The use of residential photovoltaics has increased dramatically in recent years. With battery systems becoming more affordable, the optimal operation of a photovoltaic-battery system can bring significant savings to households. Optimal control of these systems requires correct forecasts of the underlying parameters, such as photovoltaic power generation, to know how to schedule the battery. While forecasting models have become increasingly accurate due to algorithmic advances and data availability, accuracy is typically measured in generic metrics which might not align with the downstream application. This study proposes a decision-focused learning framework that integrates the optimization and prediction by training a Long Short-Term Memory photovoltaic energy forecaster on the downstream optimal scheduling of a battery system. The proposed methodology is compared against a standard two-phase approach. Across a 14-month evaluation period, the decision-focused method reduced average electricity costs across twenty buildings by 3.6% when normalized against the performance bounds defined by a perfect forecast and a baseline of no optimization. Critically, this financial improvement was achieved despite the model exhibiting a root mean squared error of 19.9%, significantly higher than the decoupled model’s 8.2%. Warm-starting the decision-focused model further improves the results, lowering the average cost by an approximate 8% reduction, while also mitigating the negative impact on statistical accuracy (with a root mean squared error of 13.7%). The findings are statistically significant at the 0.001 level following a Diebold and Mariano test across the twenty households and for each household individually. These results demonstrate that aligning forecast models with optimization goals is key for achieving cost advantages in PV-battery systems. Future research should aim to replicate these findings on other datasets, alternate forecasting models and alternate optimization algorithms.

Smart charging control of electric vehicles (EVs), considering charger and battery characteristics, is essential for optimizing energy use and power profiles of different EV models. This paper proposes an attention-augmented multi-output neural network (AMONN), trained and validated on a high-resolution experimental dataset from two EV models with 11 kW on-board chargers. Compared with the baseline models, AMONN achieved significantly higher accuracy, reducing MAPE from 28.65% to 0.49% and improving R2 from 0.9712 to 0.9995. Simulation results showed that AMONN, coupled with a greedy optimization (GO) algorithm, outperformed fixed charging profiles (4, 8, and 11 kW), reaching overall efficiencies of 94–95% and reducing energy losses by up to 60%, while considering battery temperature evolution. To validate AMONN GO based charging control in practice, we built an experimental setup with a commercial charging point (CP), the test EV, a CAN data logger, and two laptops for control and acquisition. Tests of AMONN in open-loop (OL), closed-loop (CL), and GO CL modes confirmed its practicality, achieving efficiencies of 91.5–94%. These results demonstrate the potential of AMONN-based smart charging to enhance efficiency, reduce energy losses, and improve battery temperature optimization in real-world EV charging systems.

The increasing adoption of low-voltage DC systems requires fast and reliable fault protection solutions. This paper presents a modular bypass snubber-based solid-state circuit breaker (SSCB) that enables efficient fault interruption without the limitations of conventional parallel snubber designs. The proposed topology allows voltage suppression components to be selected near nominal system voltage, reduces conduction losses, and is expandable to bipolar and bidirectional configurations.A structured design methodology is introduced for selecting and sizing key components based on system parameters and worst-case fault conditions, with a focus on interruption during the rising phase of the fault current. The operating principles, including current commutation and voltage clamping, are analyzed, and experimental results validate the effectiveness of the proposed approach.

Floating photovoltaics (PV) are rapidly scaling up solar power beyond on-land PV. Whilst offshore floating PV (OFPV) is still in pilot phase, its combination with offshore wind could enable an efficient common use of costly transmission infrastructure. This work presents a detailed, quantitative case study assessing the integration of offshore floating PV with offshore wind. Through stochastic generation expansion planning, the optimal distribution of OFPV within a representative Dutch offshore wind farm is determined. In the power collection network, OFPV is best connected to the substation, or to the wind turbines electrically nearest to it. To evaluate the economic performance of the hybrid solar-wind system, its electrical integration with the Central Western European grid is simulated. The study reveals that a considerable amount of OFPV can be integrated in a modern offshore wind farm without hindering the transmission of wind power, with the export cables being the main bottleneck in power transfer, followed by the substation transformers and the array cables. However, this is accompanied by a significant amount of OFPV curtailment. As the capacity factors of offshore wind turbines increase, the remaining transmission gap in their connections, which OFPV can utilise without any transmission expansion, narrows. Finally, cost targets are derived for which the integrated offshore solar system would break even in the analysed case, revealing challenging economic prospects. The work identifies opportunities for hybrid offshore solar-wind farms and highlights key technical and economic challenges to be addressed.

