Litcius/Paper detail

Learning model predictive control for efficient energy management of electric vehicles under car following and road slopes

Kiwon Yeom

2022Energy Reports14 citationsDOIOpen Access PDF

Abstract

In this paper, a novel control algorithm is proposed using Model Predictive Control for improving energy consumption of fully electric vehicles (FEVs) and Deep Reinforcement Learning (DRL) for understanding the driving environment in real-time, respectively. The dynamics of the FEVs and the brushless DC motor (BLDC) based powertrain system is utilized for the DRL process, and a Model Predictive Control (MPC) based optimal speed strategy for minimizing battery power consumption is developed by updating the approximation model through the repeated experience. Using the high fidelity car simulator (CarSim), the proposed algorithm is evaluated under freeway cruising scenarios with road slopes as well as preceding car following. The simulation results demonstrate in overall potential energy saving of 3.2%.

Topics & Concepts

CarSimPowertrainModel predictive controlAutomotive engineeringEnergy managementEnergy consumptionBattery (electricity)Computer scienceProcess (computing)Reinforcement learningVehicle dynamicsEngineeringSimulationTorqueControl engineeringEnergy (signal processing)Control (management)Power (physics)Artificial intelligenceQuantum mechanicsOperating systemMathematicsThermodynamicsPhysicsElectrical engineeringStatisticsElectric and Hybrid Vehicle TechnologiesElectric Vehicles and InfrastructureVehicle emissions and performance