Distributed generation in power systems is currently attracting significant attention as a means to reduce energy generation costs through the exploitation of low-cost decentralized sources. Cost reduction becomes especially relevant when a large portion of the energy supplied by a centralized generator is replaced by energy from a few high-power distributed sources. However, the effectiveness and safety of this approach critically depend on the proper allocation of distributed generators. To this end, this paper presents a novel sensitivity analysis for system losses that highlights the nonlinear dependencies on power-injecting sources. It is shown that the parameters of the nonlinear model allow easy identification of a small subset of grid buses that are suitable candidates for large-power injection. At these buses, source sizing can be determined using standard optimal power flow techniques that minimize the energy cost function. The proposed allocation strategy is validated on the IEEE 69 and 85 benchmark bus grids, and its robustness is further confirmed through simulations that account for stochastic load fluctuations. Comparison with state of the art and robust source allocation approaches, such as the Iterative Search Method, shows that for similar accuracy the proposed method is two/three orders of magnitude faster.

ASTRAEUS project aims to improve the use of Commercial-Off-The-Shelf (COTS) devices in space telecommunication applications. The goal is to develop specialized radiation mitigation techniques for both hardware and software components with a special focus on the latter. The aim is to demonstrate the feasibility and reliability of these techniques, enabling their future use in telecommunication payload processing units. The SIHFT (Software Implemented Hardware Fault Tolerance) approach will be enforced, where some proper modification to a conventional compilation toolchain, will make possible the identification of temporary fault and the adoption of fault tolerant solutions. The successful implementation of this project will de-risk the use of software-based radiation mitigation techniques and foster the adoption of high-performance while cost-effective COTS electronics in space applications.

Accurate forecasting of solar photovoltaic (PV) power is essential for grid stability and efficient energy management. This study investigates the impact of missing data, introduced through both random and block removal at varying percentages, on the performance of hybrid neural network architectures, Convolutional Neural Network–Long Short-Term Memory (CNN–LSTM), LSTM Autoencoder, and standalone LSTM models, for short-term PV power prediction. Using two real-world PV datasets representing semi-arid and desert climates, the focus is placed on non-parametric probabilistic forecasting, leveraging Kernel Density Estimation and quantile regression to generate confidence intervals that capture forecast uncertainty. Experimental results demonstrate that while all models experience performance degradation with increasing levels of missing data, hybrid architectures such as CNN–LSTM and LSTM Autoencoder exhibit greater robustness in capturing temporal patterns and maintaining forecast reliability. Notably, the CNN–LSTM model achieved a 19.35% reduction in quantile loss compared to the standalone LSTM under 30% random data removal. The findings underscore the importance of evaluating both predictive accuracy and uncertainty calibration under data scarcity scenarios to enable dependable solar energy forecasting and resilient energy management.

Abstract Accurate short-term photovoltaic (PV) power forecasts are critical for efficient grid balancing, yet training optimizers, often overlooked compared to neural network architectures, significantly influence prediction accuracy and convergence speed. Prior research primarily focuses on adjusting network architectures, typically employing a single optimizer (commonly Adam), thus leaving optimizer selection underexplored, especially under noisy and incomplete real-world PV data. This study systematically benchmarks four optimizers—Adam, Adaptive Gradient (Adagrad), Rectified Adam (RAdam), and Lookahead—across three deep-learning architectures (Long Short-Term Memory (LSTM), Convolutional Neural Network (CNN)-LSTM, and LSTM-Autoencoder) using data from two distinct PV sites. Unlike prior works, we assess optimizer effectiveness across a wide range of conditions, including varying training data lengths, sampling intervals, and missing data patterns (both random and block-wise). Using two real-world PV datasets representing semi-arid and desert climates, we analyze forecasting accuracy, convergence time, and robustness. Our empirical results demonstrate that RAdam consistently outperforms Adam by achieving up to 36% lower forecasting error under noisy and incomplete data conditions, while Lookahead offers up to 40% faster convergence in deep hybrid models. These gains translate into tighter reserve-margin planning and smoother inverter set-points, advancing state-of-the-art PV forecast pipelines. The paper concludes with optimizer-architecture recommendations for practitioners facing latency or compute constraints.

This paper outlines an innovative approach to enhance the predictability of solar photovoltaic power. By employing advanced machine learning techniques, specifically artificial neural networks, this study addresses the challenges posed by the intermittent nature of solar energy. The employed models incorporate probabilistic forecasting to provide not only precise power output predictions but also confidence intervals that signify the uncertainty in these predictions. This approach supports more effective integration of solar energy into power grids, facilitating better energy management and planning. The results indicate that our models can significantly improve the accuracy of solar power forecasting, crucial for optimizing grid operations and enhancing renewable energy adoption.

Solar photovoltaic (PV) power forecasting is a crucial aspect of efficient energy management in the renewable energy sector. This study examines the use of artificial neural networks (ANNs) to forecast solar PV power output. It considers various factors influencing power output and investigates different ANNs for prediction. Real-world PV power data is collected and preprocessed for training and testing ANNs such as recurrent neural networks, autoencoders, and convolutional neural networks. The results show that ANNs, particularly Long Short-term memory (LSTM), accurately forecast PV power output in the short term. The study also analyzes the impact of panel ageing on PV power using machine learning models, revealing effective prediction of performance degradation. Clustering the dataset into sunny and cloudy subsets, and using separate models for each subset improves prediction accuracy. The study presents a comprehensive analysis of ANNs for PV power forecasting and the influence of panel ageing, highlighting the potential of machine learning for precise and reliable predictions.