Battery electric vehicles (BEVs) have increasingly positioned themselves as a critical technology in the power system, impacting the world's energy consumption. Understanding the BEV energy dynamics can contribute to vehicle, infrastructure, and grid optimization. Currently, BEV manufacturers provide limited access to the vehicle's high energy consuming components, such as the battery and the charger. Therefore, existing public datasets consist mostly of aggregated data collected from charging points outside the vehicle, resulting in lower data resolution and not capturing the actual energy dynamics. This paper fills this dataset gap by developing a data generation method to collect datasets, including the actual energy values for the charger, the battery, and the auxiliary devices, using measurement with a second resolution. The collected dataset illustrates energy dynamics under different modes (charging, driving, parking) and environmental conditions. This dataset provides detailed technical insights that can be used to optimize smart charging, reduce operational costs, understand usage, improve the operation of high energy consuming components, build AI models, and analyze grid impact.

Due to increasing electrification, there is a need for new power converter topologies. Traditionally, deriving these topologies require skill and experience. Recently, graph theory has been employed in an effort to automate the formation process of nonisolated converters with dc ports. This article proposes an extension of this methodology, introducing ac ports and isolated topologies. This is accomplished by taking into account the directional properties of the power electronic components. As a result, a distinction can be made between unidirectional/bidirectional and dc/ac ports. Besides, other asymmetrical characteristics, such as the need for bidirectional switches, can be verified. Furthermore, the methodology provides a solution for the inclusion of components with more than two terminals. This allows for the introduction of isolation transformers. After the derivation of candidate topologies, the voltage relations and current flows are automatically checked, verifying the operational boundary conditions. The article demonstrates the proposed methodology in a PV-battery system use-case. This showcases the resulting original circuits incorporating a dc input port, dc bidirectional port, and ac output port.

The paper explores the reliability and availability of low-voltage direct current systems in industrial settings. The paper introduces a methodology that uses semi-Markov processes and universal generating operators to assess system performance. The impact of various protection devices on system availability and reliability is evaluated. Key findings highlight the benefits of combining different protection devices for cost-effective fault clearance, the impact of protection failure, and the advantages but also the sensitivities of splitting the DC bus into multiple zones to enhance system robustness. The analysis is made freely available as part of the Julia package MultiStateSystems.jl

Fast and reliable fault detection is critical to ensuring stability in low-voltage DC (LVDC) grids, especially in applications where system reliability and uninterrupted operation are paramount. Conventional amplitude-based fault detection inherently introduces delays, as it requires the fault current to exceed a predefined threshold-delays that become more pronounced for faults occurring further from the source. In contrast, current derivative (di/dt)-based detection offers a fundamental advantage: the detection threshold is surpassed at the exact moment a fault initiates, enabling significantly faster response times. This paper presents a robust and simplified method for di/dt-based fault detection in LVDC systems. Rather than relying on grid modeling or fault current estimation, the proposed approach sets the fault detection threshold based on the natural current dynamics of the grid-connected converter. Specifically, the converter's current control loop bandwidth is used to define the maximum expected di/dt during normal operation. The method is further enhanced by recognizing the role of internal slew rate limiting in converters, which introduces an even greater margin between nominal and fault-induced transients-enabling discrimination even under high-impedance fault conditions. The approach is validated through both offline simulations and realtime implementation using a solid-state circuit breaker (SSCB) prototype. Results confirm that the method provides rapid, reliable detection with minimal computational complexity, making it suitable for protection systems in LVDC grids.

This paper presents a method to determine the impact of low-voltage direct current circuit breakers on the availability of the dc bus. Low-voltage direct current is gaining traction in industry, and protection is an important aspect. New circuit breaker technologies are extensively researched, but the impact of the speed of interruption on direct current bus availability has not yet been investigated. This paper demonstrates how to determine the fault clearance probability, which is used to set up state transition diagrams for the feeder states. These state transition diagrams are solved using the semi-Markov process. The solutions of the semi-Markov process are used in the UGO method to determine the availability of the direct current bus. Furthermore, the results depend on the length of the feeder, the capacitance of the direct current bus and the minimum allowed voltage. For feeders with a fault clearance probability lower than 100%, the number of feeders also influences the availability of the direct current bus.

