Guest Editors: Bin He (Shanghai University), Yu Song (Delft University of Technology), and Yan Wang (Georgia Institute of Technology)
J. Comput. Inf. Sci. Eng. Jun 2021, 21(3): 030301 (2 pages)Paper No: JCISE-21-1137 https://doi.org/10.1115/1.4050982
Published Online: May 11, 2021
A digital twin (DT) is the real-time digital replica of a physical entity and system. It enables the seamless integration between digital models and physical devices so that the operation, monitoring, control, and upgrade of the system, as well as personnel training, can be performed in a cyber-physical mixture mode. DTs integrate technologies such as multiphysics multiscale modeling, Internet of Things, smart sensing, machine learning, and model-based control and act as a bridge between the physical world and the virtual world by mapping the whole life cycle of physical systems with realtime sensor data, and maintaining the complete digital trace. In the era of Industry 4.0, DT is becoming a powerful engine in the intelligent design of products and intelligent manufacturing. DT enables data-driven design and optimization, evidence-based sustainable design, real-time diagnostics and prognostics, plug-n-play customization, and modular improvement. This special issue aims to harvest the latest efforts in fundamental methodologies as well as their applications in DT-driven design and manufacturing. The papers provide the highlights of several topics including fundamental advances in DT technologies, DT-enabled intelligent design and smart manufacturing, and DT-enabled datadriven product sustainable design.
J. Comput. Inf. Sci. Eng. June 2021, Volume 21, Issue 3
A digital twin (DT) is the real-time digital replica of a physical entity and system. It enables the seamless integration between digital models and physical devices so that the operation, monitoring, control, and upgrade of the system, as well as personnel training, can be performed in a cyber-physical mixture mode. DTs integrate technologies such as multiphysics multiscale modeling, Internet of Things, smart sensing, machine learning, and model-based control and act as a bridge between the physical world and the virtual world by mapping the whole life cycle of physical systems with real-time sensor data, and maintaining the complete digital trace. In the era of Industry 4.0, DT is becoming a powerful engine in the intelligent design of products and intelligent manufacturing.
DT enables data-driven design and optimization, evidence-based sustainable design, real-time diagnostics and prognostics, plug-n-play customization, and modular improvement. This special issue aims to harvest the latest efforts in fundamental methodologies as well as their applications in DT-driven design and manufacturing. The papers provide the highlights of several topics including fundamental advances in DT technologies, DT-enabled intelligent design and smart manufacturing, and DT-enabled data-driven product sustainable design.
- Fundamental advances in DT technologies
The paper of Juarez et al., entitled “Digital Twins: Review and Challenges,” provides a historical view of DT technology evolution that includes product lifecycle data management (e.g., data exchange standards, ontology, and data fusion), network communication protocols (e.g., middleware), and multi-agent systems for data acquisition and transmission, as well as asset management and task scheduling. Based on the automation levels of information sharing between digital and physical worlds, digital model, digital shadow, and digital twin are differentiated. The enabling technologies for the applications such as product design, manufacturing systems, smart farm, and transportation are also compared.
Nagargoje et al. in “Performance Evaluation of the Data Clustering Techniques and Cluster Validity Indices for Efficient Toolpath Development for Incremental Sheet Forming” report their work of comparing four clustering techniques, namely, partition-based clustering (K-means), density-based clustering (DBSCAN), hierarchical agglomerative clustering, and graph-based clustering (spectral) for feature-based tool path development for arbitrary geometries in the case of incremental forming. Seven different variants of hierarchical clustering based on the linkage formation are also compared. To evaluate the quality of the clustering solutions and to find the best validity indices, the Calinski-Harabasz, Ball-Hall, Davies-Bouldin, Dunn, Det Ratio, Silhouette, Trace WiB, and Log Det Ratio are further compared. This study helps to establish that the data clustering techniques can be gainfully used to evolve algorithms for feature-based toolpath development strategies for various manufacturing processes, such as single-point incremental forming and double-sided incremental forming, by using data science and machine learning techniques.
