This thesis proposes novel electric vehicle (EV) charging strategies in DC microgrid (DCMG) for
integrating network loads, EV charging/discharging and dispatchable generators (DGs) using
droop control within DCMG. A novel two-stage optimization framework is deployed, which
optimizes power flow in the network using droop control within DCMG and solves charging
tasks with a modified Djistra algorithm. Charging tasks here are modeled as the shortest
path problem considering system losses and battery degradation from the distribution system
operator (DSO) and electric vehicles aggregator (EVA) respectively.
Furthermore, a probabilistic distribution model is proposed to investigate the EV stochastic
behaviours for a charging station including time-of-arrival (TOA), time-of-departure(TOD) and
energy-to-be-charged (ETC) as well as the coupling characteristic between these parameters.
Markov Chain Monte Carlo (MCMC) method is employed to establish a multi-dimension probability
distribution for those load profiles and further tests show the scheme is suitable for
decentralized computing of its low burn-in request, fast convergent and good parallel acceleration
performance.
Following this, a three-stage stochastic EV charging strategy is designed to plug the probabilistic
distribution model into the optimization framework, which becomes the first stage of
the framework. Subsequently, an optimal power flow (OPF) model in the DCMG is deployed
where the previous deterministic model is deployed in the second stage which stage one and
stage two are combined as a chance-constrained problem in stage three and solved as a random
walk problem.
Finally, this thesis investigates the value of EV integration in the DCMG. The results obtained
show that with smart control of EV charging/discharging, not only EV charging requests can be satisfied, but also network performance like peak valley difference can be improved by ancillary
services. Meanwhile, both system loss and battery degradation from DSO and EVA can be
minimized.