This paper implements and validates a real-time Energy Management System (EMS) for DC microgrid operation using a power hardware-in-the-loop (PHIL) test bench. The test bench replicates renewable sources, a general load, and a Hybrid Energy Storage System (ESS) comprising a battery, an emulated fuel cell, an emulated electrolyzer, and a virtual hydrogen tank. Constructed with dSPACE and Simulink, the PHIL test bench facilitates the control and monitoring of DC sources, DC loads, and controllable DC/DC converters. The EMS technique combines an Optimal Dispatching (OD) algorithm with Model Predictive Control (MPC). While OD effectively reduces the total cost of grid utilization, it falls short in responding to actual disruptions and uncertainties. To address these issues, MPC is integrated with OD. This combination of OD with MPC offers two main advantages: it ensures a reduction in grid utilization and effectively manages perturbations and sudden interruptions.

Renewable energy (RE) integrated with a hybrid electric-hydrogen storage system (HEHSS) is considered a promising solution for data centers to increase their energy efficiency and decrease their greenhouse emissions. This paper proposes a forecasting based hierarchical energy management strategy for a data center with wind energy, tidal energy and HEHSS, which is taken as a microgrid. A multi-population genetic algorithm-based day-ahead scheduling is used as the first layer and a rolling predictive control-based real-time dispatching is added as a second layer to minimize the degradation costs of HEHSS and maximize the revenue of the electricity and hydrogen trading. Also, a long-short-term-memory-based forecasting framework is developed to forecast power generation and load demand.

International audience

Abstract In this article, we address a critical concern of the growth of computer servers number and the resulting power expenditures in data centers by analyzing the statistical metrics related to the workload executed in each physical node. The aim is to build a stochastic model for power consumption estimation based on historical data. Relying on in‐depth investigation and experimental testing of the power consumption features and performance of the various workload datasets, we propose a model that considers the workload and the power consumed to be executed by a server as random variables. Based on the properties of the probabilistic distribution function of each random variable, we establish the correlation relationship between the workload and the power consumption using a non‐parametric approach. Our use of a non‐parametric method to learn a given probability model is challenging because it requires estimating the full distribution from the available data samples. The accuracy of our approach is demonstrated by estimating the energy consumption of various workloads. The experimental and simulation results show that our model outperforms many existing approaches in terms of accuracy, and it can be applied to a wide variety of workloads.

The low economy of fuel cell hybrid electric vehicles is a big challenge to their wide usage. In this article, a road, health, and price-conscious optimal cost minimization strategy based on a decision-making framework was developed to decrease their overall cost. First, an online applicable cost minimization strategy was developed to minimize the overall operating costs of the vehicles, including the hydrogen cost and degradation costs of the fuel cell and battery. Second, a decision-making framework composed of the driving pattern recognition-enabled, prognostics-enabled, and price prediction-enabled decision makings, for the first time, was built to recognize the driving pattern, estimate the health states of power sources, and project future prices of hydrogen and power sources. Based on these estimations, optimal equivalent cost factors were updated to reach the optimal results on the overall cost and charge sustaining of a battery. The effects of driving cycles, degradation states, and pricing scenarios were analyzed.

In the absence of aerodynamic pitch control, it is required to drive the wind turbine at an optimal speed for a given wind speed to extract maximum power from a wind turbine generator system. Due to unpredictable wind speed fluctuations, operating at maximum power point is a difficult task to undertake. This paper presents a maximum power point tracking (MPPT) algorithm for variable speed wind turbines. The strategy uses neural networks and genetic algorithms to learn the wind turbine's nonlinear dynamic model and achieve accurate tracking. As such, robustness to unpredictable wind uncertainties is achieved. Simulation results for different situations highlight the performance of the proposed controller under various wind speed operating conditions.

Smart grid applications, renewable energy utilization and electric vehicles (EVs) are attracting researchers due to their importance nowadays as well as in the future. An efficient power electronic converter is a main and common topic for research in this area. In this paper, a prototype of the electrical part of a power-train for EVs using an advanced multilevel converter topology is introduced, discussed and analysed. A comparison between the advanced converter, two-level and conventional multilevel converter topology is discussed as well. A switch function model is derived and discussed for the proposed converter. A mathematical model for the converter supplied by a fuel-cell (FC) and boost-converter (BC) is implemented with Matlab/Simulink. The simulation results are analysed to evaluate the converter. The evaluation is based on the harmonic analysis and power loss calculations. The converters are tested at different switching frequencies to show the effect of this variable on the converter loss. The results indicate that the proposed converter is 1.32% more efficient compared to conventional five-level DCC. Moreover, the lowest harmonic content, for all of the studied converters, is the proposed one.