Search

• A digital twin (DT) is constructed for damaged detection in structures. • There is an integration of physics-based models with machine learning. • This strategy allows interpretablibility (physics) and speed (ML). • The influence of several parameters on the DT accuracy in analyzed. • Conclusions might be helpful for a large range of applications. This work is interested in digital twins, and the development of a simplified framework for them, in the context of dynamical systems. Digital twin is an ingenious concept that helps on organizing different areas of expertise aiming at supporting engineering decisions related to a specific asset; it articulates computational models, sensors, learning, real time analysis, diagnosis, prognosis, and so on. In this framework, and to leverage its capacity, we explore the integration of physics-based models with machine learning. A digital twin is constructed for a damaged structure, where a discrete physics-based computational model is employed to investigate several damage scenarios. A machine learning classifier, that serves as the digital twin, is trained with data taken from a stochastic computational model. This strategy allows the use of an interpretable model (physics-based) to build a fast digital twin (machine learning) that will be connected to the physical twin to support real time engineering decisions. Different classifiers (quadratic discriminant, support vector machines, etc) are tested, and different model parameters (number of sensors, level of noise, damage intensity, uncertainty, operational parameters, etc) are considered to construct datasets for the training. The accuracy of the digital twin depends on the scenario analyzed. Through the chosen application, we are able to emphasize each step of a digital twin construction, including the possibility of integrating physics-based models with machine learning. The different scenarios explored yield conclusions that might be helpful for a large range of applications. [ABSTRACT FROM AUTHOR]

• A higher accuracy engine performance simulation model is developed and calibrated. • A GT-Simulink coupling simulation platform with NSGA II is established. • The SVM model is established and calibrated by the experimental data. • Integrated platform for simulation-optimization and verification is developed. • Maximum fuel consumption and NOx reduction rates reach 5.0% and 70% respectively. This paper is pioneered in developing digital twins by GT-Power software and multi-objective evolutionary optimization (MOEO) using NSGA II algorithm for an Atkinson cycle gasoline engine under China VI emissions standards. Firstly, an experimental investigation is conducted and relevant experimental data is obtained. Based on this, the corresponding 1D GT-Power simulation model is established and calibrated according to the obtained test data. Secondly, the four decision variables including the spark advance angle (SA), exhaust gas recirculation (EGR) rate, exhaust variable valve timing (VVT-E) and intake variable valve timing (VVT-I) are input into the simulation model, the optimal values of the decision variables will be determined via MOEO to minimize NOx emissions and the brake specific fuel consumption (BSFC). Thirdly, based on the data obtained from the scanning test, a machine learning method is used to build an engine performance prediction model through the support vector machine (SVM) regression algorithm. The inputs (control parameters obtained from the optimization process) including SA, EGR, VVT-I and VVT-E are imported to predict the performance output of the engine. The results show that under the engine control parameters obtained by the NSGA II algorithm, the simulation values of engine performance parameters have been greatly optimized, the decreasing extent of fuel consumption is about 5.0%, besides, the decreasing extent of NOx is about 70%. What is more, the increased EER and EEE is up to 6.21% and 2.26%, respectively. And then most of the predicted values obtained by machine learning have been optimized. For BSFC, in general, the simulation value and the predicted value are in good agreement at the smaller value, indicating that the simulation model and the regression prediction model basically achieve the same value at the lower BSFC of the engine. For NOx, the simulated and predicted values have all been optimized. Furthermore, the method and platform developed in this paper will help to carry out a series of related work in the field of vehicle energy flow distribution and optimization when changing different control strategies and optimization methods in the future. Besides, the above work provides a reliable theoretical basis and digital model support for the development of energy-saving and efficient Atkinson cycle engines, which further drives the application of Atkinson cycle engines in new energy vehicles. [ABSTRACT FROM AUTHOR]

• Digital twin for multi-timescale dynamical system is proposed. • The approach uses fuses system physics with data-based approach. • ME-GP is proposed for tracking time-evolution of system parameters • The proposed approach can accurately predict future responses Digital twin technology has a huge potential for widespread applications in different industrial sectors such as infrastructure, aerospace, and automotive. However, practical adoptions of this technology have been slower, mainly due to a lack of application-specific details. Here we focus on a digital twin framework for linear single-degree-of-freedom structural dynamic systems evolving in two different operational time scales in addition to its intrinsic dynamic time-scale. Our approach strategically separates into two components – (a) a physics-based nominal model for data processing and response predictions, and (b) a data-driven machine learning model for the time-evolution of the system parameters. The physics-based nominal model is system-specific and selected based on the problem under consideration. On the other hand, the data-driven machine learning model is generic. For tracking the multi-timescale evolution of the system parameters, we propose to exploit a mixture of experts as the data-driven model. Within the mixture of experts model, Gaussian Process (GP) is used as the expert model. The primary idea is to let each expert track the evolution of the system parameters at a single time-scale. For learning the hyperparameters of the 'mixture of experts using GP', an efficient framework that exploits expectation-maximization and sequential Monte Carlo sampler is used. Performance of the digital twin is illustrated on a multi-timescale dynamical system with stiffness and/or mass variations. The digital twin is found to be robust and yields reasonably accurate results. One exciting feature of the proposed digital twin is its capability to provide reasonable predictions at future time-steps. Aspects related to the data quality and data quantity are also investigated. [ABSTRACT FROM AUTHOR]