The electricity market is constantly growing and is facing unprecedented planning needs. Loads are no longer as systematic as in the past due to changing user habits and the likewise for the generation of energy due to the unpredictability of renewable sources. Demand response (DR) strategies are the basis of resource planning, but these methods cannot ignore the presence of accurate load models that can predict behavior. In this panorama, the prediction of the loads due to the recharging of electric vehicles, offers interesting ideas of complexity, which make it a topic of open research. In this paper we will show a method of reconstruction of charging profiles through Markov Chains, starting from distributions derived from experimental data.

An electro-mechanical model of a SF6circuit breaker for medium voltage applications, i.e. for voltages up to 36 kV and short circuit currents up 50 kA is presented. The model is developed by starting from the well known Schwarz-Advonin arc model with the addition of the mechanical effects on the arc geometry and of the “puffer”. The puffer can blow the arc during its full duration and mainly during current zero crossing to extinguish it. This action is carefully exploited by breaker designers and often is the key aspect to obtain arc extinction in high current conditions.

In the near future Electrical Vehicless (EVs) will most likely replace conventional combustion-engine based ones. This and the increase in the use of renewable energy sources will have an important positive impact on our ecosystem. At the same time it will require a serious reanalysis and possibly redesign of the structure of our distribution network. In this paper we propose a model that, at a relatively fine level, describes the behavior of communities in terms of transport requirements, derives statistic driving behavior patterns and determines the corresponding induced Electric Vehicle Charging habits. The energy consumption of every simulated EV is computed taking into account the features of the trip and the vehicle itself. The model has been tailored to data extracted from a portion of the Milan (Italy) metropolitan area, but is trivially adaptable to any area for which a similar set of data is available. The obtained results are compared to data from similar works, showing good agreement. The outputs of the model, i.e. the energy consumption of the vehicles and their time-spatial distribution within a specific area (and consequently the loads on the distribution network nodes due to charging operations), have been (and will be) used to perform impact analyses on network performances and on personal mobility.

Probe insertion is traditionally used in the frequency domain to increase robustness of the harmonic balance method and avoid the DC degenerate solution. This technique evidences several good properties, for instance when simulating oscillators that are based on high quality factor resonators or AC coupled ones. In this paper the probe insertion technique is extended to the time domain, allowing the application of the shooting method to a wider class of circuits including for example crystal oscillators. Time domain methods, such as shooting, are superior in the simulation of strongly nonlinear, and, thanks to a recent extention, mixed analog/digital circuits. Moreover, time domain probe insertion can be used to find multiple steady state solutions and check their stability properties, without the need to compute the eigenvalues of the monodromy matrix.

This paper proposes a novel electrical power system architecture capable to efficiently transfer energy from a solar generator to high power electric thrusters of a spacecraft. The architecture is based on a maximum power point tracker that acts on a DC/DC converter connected between a subsection of the solar generator and the thrusters. The other sections of the solar generator, that deliver most of the total power, are directly connected to the thrusters. The novelty of the proposed architecture is that the maximum power point tracker is always operating and, while conditioning only a fraction of the total power, is able to transfer to the thrusters an amount of power greater than the handled one.

Periodic circuits driven by multitone signals are still a challenging simulation problem despite several numerical methods being presented in the literature. In this paper, a mixed time-frequency method for the solution of this problem and suitable for both autonomous and nonautonomous circuits is presented. The method is based on an extension of the envelope following method, which allows us to reduce the number of unknowns involved in the steady-state problem with respect to previous mixed time-frequency approaches, and a suitable reformulation of the periodicity constraint that allows us to obtain a significant acceleration in the determination of the solution by reducing the time interval along which the envelope analysis must be performed. The method is first presented for nonautonomous circuits and then extended to autonomous ones.

The probe-based harmonic balance (HB) method is a well-known and largely used tool to compute the steady state behavior of autonomous circuits (oscillators). In this paper, the method is extended to analyze coupled oscillators, where the working frequencies and conditions, i.e., pulling and locking modes, have great relevance. It is shown that probe insertion can be considered as a specific matrix-bordering technique applied to the Jacobian matrix of the HB method. It transforms the original coupled autonomous system in a non-autonomous one, forcing the circuit to lock to the probes themselves. In this context, a novel approach to find the steady state solution is introduced. This approach exploits the properties of the power exchanged among the probes and the coupled oscillators. Possibly, more than one steady state solution with the oscillators working in pulling or locking modes can be found.

The shooting method is largely employed to determine the steady-state working condition of both autonomous and nonautonomous circuits. In general, the conventional shooting method employs the Newton algorithm to estimate a better approximation of the steady-state working condition. The Newton algorithm requires the computation of the Jacobian matrix and this seriously limits the use of the conventional shooting method to solve medium/large scale circuits. In this paper, an approach to efficiently determine the shooting matrix is presented. It is shown that the approach is also adequate to deal with mixed analog/digital circuits.

In this paper we present a design for reliability methodology, with the goal of reducing the impact of transistor VTHdegradation due for example to phenomena such as NBTI. It uses infrastructure IPs (I-IPs) featuring a self compensation scheme that automatically detects transistor aging effects and illustrates the design for test infrastructure used to make the SoC/System aware of the NBTI effects. This scheme is conceptually validated by using multi-level simulation and models. The discussion of possible exploitation models completes the paper.

This paper describes an approach to determine the Floquet exponents and the related eigenfunctions of linear time-varying circuits, that represent a vehicle to implement the variational model of periodic nonlinear circuits. The Floquet exponents and eigenfunctions allow us to exploit the structure of the analytical solution of the linear time-varying circuit and, as shown, yield an efficient solution. The proposed approach allows the calculation of the Floquet exponents directly and is developed from the harmonic-balance formulation adopted to find the steady-state solution of the nonlinear periodic circuit.

Some algorithmic aspects in RF circuit simulation are reviewed, emphasizing the problems related to the contributions included in this Special Session. Recent achievements as well as open problems are pointed out