Smart Wireless Electric Vehicle Charging System and Load Management Utilizing Fine-Grained Solar Power Predictions Constrained on Human Electromagnetic Exposure

Written by Adel El-Shahat and Swoon Chang

Wireless electric vehicle charging (WEVC) is considered as a potential convenient charging option for electric vehicles (EVs) for future smart grids. There are two types of wireless charging: one when the vehicle is parked and the chargers remain in a fixed geometry during charging, and another which charges the vehicle while moving from a charger imbedded in the road. This article deals with charging while moving.

A high wireless power transfer efficiency at near-field distances can be achieved by using magnetic resonant coupling, in which inductive power is transferred when a primary and a secondary coil are closely aligned and have sufficient mutual coupling. Typically, the maximum distance is less than 300mm, and proximity reduces losses. Recent research finds that particular care should be devoted to the exposure assessment of humans to the stray electromagnetic fields (EMFs) emitted by these systems. This is because of (i) high transmitted power levels up to some kW, and (ii) non-uniformity, which may cause the power levels to exceed the guidelines. It is therefore essential to investigate transformation of the WEVC mechanisms to “human health-compatible” forms.

We define human health-compatibility as the ability of a WEVC system to deliver the desired service while guaranteeing the EMF exposure below a safety level. This part of the research focuses on evaluation of the additional complication due to the human exposure constraints. In doing so, a multi-objective optimization framework will be built, which is expected to be non-deterministic polynomial-time (NP)-hard, due to (i) quadraticity/non-linearity of the constraints (e.g., driving speed range, temperature, coupling coefficient, air gap, battery lifetime, state of charging, and adjacent misalignment as operation states) and (ii) non-convexity of the problem. Due to the complexity, no existing research has provided a comprehensive analysis framework nor a testbed investigating the human EMF exposure in a WEVC system. In road tests, chargers are being built at the American Center for Mobility and in Downtown Detroit.

Additionally, incorporating a new system is difficult because there are various uncertainties to overcome, such as hourly PV generation, EV coming/departing time, customer various load demands, and required charging power, etc. These factors significantly impact the parameters of the system using the current/ traditional algorithms. So, there is a critical need for an efficient solution to incorporate those uncertainties in the system to guarantee stable and precise functioning of the EV charging system's operations and the loads' behavior. Furthermore, the system may act another way in physical applications because various pragmatic considerations have been shortened in the current algorithms. That is why it is mandatory to validate the algorithm in a physical environment.

Moreover, PV integration into the power grid is constrained by the load performance (linear and nonlinear) and EV chargers. Thus, a fine-grained prediction of dynamic demand and supply is essential for more economical and cost-effective usage by enabling dumping, storing, and managing any excess power. However, uncertainties of real load behaviors pose vital challenges in behavior control of the energy storage under the smart operation.

Another challenge for PV-EV-Battery smart charging is the synergies optimization among the PV power generation, customers’ load demand, and EV energy demand to (i) improve the PV penetration to the power grid, (ii) promote self-consumption, and (iii) provide ancillary services to the grid. In this ever-complex load management system, utilization of artificial intelligence (AI) techniques has been emerged for predictive modeling, optimization, maximum power point tracking (MPPT), and power management for smart charging. Therefore, it is very important to implement new schemes to handle these tasks.

 

Proposed Research:

The objective of this research is to build an integrated analysis and experiment framework for a WEVC system.

First, the research thrusts are identified as:

1. Modeling of representative dynamic WEVC topologies.
2. Development of an analysis framework based on multi-variate optimization and stochastic geometry.
3. Experimental evaluation of various WEVC procedures and scenarios based on the analysis framework.

Mathematical modeling based upon the stochastic geometry for a comprehensive consideration of the relative positions between the charger, EV, and nearby human users will be used for more investigations of the parametric response with emphasis on elucidating the dynamics of the variables. Example scenarios include position of human user (distance and inside/outside of the car), level of coil misalignment, car structure (e.g., frame thickness), and metal type for coils wires, etc. However samples and data from successful work from Next Energy could provide great assistance to qualified researchers.

This work will extend the state of the art in WEVC technologies by understanding:

1. The fundamental performance of a WEVC system under human exposure constraints.
2. Exact amount of EMF leaking to human users in various practical scenarios.

Then, the proposal ponders another research question concurrently with the WEVC human electromagnetic field exposure one. How can the future smart charging system can be designed to:

1. Consider uncertainties of hourly PV potentials and complex load profile?
2. Provide reliable predictions of actual operation in a physical setting?
3. Conduct the synergies optimization using AI techniques?

The proposal targets creating a prototype of smart charging system that considers the stochastic, uncertainty, variation, and fluctuation nature of PV generation, customer’s load, and EV operator behavior. The research team proposes to implement an efficient, predictive, low-cost, and adaptable optimum system via a series of intelligent functionalities to manage the EV charging and customer loads demand. The team will model, design, and analyze a dense experimental case study via the proposed smart optimum charging algorithm with the PV integration into controllable loads to set up an innovative EV-PV-Battery system. In addition, the proposal seeks to introduce a novel intelligent Maximum Power Point (MPPT) scheme to attract the maximum power from the sun utilizing AI techniques such as Genetic Algorithm (GA), Particle-Swarm-Optimization (PSO), Artificial Neural Networks (ANN), and Convolutional Neural Network (CNN).

Finally, the full system modeling and implementation will:

1. Reveal the optimized role of materials, structure, and electrical variables and using the system to improve the system efficiency. However, a good base to handle this part in the form of comparisons with successful work at Next Energy for validations and modifications if applicable.
2. Solve the problem of precise optimal smart charging and efficient power management.

Acknowledgment: This research was funded by Polytechnic FWL Research Impact Area Seed Grant, Purdue Polytechnique Institute, Purdue University.

 

This article edited by Doug Houseman.

To view all articles in this issue, please go to March 2022 eBulletin. For a downloadable copy, please visit the IEEE Smart Grid Resource Center.