MG2lab - Politecnico di Milano

🔬 New Research from MG2lab - Politecnico di Milano 🚀 We are excited to introduce our recent publication: Two-Layer Optimization Approach for Electric Vehicle Charging Stations With Dynamic Reconfiguration Of Charging Points! ⚡This innovative method optimizes both energy scheduling and dynamic power rate adaptation, addressing the growing challenge of high EV demand with an holistic Energy Management System. Key highlights: 🔃 Dynamic Reconfiguration: Adjusts the power rate of Charging Points in real-time based on EV power demand and remaining charging time, maximizing both infrastructure usage and customer satisfaction. 🖥 Two-Layer Optimization: The first layer schedules energy one day ahead, considering forecasted values for PV production, EV demand, and electricity prices. The second layer, based on Model Predictive Control (MPC), dynamically adjusts the power output of CPs in real time. 💡 Seamless Operation: The framework handles forecast uncertainties (e.g., EV arrivals and PV production) and increases profits without sacrificing EV charging times. Our research demonstrates that an optimal dynamic reconfiguration significantly improves the charging station performance, allowing for better alignment with daily energy schedules, all while boosting profits and customer satisfaction. This paper opens up new possibilities for sustainable EV charging infrastructures, especially those integrating renewable energy sources like PV systems and battery storage. 📃 Read the full paper to explore how this comprehensive framework, combining optimization and firmware improvements, could improve the operation of a charging station for both users and charging point operators. Link to the paper: https://lnkd.in/dtgFdWez #EV #EMS #ChargingStation #MOST

🔬 New Research from MG2lab - Politecnico di Milano 🚀 We are excited to share our latest paper where the Deep Model Predictive Control (DMPC) approach is applied for the first time to the optimal management of a microgrid. This groundbreaking methodology leverages neural networks to approximate predictive control, significantly reducing real-time computational costs. Unlike traditional predictive control, where complexity increases with longer horizons and larger systems, DMPC maintains constant computational complexity. 🌐 Key highlights: - DMPC eliminates the major limitation of standard MPC by drastically reducing the online computational burden. - The approximation error between DMPC and traditional MPC is minimal, ensuring efficient and reliable microgrid management. - This research sets the stage for future advancements in microgrid infrastructures, such as integrating hydrogen storage technologies and exploring fault ride-through capabilities to enhance system resilience. 💡 This paper opens new doors for real-time applications in energy management, potentially revolutionizing the way microgrids operate under varying conditions. 🌱⚡ #EnergyManagement #DMPC #Microgrid #Innovation #Sustainability #SmartGrids #NeuralNetworks #DeepLearning #MOST centro per la mobilità sostenibile --- Link to the paper: https://lnkd.in/d8vnUTbC

MG2lab - Politecnico di Milano is in Malaysia!!! 🌍 We are proud to announce that Professor Sonia Leva has been invited as a keynote speaker at the 7th International Conference on Recent Advances in Automotive Engineering & Mobility Research to deliver a presentation titled "MicroGrid and EV Charging Forecast to Minimize Impacts on the Grid". The conference was organized by the Centre for Automotive Research (CAR), Universiti Kebangsaan Malaysia from 24th to 25th September 2024. #EnergyInnovation #MicroGrid #EVCharging #Sustainability #Forecasting #SmartGrid

MG2lab - Politecnico di Milano is in Malaysia!!! 🌍 We are proud to announce that Professor Sonia Leva has been invited as a keynote speaker at the 7th International Conference on Recent Advances in Automotive Engineering & Mobility Research to deliver a presentation titled "MicroGrid and EV Charging Forecast to Minimize Impacts on the Grid". The conference was organized by the Centre for Automotive Research (CAR), Universiti Kebangsaan Malaysia from 24th to 25th September 2024. #EnergyInnovation #MicroGrid #EVCharging #Sustainability #Forecasting #SmartGrid

