Likun Chen

Parameters estimation in microgrids remains challenging due to ambiguous system dynamics brought by distributed energy resources (DERs) and scarcity of data. This study presents a modified physics-informed neural network (PINN) paradigmdesigned for parameters estimation under such constraints. This paper introduces two main innovations: First, by combining small-signal analysis with the PINN framework for ordinary differential equations, we introduce an adaptable parameter estimation paradigm applicable to different types of DERs in microgrid. Second, we introduce a modified data transformation that reduces training time by up to 82.87% compared to traditional PINN approaches at best. To validate our approach, we conducted simulation on two typical system setups based on open-source real-world microgrid using real-time digital simulation to generate data. We evaluate the proposed method by using an error margin below 5% as a key metric to confirm its robustness and accuracy for different types of DERs. The experimental results demonstrate the effectiveness and adaptability of the proposed method across varying ordinary differential equations to diverse mathematical models. Additionally, suboptimal and failed cases are analyzed and discussed to provide a comprehensive evaluation of the method’s limitations.

The steam turbine and its digital electro-hydraulic (DEH) control system constitute vital elements within thermal power generation. However, the complexity of the on-site environment and the high production costs of the equipment hinder users, especially novices, from fully understanding and mastering the operation mechanisms and production processes. In the realm of emerging technologies, the digital twin stands out as a powerful tool for enhancing industrial training and learning for students and operators in this field. This paper details the design and implementation of a web-based digital twin steam turbine system. Initially, a pioneering web-based digital twin architecture is proposed, featuring high-fidelity equipment modeling, web-based immersive 3D displays, algorithm design and networked implementation, and data-driven model synchronization. Subsequently, the functionalities and benefits of the digital twin system in facilitating industrial training are explained, covering aspects such as steam turbine cognitive learning, DEH system simulation learning, and condition monitoring. Finally, a case study in a real thermal power plant is presented to demonstrate the practicability and effectiveness of this web-based digital twin system. This research endeavors to contribute valuable insights and potential solutions to the growing field of web-based digital twin applications in industry.

To estimate the electrical dynamic characteristics and control coefficients of distributed energy resources (DERs) in microgrids, we introduce an improved physics-informed neural network (PINN) for parameters estimation of microgrid devices. The novelty of our approach lies in two key advancements: First, we introduce a data transformation methodology that significantly accelerates the training process, achieving an increase in speed of up to 82.87% compared to the original PINN. Second, we establish ordinary differential equations (ODEs) frameworks that are adaptable to both inverter-based and synchronous-based DERs, thereby enhancing the generalizability of the proposed method. These advancements enhance the application of PINN in microgrids, paving the way for instructive simulation modeling. For analysis, we established typical microgrid systems using real-time digital simulation (RTDS) to generate the necessary data. This paper explores the effectiveness of PINNs across varying ODEs, showcasing the algorithm's adaptability to diverse mathematical models. Additionally, we illustrate the accuracy of PINN in parameter estimation for both synchronous and inverterbased DERs, with our enhanced approach achieving parameter deductions with less than a 5% error margin in a training session.

This paper presents an edge computing-oriented islanding detection paradigm using multi-support vector machines (SVM) and differential entropy. The proposed methodology can detect islanding events and locate which microgrid was disconnected in a distribution system with microgrid cluster. The proposed method employs multi-SVMs in the cloud to detect and locate potential islanding events in power distribution systems. The accuracy and reliability of the results of SVMs are enhanced by introducing differential entropy for quantitative evaluation. Additionally, the method decomposes the SVM inference process at the edge, enabling edge devices with low computational power and limited resources to support data-driven algorithms without compromising accuracy and speed. An extensive study is performed on the modified IEEE123 distribution system established in real-time digital simulation. Simulation results elucidate that the proposed method can accurately and promptly detect system islanding operations with zero non-detection zone. Comparison with other methods illustrates the superiority of the proposed method in terms of discrimination accuracy, detection time, and reliability.