Skip to main content
Open Access Publications from the University of California


UCLA Electronic Theses and Dissertations bannerUCLA

Human Centered Multi-Objective Scheduling for PEV Charging when Sharing a Constrained Energy Resource Using Model Predictive Control

  • Author(s): Chynoweth, Joshua Storm
  • Advisor(s): Gadh, Rajit
  • et al.

Enabling the widespread deployment of PEVs requires significant charging infrastructure. The difficulty and cost of PEV charger installation has resulted in a persistent shortage of plug points for charging. WINSmartEVTM is a novel network system of PEV charging stations that connects multiple PEVs to a given circuit by sharing the energy resource. By significantly increasing the number of plugs per circuit, and thereby decreasing the cost per plug point installation, an increase in access to charging for PEVs is achieved. An incremental increase in the number of plugs per circuit causes an incremental decrease in the amount of energy that a user can receive during a given charging session. Dividing a circuit is a zero sum game between the number of plugs per circuit and fulfilling the users’ needs and expectations. By focusing a charge scheduling system on the users’ needs, it is possible to maximize any combination of users’ needs and expectations, and number of plugs per circuit.

A scheduling system was created that maximizes the number of PEVs that can be adequately charged from a given circuit by accounting for both users’ needs and expectations. Objective formulation was created to quantify users’ needs and expectations using multi-objective optimization. The formulation quantified 8 user needs and expectations functions categorized by 2 independent variables. Six needs were quantified as SOC based utility functions and 2 expectations were quantified as time based utility functions.

The energy allocation scheduling problem was solved using the model predictive control method. The utility functions were fit to a state space model. However, the SOC utility function and feedback functions are non-linear so a closed form solution was not attempted. The scheduling system was simulated in Matlab and compared to the default charging algorithms. The system was validated using data from 3 separate energy sharing chargers that divide energy between 4 plugs each, over an 8 months period. The results show that this scheduling system significantly shifts to the right the graph of BEVs leaving with insufficiency charge for a given number of plugs on a circuit. Therefore, for any given number of plugs the circuit is sharing, a BEV user is significantly less likely to leave with insufficient charge. Furthermore, the likelihood of BEVs leaving with insufficient charge can be kept the same while significantly increasing the number of plugs supplied from the circuit.

Main Content
For improved accessibility of PDF content, download the file to your device.
Current View