Lagrangian Decomposition based Multi Agent Model Predictive Control for Electric Vehicles Charging integrating Real Time Pricing

TL;DR

This paper introduces a distributed, real-time control strategy for electric vehicle charging using Lagrangian decomposition and model predictive control, optimizing load curves and energy prices while respecting grid constraints.

Contribution

It proposes a novel distributed control method combining Lagrangian decomposition with model predictive control for EV charging management.

Findings

Effective load curve computation by agents

Coherence between charging load and energy prices

Improved grid load management

Abstract

This paper presents a real time distributed control strategy for electric vehicles charging covering both drivers and grid players' needs. Computation of the charging load curve is performed by agents working at the level of each single vehicle, with the information exchanged with grid players being restricted to the chosen load curve and energy price feedback from the market, elaborated according to the charging infrastructure congestion. The distributed control mechanism is based on model predictive control methodology and Lagrangian decomposition of the optimization control problem at its basis. The simulation results show the effectiveness of the proposed distributed approach and the mutual coherence between the computed charging load curves and the resulting energy price over the time.