The objective of this work is to model and simulate aerial drops of fire retardants in dangerous fire environments. Specifically, the work develops a computational framework for a model problem combining: • [1.] A meshless discrete element component that tracks the trajectory of released airborne materials from a controlled aircraft, ranging from retardant powders to encapsulated packets, subjected to prevailing wind velocities and fire-driven updrafts. • [2.] A Machine Learning Algorithm (MLA) to rapidly ascertain the optimal aircraft (unmanned or manned) dynamics to maximize the fire-retardant release effectiveness (released material usage and target impact). The framework is designed to enable Digital Twin type technologies, i.e. digital replicas that run in real time with the physical system. However, it is also designed to run at much faster rates, in order to enable MLA's to optimize the planning, by running quickly on laptops and mobile systems. The overall guiding motivation is to provide a useful tool to enable rapid flight-path planning for aerial first-responders in real-time and to train pilots. Numerical examples are provided to illustrate the process. [ABSTRACT FROM AUTHOR]

To fulfill increasingly difficult and demanding tasks in the ever-changing complex world, intelligent industrial products are to be developed with higher flexibility and adaptability. Digital twin (DT) brings about a possible means, due to its ability to provide candidate behavior adjustments based on received "feedbacks" from its physical part. However, such candidate adjustments are deterministic, and thus lack of flexibility and adaptability. To address such problem, in this paper an extended concept – evolutionary digital twin (EDT) and an EDT-based new mode for intelligent industrial product development has been proposed. With our proposed EDT, a more precise approximated model of the physical world could be established through supervised learning, based on which the collaborative exploration for optimal policies via parallel simulation in multiple cyberspaces could be performed through reinforcement learning. Hence, more flexibility and adaptability could be brought to industrial products through machine learning (such as supervised learning and reinforcement learning) based self-evolution. As a primary verification of the effectiveness of our proposed approach, a case study has been carried out. The experimental results have well confirmed the effectiveness of our EDT based development mode. [ABSTRACT FROM AUTHOR]

A key aspect for achieving a high-accuracy Photovoltaic (PV) power simulation, and reliable digital twins, is a detailed description of the PV system itself. However, such information is not always accurate, complete, or even available. This work presents a novel approach to learn features of unknown PV systems or subsystems using genetic algorithm optimization. Based on measured PV power, this approach learns and optimizes seven PV system parameters: nominal power, tilt and azimuth angles, albedo, irradiance and temperature dependency, and the ratio of nominal module to nominal inverter power (DC/AC ratio). By optimizing these parameters, we create a digital twin that accurately reflects the actual properties and behaviors of the unknown PV systems or subsystems. To develop this approach, on-site measured power, ambient temperature, and satellite-derived irradiance of a PV system located in south-west Germany are used. The approach proposed here achieves a mean bias error of about 10% for nominal power, 3° for azimuth and tilt angles, between 0.01%/C and 0.09%/C for temperature coefficient, and now-casts with an accuracy of around 6%. In summary, we present a new solution to parametrize and simulate PV systems accurately with limited or no previous knowledge of their properties and features. [ABSTRACT FROM AUTHOR]

This paper presents a method for detecting damage of a bottom-set gillnet based on sensor fusion and the Artificial Neural Network (ANN) model. In this regard, time-domain numerical simulations for a 300-m-long bottom-set gillnet with an equivalent-drag-net model are extensively performed. Various wave conditions as well as numerous damaged scenarios are considered in the numerical simulations, and extensive data are collected for the training and testing of the ANN-based machine-learning scheme. In training, representative sea states, net-assembly accelerations, and location-buoy displacements are selected as the input variables. The back-propagation learning algorithm is employed for training to maximize the damage-detection performance. The output of the ANN model is the identification of the type and location of the damaged net. The damage-detection capability is significantly enhanced by employing the moving standard deviation and median filter. The well-trained ANN models are shown to accurately (96% correct) detect the damage of the net even for the sea states not included in training. This study demonstrates that the real-time automatic monitoring of the underwater net systems can be designed by using the digital-twins technology with the applied ANN model and machine-learning scheme for the detection of damages and malfunctions. • The well-trained ANN models are shown to accurately (96% correct) detect the damage of the fishing net. • The number of neurons, median filter, and moving standard deviation can improve the performance of the ANN model. • The sensor-based fishing-net monitoring system can be designed by digital-twins technology with the applied ANN model. [ABSTRACT FROM AUTHOR]

Smart technologies are advancing, and smart cities can be made smarter by increasing the connectivity and interactions of humans, the environment, and smart devices. This paper discusses selective technologies that can potentially contribute to developing an intelligent environment and smarter cities. While the connectivity and efficiency of smart cities is important, the analysis of the impact of construction development and large projects in the city is crucial to decision and policy makers, before the project is approved. This raises the question of assessing the impact of a new infrastructure project on the community prior to its commencement—what type of technologies can potentially be used for creating a virtual representation of the city? How can a smart city be improved by utilizing these technologies? There are a wide range of technologies and applications available but understanding their function, interoperability, and compatibility with the community requires more discussion around system designs and architecture. These questions can be the basis of developing an agenda for further investigations. In particular, the need for advanced tools such as mobile scanners, Geospatial Artificial Intelligence, Unmanned Aerial Vehicles, Geospatial Augmented Reality apps, Light Detection, and Ranging in smart cities is discussed. In line with smart city technology development, this Special Issue includes eight accepted articles covering trending topics, which are briefly reviewed. [ABSTRACT FROM AUTHOR]