- DT-enabled intelligent design and smart manufacturing
The paper by Feng et al., entitled “A Digital Twin-Driven Method for Product Performance Evaluation Based on Intelligent Psycho-Physiological Analysis,” proposes a DT framework that embeds human factors, e.g., psychological feedback of the customers and experts’ opinion, for evaluating product performance. The feasibility and effectiveness of the proposed framework are demonstrated by a case study where electroencephalogram data and subjective measures of participants, and physical data of the products and environment are synthesized to provide knowledge for decision-making using a novel machine learning-based method.
In the paper by Guo et al., entitled “Real-Time Prediction of Remaining Useful Life and Preventive Maintenance Strategy Based on Digital Twin,” a DT of machine degradation is built to perform preventive maintenance. The degradation of machines is modeled as nonlinear drifted Brownian motion and the evolution of probability of failure is predicted. The model parameters for drift and diffusion coefficients are obtained from the maximum likelihood estimations based on sensor data. The Bayesian approach is applied for online model update with new data. The preventive maintenance can be done by minimizing the expected maintenance cost in unit time. The method is demonstrated with the monitoring of machining centers under different working conditions.
“Multidimensional Data Modeling and Model Validation for Digital Twin Workshop” by Qian et al., proposes multidimensional data modeling and model validation methods for digital twin workshop (DTW). Five-order tensor models for representing manufacturing elements are established to unify the data from physical workshop and virtual workshop. Then, the mathematical method for verifying DTW twin model is proposed from the recessive and explicit perspective. Finally, a case study of an aerospace machining workshop is carried out to verify the operability and effectiveness of the proposed method. Results have shown that the proposed methods can effectively evaluate whether the twin model accurately provides the description of the actual behavior process of physical workshop.
In the paper by Yang et al., entitled “Posture-Invariant Three-Dimensional Human Hand Statistical Shape Model,” a posture-invariant human hand statistical shape model is presented. In the construction of the model, individual scans are aligned using a Möbius sphere-based algorithm and the postures are corrected using an articulated skeleton. Different methods for constructing the statistical model, e.g., kernel principal component analysis, principal component analysis, and independent component analysis, are studied and the resulting statistical shape model is able to approach 3D human hand shapes with high accuracy, and it can be used as part of a digital human model in human integrated DT applications.
In “Simulation-Based Hybrid Optimization Method for the Digital Twin of Garment Production Lines,” Jung et al. report their efforts by proposing a simulation-based hybrid optimization method to maximize the productivity of a garment production line. The simulation reflects the actual site characteristics, i.e., process and operator level indices, and the optimization process reflects constraints based on expert knowledge. The optimization process derives an optimal operator sequence through a genetic algorithm and sequentially removes bottlenecks through workload analysis based on the results. The correlation between workload and production is verified by analyzing the workload change trends.
The paper by Cai et al., entitled “Quality Deviation Control for Aircraft Using Digital Twin,” presents a DT for quality deviation control of aircraft final assembly based on data analysis and 3D visualization technology. Through the DT based on asset management shell technology, the multi-source and heterogeneous quality deviation data can be extracted and integrated. Furthermore, a quality deviation system is built based on the DT where the aircraft quality deviation data can be analyzed by the FP-growth association rule algorithm, and the results are provided through the system to guide the assembly site, improving the efficiency and accuracy of quality problem-solving in the physical world.
In the paper by Lu et al., entitled “Physics-Based Compressive Sensing to Enable Digital Twins of Additive Manufacturing Processes,” a sensing scheme that seamlessly integrates physical models and compressed sensing is proposed to enable DT update from a limited amount of sensor data to overcome the limitations of sensor accessibility and communication bandwidth. The physical model is reconstructed from a small number of sensor measurements at the boundaries. The reconstruction is realized by solving inverse problems with a new algorithm that takes advantage of sparsity. The new scheme is demonstrated with temperature monitoring of additive manufacturing.
