IFAC-PapersOnLine最新文献

The rapid deployment of renewable energy resources has led to the widespread use of power electronics in modern power systems. As these systems transition from being dominated by large synchronous machines to increasingly incorporating inverter-based resources (IBRs), traditional transmission simulation tools that do not model the dynamics of distribution networks with a large amount of Distributed Energy Resources (DERs) are becoming inadequate. Addressing this challenge, this paper introduces a scalable phasor-domain transmission and distribution (T&D) co-simulation framework that accurately captures system dynamic behaviors under various configurations of grid-forming and grid-following inverters based on open-source software. The main focus is on the validation of this co-simulation framework against the PSCAD Electromagnetic Transient (EMT) analysis tool for a three-phase line-to-ground fault scenario. The validation results clearly demonstrate the framework’s high fidelity and a computational time speed-up of 60 to 100 times, marking a pioneering validation effort between phasor-domain and EMT simulation in T&D co-simulation research.

This paper describes a digital sliding mode control (SMC) technique applied to a three-phase four-wire rectifier operating at a fixed frequency for ultra-fast charging of electric vehicle (EV) battery. The control algorithm employs three decoupled sliding mode controllers to achieve loss-free resistor (LFR) behavior in each phase for power factor correction (PFC). The design of the sliding mode controller is twofold. The first one is to guarantee convergence of the sliding variable to zero. The equivalent control and the discrete-time dynamic model of the rectifier are obtained by imposing sliding-mode regime in discrete-time. The second one is to stabilize the inner-loop under the obtained control law. Theoretically, the resulting inner-loop is stable with a deadbeat behavior in the inner current loop. The results are validated by numerical simulations using on a 350 kW AC-DC rectifier for EV battery ultra-fast charging applications. The numerical simulation results performed on the switched model implemented in PSIM© software are in close agreement with the theoretical analysis.

The integration of renewable energy sources into the power system is an important step towards a sustainable energy transition. This transition could subsequently introduce substantial variability that critically impacts key operational parameters, such as frequency and voltage. This variability poses significant challenges, especially within microgrid configuration, both in grid-connected and isolated modes. Therefore, ensuring the stability of these parameters I paramount of their operational efficiency, reliability, and longevity. Despite these challenges, recent advancements in the field have led to the development of numerous advanced methodologies and control strategies designed to mitigate the impact of renewable sources on microgrid frequency stability. This paper provides a comprehensive overview of these state-of-the-art technologies and methodologies, including cutting-edge technologies such as adaptive load frequency control and Time-series prediction-based approaches. It offers insights into their application, effectiveness, and advantages in frequency control during microgrid operation, contributing to the ongoing discourse on integrating renewable energy sources with enhanced grid stability. Moreover, the discussion section provides insights and barriers impacting the implementation of these technologies in the current microgrid system and the power grid in general.

This work brings together the moment matching approach based on Loewner functions and the classical Loewner framework based on the Loewner pencil in the case of bilinear systems. New Loewner functions are defined based on the bilinear Loewner framework, and a Loewner equivalent model is produced using these functions. This model is composed of infinite series that needs to be truncated in order to be implemented in practice. In this context, a new notion of approximate Loewner equivalence is introduced. In the end, it is shown that the moment matching procedure based on the proposed Loewner functions and the classical interpolatory bilinear Loewner framework both result in κ-Loewner equivalent models, the main difference being that the latter preserves bilinearity at the expense of a higher order.

In this paper, a nonlinear system Identification method, wavelet-network-based Nonlinear Auto-Regressive Exogenous (NLARX) approach, is employed for battery state estimation and lifespan estimation. More specifically, three key battery parameters and health metrics, including temperature, voltage and State of Charge (SOC), are estimated and these parameters are essential for condition or state monitoring. Further, State of Health (SOH), crucial for forecasting the battery remaining useful life, is also predicted. Two open datasets are used to train and validated the performance of the proposed method. For temperature and voltage forecasting, the NLARX model outperforms the existing Thermal Single Particle Model with electrolyte (TSPMe) for prediction horizons under 600 seconds. In SOC estimations, the NLARX method produces consistent 15-second ahead prediction results even only using a small percentage of training data, while the SOH estimation, the proposed metho provides precise SOH variation prediction for 400 post cycles with less than 10% of the batterys life for training. Extensive results demonstrates that the NLARX model’s promise for the precise prediction of key battery parameters and health metrics and it can be used as a useful tool for battery fault detection and remaining useful life prediction.

