Project Description
The objectives of this research are to develop the MATLAB code and Python scripts, and other open source forms of code, for metamodels that will be used to size Battery Energy Storage Systems (BESS's) for shipboard and microgrid applications and EMI Filters for Electric Vehicle propulsion systems. The methodology for accomplishing this task is through the development of surrogate models that correlate to a wide range of complex multi-physics decisions and corresponding results with respect to equipment size, weight, losses, cost and life. Surrogate models are derived from data that is produced by evolutionary optimization algorithms (resulting in performance versus design space Pareto fronts) or from extensive databases containing measurements and other information on constituent parts of these systems. Machine learning models are trained from this data and the resulting Surrogate models will become part of a code set that sizes the equipment according to a wide range of design space variable inputs.
Tasks and Responsibilites
The student will first develop the User Interfaces (UIs) for the metamodels and will implement prescribed code that compiles up size, weight, cost, losses and life, based upon simple physics-based equations or simple descriptions. This results in working code to demonstrate inputs and outputs. Once the metamodel code is working, the student will work with another university, University of Texas-Arlington (UTA), to merge this code with their energy storage sizing tool, which is based upon a database of energy storage cell test results. The student will work with UTA to ensure that the metamodel code selects one Battery Storage Drawer design, according to size and weight constraints and objective weightings. The student will then work with Prof. Cuzner to incorporate the sizing of a number of examples of power converter building blocks, along with their corresponding electro-thermal-physical models and make the code select converter blocks according to user design space inputs. The work will proceed to incorporate surrogate models for EMI filters for the BESS and for an EV Drive example. In the latter case, surrogate models inputs will be derived from the data of other surrogate models that have been developed by Eaton Corp. and Metamorph.
Desired Qualifications
None Listed.