Energy-harvesting systems reduce/eliminate the use of batteries by using environmental sources to collect energy. In energy-harvesting systems, microcontrollers can use different techniques to collect energy from their surrounding environments such as light or heat. By converting this ambient energy into electrical power, they can power themselves and perform a range of tasks. The collected energy can be stored in capacitors and the capacitors are used to power the systems to process tasks.  This is especially useful for large, connected networks of devices, such as the Internet of Things (IoT) and remote applications where it can be difficult to replace batteries or the costs of maintaining and replacing the batteries are high. In addition, the use of environmental sources for energy reduces energy consumption and carbon emissions and results in sustainable computing. However, to avoid power failure, and ensure task completion it is crucial to manage the collected charges. 

Objectives: In this project, students will design an energy-harvesting system that uses mini-solar cells to harvest energy from light. The amount of power generated by solar cells depends on factors such as the amount of sunlight available, the size and efficiency of the solar cells, and the power requirements of the microcontroller. Therefore, it is important to carefully design and optimize the solar cell and microcontroller system to ensure that it can operate reliably under various lighting conditions. The students will design circuitry to harvest and store charges using capacitors for a batteryless microcontroller.  In addition, the students will design an algorithm to schedule the tasks that will run on the microcontroller and manage the charges to prevent power failure Finally, students will simulate the algorithm and run it on the actual hardware, measure energy consumption, find the task completion rate, calculate the failure rate, and compare the results with state-of-the-art algorithms.

The solid-state transformer (SST) has attracted wide attention for microgrid and smart grid utilization. SSTs are highly energy-efficient and have a smaller footprint than conventional transformers due to their substantially reduced size, weight and cost. They also include smart functionalities that allow them to respond better to grid disturbances. These features can make SSTs essential microgrid assets. The high frequency (HF) transformer is recognized as the key element of the SST. High frequency and high voltage operations exert substantial electromagnetic and thermal stresses into these devices, reducing their lifespan and increasing the cost of their operation. The efficient and affordable design of HF transformers is critical for achieving the main requirements of SSTs: high density, minimal losses, voltage regulation, and electric isolation. Thus far such design has been a bottleneck for the mainstream penetration of SSTs in electric distribution systems and microgrids due to reliability and operating life concerns.

Objective: The goal of this project is to assess the effectiveness of the synergistic combination of novel multiphysics and microgrid modeling and simulation tools for the optimal design of HF transformers for microgrid application, considering the stresses produced by the extensive inclusion of power electronic-interfaced sources, loads and storage units during steady state and transient conditions. To reach this goal, this project aims to develop, implement, and thoroughly test a modeling approach for accurate dynamic simulation of the microgrid system and its interaction with a detailed physics-based HF transformer model.

The time-varying neuron membrane voltage is fundamental to biological neural network functionality. A neuron is readily modeled as a nonlinear circuit, where electrical current inputs model the effects of other interconnected neurons and the circuit output is the membrane voltage. One area of interest is how to find reduced-energy stimulation currents to achieve a desired response or to suppress an undesired response using circuit models, with implications to treatment of neurological disease. 

Objective: This project aims to explore optimization techniques to compute reduced-energy currents that achieve the desired response in a neuron circuit model. Both intracellular and extracellular currents will be considered. In laboratory work, intracellular currents are injected into a neuron, while extracellular currents are locally applied and do not require penetration of the neuron cell and are thus suited for clinical applications. Previous work at WMU has primarily focused on using optimal control theory; while effective, more computationally-efficient techniques are  needed for real-time control of neural dynamics and for reducing energy consumption.