and reliably assessed under blizzards, such that instability can be detected and emergency response can be made before collapse occurs. In this study, two early warning strategies are proposed for instability monitoring of space grid structures based on local/global vibration information, such that an alert will be sent to the maintenance crew before buckling occurs. The result shows that the strategies are able to predict and identify dangerous members and hence send early warning of instability in space grid structures before buckling occurs.
-
Fu, Y., Gomez, F., Spencer Jr, B.F. (2018). “Instability Monitoring of Space Grid Structures under Blizzards”, Proc. 7th World Conference on Structural Control and Monitoring, Qingdao, China.
Research on the Performance of CFRP-strengthened Tubular Gap K-joints
An efficient technique of Carbon Fiber Reinforced Polymer (CFRP) application was proposed to promote the joint capacity of general tubular K-joints fabricated from Circular Hollow Section (CHS) members. Using this technique, in order to understand the static performance of CFRP-strengthened CHS joints, a systematic investigation was carried out by means of both experiments and the finite element method. Three CHS gap K-joints strengthened with CFRP sheets were tested under static axial force in braces, whilst one additional joint was served as a reference joint without CFRP. A series of finite element models were developed and validated for the joints with and without CFRP reinforcement. A parametric study was conducted to evaluate the effect of variables (length, layers, and mechanical properties of CFRP) on load-bearing capacity. Finally, formulas were proposed for calculating the ultimate load-bearing capacity of CHS gap K-joints with CFRP composites, and their calculation results matched well with the experimental and numerical results respectively
-
Fu, Y., Tong, L., He, L., Zhao, X.L. (2016). “Experimental and Numerical Investigation on Behavior of CFRP-Strengthened Circular Hollow Section Gap K-Joints.” Thin-Walled Structures, 102, 80-97.
Study on the Innovative Application of Bamboo-Cable Composite Structures
In the 2010 Shanghai Expo, people were deeply impressed by the Sun Valley. However, due to large energy and labor consumption, the Valley is not consistent with the Expo slogan of sustainability. What if the steel is replaced with bamboo? Possessing excellent mechanical properties, bamboo has been nowadays recognized as one of the most sustainable potential structural materials. However, the irregularity in cross-sections and the inefficient joint configuration could be bottlenecks in developing future large-span bamboo structures. A novel spatial composite structure is proposed with the methodology of bamboo-cable structural systems which consist of bamboo, steel elements, and adhesive construction materials. Meanwhile, key technical difficulties involved with this application are carefully investigated and analyzed, which we target to address in the near future. Additionally, several tentative structural styles are presented in order to explore the application of this bamboo-composite structure.
-
Fu, Y., Hu, L., Hu, Y., He, X., (2014). “Study on the Innovative Application of Bamboo-Cable Composite Structures.” XXV International Union of Architects World Congress, Durban, South Africa.
Congratulations to Professor Fu on becoming a committee member of the American Society of Civil Engineers, Engineering Mechanics Institute - Structural Health Monitoring & Control! (link)
IoT Sensor Prototypes
Xnode:
The Xnode Smart Sensor is designed as the next-gen solution for high fidelity distributed sensing. This modular and versatile sensor platform enables wireless data acquisition and processing for data-intensive applications (high resolution, high sampling rate) such as structural health monitoring, manufacturing and industrial equipment monitoring, and seismic sensing. In practice, Xnodes are usually used in form of Wireless Sensor Network (WSN), typically 1 gateway nodes and several sensor nodes. Each sensor node has 8 available channels for data collection, of which the first 3 channels are typically used for acceleration measurement. Raw data and local-processed data are collected by the gateway node and then sent back to the cloud server for applications via cellular communication.
RoomWisor:
This project aims to create an innovative framework for evaluating occupants' thermal comfort concerning smart building applications, aligning with global initiatives to reduce operational carbon emissions. The pivotal role of Roomsensors in this endeavour is to enhance energy efficiency and occupant comfort. These sensors are designed to accurately measure various indoor parameters, including light intensity, carbon dioxide concentration, humidity, and temperature. Their adeptness extends to detecting human presence and activity through the use of microphone and camera sensors. The system employs Arduino and Raspberry Pi for seamless data collection and transmission, ensuring integration with the data storage and visualisation infrastructure.
LiftNode:
A multifunctional integrated sensor is developed for construction monitoring, including vibration, vision and strain sensors. STM32H743 with 480 MHz will serve as the main micro controller unit (MCU) for data acquisition and host complex application, and a 4G module with GPS navigation is utilized for wireless data transmission. On MCU the FreeRTOS operation system is installed for scheduling multi-task at run time, including power management, heterogeneous sensing synchronization and data transmission, to achieve high quality measurement, and light AI algorithms can be also deployed for various monitoring requirements, such as anomaly detection, data imputation and denoising.