Prediction of charging energy and power profiles is crucial for optimal scheduling of different EV models. This paper proposes the idea of using advanced machine learning (ML) techniques to predict the operating efficiency of on-board (11 kW) and off-board (50 kW) EV charging systems. An experimental setup is developed to control the charging power, measure and collect datasets under several operating conditions using a real EV model. The collected dataset is used to train, validate, and compare ML techniques such as linear regression (LR), random forest (RF), artificial neural network (ANN) and conditional generative adversarial network (cGAN). The evaluation metrics used for this comparison are mean absolute error (MAE), mean square error (MSE) and the coefficient of determination (R2). The results demonstrated that the RF and ANN performed better than the LR and cGAN models in both charging systems.

Offshore floating photovoltaics, tidal turbines and wave converters face similar challenges in terms of grid integration: electrical power must be transferred over long distances through a reliable and efficient grid connection. Whereas AC power collection systems are considered the industry standard for large-scale photovoltaics and (offshore) wind systems, a DC power collection grid may be more suitable for offshore floating photovoltaics. This work provides a qualitative discussion on the advantages and challenges tied to the grid integration of offshore floating PV systems through DC collection grids. The proposed advantages include reduced transmission and power conversion losses, improved power density, reliability, power quality, efficient integration with energy storage and high-voltage DC links, and flexibility in power flow control. Whereas many of these advantages apply onshore as well, this work argues that reduced transmission losses, improved power density and reliability benefits are more significant offshore. To unlock this potential however, challenges such as high capital costs, adequate protection, dynamic grid stability and lack of standards must be addressed.

Deep learning models have gained increasing prominence in recent years in the field of solar pho-tovoltaic (PV) forecasting. One drawback of these models is that they require a lot of high-quality data to perform well. This is often infeasible in practice, due to poor measurement infrastructure in legacy systems and the rapid build-up of new solar systems across the world. This paper proposes SolNet: a novel, general-purpose, multivariate solar power forecaster, which addresses these challenges by using a two-step forecasting pipeline which incorporates transfer learning from abundant synthetic data generated from PVGIS, before fine-tuning on observational data. Using actual production data from hundreds of sites in the Netherlands, Australia and Belgium, we show that SolNet improves forecasting performance over data-scarce settings as well as baseline models. We find transfer learning benefits to be the strongest when only limited observational data is available. At the same time we provide several guidelines and considerations for transfer learning practitioners, as our results show that weather data, seasonal patterns, amount of synthetic data and possible mis-specification in source location, can have a major impact on the results. The SolNet models created in this way are applicable for any land-based solar photovoltaic system across the planet where simulated and observed data can be combined to obtain improved forecasting capabilities.

Residential photovoltaics have seen significant uptake in recent years. With battery systems becoming more affordable, the optimal operation of a photovoltaic-battery system can bring significant savings to households. Optimization of such a system requires us to make accurate forecasts of the photovoltaic power generation, in order to properly know when to charge and discharge the battery, given how much energy was generated. Forecasting models have become more and more accurate, due to algorithmic advances and data availability, but forecast accuracy is typically measured in generic statistical metrics, which might not align well with the goals of the downstream application. Therefore, we propose an integrated approach to residential PV forecasting, in which the optimization algorithm is solved online during the training of the forecaster. Our results show that applying this methodology decreases the cost when applying the forecast to the overall energy cost minimization of a photovoltaic-battery system. These findings show that using generic statistical metrics for forecast evaluation will not necessarily yield the optimal forecast for any given scenario and that coupling the forecast of a parameter with the optimization problem itself adds value for the user, providing a relative cost decrease of c. 9% over a two-stage approach.

The article proposes a methodology to detect real-time power mosfet degradation, in variable mission profile applications, using externally measurable electrical parameters. This complements the work done for fixed operation conditions in current literature. To achieve this, the damage and temperature sensitive drain to source resistance is accompanied with a gate resistance measurement only sensitive to temperature. Together, they allow for the detection of, and the distinction between, bond wire and die attach solder layer degradation. A dual extended Kalman filter is used to filter the measurement data and to estimate the change in thermal model. The article shows the measurement circuits together with proof of concept lab results in a solar photovoltaic use case. The main aim is to show that the resistance measurement can be compensated for mission profile temperature variations and that the thermal resistance can be estimated, reflecting bond wire and die attach solder layer degradation.