🔋🌟 Insight from @MG2lab🌟🔋 We’re excited to present you a feature inside the MG2Lab that could significantly enhance our understanding of Energy Management Systems (EMS) in real-world scenarios.   🔍 Inductances in real-world microgrids: Real-world microgrids often feature assets spread across diverse locations, resulting in increased impedances within electrical connections. To better simulate these conditions, our MG2Lab is equipped with a set of inductors that can be used to mimic longer electrical lines inside a microgrid.   🔗 New possible topologies: In all the previous studies we presented you, the inductors were bypassed through electrical switches. Moving forward, we’re excited to explore how incorporating inductors can impact microgrid performance. In particular, the six inductors have a 240 µH inductance, corresponding to a distribution line of an equivalent length of 1 km at low voltage and 50 Hz frequency. They can be inserted on the distribution lines, creating a meshed topology, on the derivation lines, creating a radial topology, or in a mixed fashion.   📊 Preliminary studies: To understand the potential of implementing the inductors inside MG2Lab, we are conducting some preliminary studies and experimental activities. 1.       The impact of the inductors in the microgrid power flow, especially in terms of reactive power 2.       The effect of topology on frequency and voltage stability Currently, we’re experimenting with a meshed topology to gain insights into these dynamics.   🌱 Future research questions: With a deeper understanding of inductors’ effects, we aim to refine our EMS to better account for power losses and topology variations. Our goal is to develop a more robust and reliable EMS that enhances microgrid efficiency and resilience.   Stay tuned for more updates as we continue to advance our research and bring innovative solutions to the forefront of energy management! #MG2lab #Research #Microgrids #Topology #RealWorld

🔋🌟 New Research from MG2lab - Politecnico di Milano: “Impact of PV and EV Forecasting in the Operation of a Microgrid” 🌟🔋 We're excited to present our latest paper, which dives deep into the predict-then-optimize approach for optimizing the operation of microgrids, particularly focusing on the impact of forecast accuracy for both load and solar predictions. 🔍 Analysis of Forecasting Techniques: We compared various EV load forecasting techniques, including LSTM and persistence, within a hierarchical predictive control strategy. The study included both experimental tests at MG2lab - Politecnico di Milano's facility and simulations using historical data from 2023. ⚡ Operational Validation:  The study validated the operation of a multi-good microgrid using both PHANN PV and LSTM EV load predictions. These methods were combined with an EMS (Energy Management System) to minimize operational costs, proving effective for real-world applications. 📊 Quantitative Findings: - Forecast Accuracy Matters: Higher accuracy in combined forecasts led to better EMS performance and operational efficiency. - Complex Interactions: Interestingly, despite LSTM's superior accuracy in EV load prediction, it did not always yield the best EMS performance, especially when coupled with PHANN PV forecasts. Moreover, during periods of high electricity prices, the persistence method outperformed LSTM in cost-effectiveness. 🌱 Future Research: Our future research will focus on developing customized loss functions for training and evaluating forecasting models. These will go beyond generic statistical metrics to capture the nuances of downstream decision-making, aiming to enhance the integration and synergy between forecasting and optimization processes. #MG2lab #Research #Microgrids #Forecasting #PredictThenOptimize #EnergyManagement #LSTM #PHANN #Sustainability --- Link to the paper: https://lnkd.in/d_nfXaeh

🚗🔋 Exciting Research Update from MG2lab! 🔋🚗 We're thrilled to share our latest research paper, which makes significant strides in understanding the integration of electric vehicles (EVs) with the power grid. Here are the key highlights: 🌟 Innovative Smart Charging and V2G Strategies: Using the “oemof” modelling framework, our study implements smart charging and vehicle-to-grid (V2G) strategies. This allows for shifting EV charging to periods of renewable overproduction and utilizing EV batteries for grid storage. 📈 Key Findings: - Electricity Demand: Full electrification of passenger transport by 2050 could increase electricity demand by up to 28% and add 50–250 MW to the current 652 MW peak demand. - Peak Reduction: Implementing smart charging for 50% of EVs reduces peak charging by 34%, shifting demand to nighttime renewable surplus periods. - Import Reduction: Enabling V2G for 50% of EVs decreases reliance on energy imports by 60%. 🔧 Flexible Methodology: Our advanced modelling methodology adapts to various energy mix scenarios, providing new capabilities to explore sustainable EV integration strategies. 📊 Broader Implications: Beyond transport electrification, future research should consider the electrification of other sectors like heating and industry to fully understand the impact on the electricity grid. Integrating these results will offer a comprehensive view of future energy demand and supply, aiding policymakers in making informed decisions about the future energy mix and grid infrastructure for a low-emission economy. #MG2lab #Research #ElectricVehicles #SmartCharging #V2G #RenewableEnergy #Sustainability 🌱🔋 --- Link to the paper: https://lnkd.in/dGYUmzQj