In this paper, we provide an automata-based framework for verifying diagnosability property of Cyber-Physical Systems leveraging a notion of so-called hybrid barrier certificates. Concretely, we first construct a so-called (δ,K)-deterministic finite automata ((δ,K)-DFA) associated with the desired diagnosability property, which captures the occurrence of the fault to be diagnosed. Having a (δ,K)-DFA, we show that the verification of diagnosability properties is equivalent to a safety verification problem over a product system between this DFA and the dynamical system of interest. We further show that such a verification problem can be solved via computing hybrid barrier certificates for the product system. To compute the hybrid barrier certificates, we provide a systematic technique leveraging a counter-example guided inductive synthesis framework. Finally, we showcase the effectiveness of our results through a case study.

This paper proposes a data-driven method using a Fractional-Order Proportional-Integral-Derivative (FOPID) controller. The proposed method simultaneously obtains FOPID controller and reference model parameters to achieve tracking performance and specified robust stability from only one-shot closed-loop input-output data. The proposed control law is designed by solving an optimization problem, subject to the constraint condition of using the maximum value of the sensitivity function. Therefore, the proposed method provides trade-off design between tracking performance for the reference input and robust stability by selecting robust stability. By comparing numerical example results obtained for FOPID and integer-order controllers, it is shown that the use of the FOPID controller is effective in improving tracking performance for reference output.

This study compares methodologies for fault classification in reciprocating compressors, focusing on traditional Machine Learning (ML) with classical feature extraction processes and one-dimensional Convolutional Neural Networks (1D-CNN) in Deep Learning (DL). Both techniques demonstrated viability by employing a dataset of compressor vibration signals encompassing ten fault classes. While ML achieved a classification accuracy of 86%, DL reached 90.709%, highlighting its superior learning and generalization abilities, although with longer training times. These findings suggest that, despite ML being effective when relevant prior knowledge is available, DL, particularly with 1D-CNN, offers enhanced fault classification performance for this study case at the expense of additional processing resources.

Smart industry and Industry 4.0 are terms which are often used interchangeably. They characterise industry that capitalises on optimising processes through the successful integration of advanced digitalisation and manufacturing technologies, while applying sound organisation and human factors management principles. Equipping the current and future generation professionals with the necessary skills and personal qualities needed for the transition to Industry 4.0, and its extension to Industry 5.0 has been targeted by academic and professional education. Lessons learned from existing studies point to problem-based learning as an effective mechanism for the internalisation of interdisciplinary concepts, methods, and technologies. This paper outlines the formulation and experience gained from educational activities within the context of a smart industry postgraduate MSc course. The aim was to bring together methods for process and data integration, technologies such as machine learning, and management aspects, targeting domains relevant to smart industry. An educational activity was designed relevant to risk prediction within the asset management of wind farms. With scenarios of diverse criticality assumptions, marking the importance of Industry 5.0, results obtained from the educational activity show that students excelling in individual dimensions of smart industry are valuable contributors in a team setting, but a sound holistic understanding and competences across all three pillars of smart industry are needed for best learning objectives.

Printed Circuit Boards (PCB) manufacturing is a critical part of volatile supply chains for a wide variety of products and high value assets. PCBs are expected to exhibit zero defects and be subject to zero-maintenance. However low the defect rates, defects are highly disruptive and costly. Such defects can be introduced by a multitude of reasons, including faulty parts or sub-standard manufacturing processes. While sophisticated and dedicated quality inspection systems are typically in place in production environments, they still leave room for erroneous quality control outcomes. Besides in-line or post-production quality inspection, manufacturers can exploit experience gained from historical records of past inspections to predict future defect rates. This paper presents the development of a predictive quality modelling approach, which capitalises on such historical data and domain knowledge, to predict defect rates in new production orders. Employing appropriate encoding of knowledge through data pre-processing and applying regression type of machine learning, the proposed approach is validated on a real case study from an electronics manufacturing company. The developed approach can positively contribute towards optimising consequent maintenance and warranty services and become part of a zero-defect production strategy.