“Digital Twin-Driven Rapid Customized Design of Board-Type Furniture Production Line” by Yan et al. proposes a rapid customized design method for developing new board-type furniture production lines based on the DT technology. A production line design platform is developed based on the DT model, which can parallelize the design process and reduce the design cycle. In addition, five key enabling technologies are introduced to provide the theoretical fundamentals for implementing DT-based manufacturing system design. Finally, a board-type furniture production line is presented as a case study to verify the effectiveness of the method.
In the paper by Zhang et al., entitled “Tool Wear Online Monitoring Method Based on DT and SSAE-PHMM,” a DT framework for online manufacturing tool wear monitoring is presented, where sensor data and digital models of tool wear are integrated. A stack sparse auto-encoder is employed for feature extraction from vibration signals collected by sensors. A parallel hidden Markov model is developed to recognize the tool wear states based on multidimensional feature vectors so that the classification can be more accurate. High-fidelity virtual environment is used for online visualization and provides qualitative basis for tool wear monitoring.
- DT-enabled data-driven product sustainable design
He et al. in “Digital Twin-Driven Remaining Useful Life Prediction for Gear Performance Degradation: A Review” review DT-driven remaining useful life prediction methods for gear performance degradation, from the view of DT-driven physical model-based and virtual model-based prediction method. From the view of physical model-based one, it includes a prediction model based on gear crack, gear fatigue, gear surface scratch, gear tooth breakage, and gear permanent deformation. From the view of DT-driven virtual model-based one, it includes nondeep learning methods and deep learning methods. In the end, they combine the physical model and virtual model of the gear to build a DT model of gear performance degradation and life prediction.
The paper by Gao et al., entitled “A Deep Lifelong Learning Method for Digital Twin-Driven Defect Recognition With Novel Classes,” presents a lifelong learning approach that can recognize new classes in classification problems to detect surface defects, given that traditional classification methods can only recognize pre-determined classes. A support vector data descriptor as the detector can detect new classes. The information of features is channeled to a neural network classifier as the recognizer, where a weight imprinting or semantic embedding scheme is applied so that the features of new classes are embedded in additional weights of the neural network and the feature space is expanded on the fly. The method is also experimentally compared with other incremental classifiers.
“Digital Twin-Driven Controller Tuning Method for Dynamics” by He et al., proposes a DT-driven proportion integration differentiation (PID) controller tuning method for dynamics. The structure and operation mechanism of the DT model for PID controller tuning are described. Taking the advantages of virtual real mapping and data fusion of the DT model, combined with the online identification of the controlled object model, the problems of real-time feedback of an actual control effect of the controller and the unreal virtual model of the control system caused by time-varying working conditions are effectively solved. Genetic algorithm is integrated to tune the PID controller parameters to improve the efficiency and accuracy. Finally, the controller tuning for gear transmission stability is taken as an example to verify the practicability of the proposed method.
The paper by Demirel et al., entitled “Digital Twin-Driven Human-Centered Design Frameworks for Meeting Sustainability Objectives,” demonstrates a simulation-based computational design methodology, digital human-in-the-loop (D-HIL). The D-HIL enables human, product, and systems data to be coupled with computational toolkits, bringing opportunities for facilitating ergonomics decision making to be part of the DT-driven design. In the D-HIL, a prototyping toolbox and the human error and functional failure reasoning framework are presented as means for realizing sustainability goals through DT-based design.
Through worldwide dissemination of the special issue’s call for papers, we received a large number of submissions. After a minimum of two rounds of reviews, a total of 15 papers were accepted as this special issue for publication. The guest editors would like to thank all contributing authors for their excellent work. We are very grateful to the reviewers for their precious time and efforts to help finish the review process and for offering constructive comments to authors. Particularly, we would like to express our sincere appreciation to Prof. S. K. Gupta, the Editor-in-Chief of JCISE, for providing the opportunity and his support of this special issue. We also thank Ms. Amy Suski for editorial assistance. Without all of you, this special issue would not have been possible.
Copyright © 2021 by ASME