🔋🔬 Exciting Research Update from MG2lab - Politecnico di Milano! 🔬🔋 We're thrilled to share our latest paper, which introduces a framework for joint SOC-SOH estimation in lithium-ion batteries, taking into account ageing effects to enhance predictive accuracy. 📊 Innovative Model: Our new model integrates State of Health (SOH) into the inputs, significantly improving State of Charge (SOC) estimations. This advancement addresses the critical need for accurate battery management as batteries age. 🔧 Optimized Hyperparameters: We explored a wide range of hyperparameters, including optimizer rate, RNN layers, Fully Connected layers, and activation functions. Our analysis revealed near-optimal configurations for an efficient network structure. 🧠 Top-Performing Model: Through rigorous evaluation, the BiLSTM emerged as the most suitable model for both SOC and SOH estimation. Our framework achieved impressive results: - SOH Estimation: Average MAE of 0.3% and RMSE of 0.42% - SOC Estimation: Average MAE of 1.42% and RMSE of 2.31% 🔍 Future Directions: - BMS Integration: Focus on integrating this framework into a Battery Management System (BMS) microcontroller. - Neural Network Compression: Explore neural network compression for microcontroller compatibility. - Transfer Learning: Apply transfer learning techniques to validate the model across new datasets and battery types. - Upscaling: Test the model on battery modules/packs to further validate its effectiveness. Stay tuned for more advancements from MG2lab - Politecnico di Milano! #MG2lab #Research #BatteryTechnology #SOC #SOH #BiLSTM #BatteryManagement #Innovation 🌱🔋 --- Link to the paper: https://lnkd.in/dU8S9q2t

🚗🔋 Exciting Research Update from MG2lab - Politecnico di Milano! 🔋🚗 We're thrilled to share our latest research paper, which presents a groundbreaking approach to electric vehicle (EV) charging load forecasting. Here's a snapshot of our key findings: 📊 Advanced Forecasting Model: We applied the LSTM model with an attention mechanism to predict EV charging loads 24 hours in advance, using a real-world dataset of public EV charging equipment. 🔍 Performance Highlights: - Our LSTM with attention mechanism outperformed state-of-the-art Deep Learning models, excelling in metrics like MAE, MSE, and RMSE. - It demonstrated comparable SMAPE results to simpler LSTM and GRU models but with better training stability. - The attention mechanism significantly improved the results of Sequence-to-Sequence (S2S) models, justifying the added complexity. 📉 Limitations: The study's dependence on limited data constrains validation. The lack of expansive open datasets hampers progress, but future research aims to incorporate additional inputs to enhance precision. 🌦️ Future Directions: We plan to include exogenous variables like weather forecasts, ambient temperature, and events (holidays, mass events, maintenance) to refine our models further. This will help us better predict EV charging station occupancy trends and associated power supply. Stay tuned for more groundbreaking advancements from MG2lab - Politecnico di Milano! #MG2lab #Research #EVCharging #DeepLearning #LSTM #AttentionMechanism #Sustainability 🌱🚀 --- Link to the paper: https://lnkd.in/dJM4J2zU

🚗🔋 Exciting Research Update from MG2lab - Politecnico di Milano! 🔋🚗 We're thrilled to share our latest research paper, which presents a groundbreaking approach to electric vehicle (EV) charging load forecasting. Here's a snapshot of our key findings: 📊 Advanced Forecasting Model: We applied the LSTM model with an attention mechanism to predict EV charging loads 24 hours in advance, using a real-world dataset of public EV charging equipment. 🔍 Performance Highlights: - Our LSTM with attention mechanism outperformed state-of-the-art Deep Learning models, excelling in metrics like MAE, MSE, and RMSE. - It demonstrated comparable SMAPE results to simpler LSTM and GRU models but with better training stability. - The attention mechanism significantly improved the results of Sequence-to-Sequence (S2S) models, justifying the added complexity. 📉 Limitations: The study's dependence on limited data constrains validation. The lack of expansive open datasets hampers progress, but future research aims to incorporate additional inputs to enhance precision. 🌦️ Future Directions: We plan to include exogenous variables like weather forecasts, ambient temperature, and events (holidays, mass events, maintenance) to refine our models further. This will help us better predict EV charging station occupancy trends and associated power supply. Stay tuned for more groundbreaking advancements from MG2lab - Politecnico di Milano! #MG2lab #Research #EVCharging #DeepLearning #LSTM #AttentionMechanism #Sustainability 🌱🚀 --- Link to the paper: https://lnkd.in/dJM4J2zU