Cooperative EV management in a residential parking: Charging satisfaction and current harmonics mitigation

This paper proposes a Cooperative Energy Management System (CE-EMS) for electric vehicle (EV) charging in a residential parking lot. This type of parking could be an important lever for promoting EVs. However, residential EV charging introduces additional problems to the low-voltage electrical grid, such as power demand peaks and current harmonics that can degrade the distributed energy quality. To mitigate these disturbances, an energy management system coupled with an active anti-harmonic filter is required. This paper proposes a residential parking management strategy based on effective cooperation between parked vehicles to meet, at a lower cost, their energy needs while ensuring the anti-harmonic filter power supply, thus reducing the pressure on the main grid. This management issue is posed as a multi-objective optimization problem aimed at minimizing the parking lot's energy bill, the cost of battery degradation, the main grid peak-to-average ratio (PAR), the EV charging dissatisfaction, and harmonics generated by the vehicles being charged. This problem is solved using the particle swarm optimization (PSO) algorithm. Unlike public parking, residential parking has the particularity of being operated by a limited number of privileged users. This allows for the creation of a database for predicting important data to improve the EMS's performance. Indeed, based on GRU neural networks, a predictive model is developed for estimating arrival and departure times as well as the initial state of charge for each EV.statistical test was applied to the forecasting metrics to confirm their reliability. The strength of the developed management algorithm lies in the high level of cooperation between EVs in energy exchanges. Indeed, the CE-EMS establishes bidirectional energy flows between vehicles (V2V), controls their exchanges with the main grid (V2G and G2V), and promotes their participation in powering the harmonic filter (V2F). Simulations conducted over a 24-hour horizon validate the CE-EMS's effectiveness. Two comparative studies are developed: the first highlights the contribution of the predictive layer in improving EMS performance; the second emphasizes the benefits of vehicle cooperation to meet their energy needs and compensate for harmonics. Comparative results show that the CE-EMS approach enables a 30.05% reduction in grid-related cost and a 52.4% decrease in the Peak-to-Average Ratio (PAR) compared to the NC-EMS strategy. This optimization is also accompanied by a 26.6% reduction in battery degradation cost relative to SLMSP. The minimum state of charge (SOC) reaches 80%, ensuring satisfactory vehicle charging. Harmonic reduction is achieved through vehicle participation in V2F mode, with an energy contribution of 9.7%.In addition, a sensitivity analysis was conducted to confirm the robustness of the optimization system under varying input conditions. Furthermore, a statistical test was applied to validate the reliability of the obtained performances. Finally, the study also discusses the limitations and perspectives for practical deployment, emphasizing the transition from simulation to real-world application. • Proposal of a Cooperative EMS (CE-EMS) A novel energy management system is developed for residential electric vehicle (EV) charging, designed to minimize costs, enhance user satisfaction, and mitigate power quality issues. • Multi-Objective Optimization Framework The system formulates the energy management problem as a multi-objective optimization using the MOPSO (Multi-Objective Particle Swarm Optimization) algorithm, targeting: ∘ Energy cost reduction ∘ Battery degradation minimization ∘ Peak-to-Average Ratio (PAR) reduction ∘ Charging dissatisfaction mitigation ∘ Harmonic distortion reduction • Integration of Forecasting Layer (GRU) A predictive layer based on Gated Recurrent Unit (GRU) neural networks forecasts EV arrival/departure times and initial SOC to improve scheduling and system responsiveness. • Bidirectional Energy Flows and V2X Strategies The CE-EMS incorporates G2V , V2G , V2V , and V2F strategies for intelligent, distributed, and cooperative energy flows: ∘ V2F : EVs supply energy to a harmonic filter. ∘ V2V : Vehicles share energy with each other. ∘ V2G : Grid support via vehicle discharge. ∘ G2V : Grid-to-vehicle charging. • Power Quality Improvement via Active Filtering Harmonic mitigation is handled through a Vehicle-to-Filter (V2F) approach, where parked EVs with sufficient SOC contribute to powering the active harmonic filter (aligned with IEEE 519 and 1547 standards). • Forecast-Based Prioritization of Charging Vehicles are ranked and prioritized for charging based on urgency, predicted departure time, and SOC, ensuring maximum user satisfaction even under limited resource conditions. • Performance Metrics and Comparative Results The CE-EMS demonstrates: ∘ 30.05% grid cost reduction vs. NC-EMS ∘ 52.4% PAR reduction vs. NC-EMS ∘ 26.6% lower battery degradation cost vs. SLMSP ∘ 9.7% V2F energy contribution ∘ Full charging of all 6 vehicles with minimum SOC ≥ 80% • Comparative Analysis with SLMSP and NC-EMS Two benchmark strategies are evaluated: ∘ SLMSP (non-predictive, static): high dissatisfaction and battery wear ∘ NC-EMS (non-cooperative): high grid dependency and poor cost/PAR balance CE-EMS clearly outperforms both. • Simulation and Co-simulation Environment The system is implemented using a Python–MATLAB integration: ∘ Python handles GRU-based forecasting ∘ MATLAB handles MOPSO optimization • Comprehensive Constraints and Realistic Modeling The model includes: ∘ SOC limits ∘ Power balance ∘ Charging/discharging exclusivity ∘ Realistic SOH degradation modeling ∘ Time-of-Use (TOU) tariff-based cost modeling