A spreadsheet tool and underlying model was developed to aid non-expert users in sizing off-or on-grid photovoltaic systems with battery back-up for office applications in Africa.The tool offers the user a number of choices which help in the decision process.The model is based on the concept of energy equivalence and extended by taking into account nonideal behaviours of photovoltaic system components, modelled as efficiency deviations.The spreadsheet tool uses freely available data, such as PV system component manufacturer's data sheets, as well as climatic data from the NASA SSE database.The irradiation data on the plane-of-array is calculated for the chosen location using the well established HDKR irradiation model.The obtained irradiation and energy output were compared to free services from Soda-is and the HOMER sizing software.The predicted energy output comparison of identical off-grid photovoltaic systems per location for two sites in Africa were within 14% of the value given by HOMER.A "performance ratio"-equivalent efficiency was calculated as part of the model.This intuitive approach to photovoltaic system sizing reduces the learning curve for non-expert users.The predicted "performance ratio"-equivalent efficiency can be used as a predictive analytic tool for grid-connected and off-grid photovoltaic systems.

Many options are available when it comes to protecting low voltage direct current grids. As each type of protection device has different tripping characteristics, they all have their own unavailability risk, which is the cost of unavailability. This paper proposes a method that uses transition frequency densities to determine the unavailability risk. These transition frequency densities are determined through a semi-Markov process and are used to calculate the unavailability risk of a simple use case as presented in a numerical illustration. Fuses are the most cost-effective type of protection device for the presented numerical illustration, however, cannot be used under all circumstances. Solid state circuit breakers are to be used in highly critical systems, mechanical circuit breakers when the cost of outage is relatively low and hybrid circuit breakers in between.

EV charging technologies could improve grid flexibility, reduce the charging infrastructure need, and increase revenues when their operating efficiency is higher. This paper presents this operating efficiency measurement of AC, DC, and V2G charging using different EV models. The common trend for all the charging technologies and EV models is their low efficiency at lower charging power. In AC charging, the operating efficiency is lower than 85% when operating at lower than 25% to 40% of the nominal power depending on the EV models. In DC charging, the efficiency is around 85.21% to 89.41%. The operating efficiency of V2G is lower than 85% when operating around 10% to 15% of the nominal power. The charging efficiency is higher than the discharging one. The average round-trip efficiency is around 76.12%. Several solutions to maximize this operating efficiency are discussed based on the measurement results.

This paper aims to assess the impact of a volumetric and a capacity-based network tariff, as well as the impact of a substantial electricity price increase on the decision of a household to invest in a PV-battery system. Therefore, a convex optimization model is implemented which returns the optimal sizing and operation from the households' perspective by minimizing the equivalent annual cost. Based on the analysis of the optimal PV-battery system for 200 households under four scenarios, this study found that the investment driver of a household changes from minimizing grid withdrawal to maximizing grid feed-in when the feed-in remuneration increases, as well as the maximization of the installed PV capacity. Additionally, the price increase leads to a net profit as opposed to a reduced cost. The shift from a volumetric to a capacity-based tariff leads to a smaller gap between the consumers' and prosumers' contribution to the distribution grid costs, increasing fairness. However, the contributions could be insufficient to ensure adequate cost recovery, requiring possible adjustment of the tariff height by the DSO. Finally, policy makers need to be aware that a capacity-based tariff leads to a lower reduction of carbon emissions as opposed to a volumetric tariff.

Snubber circuits play a critical role in enhancing the reliability and performance of low voltage DC solid state protection devices. These circuits mitigate transient voltage spikes and oscillations, protecting sensitive semiconductor components from damage due to overvoltage and ensuring stable operation. This review provides a comprehensive overview of snubber circuits specifically designed for solid state protection applications. It explores various snubber configurations, highlighting their operational principles and protective capabilities. The differences between these snubber networks are discussed in detail, focusing on their effectiveness, complexity, and suitability for different protection scenarios. Additionally, the review delves into various aspects related to the application of snubber circuits in protection devices such as efficiency, and the impact on overall circuit performance. By examining these factors, the paper aims to guide the selection and design of a snubber circuit for application in bipolar bidirectional low voltage DC grids.